o m
* r» « r» r-. to
i- f r- f r- CM
I I I I I I I I I
CM
O>
o>
f»
5 3 CM T? CM
^TCOW'-CM
IO l«- K 00 i-
CM in q cp
-------
o
z.
^£
0
Of
i
LU
Si
LU
§
O
O
i i
1
^f
O LO
CO !"»
C£ Ct\
LU r
CX 1
CO
1 v^
«t cn
o
I co
1 1 1
2^ i t
I-H Q;
O
LU CJ3
CD LU
Z. 1
1C 0
o
1 fv
z: i
1 1 i j |_.t_p
o; z:
LU O
ex z:
CM
LLJ
CQ
*^.
I
iH
cd
I* 4-1
3 C
C 0)
X o3
iH '!->
r-H rX3
cd rt
4^ C*
O O
H C
LO
[**
1
oo
O
rH O
f* ^j
U g
o S
r* O
00 Z
4-1
c
o
u
V"*
0>
ex
g
rH
4-J
m
S
^j
V
c
o
o
(JJ
. o
o
c
1 t
r"**« ^J T*^ r^ l*O fO
J^x. Q^ ^^ \XD ^t" ^^
T( r-H CO O OO O^
rH rH rH r-«
C^~J ^Q ^J ^Q t^^ ^\J
OO t^ O^ "*J* f1*^ CO
CTi O O OO vO OO
rH CJ
v£3 fO ?H to <^> ^^
OO O>1 O^ QO vO to
r^» oo ""^ \o ^^ oc
evi
10
4->
U ?*
C CD 03 OO
H 6 Ci. C
O *H
rH O r-i ^H 4->
SC 4-> B OO1-*-4 £
CO rH f- C 3 oo oo cr. to
1 1
<7) CM or o 01 oo
O vO t . LO OO t*- OC
t~- OO f~- LO vO OO rH
rH
t/2
.0 ^
o cd
HOC X
rH *rH O ^O1 *H
JQ f± -H 4-)
3 3 4-» (U Irt
C, t U (J CD
> o 2 3 o
C O -P VH <4-l
2 00- C W
4-> O C 4-> ClT
cd t/i U -H cd !i in
4-> 3 (U
(/) r-i .H 4-) * 10 4-1 -H
rHr-i
OJ'H (/IrH JH CrH (JjC C
T3 J> C -H 4-> Cd Cd -H V) -H
CdrHCd4->CCCiJr-l.HC
t-1 CO H CJ UH d)
C^ D^
M
-c;
« o
18 -H
0 <8
'M S
Q) CU
iqj
JJ
0 C!
M CD
^J K
cu 04
§ o
o "3
fe fc.
«a
H
£2
o ce:
CJ CO
CJ
CD Co
S «
0 \
O "^
C Q
H Cn
s S
CU
CJ
o
in
-------
possible agglomeration benefits. Several empirical studies using
pre-1960 data tend to support this theory, but a recent study by
Petrulis (1979) using 1967 to 1973 does not. Nonmetropolitan regions
of the nation were gaining manufacturing employment because of an
increased share of both slow growth and fast growth industries.
The growth of other economic sectors in rural areas also was not
anticipated by previous concepts because it was assumed that economies
of scale in firms and economies of agglomeration of places would domin-
ate the location matrix. Without economies of size, small but efficient
firms would be located in dispersed patterns. Data in Table 2, however,
suggest that other sectors are growing in rural areas. Unfortunately,
future rural/urban patterns of the location of sectors are difficult to
predict because too many factors are changing. Agglomeration economies
which might now exist in metropolitan areas may become obsolete in the
future due to rapid advances in communications technology. Counter-
vailing diseconomies may surface due to social and other changes such
as growing crime rates and an increased sensitivity to pollution and
congestion. Technological advances may support the dispersion of in-
dustries. The cost of electric power in the future may give new areas
an industrial advantage. It should be clear that the diversification
rural areas have undergone in their industrial mix bodes well for the
future, and that rural economies are no longer as dominated by the
vagaries of agricultural prices and weather and should expect to
participate in long-term expansions of the national economy as well as
structural changes which could add to their growth potential.
DEFINITIONS AND DATA
Economic structure is defined as the set of behavioral relation-
ships within and among regions of the nation and the world. A class-
ification of structural changes may be useful for conceptualization
even though it is beyond the scope of this paper. Periodic structural
changes may come about due to threshold effects or alterations in
expectations which accompany cumulative change. At some size, a growing
town starts to perform the function of a regional trade center. Cyclical
structural changes can be brought about by national business cycles.
Behaviors in each phase of the cycle may be different, cycles may be the
mechanism for industry location shifts, and the accelerator may operate
during these cycles. Finally, the secular changes in structure discussed
below may appear over time in response to technology, institutions,
ethics, or tastes. Truly dynamic models would include a specification
for each type of structural change. An easier albeit difficult task is
to specify only secular structural changes.
The data used in this discussion are drawn from the Bureau of
Economic Analysis and the decennial census. The Bureau of Economic
-------
Analysis (BEA) data for both employment and population are estimates.
Employment is presumed to reflect the place of work while population
is reported by place of residence in the BEA data.
Counties are classed by metropolitan status, and nonmetropolitan
counties are further classified by urban population size and adjacency
to metropolitan areas. This classification minimizes errors in the data
due to reporting, and standardizes for shopping and work commuting pat-
terns. The urban to rural continuum of counties allows rural counties
to be highlighted even though the data presentation is made complex.
The data with which we have to work merely indicate manifestations
or symptoms of changes in behavioral relationships. For pedagogic pur-
poses we choose to isolate five indicators of structural change for
discussion. They are the relationships between (1) the national economy
and the basic sectors in rural areas, (2) basic activities and service
activities, (3) employment and population, (4) migration and income,
and (5) transportation costs and the regional distribution of activities.
These can be viewed as coefficients of a very simple regional forecast-
ing model. The basic sectors of rural areas will be tied to that of
the national economy. Economic base concepts dictate some relationship
between basic activities and service activities in an area. The number
of people in a region is a function of employment, labor force partici-
pation, and income. Finally, the distribution of service activities
within a region should be dependent upon transportation and access costs.
LINKAGE TO THE NATIONAL SYSTEM
Rural counties have become closely linked to the national economy.
They are affected by long-term expansions of the national economy as
well as cyclical fluctuations. This is illustrated by the changes in
employment in basic economic sectors of rural areas (excluding farmers
who do not respond to the same cyclical pattern). The proportion of
nonmetropolitan counties with more than a 10 percent change in basic
employment each year is shown in Figure 1 for the period 1969-1976.
The proportion of counties in each urban class increasing more than
10 percent is in the upper part of the graph, and the proportion de-
creasing is in the lower frame. Superimposed over the bar graph is an
index of U.S. industrial production. Three implications are apparent.
The first implication is that employment change is common to most
nonmetropolitan counties regardless of their size or location. Large
proportions of counties exhibit more than a 10 percent change in basic
employment. Between 1974 and 1975, 34 percent of the adjacent rural
counties with urban populations greater than 20,000 had a change in
basic employment greater than 10 percent 31 percent declined and
3 percent gained. In 1974-75, the total proportions increasing and
-------
FIGURE 1. PERCENT OF RURAL U.S. COUNTIES INCREASING OR
DECREASING IN EACH YEAR MORE THAN 10 PERCENT IN WAGE
AND SALARY EMPLOYMENT IN BASIC ECONOMIC SECTORS a/, BY
RURAL CLASS, 1969-1976
Percent of
counties
30
25
20
4)
U)
(0
b 15
c
M
10
5
0
5
10
(fl
a 15
o
Q
20
25
30
-
-
_
-
_
1
f
r
|
ff
r
r
j[
x
^
r
^
L
I
1
*-"
Legend
Urban population
Greater Chan 20,001) '
'__ f° ,'
s
1
1
r
i
-
L
-
<- ^
^
\
\
\
d Ad], to metro area
^M
i
. . J
X
L _L
I '
production b/
-
-
\
-
X
~"
-
-
_
160
150
140
130
120
110
100
90
80
70
60
50
70-71
72-73
74-75
75-76
a/ Basic sectors are farm, raining, manufacturing, heavy construction, railroad, motor freight,
water transport, and federal wage and salary employment.
b/ 1967 = 100.
-------
decreasing ranged from 26 percent of the urban nonadjacent counties
to 41 percent of the rural counties adjacent to metropolitan areas.
Over the whole period, many if not most counties have been subject to
rapid changes in basic employment.
Second, year-to-year variability is common. During 1974-75, 22
to 31 percent of rural counties registered decreases greater than 10
percent in basic employment. The very next year, from 9 to 31 percent
had increases. Note that the temporal aggregations commonly used tend
to mask this type of variability. Population changes and mobility are
probably more dynamic than revealed by most analyses. Thus, not only
are changes in basic employment widespread, but also they are highly
variable from year-to-year.
The third and most important implication is that changes in rural
areas are firmly rooted in the health of the national economy. The
association between national industrial production and changes in the
basic employment of rural counties is clear. By the same token, rural
parts of the nation should expect to participate in long-term expansions
of the national economy.
Changes in production technology, communications, and transporta-
tion on balance may increase the comparative advantage of rural areas
in the basic sectors of the nation's economy. That would mean more
rapid growth than that of the nation as a whole. In terms of struc-
tural change and the timing of that change, it may be important to note
that the process of removing older and inefficient productive capacity
from urban areas to less urbanized areas may be associated with the
national business cycle.
THE MIX OF SERVICE EMPLOYMENT
The ratio of service employment to total employment over time
provides some evidence that the basic to service employment mix is
changing in rural areas. Historical census data are reported in
Table 3. The proportion of service activities to total employment has
been increasing in the nation and in rural areas for several decades.
In 1940, 46 percent of total employment was in selected service sectors
in the nation. By 1970, the proportion was 61 percent. The service
ratio in selected Northern Plains states and counties dramatizes the
fact that the change is consistent with that in many rural areas and
is not concentrated in urban centers alone. Note that by 1970 the
service to employment ratio was higher in Montana, Wyoming, and
North Dakota than in the nation, even though it was somewhat lower in
1940. The same pattern is evident in selected rural counties of these
three states even though the economic activities there were dominated
by an agricultural base in this period.
-------
TABLE 3. PERCENT SERVICE EMPLOYMENT, SELECTED AREAS 1940-1970 a/
Area
United States
Montana
Rosebud
Wyoming
Campbell
North Dakota
Mercer
1940 :
46
41
34
44
30
37
23
Year
1950 :
49
47
33
49
35
41
30
1960 :
55
56
51
55
44
53
41
1970
61
63
56
63
45
65
51
a/ Service is other transportation, utility services, wholesale and
retail, finance, insurance, real estate, business and personal services,
and government.
Source: Regional Employment by Industry, 1940-1970, U.S. Department of
Commerce, Bureau of Economic Analysis, n.d.
-------
The proportion of total employment in the same service sectors has
continued to increase through the 1970's in all nonmetropolitan areas
(Table 4). No pattern of change is discernible among the various
urbanization classes. All areas show an increase during this period
of about the same magnitude.
The important point is that the service employment mix is high in
rural areas and increasing, and that service activities are geographi-
cally dispersed much more than in the past. Service activities have
been characterized as being uniquely suited to centralized urban loca-
tions. These data suggest that the degree to which service employment
is centralized in urban places can easily be exaggerated because large
differences among rural and urban areas are not evident.
The secular increase in service employment can be explained by
income increases, the cost of access and transportation, changes in
service technology and innovations, and the composition and technology
of basic industries. We suggest three implications of these data for
purposes of modeling the economies of rural areas. First is the obvi-
ous increase in service employment per unit of basic employment in a
region. That means service employment may increase even though the
economic base of a region remains stable. Second, service employment
seems to be dispersing into rural areas rather than remaining central-
ized in regional trade centers. Third, we might suggest that an impor-
tant new function for the old trade centers is that of providing excess
service capacity which is used to absorb rapidly changing demands from
year-to-year in the surrounding hinterland.
LABOR FORCE PARTICIPATION RATES
One of the more dramatic structural changes in rural areas is the
proportion of the population in the labor force. The labor force partic-
ipation rate of the people living in nonmetropolitan areas has increased
to 60.7 percent of the working age population in 1979 compared with
64.2 percent of those in metropolitan areas (Bureau of Labor Statistics,
1974-78). Total labor force participation of farm people was 61.2 per-
cent by early 1979. Many of the recent changes in labor force partic-
ipation are due to the entry of women into the labor force. Brown and
O'Leary (1979) report that between 1960 and 1970, 89 percent of the
increase in nonmetropolitan employment was comprised of women. In 1978,
43 percent of farm women in nonmetropolitan areas were in the labor
force. It is not surprising that off-farm income contributed 57 percent
of the personal income of the farm population in 1977 (Carlin and Ghelfi,
1979).
Labor force participation rates for counties over the 1969-76 period
are not available. However, a superior measure for purposes of modeling
is employment relative to the population. This ratio indicates the
-------
»s^
fTJ 1
'
1*^
p^
1
cr>
CT>
i i
*
co
CO
i
o
^^
^c
CD
ID
CO
CO
C£
UJ
NX
01
o
3
>
o;
_i
co
0
2:
i i
CD
e£
3
U.
O
CO
m
O
"-}
UJ
o
i <
f^f
Ul
CO
1
UJ '
Q:
UJ
r>
^
1. 1
CO
r-
vO
en
iH
LO
f-^
0>
iH
, .
^^-
^.
O"j)
rH
to
f*^.
en
rH
Csl
t--
cn
rH
. .
rH
en
rH
O
en
m 9
O*i
vO
cn
co
CO
rH
o
i
0
o
t**- C*J C^J
^O ^D v£J
t~- CM Csl
vO vO vO
LO O O
\O vO vO
LO en o
vO LO vO
LO en en
VO LO LO
*t oo en
VO LO LO
IO vO t--
VO LO LO
i-H LO \O
VO LO LO
^1
H -H TH
P P P*
CO CO CO
rH iH l-H
333
o o o
PL< f\ 0<
0 O 0
o o o
O O O
^ M M
o en o
o en LO
c/i o en CM
cd -
OJ rH
MI
< d o
§1:1
P OP
H Mi O
rH 0) « CO
O P O CO
ex co LO co
O 0) CNl J
p t3
JU
3
vO \O l*^
LO LO LO
\O f*^ OO
LO LO LO
»* * LO
LO LO LO
"Sfr ** LO
LO LO LO
«* LO LO
LO LO LO
to ^t LO
LO LO LO
(M tO LO
LO LO LO
O CM Tj-
LO LO LO
^,
^L
CO
co
d)
J-4
cd
O
Mi C C C
P O O O
CU *iH *rH *rH
cd co co
O rH rH rH
P 3 3 3
P O O O
ei ex ex ex
rt co cd co
p-^^jQ JL^) jrj^*t
'O J-j fn fn
cd 3 3 3
co o cr> o
cu o en o
H o en LO
4»J * * ^
So en eg
CM i-H
0
o
TO C O C
p co P g
rH P P
o
CX r< 0
O 0) O CO
Ml P LO CO
P CO - <1)
«> CO CM J
i"
P
o
«z
00 00 O
LO LO VO
oo oo o
LO LO VO
\O vO 00
LO LO LO
vO VO OO
LO LO LO
LO vO 00
LO LO LO
LO vO t-~
LO LO LO
tO LO t-^
LO LO LO
CM «* VO
LO LO LO
*,
3|
t/)
TO
rt § 0 §
H -H -H
p p p
cd cd cd
i-H rH i-H
333
&&§<
ex ex ex
3 co ro
'S'6'£
333
o en o
o en o
o en to
M M *
o en CM
p P
Mi O
cu o co
P LO CO
cd 4J M C7v
Cd CJ 4J rH
4-1 Cd CU
CO M S J=
a) .u u
d CM
rH O -H Cd
S ° c S
M 60 0 >
C -H <|
T3 -H 4J Q
C T3 Cd CO
Cd rH rH ^
H 3 -
« 3 a co
0) rJ3 O P3
O PH W
C rH
MM ,£ CM
30) 4J r-.
CO C CM
CO) MH
H 60 O &
* * CO -3
cu C cj
0 O iH >
C -H 4J M
(0 4J CO 0)
fl cd -H §
H 4-1 M0
<4H M 0) -H
O 4J N
« a. o
cu co cd c
cd cd cd o
MM ,C 1-3
4-1 4J U
TJ
rH M O C
HO) -H Cd
cfl JZ B
4J 4J O«<
0) O C C
M 0 S
o o
T3 CO W M
C C «
CO O T)
H c -a
0) 4J Cd iH
rH Cti ^
Cd O rH Cd
co -H cd p
0) C 'rl
rH 3 a >
OS O CO
MB CO CU
& 0 = C
O -H
^) s K
TJ « iH
3 CO
rH Q) T) «
U CJ 0)
C vH C T3
H > -HO)
M MH M
CO 0) 1) [V|
M CO TJ
4-1 M CO ,O
CJ CU CO
cu jz 4J cd :
CO 4J fj rH >
O a) 0 o
H C M 4-1 r-l
> cd cu a
M > 3 >
0) rH O O CO
CO cd 60 U 0)
C -H
O rH 4J
^ co cd *>., c
cd M O ^3 3
0) o O
CX rH CJ
.
V
W
'H
&t
r-n]
fQ
c
^
0
H
Q
CJ
8
"3
>M
o
3
UJ
o
3
CQ
CD
^H
4J
^
0
^
31
4J
H
CO
(H
0)
£i
p^
g
Q)
4J
OJ
4J
CO
IB
m
-u
q
0
S3 to
(!)
v -i-i
tq (H
H Q)
to to
M IB
"S "O
IB <1)
U O
Q 0)
*~~i E
fQ O
"H O
CJ
0 MH
to 0
O q
MH cu
M 4J
Cente
Depar
(U
0
M
3
O
CO
-------
potential for labor force entry of those who are not currently in the
labor force. Furthermore, a high employment to population ratio can
be interpreted unambiguously as a high labor force participation rate.
Employment per 100 population is high in the nation and has in-
creased gradually over the period (Table 5). The increase has been
interrupted by periodic recessions. Rural counties not adjacent to
urban centers also have high employment ratios, and the ratios are
tending to increase over the recent period. The data do not show a
more rapid increase in the employment ratio in rural counties than in
urban areas as in the past.
The major conclusion from these data is that the employment ratio
already is high in rural areas, but is continuing a gradual increase.
Further employment increases in rural areas will require either an
increase in labor force participation, inmigration, an upgrading of
jobs, or a combination of all three. An increase in labor productivity
through job upgrading could be an important contribution to rural devel-
opment, and would substitute for inmigration. It is clear that in areas
with rapid increases in employment, migration will be the major source
of labor because participation rates are already high in rural areas.
Thus, rural areas will be competing directly with urban areas for addi-
tional labor, especially in periods when the working age population is
increasing slowly.
INCOME CHANGES
Increases in wages of people in rural areas will be necessary to
attract inmigrants as employment expands. Elsewhere we have hypothe-
sized that the indigenous population would respond rapidly to higher
wages by entering the labor force (Bender, 1979). As labor force
participation rates increase in rural areas, new employment opportun-
ities will require even higher wage increases in order to attract
migrants. And if the labor force participation rates of people in
rural areas peak, then wages will have to rise at least to levels
competitive with those in labor sheds of surrounding areas.
Average earnings and per capita personal income both tend to be
increasing more rapidly in rural parts of the nation than in urban
areas. The patterns of the two measures are approximately the same
except that net farm income (after inventory adjustments) leaped 79
percent from 1972 to 1973, then returned to 1972 levels by 1976. This
farm entry distorts the rural income data. Average earnings of wage
and salary workers are used instead to illustrate the decline in in-
come differentials among rural and urban parts of the nation.
-------
vo
r-.
cn
r
\
CT>
r
«
CO
CO
1
CJ
z
CQ
a:
>.
CQ
z
o
1 1
1
o cn (sj
rt >
CJ rH
(H O
"^ § ^ C
fl .^ w
Co +^ C3 11 ft
P OP
H SH O
rH CD » (/)
ap o in
CO LO CD
O 3- tO tO
CD r*^ ^^
rf tO tO
rH OO " o cn o
CD o cn o
H o cn to
g o cn CM
g (S| rH
u
c2
JO cB
^ r^i
4>3
rH O
O 0) O t/>
P CO CD
CD, CD (S) J
C C3
Q
Z
«^
I
O
rH
P
£
O
P
CD
O
CO
rt
P
O
CD
H
4->
O
U
P*
P
iH
rH
O
CX
0
P
g
Q
2J
"* * tO
LO rH OO
<<» ^r to
vO rH OO
<3- * tO
LO rH 00
^^ ^^ r^^
rt «* ro
^ cn t--
^3- to to
rt- cn r--
* to to
"* cn t~»
»* to to
^»
o|
in
cO
%
rtgs§
H TH -H
P P P
cO cO cO
i 1 rH rH
ill
aaa
CO Co CTj
e-e-e
333
O O> O
o cn o
o en to
o cn rsT
fs] rH
Jo ro
P .*^
^.J
>H 0
CD O (/)
P to [/)
CJ
^
m
!-H
o
'O M
d CO
cO a
0 >
rJ 4-l -rl
O M
CO d
O J3
H O
4-1 1-3
CQ
H *T3
M d
01 CO
4J
o >
cfl d
co 1
vu w
rd M
O PQ
O T)
S >
O Cfl
d Q
o
o
M CO
0)
cfl p"{
rH
Cfl W
H
cj 'O
O 01
CD M
d ^
H ,£3
0> =
d «
°
0) CTl
CO >
CO CO
cfl 0)
i-H -H
O 4-1
>> §
4J O
d o
o o
O rl
4J
"cfll d
o
V
_to
to
*io
c
l^
O
c
P
o
fel
^H
0
3
to
(U
3
cq
0)
c;
4J
E
O
s; 4J
^S §
g
SH -4J
d) M
-U ft)
Q) Q)
O C$
,.
0)
o
y
o
CO
-------
Annual earnings of workers increased 64 percent between 1969 and
1976 in the most rural counties of the nation, from $4,300 to $7,109
(Table 6). The corresponding increase in large metropolitan centers
was 58 percent, from $7,299 to $11,521. The differential between these
two extremes remains wide. In part, it may reflect cost of living and
quality of life. But, the differential is slowly closing.
The relative earnings increases in rural areas may be due to
changes in the industry mix of employment, productivity of workers, or
the supply of labor. Regardless which is the cause, earnings increases
should tend to reduce emigration and increase immigration.
Relative earnings in counties containing no urban center and not
adjacent to an urban area were 67 percent of the national average in
1969 and 69 percent in 1976. Earnings in the most rural county groups
which were adjacent to an urban center increased 4 percentage points
to 73 percent of the national average by 1976. Earnings in the largest
metropolitan areas declined from 113 to 112 percent of the national
average over the period.
It is noteworthy that personal income per capita in the largest
urban areas declined from 117 to 114 percent of the national average,
a larger loss than reflected in earnings. Personal income includes
proprietor's incomes, interest, and transfer payments. Furthermore,
the income differentials adjusted for cost of living would be less
than those for earnings (Hoch, 1972a, b). The slow but persistent
increase in relative incomes in rural places will tend to hold and
attract population, and earnings should increase relative to the nation
as an inducement to emigrant labor.
ACCESS AND TRANSPORTATION COSTS
The continued participation of rural areas in the geographic
dispersal of service activities will depend upon the distribution of
services within functional economic regions the spatial structure
of economic activities. Access and transportation costs and scale
economies usually are isolated as the most important determinants of
the spatial location of activities. The question is whether high
relative petroleum prices will cause basic sectors to shift to urban
areas, and whether service activities will tend to centralize in
urban places.
The fundamentals of aggregate energy economics in the national
system after all trade offs have been accounted for are that labor
will substitute for energy, that the rate of capital formation will
decline, and that labor productivity will decline as energy prices
increase (Ozatalay, Grubaugh, and Long, 1979; Berndt and Wood, 1979).
-------
LU
>
i i
i
r^
i
LU -^.
OH rrj|
oo ^0
a: £
LU 01
jf~ f
Q£ 1
o cr>
3 vo
^r^
a:
*s
' >-
< r-
00 5
_, =3
Q 0
rf °
n|f>J
t^
cn
rH
rH
t^
CJ1
r-n;
" *
O
r--.
cn
rH
cn
\o
cn
rH
W
W
CO
rH
O
X
p
3
o
u
CM cn to
rH cn cn
rH
CM cn CM
rH cn cn
rH
CM cn CM
rH cn 01
I-J
T~n
to cn CM
rH cn ai
rH
to cn CM
rH cn cn
rH
to cn o
o cn o
A «S M
o cn o
o cn LO
t/j o cn csj
CD rH
JH O
^g^s
C42 S
3 P O 4^
P O -P
H rH O
rH 0 > 10
ap o w
CO LO CD
O^CMJ
P CJ
JU
2
oo oo to
oo r-- r--
oo r-- CM
oo r-- t-~
OO r-- rH
oo t-. r-
oo \o o
OO f^ t--
r-- \o o
00 t^- 1^
r~- vo o
oo r^- r^-
r-~ LO o
oo r-- t--
oo LO 01
00 t-- \D
^k
VI
cO
S
cO
O
^ C C C
P O O O
CD -H -H -H
g 4J +J 4_)
cO cO cO
O rH rH rH
P 3 3 3
PH PH PH
P O O O
CJ PH PH PH
CD
co co 3 3
^-.434P4P
Tj JH rH fH
CO 3 3 3
to o cn o
CD O O1 O
H O O1 LO
4-> > > »
g O Ol CM
3 CSJ rH
u
C
SCO
p ro P c
H 45 3
rH 4-1 4±
o +3
PH rH 0
O 0) O V)
in P LO t/5
P CO > CD
CD. CD CS] nj
5 u
O
2
>s
rJ
2
4-»
O
p"
o
4->
P
c
dj
u
CO
r-,
T3
rt
p
g
w
CD
H
P
O
0
c
c3
p
H
rH
O
PH
2
4-»
J-3
o
£
vo t^ cn
oo r>- \o
vo vo cn
OO t^- vO
LO LO OO
oo r-- \o
LO rr t-~.
oo r-- \o
& *t r~-
00 t-^ O
"3- * t
oo r-- vo
to «* \o
oo r~- vo
to r± t~-
oo r-- \o
Si
al
in
cO
CD
rH
CO
ggg
H -H -rH
P P P
CO cO cO
rH rH rH
PH PH PH
000
PH PU PH
C C C
cO cO cO
P-S-P
rH rH ,H
333
o cn o
o cn o
o cn LO
^ #« *
o cn CM
CM rH
Is S
P f
"P
rH 0
CD O W
P I/) Cfl
CO -CD
CD CM _q
JH
CJ3
O
O
rH
O
O
rH
O
O
rH
0
o-
rH
0
o
rH
0
O
rH
O
o
rH
O
O
rH
W
CD
H
4-)
o
o
CO
£3
rH
rH
CO
l f
cO
P
Q
H
m
r-
ON
,_(
43
13 CJ
Pi H
cd cd
S
o
iJ
Pi
f3 W
H >
4J CM
cd i~-
rH CM
ft PH
O W
P-< <
0) «
43 t-i
4J CO
<4-H §
O -H
N
CO
O fi
.-4 _S
Hi fj i
4J O
M >-}
rl
M 13
0) 13
4J Cd
fj
cd **
u (3
cd S
6 2
PQ
O
H T3
. 0 -H
,!
TH {3 cd
OOP
& CJ
fd >
<4-l CO
0 13 N O 13
43 CO QJ
S V-i
CO PH
01 pi
OQ-rl f*,
cd 43
r-l 13
cu Q) :
> C -
cd -H O
IM t-»
>»
4J TJ -H
pj
3 CO «
o co co
cj cd » 3
43 4-> O
00 O U
H 3
Q) o O
tS 0 rl
4J
0)
^.^, s
Cd 43 (3
0
S3
V
to
H
to
3i
M
*0
=2
O
H
1
§
O
o
£
8
3
IB
0)
M
3
CQ
Q)
.r*
**-H
M
r\
s
<*-!
31
+J
>rH
to
M
0)
H
q
s
CD
4J
IB
p
CO
IB
q
18
U
q
o
§ to
Q)
«. -H
W rH
H (U
to to
31
r~| (TJ
(8 -U
q tB
^ T3
(B Q)
U 0
S ^3
M i
(B 0
rH 0
0 'M
CO 0
M 4J
o q
IM
-------
The net change in the aggregate system due to energy prices is sum-
marized by the elasticities of substitutions cited above. Examples
of specific cases may be misleading for that reason. In general,
higher energy prices are consistent with the experience of the nation
in the past decade. More labor is being used as energy prices increase,
and the productivity of labor is declining. Since capital and energy
are complements, higher energy prices retard the rate of capital
formation.
The rural economy is only a subset of the national system and the
same general effects may not be felt in rural areas. At this time it
is a subject of speculation. However, in the farm sector, it would
imply a slower adoption of energy consuming technology, hence, a slower
release of labor from farming than in the past. This would tend to
stabilize employment in a basic economic sector of rural areas. Food
manufacturing is a uniquely rural industry and possibly would be
affected in the same way.
The effect on the location of manufacturing activities in rural
areas is subject to many more trade offs. One of the original reasons
that manufacturers located in rural areas was that their transportation
costs were relatively unimportant. Higher gasoline costs may change
that matrix somewhat. On the other hand, there may well be a regional
redistribution of some industries to areas which have less expensive
process energy costs.
Possibly the greatest effect on the structure of rural economies
will be in the service sectors. Increased gasoline costs should reduce
the number and length of shopping commutes by rural people. The re-
sulting increase in the demand for services locally even though prices
might be higher would complement the long-term trend of increasing
service activities in rural parts of the nation at the expense of
regional service centers. As mentioned earlier, the major function
of regional service centers in this case may be to provide enough
excess capacity to absorb the variable service sector demand of its
hinterland.
SUMMARY AND DISCUSSION
The population turnaround in rural areas calls attention to the
possibility of structural changes in rural economies, changes which
may have caused or which accompanied the population turnaround. Inte-
gration of rural economies into the national system, greater dispersion
of services into rural areas, increased demand for service activities,
higher labor force participation rates, and higher relative incomes
-------
are all reflected in data for rural areas. These changes appear to be
caused by a complex set of forces which we failed to take account of
in regional models the technology of production (especially in ser-
vices) , advances in communication, changes in transportation systems
and the relative cost of transportation, the functions of towns and
cities, and changes in values, ethics, and preferences. Although these
data may alert us to structural changes which should be a part of re-
gional modeling of employment and population, there is a long step be-
tween the recognition of the importance of structural changes and the
inclusion of them in economic models.
A regional economy is a subset of the nation. These data suggest
treating it explicitly in that way rather than as an isolated economy
with a constant structure. To a large extent, the behavior evident in
a county economy is dominated by and a reflection of what is happening
in its national cohorts. Thus, one would expect increases in relative
incomes, labor force participation, and service activities even though
there may not have been a change in the economic base or a new surge
in consumer demand. Too often these effects are ascribed to rapid
changes in the economic base of growing areas. A major deviation of
behavior in a county from the norm of its national cohort would be
expected if background conditions such as rapid changes in its economic
base were occurring. These differences between conditionally expected
events in national cohort counties and the observed values of a sample
of object counties could define the adjustments being made in a rapid
growth situation. A large increase in the economic base would precip-
itate a rise in wages and salaries, participation rates, and service
employment greater than that of other counties not experiencing that
change. Thus, long-term structural changes and business cycles would
become a part of the norm of the system.
Nerlove (1979) describes the ideal toward which we might strive
and provides some general insights into initial attempts at modeling
the dynamics of economic change.
"... An econometrically relevant dynamic theory
would characterize response paths of economic agents
who are optimizing their behavior under dynamic con-
ditions and forming expectations of the future on the
basis of all information available to them. Such a
theory would not, in general, involve the notion of a
long-run equilibrium toward which adjustment is being
made nor simple forms of stationary expectations."
His application is on perennials and livestock in developing agricul-
tural economies. It is not the purpose here to review his work, but
only to suggest that his summary of the state-of-the-art is important
51
-------
at this stage. Nerlove's parting remark is "Sophisticated econometric
techniques and high-powered economic theory are complementary, not
antithetical, to case studies and common sense." It appears that there
is a great deal of room for all types of research in this area if we
are to come to grips with the set of forces affecting rural areas.
The Economic Development Division's objective in this research is
to more fully understand the process of adjustment to rapid growth in
rural areas. We are planning additional work this year as a part of
the Energy Impact Project. The first step is the selection of a sampl-
ing frame of counties. That involves screening for metropolitan status,
size of urban population, adjacency to large metropolitan areas, and
anomalies such as large minority populations, limitations on land use,
presence of Federal establishments, concentration of state and local
employment, retirement and tourist areas, and universities. From the
screened sampling frame will be selected counties which will be used
as national norms, and counties with changes in their economic base
which appear to be desirable for estimating coefficients which apply
to rapid growth areas. Coefficients will be re-estimated for the cur-
rent equations in the COALTOWN model. At some point, new equations
and a new model structure will be conceived.
We are working with undeleted BEA data for counties for the period
1969-1977. The data are commonly used for this purpose but data short-
comings are evident. Very early in our efforts this year, we will be
faced by the serious trade off between the complexity of the model and
its specification, and the complexity of the econometric problems faced
and techniques used.
52
-------
References
Alonso, William
1977 "Surprises and rethinkings of metropolitan growth: A comment."
International Regional Science Review 2(2):171-174.
/
Ashby, Lowell D., and David W. Cartwright
n.d. Regional Employment by Industry, 1940-70. Washington, B.C.:
U.S. Department of Commerce, Bureau of Economic Analysis.
Beale, Calvin L.
1975 The Revival of Population Growth in Nonmetropolitan America.
ERS-605. Washington, B.C.: U.S. Department of Agriculture,
Economic Research Service.
1977 "The recent shift of United States population to nonmetropolitan
areas, 1970-75." International Regional Science Review 2(2):
113-122.
Beale, Calvin L., and Glenn V. Fuguitt
1978 "The new pattern of nonmetropolitan change." Pp. 157-177 in
Karl E. Taeuber, Larry L. Bumpass and James A. Sweet (eds.),
Social Demography. New York: Academic Press.
Bender, Lloyd D.
1979 EDO Local Impact Model: A Progress Report. Paper presented
to Great Plains Coordinating Committee 8 meetings, San Antonio.
January 10-15.
Berndt, Ernst R., and David 0. Wood
1979 "Engineering and econometric interpretations of energy-capital
complementarity." The American Economic Review 69(3):342-354.
Bluestone, Herman
1979 Income Growth in Nonmetro America, 1968-75. Rural Development
Research Report 14. Washington, D.C.: U.S. Department of
Agriculture, Economics, Statistics, and Cooperatives Service.
Brown, David. L., and Jeanne M. O'Leary
1979 Labor Force Activity of Women in Metropolitan and Nonmetropol-
itan America. Rural Development Research Report 15. Washington,
D.C.: U.S. Department of Agriculture, Economics, Statistics,
and Cooperatives Service.
-------
Burns, Leland and Robert Healy
1978 "The metropolitan hierarchy of occupations." Regional Science
and Urban Economics 8(4):381-393.
Carlin, Thomas A., and Linda M. Ghelfi
1979 "Off-farm employment and the farm sector." Pp. 270-273 in
Structure Issues of American Agriculture. Agricultural Eco-
nomics Report 438. Washington, B.C.: U.S. Department of
Agriculture, Economics, Statistics, and Cooperatives Service.
Hines, Fred K., David L. Brown, and John Zimmer
1975 Social and Economic Characteristics of the Population in Metro
and Nonmetro Counties, 1970. Agricultural Economics Report 272.
Washington, D.C.: U.S. Department of Agriculture, Economic
Research Service.
Hoch, Irving
1972a "Income and city size." Urban Studies 9(3):299-328.
1972b "Urban scale and environmental quality." Pp. 235-284 in
Ronald G. Ridker (ed.), Commission on Population Growth and
the American Future. Research Reports of the Commission on
Population Growth and the American Future. Ill: Population,
Resources, and the Environment. Washington, D.C.: U.S.
Government Printing Office.
Latham, William R. Ill
1978 "Measures of locational orientation for 199 manufacturing
industries." Economic Geography. Vol. 5:53-65.
McCarthy, Kevin F., and Peter A. Morrison
1977 "The changing demographic and economic structure of nonmetro-
politan areas in the United States." International Regional
Science Review 2(2):123-142.
Nerlove, Marc
1979 "The dynamics of supply: Retrospect and prospect." American
Journal of Agricultural Economics 61(December):874-888.
,
Ozatalay, Savas, Steven Grubaugh, and Thomas Veach Long II
1979 "Energy substitution and national energy policy." The American
Economic Review 69(2):369-371.
Petrulis, M.F.
1979 Growth Patterns in Nonmetro-Metro Manufacturing Employment.
Rural Development Research Report 7. Washington, D.C.: U.S.
Department of Agriculture, Economics, Statistics, and Coopera-
tives Service.
-------
Thompson, Wilbur R.
1967 "Toward an urban economics." In Leo F. Schmore and Henry Fagin
(eds.)> Urban Research and Policy Planning.
Schwarzweller, Harry K.
1979 "Migration and the changing rural scene." Rural Sociology
U.S. Department of Commerce
1979 Regional Economic Information System. Washington, B.C.: U.S.
Department of Commerce, Bureau of Economic Analysis. Unpub-
lished data.
U.S. Department of Labor
1974-78 Employment and Earnings. Vol. 21-26. Washington, D.C.:
U.S. Department of Labor, Bureau of Labor Statistics.
Vining, Daniel R. Jr. , and Thomas Kontuly
1978 "Population dispersal from major metropolitan regions: An
international comparison." International Science Review
3(l):49-73.
55
-------
A MANAGEMENT INFORMATION SYSTEM FOR ASSESSING
LOCAL IMPACT OF ENERGY DEVELOPMENT
William F. Hahn
William R. Schriver I/
INTRODUCTION
The Construction Labor Demand System (CLDS) is an outgrowth of
continuing efforts under four Secretaries of Labor to obtain more
accurate information about the construction industry and to enhance
planning and policy-making relative to construction both in and out
of government. These include methods of monitoring industry activity
and labor market conditions, as well as a capability to conduct research
on construction and construction labor markets. This information can
be used to influence industry activity in beneficial ways. The concept
was initiated during the period of economic controls in 1972 when there
was severe wage and cost inflation in construction. Those first track-
ing systems have evolved into a sophisticated system which is being
used as an important economic forecasting tool.
CLDS is a management information system that has been designed to
provide users in the Department of Labor (DOL) and others with short-
term forecasts for 1 to 5 years of the volume, type and regional loca-
tion of future construction activity and the associated on-site
requirements by individual crafts. Separate forecasts are developed
for each state in the U.S., for each of 35 types of construction and
29 construction occupations, and for privately and publicly owned
facilities. The function of the system is to estimate construction
activity by type in geographical regions, primarily states, translate
the estimated construction volume into labor requirements by converting
project characteristics (dollars, megawatts, etc.) into craft labor
time and forecast both construction activity and labor demand.
J7 William F. Hahn is with the U.S. Department of Labor,
Washington, D.C. William R. Schriver is with the University of
Tennessee, Knoxville, TN.
56
-------
CONSTRUCTION, JOBS AND LOCAL IMPACT
Thousands of new jobs will be created annually during the next
decade as a consequence of the development, transportation, and pro-
cessing of fuels and the development of new electrical generating
capacity and other heavy construction, both of which are required for
continued economic growth. These jobs will involve construction, main-
tenance, and operation of facilities. Individual projects in many cases
will be so large in size that they will have drastic impact on local and
regional labor markets.
The CLDS program has unique capabilities with respect to assessing
the local impact of major development particularly in the energy area.
One of the objectives in initiating the CLDS effort was to develop, for
the participating Federal agencies, a capability to assess the labor and
other resource requirements of alternative U.S. energy development
scenarios. To accomplish this goal, the CLDS staff has initiated the
development of a continuously updated computerized file of all announced
and on-going energy projects in the U.S. (electric generating plants,
transmission lines, and fuel extraction, processing, and transportation
facilities). Under authority granted to CLDS by OMB, each energy con-
struction project is monitored on a periodic basis; project stocks are
converted into time-phased flows of labor requirements construction,
operation and maintenance.
The implications for rural job development are enormous and are
discussed in detail in the last section. The Construction Labor Demand
System is described first so that the reader can assess its applicability
to rural job development.
Project Data Base
Detailed and accurate information on construction projects is the
nucleus of the CLDS system. It plays a critical role for assessing the
construction situation and labor markets but also provides valuable in-
sight to future levels of activity for those types of construction
featuring long-lead times and lengthy construction periods.
Project data are entered into CLDS from private and government
sources. The most important source in terms of quantity of project
records is the monthly statistical file from the F.W. Dodge Division
of McGraw-Hill Information Systems Company, Incorporated. A monthly
magnetic tape from Dodge (Dodge Construction Potentials) provides
information for 267 structure types of projects about to begin construc-
tion. From Dodge sources, 125,000-175,000 new projects are entered into
CLDS each year.
57
-------
Information on planned energy-related construction projects is
gathered from other sources. The data source for fossil-fueled and
hydroelectric units is the Energy Department's Generating Unit Refer-
ence File. This file, updated semiannually by CLDS questionnaires to
the utility companies, provides all of the plant and location charac-
teristics necessary to generate manpower requirements.
The data source for all nuclear electric generating facilities,
either presently under construction or planned for future construction,
is the Construction Status Report (Yellow Book) of the Nuclear Regula-
tory Commission's Office of Management and Program Analysis. This file,
updated monthly by NRC, provides all plant and location characteristics
necessary to generate manpower requirements.
For energy projects other than powerplants, CLDS has a cooperative
arrangement with the Department of Energy (Energy Information Adminis-
tration, EIA), whereby we maintain a computerized file containing a
record for each planned (announced) energy project in the U.S. (approx-
imately 90 different types of fuel extraction, processing, and transpor-
tation construction projects). EIA field offices collect the initial
project data from secondary sources and report the data to CLDS. DDL
then contacts project owners by telephone to obtain missing data on the
projects which are required to convert stock of projects into flows of
on-site construction (and operation and maintenance)labor requirements.
Labor Conversion
The construction activity-labor requirements transformation
utilizes survey and engineering derived data to convert a stream of
construction activity by final use characteristics into labor require-
ments over time.
For non-energy construction, CLDS translates given levels of
activity into man-hour estimates by applying base-year coefficients
of labor craft time per dollar to each type of construction. The
estimated real value of construction (nominal value deflated by an
appropriate construction cost or price index) is multiplied by a labor
conversion factor to obtain estimated work-hours for each on-site
occupation. For energy projects, the labor requirements flow and
duration for specific projects are a function of type, output, cost
and engineering criteria.
The construction period is subdivided into deciles and labor
requirements are estimated separately for each decile, since labor
involvement for any given construction trade varies over the duration
of the project as different construction stages ensue. If a given type
of construction project of a certain dollar volume is estimated to take
-------
20 months to complete, the CLDS program would automatically estimate the
labor requirements by craft for each two month period of construction.
Factors for accomplishing the conversion are derived from BLS
surveys of labor and material inputs for selected types of construction
and on the basis of engineering evaluation of construction resources
needed to build facilities. Approximately 250 separate conversion pro-
files are included in the system.
Forecasting
Future and as yet unknown construction initiations are generated
from models developed by CLDS for energy-related construction, and for
CLDS by the Institute for Defense Analyses (IDA) for other types of
construction. The logic and assumptions underlying both the energy and
nonenergy components are similar enough to allow aggregation when desired
into one overall forecast for a designated geographical area.
Briefly, the forecasting method for nonenergy construction is as
follows: The first step is utilization of a set of forecasted regional
economic activity and demographic variables. The projected values re-
present a capsule summary of the anticipated health of a region, its
rate of economic growth, the composition of its industry, and the age
and social structure of its population.
The National Regional Impact Evaluation System (MIES) model
developed by the Bureau of Economic Analysis of the Commerce Department
is the source of the economic and demographic forecast. It is a set
of 51 state models estimated from time series data using a common
theoretical structure. Each state model is composed of 69 behavioral
equations and generates the state forecasts which provide the regional
inputs to our construction model. An interagency agreement between DOL
and DOC provides for access to NRIES by CLDS and also makes provisions
for periodic modification by the NRIES staff of the model to meet CLDS
requirements.
A model in CLDS relates construction activity in each state to
local and national variables which are thought to influence the level
of activity. The model converts measures of construction activity and
demographic characteristics into estimates of construction starts for
each type of construction included in the system. For example, private
school construction is related to the size of the school population and
disposable income. These variables, plus tax revenue per capita as a
measure of fiscal efforts, are used to explain demand for public schools.
Hospital construction is related to income, fraction of the population
who are elderly, and Federal grants in aid.
59
-------
Three steps are required to forecast nonelectric energy construction
activities in terms of energy supply facilities. First, the forecasted
energy supply (production) including energy transportation is generated
by type and region over time from selective energy supply models such as
Midterm Energy Forecasting System in the U.S. Department of Energy,(the
basis for the Annual Report to Congress, 1978 by EIA of DOE). Second,
new additions of required energy capacities by region are estimated,
based on the forecasted energy supplies.
Required information includes capacity utilization, existing capa-
cities, facility retirements and known projects planned by producers. An
Energy Pseudo-Project Generator in the CLDS model prescribes construction
requirements to meet energy needs. It generates information on calendar
month of construction start, project size, project type, and location.
IMPLICATIONS FOR RURAL DEVELOPMENT
A casual analysis of the energy project file reveals one fact of
major significance energy projects are predominantly rural in situs.
When the manpower requirements profiles for energy construction are
compared to the labor force in many counties in which future plants
are scheduled for construction, the peak construction work force, in
many instances, will exceed the county's entire nonagricultural labor
force. Boomtown phenomena will be commonplace without adequate planning
and intervention.
All too often craftsmen from distant parts of the nation (and
Canada) immigrate to the site of major energy projects, attracted by
potential annual earnings often exceeding $30,000, while local young
people entering the labor force continue trekking to distant cities
seeking jobs with much lower earning capacities than those of the new
jobs in their native counties.
Although unionized projects and open shop projects vary with
regard to utilization of locally trainable labor, either arrangement
can lead to an inefficient utilization of the local labor supply unless
special labor recruitment and training activities are undertaken.
Employment in the construction industry is inherently unstable
due to certain immutable characteristics of its product market. Each
project is essentially an immobile, customized, unique product; there-
fore, productive factors must be shifted among the different owners'
sites. This is translated into a continuous assemblage and disassem-
blage of work forces by constructors as jobs progress from one stage
to another and end at one location and begin at another. Frictional
(as well as seasonal) unemployment is indigenous to the industry and
60
-------
is capitalized into the wage structure. Construction workers are
accustomed to commuting long distances to or temporarily relocating
near project work sites. Thus, the mobility of inputs rather than
outputs in the construction industry induces local expansion from the
influx of construction workers directly and indirectly as a consequence
of their consumption of local goods and services.
CLDS can serve as an early warning system to identify geographic
areas, specific projects, and even specific construction crafts where
skill supply/demand imbalances are likely to occur as a consequence
of energy development (and, for that matter, as a consequence of other
types of construction). Local impact will be lessened by developing
special cooperative recruitment programs and innovative apprentice or
other training program arrangements (such as accelerating the first and
second periods of apprenticeship by utilizing specially structured
programs in local area vocational schools or technical institutes).
Each local person trained in a critical craft reduces by one the movers
to the project, but the dampening effect on overall expansion is much
greater. (This point is subsequently treated in greater detail.) The
average cost of employment to the work force is decreased since fewer
workers have to maintain two residences, and social costs to the com-
munity are substantially reduced. Inflationary pressure on construc-
tion wages may be reduced because fewer movers have to be attracted.
When construction on the project is completed and operation of
the facility is begun, a different and much smaller work force is
required, often only 5 or 10 percent of peak construction employment.
Some of the local construction craftsmen may become employed as main-
tenance craftsmen, but most of the operational work forces consist of
managers, engineers, technicians, clerical workers, and security
personnel.
The new local entrants into the construction trades would have to
find new employment, as do the other craftsmen, when construction is
completed; their alternatives to use their newly acquired human capital
are the same (although utility functions may have quite different shapes),
They may become movers within the construction industry; they may work
for lower wages in local residential construction or local maintenance;
or they may move to other areas to work at related jobs outside the con-
struction industry. In any case, their relative wage positions would
have been improved.
Thus far, only the direct job openings created in the construction
industry have been considered. There are also obvious secondary employ-
ment effects which may be of interest to rural job planners, although
these jobs may generally be less skilled. The exogenous shock of a
new energy project induces two additional effects beyond direct growth
61
-------
in the construction industry. Output is increased by those local
industries producing input goods and services for the construction
industry and, as a consequence, employment in those industries is
increased (and a likewise ripple through their input industries is
initiated). CLDS (as do input/output results) will estimate time-
phased profiles for 50 types of materials and equipment inputs for
each energy type. An analysis of the structure of the local economy
could then be made to estimate the indirect changes in employment by
industry and by occupations associated with the expected change in
output over time.
Another type of secondary employment that derives from the shock
of an energy project is income-induced employment. The share of the
construction workers income spent locally initiates a wave of economic
activity which expands throughout the economy, the magnitude being
determined primarily by the amount of income spent locally and its
velocity. As local firms expand their output in response to consump-
tion by the construction workers, their employment is expanded.
Although CLDS will not directly estimate income-induced or
indirect employment from energy construction projects, systematic
estimates could be made by merging the output of CLDS into the
Industrial-Occupational Labor Demand Model developed at the University
of Utah for DOL. This latter modeling effort involved an empirical
investigation of actual local labor market situations, making use of
an econometric modeling technique that incorporates a dynamic indus-
trially disaggregated formulation of the traditional economic base
concept.
In conclusion, CLDS in tandem with other regional models such as
DOL's Region VIII Energy Impact Model, can be used to estimate direct
and secondary job openings that are likely to result from the introduc-
tion of an exogenous energy project. The results may be used to lessen
boomtown impacts and serve as a major strategy for job creation and
training in rural areas by coordinating the training of local people
with local job openings.
62
-------
PROFILE OF INMIGRATING WORKERS AT LARGE CONSTRUCTION
PROJECTSIMPLICATIONS FOR MODELING
Suresh Malhotra
Diane Manninen I/
INTRODUCTION
This paper focuses on the work force composition at nuclear power
plant construction sites and attempts to determine the extent to which
occupational groups exhibit differences with respect to variables which
are critical to socioeconomic impact assessment. These findings have
important implications for socioeconomic impact assessment and provide
valuable insights for the development of improved forecasting procedures.
The results presented here represent the first phase of a study
(Malhotra and Manninen, 1979) to develop methods by which information
regarding the local labor profile prior to plant construction, antici-
pated construction worker requirements, and various regional character-
istics can be used to predict:
1. The number of workers who will move to the area to work
at the construction site; and
2. The residential location pattern of inmigrating workers.
This paper examines four variables critical to estimating socioeconomic
impacts migrant proportions, relocation of dependents, intention to
remain in the area, and demographic characteristics of movers and
presents a brief summary of our findings.
JL/ Battelle Human Affairs Research Centers, Seattle, Washington.
This paper is based on a study in progress. The study is being sup-
ported by the U.S. Nuclear Regulatory Commission, Contract EY-76-C-06-
1830, FIN B 2265-9. The opinions expressed in this paper represent
those of the authors and do not necessarily reflect the opinions of
the Nuclear Regulatory Commission.
63
-------
The discussion is divided into three sections. The first section
introduces the data set used in the study and outlines the analyses
performed. The second section presents the analysis results and dis-
cusses the implications of these findings for estimating socioeconomic
impacts of large energy development projects. The final section sum-
marizes the implications of these findings for the development of
improved forecasting procedures.
ANALYSES PERFORMED
The data used in this study include the responses of over 49,000
workers from 28 surveys conducted at 13 nuclear power plant construction
sites. These data were obtained in two ways. Worker surveys were
conducted at four nuclear power plant construction sites specifically
for this study. In addition, we were able to increase the number of
sites in our analysis by including data from similar surveys which were
conducted by utilities at nine additional sites.
These 28 surveys exhibit a wide range of variation in a number of
important dimensions. For example, considerable variation can be
observed with respect to regional distribution, the existence of other
power plant construction activity in the region, local community
characteristics, stage of project completion, utility and contractor
arrangements, extent of unionization, and various power plant charac-
teristics. Consequently the data can be expected to yield estimates
of critical socioeconomic variables under a variety of situations. This
increases the usefulness of such an analysis, since an examination of
these data is more likely to yield results which are generalizable to a
variety of power plant construction sites.
The variables which were included in the analysis are defined as
follows:
1. Migrant proportion the proportion of workers who moved
to the area to work at the site; 2/
2/ Workers were classified as movers and nonmovers in one of two
ways, depending upon the survey instrument used. Some questionnaires
asked workers whether or not they had moved to the area to work at the
site. Workers who responded "yes" to this question were considered
to be movers. Other questionnaires asked for the worker's current
address and their address before beginning work at the site. A com-
parison of these responses was used to classify workers as either
movers or nonmovers.
-------
2. Relocation of dependents the proportion of movers with
family present;
3. Intention to remain in the area the proportion of movers
who are temporary (movers who intend to leave the area
before or upon completion of the project); J>/ and
4. Demographic characteristics of movers marital status,
family size, number of school-age children and income.
An accurate estimate of the number of immigrating workers is
critical to socioeconomic impact assessment because this projection
is the single most important determinant of the magnitude of most
other impacts. Similarly, consideration of information regarding
relocation of dependents, intention to remain in the area, and the
demographic characteristics of movers may be important because
household composition and family size influence movers' demands for
housing, consumer goods, and public services.
We examined the variation in each of the above variables across
sites both overall and for various subgroups of workers with a
view toward identifying patterns of association between variables
which would be useful in the specification of the models to be es-
timated in the multivariate analysis portion of the study. No
attempt was made to explain the observed patterns of variation
during this portion of our analysis.
Differences in migrant proportions among various occupational
groups have been observed in past studies (Mountain West Research, Inc.,
1975). However, no attempt has been made to examine differences in
variables such as migrant proportions, intention to remain in the area,
and relocation of dependents among various worker groups in any system-
atic way. The intent of this analysis was to determine the extent to
which workforce composition could be an important factor in understand-
ing the observed variation in the variables of interest across sites.
_3/ Two different measures were defined to examine this variable.
First, based upon workers' responses to a direct question regarding
their intention to remain in the area, we classified movers as either
permanent or temporary movers. Temporary movers were defined to be
those movers who expected to leave the area either before or on com-
pletion of the project. Second, as an indirect measure, we examined
the proportion of movers who maintain a permanent residence elsewhere
as an indication of intention to leave the area upon completion of
the project.
-------
The importance of considering work force composition stems from
the fact that both the demand for labor at the construction site and
the availability of labor within commuting distance vary for different
craft groups. For instance, crafts with more specialized skills are
required for jobs of rather limited duration, whereas other crafts are
required throughout the entire construction phase. Moreover, other
opportunities for employment in the area during and upon completion of
the project vary considerably for different crafts. Since it is likely
that these factors influence workers' relocation decisions, our analysis
attempts to measure the extent to which differences in migrant propor-
tions, relocation of dependents, and intention to remain in the area
are observed among various craft groups.
First, we divided workers into two broad groups construction
and nonconstruction workers, kj Each of these groups was then further
subdivided. In the case of the nonconstruction group, workers were
divided into two subgroups. The first group included all managers,
engineers and supervisors. The second group included the remaining
nonconstruction workers (i.e., the clerical, security, and medical/
nursing staff).
We divided construction workers into three groups based on the
relative scarcity of labor among the different construction crafts.
This classification of crafts into scarcity groups was based on two
factors: (1) the size of the craft labor markets, and (2) the demand
for their services at nuclear power plant construction sites. We
defined our proxy measure of relative scarcity of labor to be the
ratio of the number of manhours of labor required in each craft to
construct a typical nuclear power plant to the total number of union
members in the construction craft in the nation. 5/
This ratio was used to classify all construction crafts except
laborers and teamsters into two scarcity groups: a scarce craft group
and a common craft group. The scarce craft group includes pipefitters,
ironworkers, asbestos workers, and boilermakers. Carpenters, electri-
cians, operating engineers, sheetmetal workers, bricklayers, and cement
4_/ Information on workers' crafts was not available for four of the
surveys in our sample. Therefore, it was necessary to exclude these
sites from our analysis of variation among various subgroups of workers.
5j We recognize that this measure of relative scarcity of labor is
rather crude. Ideally, this measure should have been based upon re-
gional estimates of the demand and supply of the various construction
crafts. Unfortunately, data to develop such a scarcity measure were
not available.
66
-------
masons comprise the common group. Laborers and teamsters formed the
third scarcity group an abundant craft group. We treated laborers
and teamsters differently from the other crafts because these jobs do
not require very specialized skills. These occupations do not have
lengthy apprenticeship programs, as is the case with other construction
crafts. Therefore, it is relatively easy to meet an increase in the
demand for laborers and teamsters by attracting workers from other
occupations.
In addition to our examination with respect to various worker
groups, we also examined possible variation in relevant variables with
respect to region. Among the 13 sites included in our study, four were
located in northern states and nine were located in southern states.
We examined the data in an effort to determine the extent to which
systematic differences exist between northern and southern sites.
RESULTS
Migrant Proportion
Table 1 presents a comparison of the absolute ranges and typical
values both overall and for various worker groups for the 28 sur-
veys included in this study, bj Large variation was observed in over-
all migrant proportion, and in the migrant proportions of most worker
groups. Typically, overall migrant proportions ranged from 17 to 34
percent; however, there was considerable difference between the migrant
proportions of the construction and nonconstruction groups. In general,
migrant proportions among nonconstruction workers were two to three
times higher than the migrant proportions among construction workers.
Migrant proportions among nonconstruction workers typically ranged from
40 to 58 percent, whereas among construction workers migrant proportions
typically ranged from 11 to 29 percent.
Differences were also observed when these two groups were further
subdivided. The high migrant proportions observed among the noncon-
struction group essentially reflect the extremely high migrant propor-
tions of the management subgroup. In fact, migrant proportions among
clerical workers were similar to those observed among construction
workers. Migrant proportions among the management group typically
ranged from 58 to 76 percent. Migrant proportions among the clerical
group, on the other hand, typically ranged from 21 to 36 percent.
6/ We defined the range of typical values to be the smallest range
of values which contained at least 75 percent of the observations of
a particular group.
67
-------
TABLE 1. VARIATION IN MIGRANT PROPORTION, OVERALL
AND FOR VARIOUS WORKER GROUPS
Overall
Construction
Nonconstruction
Management
Clerical
Scarce
Common
Abundant
Migration
Absolute
Range
15-50
7-52
28-68
34-86
10-46
6-69
7-51
3-23
Proportion
Typical
Values
17-34
11-29
40-58
58-76
21-36
17-45
9-28
5-12
-------
Within the construction group, the differences in migrant propor-
tions among the three scarcity groups were not as extreme. Nevertheless,
the differences were significant. Migrant proportions among workers
from scarce crafts, which typically ranged from 17 to 45 percent, were
higher than those among workers from the common crafts, which typically
ranged from 9 to 28 percent. Migrant proportions among these two groups
were considerably higher than migrant proportions among workers from
the abundant craft group, which typically ranged from 5 to 12 percent.
Furthermore, it is interesting to note that in the case of the abundant
group, little variation in migrant proportions was observed from site
to site. Much larger variation, however, was observed in the case of
the scarce and common craft group.
These differences in migrant proportions are not surprising con-
sidering the likely availability of these groups in the area immedi-
ately surrounding the construction site. Consider, for example, the
extreme difference between the management and clerical groups. Manage-
ment workers are generally better educated and have relatively high
salaries. These workers are not likely to be readily available in the
area, especially in the case of more rural sites. Instead, many of
these workers are regular employees of the utility or prime contractor
and are moved to the area specifically to work on the project. Clerical
workers, on the other hand, have jobs that require less education and
training and, as a result, most of them are hired locally.
Construction workers also are more likely to be available locally.
This is especially true among certain worker groups such as teamsters
and laborers, since these jobs do not require extensive prior training.
There are, of course, several rather specialized construction crafts,
such as those represented in the scarce craft group, and the local
labor supply may not be able to provide a sufficient number of such
workers. Therefore, it is not surprising to observe greater numbers
of movers among crafts which are relatively scarce.
As can be seen in Table 2, regional differences in migrant propor-
tions were not very pronounced. Although we did observe some evidence
of higher overall migrant proportions in the South, the difference
between the two groups was not significant. A much more noticeable
difference was observed when construction and nonconstruction workers
were examined separately. Little difference was observed in the case
of the construction group. However, in the case of the nonconstruction
group, we observed a more obvious regional difference (migrant propor-
tions typically ranged from 40 to 50 percent in the North and 52 to 56
percent in the South). Furthermore, in a further disaggregation of
these two broad groups we observed that this difference was primarily
due to a very pronounced difference for the clerical group. Migrant
proportions among the clerical group typically ranged from 10 to 22
percent in the North compared with 26 to 30 percent in the South.
69
-------
^
o
CD OO
LU ' Q-
O
03 C?
Z ! 0£
O LJ
^"^ ^^
(-' -O:,
o; o
O ~3-
a.
g CO
<*- 0
z S
o? >
CD Q;
£ £
5 Q
2
CO
3 O>
r- C
O *O
woe:
^
**
r- O>
s§
jQ C£
CM CM
CO CM
1 1
CM r-
r>- CM
«± in
i i
r >^
CM i
UJ CO
CM CM
1 1
CM
CM i
V£> in
CM CM
1 1
in CM
r r
c
o
.,
4-)
U
f i.
10 4->
i- I/)
0) C
> o
0 0
vo
in
i
CM
in
co
10
i
CO
co
o
o
o
in
00
CM
c
o
r
0
3
<~_
4^
I/)
C
O
o
c
o
z
CO O
r^ co
i i
VO CM
in o
r**> co
i i
in co
O CM
r>. CM
i i
o
o r^»
r-. CM
«* 0
co r
4J
c
Q) r~
^S 'O
0) U
(0 $1
C O)
s: o
CT>
CO
1
co
CM
CT»
vo
t
in
00
CM
i
CM
CM
CO
1
CM
CM
-------
Although interesting, the reason for this regional difference in
migrant proportions among clerical workers is not obvious. It is per-
haps related to the differences in local availability of such workers
in the two regions. Alternatively, the pattern could reflect differ-
ences in the hiring practices of the utilities or prime contractors.
The results of this analysis have clearly demonstrated a system-
atic variation in migrant proportions across worker groups. This
implies that workforce composition could be an important factor in
explaining the variation in overall migrant proportion across sites.
Table 3 presents the proportion of the total workforce in each of ten
craft groups at the time of the survey for the 24 surveys for which
information on craft was available. As can be seen in the table, work-
force composition varies considerably from site to site and at the same
site at different stages of project completion. Because migrant pro-
portions vary considerably for different worker groups and because
workforce composition varies across sites, the variation in overall
migrant proportions may, to a large extent, reflect differences in
workforce composition. Thus, any attempt to explain the observed vari-
ation in migrant proportion across sites should explicitly include a
consideration of workforce composition at the site at the time of the
survey.
In addition, the results of this analysis can be used to identify
the underlying factors which explain the variation in migrant propor-
tions across different worker groups. While many such factors exist,
in general, variables can be classified into the following two groups:
those which reflect the availability of labor in the area and those
which reflect the relative attractiveness of employment opportunities
offered by the construction project.
The availability of labor varies considerably across craft groups
as well as across sites, depending upon factors such as the urban/rural
nature of the surrounding area. If there are several large cities with-
in daily commuting distance of the site, one might not observe large
numbers of workers moving to the area to work at the site. This may
be particularly important among workers from the scarce crafts whose
availability may be somewhat limited in rural areas. The attractiveness
of employment opportunities for various craft groups depends upon the
nature of labor requirements. This includes a variety of factors such
as the total number of workers required, and the expected duration and
continuity of employment. There is considerable variation in these
labor requirement variables both across sites and across crafts. Since
it is these variables which determine the expected value of benefits
associated with employment at a particular site, it is therefore likely
that these variables will also influence workers' relocation decisions.
Thus, this analysis suggests that any model which is developed to
explain the observed variation in migrant proportions should include
71
-------
TABLE 3. WORKFORCE COMPOSITION AT TIME OF SURVEY BY CRAFT
GROUPS (PROPORTION OF TOTAL WORKFORCE)
Survey
Identification
Number
1.0
2.0
3.0
4.0
8.0
9.1
9.2
9.3
9.4
10.1
10.2
10.3
11.1
11.2
11.3
11.4
12.1
12.2
12.3
12.4
13.1
13.2
13.3
13.4
Pipefitters
5.9
18.8
18.0
16.9
22.3
6.2
10.3
11.0
14.8
11.4
25.8
17.6
5.5
9.2
8.5
24.4
5.1
9.2
13.3
21.4
8.1
7.8
9.7
9.7
Ironworkers
14.0
11.9
17.6
15.5
2.4
7.1
12.2
3.4
12.1
8.1
6.8
5.4
12.5
13.2
2.0
1.9
1.5
10.9
8.8
6.3
7.9
6.4
Vl.l
12.5
Boi lermakers
4.3
2.9
4.3
3.7
2.4
.2
2.9
3.1
5.6
2.8
4.3
2.8
.0
3.5
2.3
.0
.0
.2
1.4
2.0
.6
.1
1.3
1.8
Operating
Engineers
i
6.6
8.4
6.7
3.7
2.7
18.0
5.1
3.9
5.7
6.0
8.0
9.3
18.8
7.8
11.3
4.7
18.9
10.0
6.1
6.3
29.7
25.5
23.5
18.8
Electricians
5.1
6.6
5.6
13.2
11.5
8.9
9.1
13.8
13.0
8.9
10.6
16.5
3.6
9.3
17.4
18.5
3.1
5.1
11.1
17.5
3.8
6.8
5.3
5.2
Carpenters
15.3
16.1
13.4
13.2
7.3
11.7
20.3
23.1
15.5
24.6
12.8
12.4
20.2
15.5
18.7
8.7
7.7
20.4
19.2
10.5
7.7
7.9
8.4
9.2
Other
Construction
1.8
3.5
8.6
12.1
9.2
5.6
4.7
7.3
6.3
6.6
7.2
5.8
3.9
6.4
3.3
8.7
8.4
1.3
5.8
7.8
8.2
6.5
4.7
4.2
Laborers
20.6
18.9
16.8
14.9
16.7
22.2
10.6
17.1
13.8
19.6
14.4
14.5
18.8
25.6
15.6
11.8
35.5
24.3
17.0
14.4
15.4
22.5
16.8
17.9
Clerical
10.8
3.5
4.1
3.7
9.4
7.8
7.0
5.1
3.8
4.1
4.0
5.4
5.4
3.1
5.6
2.9
8.4
6.5
4.8
3.7
5.2
5.2
5.8
5.7
Management
15.6
9.5
4.8
3.1
16.0
12.5
17.8
12.0
9.4
7.8
6.0
10.3
11.4
6.5
15.4
18.5
11.5
12.2
12.5
10.2
13.4
11.2
13.3
15.0
72
-------
variables which capture both the availability of labor in the area
and the relative attractiveness of employment opportunities at the
construction project. These factors typically have not been con-
sidered in empirical studies of this nature.
Relocation of Dependents
Another important issue in the estimation of construction-related
impacts involves movers' decisions regarding the relocation of their
families. The presence or absence of a family influences community
impacts in several ways. In particular, household composition and
family size influence the types of housing, consumer goods, and public
services demanded by movers.
The variation in the proportion of movers with family present
both overall and for various worker groups is presented in Table 4.
We observed a very pronounced regional variation with respect to the
relocation of dependents, with the proportions of movers with family
being much higher in the South than in the North. Overall proportions
of movers with family present typically ranged from 52 to 53 percent
in the North compared with 58 to 74 percent in the South. This pattern
was apparent overall, as well as for various worker groups.
We also observed differences in the relocation of dependents among
various worker groups. Nonconstruction movers were much more likely to
relocate families than were construction movers. Among nonconstruction
movers, the proportion of movers with family present typically ranged
from 64 to 76 percent in the North and from 70 to 85 percent in the
South, whereas among the construction group proportions typically ranged
from 46 to 54 percent in the North and from 51 to 59 percent in the
South. In examining differences between the scarce craft group and the
common and abundant craft group, we observed a significant difference
in the case of the northern sites. Typically, proportions ranged from
41 to 48 percent for the scarce craft group compared with 51 to 56
percent for the common and abundant craft group. However, a similar
pattern was not observed in the case of the southern sites.
The likely explanations for the differences between worker groups
with respect to the relocation of dependents lie in the differences in
employment opportunities for various worker groups. In the case of the
nonconstruction group, movers consist primarily of management rather
than clerical workers. These workers are often hired by the utility
or prime contractor for the duration of the project. Thus, while their
tenure at the site is limited, it is not necessarily short. Employment
at the site is quite stable and could extend for a period of up to ten
years. In addition, the employment contracts of management workers are
likely to cover the cost of moving their families. The situation,
73
-------
TABLE 4. VARIATION IN THE PROPORTION OF MOVERS WITH FAMILY PRESENT
BY REGION, OVERALL AND FOR VARIOUS WORKER GROUPS a/
Migrant Proportion
Absolute
Range
Typical
Values
Overal1
North
South
Construction
North
South
Nonconstruction
North
South
Scarce
North
South
Common and Abundant
North
South
51-53
51-86
35-54
44-85
52-76
50-87
25-48
53-83
51-64
40-89
52-53
58-74
46-54
51-69
64-76
70-85
41-48
53-69
51-56
49-69
ji/ Movers without families include single movers as well as
married movers who do not relocate their families.
-------
however, is quite different in the case of the construction group.
Because of the more limited duration of employment opportunities at
the site for construction workers, they are less likely to relocate
their families. This is most obvious in the case of the scarce craft
group.
The reasons for the regional difference with respect to the relo-
cation of dependents are not readily apparent. However, several factors
have been identified which could contribute to the explanation of this
difference. For instance, our data indicated that a relatively higher
proportion of movers was married in the South than in the North
approximately 75 percent in the North compared with approximately 85
percent in the South. The observed variation in the relocation of
dependents could also be related to regional differences in the housing
type chosen by movers. In our analysis of housing type, we observed a
much higher proportion of movers living in mobile homes in the South
than in the North approximately 20 percent in the North compared
with approximately 40 percent in the South.
Thus, we observed a rather large variation in the proportion of
movers with family present among the surveys included in our study.
In most cases, the proportion of movers with family present was rela-
tively large (ranging from 45 to 70 percent among construction movers
and from 65 to 85 percent among nonconstruction movers). Since the
proportion of movers who relocate their families has an influence upon
a number of important impacts (i.e., housing requirements, school en-
rollments, other service demands), it follows that careful consideration
should be given to this factor in socioeconomic impact assessment.
The large variation in the relocation of dependents which has been
observed across sites, as well as across various worker groups, has
identified a clear need to develop improved forecasting procedures.
However, our analysis has shown that the observed variation in the
proportion of movers with family present is a function of several
factors. In addition to certain labor requirement variables (vari-
ables which reflect differences in the expected duration and continuity
of employment at a site), these factors might include differences in
marital status, availability of different types of housing, and the
quality of schools and other services in nearby communities. An ade-
quate explanation of the observed variation would, therefore, require
a multivariate analysis. An examination of the data included in this
study suggests that such an analysis is likely to provide a means of
improving the accuracy of predicting the proportion of movers who will
relocate their families at future construction sites, thereby improving
the general accuracy of the impact assessments.
75
-------
Intention to Remain in the Area
Some workers will move to the area to work on the project with
the intention of remaining in the area after employment opportunities
at the project have been exhausted. Other workers will move to the
area to seek employment while it is available, or perhaps for shorter
durations. It is likely that these subgroups of movers will differ in
their utilization of goods and services. For instance, movers who ex-
pect to remain in the area for a short period of time are likely to
seek temporary housing (i.e., apartments, mobile homes, hotels/motels)
and may or may not relocate their families. A certain number of such
workers may live in the area only during the workweek, returning to a
permanent residence on weekends.
Recognition of these differences raises questions of (1) whether
or not a significant number of movers at a site consider their move to
be temporary, and (2) whether or not the size of this group varies from
site to site. Answers to these questions are used as a means of deter-
mining the importance of considering intention to remain in estimating
socioeconomic impacts.
Table 5 presents the variation in intention to remain in the area,
overall and for various worker groups, for those surveys for which in-
formation on intention to remain was available. Tj In general, more
than half (50 to 59 percent) of all movers were classified as temporary
movers (movers who intended to leave the area either before or on
completion of the project).
As was true with respect to migrant proportion and relocation of
dependents, we also observed differences with respect to intention to
remain in the area for various worker groups. Most pronounced were
the differences between the construction and nonconstruction groups.
Among the construction group, only 41 to 49 percent of the movers were
classified as temporary movers, whereas, among the nonconstruction
group, 59 to 74 percent of movers were considered to be temporary.
The rather limited number of movers in certain subgroups of workers
(i.e., the clerical and abundant categories) limited the comparisons
which could be made by further disaggregating the construction and non-
construction groups. However, our analysis did indicate that scarce
movers were less likely to remain in the area upon completion of the
project than were movers in the common and abundant group. The
Tj Temporary mover proportions were based on information from only
four surveys, while information on the proportion of movers with a
permanent residence elsewhere was available for seven surveys.
76
-------
TABLE 5. VARIATION IN INTENTION TO REMAIN IN THE AREA,
OVERALL AND FOR VARIOUS WORKER GROUPS
Overall
Construction
Nonconstruction
Scarce
Common and
Abundant
Temporary
Mover
Proportion
50-59
41-49
59-74
45-56
27-49
Transient
Mover
Proportion
8-18
8-18
10-21
8-17
7-18
Proportion of
Movers with
Permanent
Residence
34-45
40-59
7-24
39-68
24-49
Worksheet
Mover
Proportion
15-28
15-37
2-11
14-45
15-29
77
-------
proportion of movers who were temporary typically ranged from 45 to 56
percent for the scarce crafts compared with 27 to 49 percent for the
common and abundant crafts.
As an alternative measure of intention to remain, we examined the
proportion of movers who maintain a permanent residence elsewhere. These
proportions were similar to the proportions of movers who were considered
to be temporary among construction movers (40 to 59 percent). However,
the use of a permanent residence as an indication of intention to remain
was not appropriate in the case of the nonconstruction group.
In an attempt to identify those movers who moved to the area for
a much shorter duration, we examined two other subgroups of movers
transient movers j}/, and workweek movers. 9/ Approximately 10 to 20
percent of movers among both construction and nonconstruction movers
moved into the area for specific jobs and did not stay in the area for
the entire construction period. Typically, 15 to 28 percent of all
movers lived in the area only during the workweek, with workweek movers
proportions being higher among construction movers (15 to 37 percent)
than among nonconstruction movers (2 to 11 percent).
Again, these observed differences in intention to remain in the
area are not surprising given the differences in employment opportunities
among the various worker groups. Most nonconstruction movers consist of
management rather than clerical workers. These workers typically are
transferred upon completion of the project. Furthermore, the long-term
employment potential for these workers in this area (especially a rural
area) is rather limited. As a reflection of these factors, we did
observe that most nonconstruction movers expect to leave the area upon
completion of the project.
In the case of construction workers, the nature of employment
opportunities, both at the site and in the area after completion of the
project, differ considerably from those experienced by nonconstruction
workers. Construction workers may be more likely than nonconstruction
workers to find employment in the area after completion of the project.
As a reflection of this, we did observe lower proportions of temporary
movers among the construction group.
8/ Transient movers are a subset of temporary movers. Transient
movers are those workers who move into the area for short durations ir-
respective of the availability of continued employment at the site.
9f Workweek movers are a subgroup of movers who maintain permanent
residences elsewhere. These workers live in the area during the work-
week and return to their permanent residences on weekends.
-------
Furthermore, we observed that a relatively large proportion of
construction movers maintain permanent residences elsewhere. This is
also consistent with the nature of employment in the construction
industry. The construction industry is characterized by a high labor
turnover. Employment for construction workers is seldom permanent.
Rather, workers are hired for specific jobs of a limited duration. As
a result, it is often necessary for construction workers to move from
site to site or to commute rather long distances each day to maintain
steady employment. Because of the temporary nature of employment oppor-
tunities, many workers maintain permanent residences, live in temporary
housing near the construction site during the workweek, and return to
their permanent residences on weekends.
Intention to remain in the area is important to socioeconomic impact
assessment only to the extent that the bundle of goods and services con-
sumed by temporary movers differs from the goods and services consumed
by permanent movers. However, the major difference in the consumption
patterns between these two groups is likely to stem from differences in
the relocation of dependents. Examination of our data did indicate that
permanent movers are more likely to relocate their families than are
temporary movers (the proportion of movers with family present typically
ranged from 50 to 60 percent among temporary movers compared with 65 to
75 percent among permanent movers). 10/ However, this difference between
permanent and temporary movers was not as large as one might have ex-
pected. This difference can, in part, be attributed to the nonconstruc-
tion group. Despite the fact that the vast majority of nonconstruction
movers regard their move as temporary, most nonconstruction movers relo-
cate their families. Moreover, while employment opportunities for
construction workers at nuclear power plant construction sites may be of
limited duration, they are not necessarily short. Indeed, employment for
some worker groups may extend for a period of 5 years or more. Therefore,
there may be little justification for distinguishing between temporary
and permanent movers for the purpose of socioeconomic impact assessment.
Transient movers and workweek movers, on the other hand, are more
likely to exhibit different consumption patterns from those of other
movers. Thus, it might be argued that these groups are worthy of special
consideration in impact assessment. However, in most cases, these two
groups constitute a fairly small proportion of all movers. In terms of
absolute numbers, it is likely that transient movers and workweek movers
10/ These figures may appear low for permanent movers. However, one
will recall that the group of movers who did not relocate their families
included single movers as well as married movers who did not relocate
their families.
79
-------
would constitute a large group only at a site with a large total workforce
and high migrant proportion. The results of this analysis, therefore,
suggest that in most cases a consideration of intention to remain will
not greatly improve the accuracy of socioeconomic impact assessments.
Demographic Characteristics of Movers
There was little variation in most demographic characteristics of
movers across sites and surveys (Table 6). However, we did observe an
interesting regional difference with respect to marital status. This
difference was observed among the construction group with over 85
percent of movers being married in the South compared with 75 percent
in the North. No regional difference, however, was apparent among the
nonconstruction group.
Family size varied somewhat across the various worker groups. The
average family size of construction movers (3.4) was somewhat higher
than that of nonconstruction movers (3.1). This difference was also
reflected in the average number of school-age children. Averages were
higher among construction movers (0.9) than among nonconstruction movers
(0.7).
Information on income was available for only four surveys, and as
a result, the conclusions drawn on the basis of these data were somewhat
limited. However, a few general patterns were apparent. The overall
median income among movers was approximately $21,000 a year, with the
average among the nonconstruction ($21,700) group being higher than
among the construction group ($20,500). This, of course, reflects the
higher average salaries of the management subgroup, which dominates the
nonconstruction mover group. Similarly, it is not surprising to note
that the median family income among the scarce movers ($21,000) was
higher than that among the common and abundant mover group ($19,300).
This undoubtedly reflects the more specialized skills of craft which
comprise the scarce craft group.
The relatively small variations which we observed in most mover
characteristics would tend to imply that survey results (with only a
few modifications to account for the special characteristics of the site
in question) can quite safely be used for forecasting mover characteris-
tics at future construction sites. These data could be used to derive
multipliers which, in turn, could be used to estimate total family size
(or perhaps number of school-age children) associated with an inmigrat-
ing construction workforce.
80
-------
TABLE 6. AVERAGE MOVER CHARACTERISTICS, OVERALL
AND FOR VARIOUS WORKER GROUPS a/
Percent
Married
Average
Average Number of Median
Family School-Age Family
Size Children Income b/
Overall
North
74.7
(4.9)
3.26
(.15)
.81
(.16)
21,000
(1,143)
South
Construction
North
South
Nonconstruction
North
South
Scarce
North
South
Common and Abundant
North
South
85.1
(3.4)
74.8
(4.0)
86.7
(4.0)
75.9
(10.8)
76.0
(4.6)
74.4
(7.4)
85.9
(4.4)
75.4
(5.4)
87.5
(3.9)
3.39
(.14)
3.10
(.16)
3.42
(.20)
3.35
(.16)
.90
(.16)
.67
(.16)
.94
(.24)
.89
(.18)
20,500
(1,365)
21,700
(1,503)
21,000
(1,444)
19,300
(741)
a/ Mean values with standard deviation in parentheses.
"57 Incomes are in 1978 dollars.
81
-------
IMPLICATIONS FOR MODEL DEVELOPMENT
In our discussion of the results of this analysis, we highlighted
several implications with respect to impact assessment and the develop-
ment of improved forecasting procedures. Implications specific to each
of the four variables migrant proportions, relocation of dependents,
intention to remain in the area and demographic characteristics of the
movers were discussed individually in each section. In this concluding
section, we focus on the more general implications of these findings for
model development.
Foremost, in our examination of these data, we observed a large
variation in both overall migrant proportion and the relocation of
dependents across sites. This suggests that the practice of making
forecasts at one site based solely upon the observed values at another
site (or even the average across several sites) can be a dangerous
practice. In addition, our examination clearly highlights the need for
a multivariate analysis to improve our understanding of the factors
underlying the variation in these variables across sites as a necessary
first step in the development of improved forecasting procedures. The
results of this analysis indicate that workforce composition, the avail-
ability of labor in the area and the nature of labor requirements could
be important factors in explaining the observed variation in these vari-
ables across sites.
A second major finding of this analysis is that one also observes
considerable variation in migrant proportion and relocation of depen-
dents among different workers groups. This suggests that occupation
may be an important dimension in examining these issues. The importance
of this finding is that it allows one to perform a multivariate analysis
using craft-specific worker groups as the unit of analysis.
Past studies of this nature have been limited by the availability
of data. Construction worker survey data are difficult to obtain and,
as a result, data are not available for a large number of sites. Indeed,
past studies have often been limited to at most 14 sites. If the site
is used as the unit of analysis, the data do not provide a sufficient
number of observations to conduct a multivariate analysis. However, the
results of our examination indicate that it is possible to identify 7 or
8 major craft groups with sufficient numbers of workers in each craft
group to examine variation with respect to each craft group at each site.
Adopting this approach provides a sufficient number of observations for
a multivariate analysis, thereby overcoming the data limitations which
have hindered past studies of a similar nature.
An additional advantage of using craft as the unit of analysis
stems from the fact that labor requirements are craft-specific. Thus,
82
-------
it will be possible to consider the importance of a number of craft-
specific labor requirement variables such as income potential associated
with employment at the site, labor requirements for project construction,
local availability of labor and competing demand for labor for each craft
group, in addition to the importance of regional and project characteris-
tics, in explaining the observed variation across sites and crafts.
Another rather interesting finding is the very pronounced difference
with respect to both migrant proportion and relocation of dependents
which was observed between construction and nonconstruction workers. In
most cases, however, this variation was seen to reflect basic differences
in the nature of employment opportunities between the two groups at con-
struction sites. This suggests that the factors underlying such reloca-
tion decisions are not the same for these two worker groups. The
implication of this for purposes of model development is simply that a
separate model should be developed for each group. Combining these two
very different groups in the same model is likely to result in a reduc-
tion in the overall predictive power of the model.
Adopting a craft-specific approach to model development requires
that an accurate estimate of craft-specific labor requirements be avail-
able prior to project construction. Accurate estimates of labor
requirements are necessary in the development of forecasting procedures,
as well as in the application of these models for estimating impacts at
future construction projects. Our study, however, has indicated that
in most cases pre-construction projections provided by utilities are
poor indications of the actual utilization of labor. Improvements in
this regard are critical to improvements in impact assessment, because
even the best forecasting procedures will fail to perform well if they
are based upon inaccurate estimates of manpower requirements.
In conclusion, our examination of migrant proportion and relocation
of dependents has identified considerable variation across sites and
across different worker groups. These findings imply the need for a
craft-specific multivariate analysis as a possible means of improving
forecasting procedures. A consideration of intention to remain in the
area, except perhaps in very special instances, is not extremely
important in making socioeconomic impact assessments. Demographic
characteristics of the movers are quite invariant across sites. They
are, therefore, amenable to the development of multipliers that can be
readily employed to make impact assessments after the migrant proportions
and proportions of workers who will relocate families have been estimated.
-------
References
Malhotra, Suresh, and Diane Manninen
1979 Socioeconomic Impact Assessments: Profile Analysis of Worker
Surveys Conducted at Nuclear Power Plant Construction Sites.
Preliminary report to the U.S. Nuclear Regulatory Commission
(August).
Mountain West Research, Inc.
1975 Construction Worker Profile: Final Report. Washington, B.C.:
Old West Regional Commission.
-------
AN ANALYSIS OF THE AGRICULTURAL HIRED LABOR MARKET FOR THE
NORTHERN GREAT PLAINS WITH EMPHASIS ON THE EFFECTS OF ENERGY DEVELOPMENT
Dale J. Menkhaus
Richard A. Adams I/
INTRODUCTION
Residents of the Northern Great Plains (defined as Montana,
North Dakota and Wyoming) have expressed concern over adverse conse-
quences of the rapid development of the mining sector in the region.
Environmental, social and economic effects are of particular interest
to affected parties, as well as to researchers and public policy
makers. This interest has been manifested in numerous studies related
to the environmental and socioeconomic effects of energy 2/ develop-
ment on cities and towns in affected areas (Bradley £t_ alL , 1979;
Leistritz, 1973; Northern Great Plains Resources Program, 1975).
However, there is a scarcity of information concerning the effects
of growth in the mining sector on agricultural factor markets.
Numerous effects of increased mining activity on farmers and
ranchers may be identified. Among the more important is the potential
competition for factors of production (Leistritz, 1973). Land and
water are two obvious factors for which the mining and agricultural
sectors compete. Another factor market being affected by energy
development is labor, where wages have risen sharply in energy. A
survey by Conklin (1977) of selected livestock producers within an
energy-impacted region of Wyoming (Powder River Basin) reveals concern
among such producers over the ability to pay wages competitive with
_!_/ Associate Professors, Division of Agricultural Economics, Univer-
sity of Wyoming.
"I] The terms "energy" and "mining" are used interchangeably in this
study to refer to the aggregate of coal, oil, natural gas and uranium
extraction.
-------
those realized in mining. While differences in agricultural and mining
wages have persisted over time, the recent rapid expansion of mining
employment has increased the magnitude of the difference. Even in the
absence of a closed economy, a large wage difference may create an
incentive to transfer between sectors.
This paper investigates the nature of the market for hired agri-
cultural labor for the Northern Great Plains., with particular reference
to adjustments that may be attendant to energy development. Specific
objectives include: 1) review alternative model specifications and
specify a labor market model which defines the structure for hired
agricultural labor for the Northern Great Plains; 2) assess the
importance of wages in agriculture and mining in explaining the
structure of hired labor to agriculture in the Northern Great Plains,
with emphasis on the plausibility of mining wages as an incentive for
the transfer of labor out of agriculture, as suggested by affected
parties in the region; and 3) compare the results with earlier regional
supply and demand studies. The focus is directed toward measuring the
impact of energy development on a particular segment of the economy
in the impacted region as reflected in a particular factor market,
labor. Emphasis is on understanding the causal relationship within
that factor market and on examining the structure which generates the
data observed at the macro or aggregate level.
THE PROBLEM SETTING
There is a scarcity of published elasticity measures available at
the regional level. Adjustments at the regional level, such as the
Northern Great Plains, may indeed be more substantial than national
developments. Specifically, the rural nature of the Northern Great
Plains, when coupled with the imposition of a growth sector of the
magnitude of energy, suggests a need to estimate the (changing)
structural relationships in the hired labor market at the regional
level.
To demonstrate the growth of the mining sector within the Northern
Great Plains, evidence gathered from the three-state region of Montana,
North Dakota and Wyoming is presented in Table 1 for the period
1964-78. Employment data for agriculture and mining as well as wage
rates are presented. As is evident from the table, the out-migration
of hired workers from agriculture continues (1978 is the exception)
but the rate appears to have slowed. During this same 15 year period,
mining employment within the three state region has more than doubled,
with particularly rapid increases in the last three years. Wage rates
86
-------
TABLE 1. AGRICULTURAL AND MINING EMPLOYMENT AND WAGES IN THE
NORTHERN GREAT PLAINS a/, 1964-1978
Year
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
£/
Sources
Employment
Agriculture
- Number
42,000
31,000
32,000
31,000
27,000
26,000
26,000
28,000
27,000
27,000
29,300
26,000
25,300
21,300
23,200
Data for North
Wage
Mining
17,900
18,000
18,300
16,900
18,100
19,500
19,700
18,100
20,000
20,400
25,000
26,500
30,000
34,800
40,000
Agriculture Mining
- dollars
1.26
1.27
1.37
1.40
1.51
1.63
1.68
1.80
1.93
2.14
2.42
2.65
2.90
3.08
3.46
per hour
2.97
3.06
3.18
3.29
3.41
3.63
3.87
4.03
4.46
4.75
5.49
6.16
6.80
7.91
9.65
Difference
1.71
1.79
1.81
1.89
1.90
2.00
2.19
2.23
2.53
2.61
3.07
3.51
3.90
4.83
6.19
Dakota, Montana and Wyoming.
. U.S. Department of Agriculture; Agricultural Statistics and
U.S. Department of Labor, Bureau of Labor Statistics,
Employment and Earnings.
87
-------
have increased in both sectors, but mining wages increase at a more
rapid rate, as measured by the differences in wages.
This descriptive evidence may serve to lend plausibility to the
assertion by agricultural interests that growth in the energy industry
with attendant multiplier effects as manifested in the service sectors
is accelerating the transfer of labor from agriculture. Associated
adjustment problems may be particularly acute in areas immediately
surrounding energy development, given that the economic base in many
of the rural areas in which energy expansion is occurring is relatively
simple and is historically dominated by the agricultural (livestock)
sector. The Powder River Basin of Wyoming, the principal coal pro-
duction area in the Northern Great Plains, is typical of the economic
structure of much of the Northern Great Plains area: livestock
oriented with no industrial base other than energy.
The following section outlines conceptual models for hired
agricultural labor for the Northern Great Plains based upon theoretical
constructs outlined by Magee and Borts and upon the formulation devel-
oped by Schuh (1962), Schuh and Leeds (1963), and Tyrchniewicz and
Schuh (1966) and adopted by Hammonds, et al. (1973). Modifications
intended to examine the specific importance of energy development in
the Northern Great Plains are also discussed. A discussion of data
problems, a statistical model and estimated relationships are then
presented.
THE MODEL
At least two approaches may be taken in analyzing the impacts of
energy development on agricultural hired labor. These include:
1) model the wage difference and attempt to determine factors which are
are important in explaining the wage difference, and 2) model the
agricultural hired labor system incorporating a measure which reflects
the growth in the mining sector.
The relationship between agricultural employment and wage
differences (across agriculture and alternative employment) is hypo-
thesized to be an important aspect of agricultural labor supply
(Tyrchniewicz and Schuh, 1966). That is, the magnitude of the
difference in factor returns affects the ultimate labor supply to
agriculture. This suggests that current energy development patterns
in the three-state region and other rural states which continue to
alter the rate of return across rural economic sectors may affect
agricultural employment levels and wages.
-------
The existence of a substantial wage difference between the
agricultural and mining sectors in the Northern Great Plains may be
explained by general equilibrium theory related to factor market
distortions and factor price differentials (Magee, 1976 and Borts,
1960). As noted, the wage difference itself is a measure of the
difference in the returns to a factor (labor) in the agricultural and
mining industries. This may be brought about by market imperfections
or distortions. Following Magee (1976), distortions in the agri-
cultural and mining labor market may include:
(1) "monopoly" power of firms within the mining sector,
enabling them to share monopoly profits with labor;
(2) union pressure on wages in the mining sector, both
direct (where unions exist), and indirect (where
management seeks to avoid labor disruptions by
meeting or exceeding union wages); and
(3) disguised unemployment in agriculture (e.g., a hired
man is retained year-round although his services may
not be fully utilized in the off seasons).
The existence and significance of these trends concerning factor
market distortions may also be viewed within the conceptual framework
presented by Borts (1960). Differences in returns to factors of
production, such as labor, may be explained by relative changes in
final demand experienced by each industry. Thus, the two prominant
rural industries within the Northern Great Plains may be viewed as
competing for the same input, labor. However, one industry (agri-
culture) has been characterized by a relatively constant demand,
while the other is experiencing rapid growth. Such a situation then
results in a differential in the rate of return to the factor of
production as manifested in the disequilibrium in wage rates currently
being observed.
Other causes, which may be termed non-distortionary, may also be
responsible for the observed wage difference between mining and agri-
culture. These include: 1) transfer costs associated with moving
from agriculture to mining employment; 2) a preference on the part of
some agricultural labor to remain employed in agriculture; and 3)
perquisites which may be included in the agricultural wage package but
not reflected in the agricultural wage rate.
A model incorporating these constructs has been presented by
Conklin, £t al. (1978). Data are difficult to obtain for a time series
of sufficient length of conduct empirical analyses with respect to
89
-------
distortionary effects. Therefore, while appearing to be promising
from a theoretical standpoint, little empirical evidence is available
which proves or disproves the usefulness of a model using the above
framework to explain wage differences between mining and agriculture.
The second approach, pursued in this paper and reported by Adams
and Menkhaus (1980) , follows that developed by Schuh (1962) , Tyrchniewicz
and Schuh (1966) and Hammonds, et al. (1973) with minor deviations to
address the influence of the mining sector on agricultural hired labor.
Specifically, the growth of the mining sector, as a basic sector with-
in the context of a regional economy, is assumed to generate sub-
stantial economic activity within the region. The resultant economic
growth, particularly as manifested in rising wages, may affect the
agricultural labor market, even in the absence of direct substitution
between mining and agricultural employment. Within this model, the
quantity of hired labor demanded is assumed to be a function
of the real agricultural hired labor wage, index of prices received by
farmers deflated by the index of prices paid by farmers, and an index
of productivity (technology). The quantity of hired labor available
(supplied) to agriculture is hypothesized to be a function of the
real agricultural hired labor wage, the regional civilian labor force,
real mining wage ^3/ and a trend variable. Consistent with the Schuh
(1962) formulation, each equation in each model is specified in the
Nerlovian framework to facilitate estimation of both long-run and
short-run elasticities. The empirical content of this model appears
to be somewhat stronger compared to the former approach. Specifically,
data are available and the model is theoretically sound.
THE STATISTICAL MODEL
A statistical model for the regional labor market incorporating
the Schuh-Hammonds framework may be summarized in the following two
equations:
_3/ This variable represents a deviation from the models developed
by Schuh and his associates (1962, 1963, 1966). Schuh used "corrected"
non-farm income to represent alternative employment opportunities.
Since the main emphasis of this study is on the impacts of the mining
sector on agricultural hired labor, alternative income opportunities
are represented by the mining wage.
90
-------
(i)
(2)
where: Y's = endogenous variable;
X's = exogenous variables'
Y- = agricultural hired labor force for Montana, North
Dakota and Wyoming in year t in thousands (U.S,
Department of Agriculture) ;
Y_ = average hourly agricultural wage in year t deflated
by the Consumer Price Index (1967=100) for the three
state region (U.S. Department of Agriculture);
X- = ratio of the index of prices received over the index
of prices paid by U.S. farmers in year t (U,S.
Department of Agriculture, n.d.);
X_ = farm productivity index of output per unit of input,
for the Northern Great Plains in year t (U,S,
Department of Agriculture, 1977);
X_ = average of Montana, North Dakota and Wyoming hourly
mining wage in year t deflated by the Consumer Price
Index (U.S. Department of Labor 1977 and 1979);
X = the first difference of the civilian labor force
in Montana, North Dakota and Wyoming in ten thousands
in year t (U.S. Department of Commerce, n.d.);
X- = time trend variable;
H.. and \i~ = random disturbances;
a's and 8's = structural parameters;
$ = 1 - Y-i > where Y-, = the
for
a, = 1 - Yn» where j^ = the coefficient of adjustment
$ = 1 - Y-i > where Y-, = the coefficient of adjustment
for demand;
for supply;
1964-1978.
DATA
The data for agriculture and mining wages for the region are
represented by those observed in Montana, North Dakota and Wyoming.
The selection of this three state region is motivated by the signifi-
cant position this region has assumed in national coal production.
Consistent with the Schuh and Hammonds, et al. (1962, 1963, 1966)
formulations, wage data are deflated.
91
-------
In the model, agricultural employment represents the hired farm
labor force in Montana, North Dakota and Wyoming, which are the states
most affected by energy development in the Northern Great Plains.
Similarly, the civilian labor force includes non-agricultural employees
in the three state region. Finally, national data for prices received
and paid by farmers are used to represent the income to farming and
ranching in the region. Data were collected for 1964-1978, the
period encompassing the expansion of energy development in the Northern
Great Plains region.
STATISTICAL CONSIDERATIONS
There are problems attendant to empirical application of the
above model, especially with time series data. Griliches (1961)
suggests that the use of a lagged independent variable leads to
serial correlation in the error term. Serial correlation of the
error term then leads to inconsistent parameter estimates in auto-
regressive models (Griliches, 1967). The traditional Durbin-Watson
statistic is also invalid in such models. Durbin (1970) has addressed
this latter issue, but the problem of inconsistent parameter estimates
must still be faced if the neo-Durbin test reveals the existence of
serial correlation among the residuals.
An approach to handling serial correlation is to assume the
existence of serial correlation and adjust for its presence, using an
appropriate estimation procedure (Johnson, 1961). One method for
calculating regression parameters in the presence of first order auto-
regressive disturbances is the Cochrane-Orcutt iterative technique.
Fair offers further input into the estimation of simultaneous equation
models with lagged endogenous variables and first order serially
correlated errors, such as the model assumed here. More specifically,
Fair suggests care must be taken in forming the instrumental variables
in two-stage least squares regressions. To arrive at consistent
estimates, predetermined variables, predetermined variables lagged
once, endogenous variables lagged once, and the dependent variables
lagged once must be included as first state predetermined variables
(Fair, 1970, p. 508).
The Cochrane-Orcutt iterative technique, incorporating the
suggestions by Fair, was employed to estimate the economic model. The
computer program which includes the software to handle this estimation
procedure is available through Synergy, Inc.
Another statistical problem in Nerlovian-type models is that the
-------
lagged dependent variable has a tendency to pick up the effect of
omitted variables, thus leading to problems of specification bias.
Usually, this problem is handled in part by introducing a trend
variable into the statistical model (Schuh, 1962, p. 313; Hammonds,
et al., 1973, p. 245). Unfortunately, the introduction of such a
variable compounds the already severe collinearity problems associated
with models of this type. High intercorrelation among the independent
variables was observed by Tyrchniewicz and Schuh (1966, p.547) and
Hammonds, et al. (1973, p. 245). The authors of the latter study
removed the trend variable in the demand equation where there was
high collinearity between the trend variable and the productivity
index. Further, Schuh and Leeds (1963, p. 301) found that the trend
variable was significantly different from zero at the 5 percent level
in only two of nine regions in their study of the regional demand for
hired agricultural labor. Schuh (1962, p. 312) found the trend
variable in the demand equation to be insignificant. In the model
outlined above, high intercorrelation (.98) exists between the trend
variable and the real agricultural wage, which appears in both the
supply and demand equations. As a result of this high interdependence
between the trend variable and real agricultural wage and supported by
results of previous studies, the trend variable was not included in
the demand equation. As a result of omitting a trend variable,
caution should be exercised in interpreting the direct price elastic-
ities obtained from the estimated demand relationship due to possible
specification bias.
The omission of the trend variable from the demand equation may
not be particularly serious from an economic standpoint, since the
productivity index variable included in the demand equation should be
measuring the influence of technology. However, in the supply
equation the inclusion of a trend variable is deemed important in
terms of explaining the secular movement in the flow of labor out of
agriculture due to changes in tastes for employment in specific
industries, aspects of industrialization and the level of education
(Tyrchniewicz and Schuh, 1966, p. 542 and 551). Even though collin-
earity exists between trend and the real agricultural wage, there
appears to be theoretical justification to retain the trend variable
in the supply equation.
High collinearity also exists between the civilian labor force
and real agricultural and mining wage, .97 and .95, respectively.
Following Ktnenta (1971, p. 390), the civilian labor force and non-
farm income data were transformed to first differences to reduce
collinearity. While this transformation reduces the degree of multi-
collinearity, it introduces autoregression in the disturbances, which
may otherwise be independent. It was felt the use of the first
-------
difference transformation was preferable to omitting a variable
whose inclusion was again justified on theoretical grounds, since
the estimation procedure adjusts for serial correlation.
RESULTS AND IMPLICATIONS
The model estimated and reported here may be viewed as dynamic
given the presence of distributed lags. The model thus allows for
the estimation of both short- and long-run structural relationships
via the use of the coefficient of adjustment. Alternative formulations
of the model were tested, most notably a log-log version. The
reported model (linear) was deemed "better" due to slightly more
robust statistical estimates. In addition, there is no reason to
believe that the assumption of constant elasticities is valid.
_A priori expectations concerning the signs of the variables
included in the above model are suggested by theory. One may hypo-
thesize that the quantity demanded is inversely related to the labor
wage and the productivity index, and directly related to the price
indices ratio and the lagged value of the dependent variable. Quantity
supplied is expected to be directly related to the agricultural wage
and the civilian labor force, and inversely related to the "corrected"
non-farm income, mining wage and trend. The simultaneously estimated
regression coefficients for the above economic model are presented in
Table 2. The results appear to be generally consistent with those
expectations.
The statistical results for the demand equations are more robust
than for the supply relationships, as is evident from Table 2.
Specifically, the agricultural wage, the price indices ratio and the
lagged dependent variable are significant at the 5 percent level and
display signs consistent with expectations except for the productivity
index. Also, the R^'s associated with the demand equations are
greater than those observed for the supply relationships. (The R^ in
both equations may be artificially inflated due to the presence of the
lagged dependent variable.) On the supply side, the results are less
robust but do display general consistency of signs. Only the change
in the regional civilian labor force, which is insignificant at the
10 percent level, displays an incorrect sign. The regional agricul-
tural wage and the mining wage are significant at the 10 percent
level, with the time trend being significant at the 5 percent level.
The significance of several variables bears closer examination
in terms of economic implications. Within the demand equations, the
94
-------
^H|
oo %
O\ *
"~ i
" 3
CO w
(1
1 gj
°- 3
1 H
UJ
C£
C C
H to
Z ^ ^*
x; * in o
"- 00 -H X
UJ c > n
-r- IB -H *-3
. J= U
1 0
ex
O
2
t*«
T>
B
2
H
0>
0
O
tb
tJV
o
i
m
0
00
00
d
i
i
i
LjJ CO OJ
;£ C (Ml
H- c
H
i
col
I
z.
UJ
»__i
1
0 ox
U- o >
li W -H
III fX, 4J
LJJ
o
0
O « at to
1 1 0) > OJ
. o *H u
CO .rt 0) vH
CO u u )-<
1.1 CU
to
a
c
H
CO
01
M
3
00
H
Ud
i (
a>
01
0
i-H
Ol
JJ
O
0)
N
g
O
JJ
0)
01
H
^
jj
ctj
CJ
H
H
00
-H
CO
«
CO
0)
3
rH
to
4J
0)
CO
0)
01
JJ
fl
0)
cO
a
fl
H
CO
OJ
3
00
H
U-i
-I
01
0)
m
a>
jj
cfl
O
O)
N
e
o
M
M-4
JJ
C
O)
0)
H
T)
JJ
C
CO
O
H
U-i
H
fl
00
W
K
cO
3
jj
CJ
cfl
O
*o
cfl
0)
4J
to
fl
H
(A
0)
4J
§
H
CO
0>
C
o
o
0)
to
CO
0)
cfl
01
4J
0>
o
c
H
CO
CO
H
M-4
O
0)
00
4J
CO
t)
c
o
o
0)
cn
01
o
T)
H
r-l
cfl
H
jj
O
H
4J
CO
4J
O
c
0)
CO
CO
01
3
t
cfl
T)
fl
CO
CM
04
JJ
Ol
H
r- ij
0)
jj
tO
0)
JJ
H
jj
3
U
0)
fl
cO
M
x:
o
o
0>
x:
0
01
CO
3
0)
JJ
o
JJ
0)
3
o
3
Q.
3
13
0)
CO
(0
H
XI
a>
M
(0
to
O
H
JJ
CO
H
4J
cfl
JJ
Cfl
JJ
0)
x:
4J
C
o
H
JJ
H
o
cfl
C
M
(0
0)
tH
XI
cfl
H
CO
to
3
O
fl
OO
o
TJ
C
01
o
to
0)
3
rH
CO
CO
o
S
O
01
3
t-t
tO
0)
x:
c*
o
Jj
tfl
(U
0
o
to
H
01
to
o
CO
CO
cO
o
VJ-I
4J
CO
3
TJ
C3
O
JJ
t)
0)
tfl
3
CO
CO
CU
3
o
0)
U
o
a cJ
H
JJ CO
3 -H
U JJ
^ CO
O JJ
1 (0
0)
c c
cfl O
V-i CO
x; w
U (IJ
o 3
U 1
C
0) -H
x: xt
Ol d
CJ
C 01
H 42
a)
o oi
o x:
CJ) U
^ co
V4
01
3 -o
cr 01
x: o
u a
cu ai
(0
3
cfl
o
C
(0
E
-------
lagged dependent variable (and resultant coefficient of adjustment)
are significant, which indicates that substantial adjustments are
occurring within the regional agricultural hired labor market and
conforms to the distributed lag hypothesis. However, the lagged
dependent variable in the supply equation, while having the correct
sign, is not significant, an observation consistent with Hammonds,
et al. Also, the significance of the mining wage (10 percent)
within the supply equation provides some support to the hypothesized
importance of this variable in the agricultural labor market. The
implication of this relationship is that continued change in the
mining wage does appear to have an effect on the supply of labor to
agriculture in this region. Within a regional context, the significance
of the trend variable lends support to Tyrchniewicz and Schuh's
assertion that this variable may be an important shifter of the
supply of labor to agriculture; that is, the supply of hired agri-
cultural labor may be influenced by factors such as tastes, education
and industrial structure, in addition to traditional economic
variables. The trend variable may also be capturing the effects of
consistent measurement error in some of the variables (Tyrchniewicz
and Schuh, 1966, p. 551).
Based upon the estimated structural relationships presented in
Table 2, short- and long-run supply and demand elasticities were
calculated (Table 3). Specifically, demand elasticities were cal-
culated with respect to the regional agricultural hired labor wage
at mean levels and for 1978 to investigate any shifts in the under-
lying relationship. For the supply equation, elasticities with
respect to agricultural wage as well as the mining wage were derived.
The regional demand elasticity (Table 3) for 1978 was approximate-
ly 30 percent greater than that calculated at the mean level (-0.76 as
compared to -1.06). The long-run elasticities increased over short-
run values within periods as well as between mean and 1978 levels,
i.e., the long-run elasticities are approximately 40 percent greater
in 1978 than for the mean. Comparing these elasticities with those
reported by Schuh and Leeds (1963) for the Mountain States and
Hammonds et^ al. (1973) for Oregon, the absolute levels observed fall
within the range observed in these earlier studies; i.e. that is, higher
than those for Schuh and Leeds but lower than those of Hammonds ejt al.
(1972). These differences, as compared with those of Hammonds et al.
(1973), may be attributed to the livestock orientation of the agri-
cultural sector in the region, which limits the range of substitution
possibilities on the input side. The higher values observed in this
study when compared to those estimated by Schuh and Leeds (1963) may
be attributable to difference in time periods and the larger regional
difference in time periods and the larger regional definition in the
Schuh study.
96
-------
UJ
'lI
o
GO
^
o
a
a
a
I
I
Qi
Q-
Q
a:
co
VI
0
I
5
M
&
LU
Qi
CsL
01
o
a.
«
O
U-
3 I
oo
,
fil
(^ « « <5
6 M C8 CU ^
OJ O r-( r-l 3
I
V)
97
-------
The supply elasticities are perhaps of greater empirical import-
ance. The values presented in Table 3 for the agricultural wage are
elastic both at the mean values and for 1978. Such an increase over
time is consistent with both Schuh's and Hammonds's observations.
(Note that the long run regional supply elasticity is calculated from
an insignificant coefficient of adjustment.) The elasticity with
respect to mining wage, which is inelastic, displays the greatest
increase over the time period for any of the elasticities calculated,
increasing by more than 50 percent between mean values and 1978, This
increase in the response of supply of agricultural labor to changes in
the mining wage adds support tothe effect of this variable on the
adjustment process with respect to equilibrium levels of agricultural
supply and demand. It should be noted that the elasticity of labor
supply with respect to mining wages is perhaps a gross effect, given
the high multiplier effect of such a basic industry. Further, the
elasticity need not and probably does not imply direct substitution.
However, the conclusion that agricultural labor is affected by mining
does appear plausible in view of the results of this analysis.
SUMMARY AND CONCLUSIONS
The rapid development of the mining industry within the Northern
Great Plains over the past decade is well documented. Motivated by
rising energy prices, coal, uranium, oil and natural gas producers
have accelerated the development of these natural resources and the
number of workers employed within the region has increased and con-
versely reduced the unemployment rate to one of the lowest in the
U.S.
While such development has been beneficial in terms of per capita
incomes and state government revenues, certain adjustment costs are
being felt. In agriculture, one potential adjustment involves the
supply of labor to agriculture and the attendant wages paid in
agriculture. More specifically, agriculture faces the prospect of
reduced supplies of labor as workers are drawn to the relatively high
wages in expanding sectors. One means of alleviating this outflow
of labor from agriculture is to offer competitive wages within
agriculture. Higher wages in agriculture may involve reductions in
the returns to farmers and ranchers, given that individual farmers
are usually not able to pass such costs on directly to the consumer.
The magnitude of any shift of labor from agriculture to mining
is uncertain. No current data exist with which to accurately determine
98
-------
the number of workers shifting directly from agriculture to energy.
It is plausible that some agricultural labor is being attracted to
energy, partially motivated by the existence of a substantial wage
difference between agriculture and mining. In addition, labor
remaining in agriculture may have some perception of the wages within
the energy sector and hence demand some adjustment in the agricultural
wage. The results of the model developed in this study suggest that
the supply of labor to agriculture has become more responsive to the
mining wage over time. However, one need not assume any direct
substitution between mining and agriculture to suggest that mining
growth is altering the structure of the agricultural hired labor
market, given the multiplier effect associated with a basic sector
such as mining. Thus, the mining wage variable may serve to summarize
(a) the derived regional labor demand in service sectors which is
due to the incremental multiplier effects of changes in income in all
basic sectors and (b) any direct substitution.
The hired labor market model and the data sets used in estimating
this model are generally consistent with previous approaches and
represent the most recent data available for this region. In addition
to offering an updated examination of the functioning of a regional
labor market, the results provide some quantitative support to the
hypothesized effects of growth in the mining sector on the supply of
labor to agriculture. The magnitude of the structural elasticities
with respect to the change in the mining wage, while perhaps a gross
effect, also implies an increasing sensitivity over time.
99
-------
References
Adams, R.M., and D.J. Menkhaus
1980 "The effect of mining on agricultural hired labor in the
Northern Great Plains." American Journal of Agricultural
Economics. (forthcoming.)
Borts, George H.
1960 "The equalization of returns and regional economic growth."
The American Economic Review 50(3).
Bradley, Edward, James J. Jacobs, and Andrew Vanvig
1979 Impact of Coal Development on Ranchers and Farmers in Wyoming's
Powder River Basin. Research Journal 146. Laramie: Agricul-
tural Experiment Station, University of Wyoming.
Cochrane, D., and G.H. Orcutt
1949 "Application of least-squares regression to relationships
containing auto-correlated errors." Journal of the American
Statistics Association. March.
Conklin, Neilson C.
1977 The impact of of energy development on Wyoming cattle producers
with reference to cost and availability of labor. Unpublished
M.S. thesis. Laramie: Division of Agricultural Economics,
University of Wyoming.
Conklin, N.C., R.M. Adams, and D.J. Menkhaus
1978 "Agricultural employment and intersectional linkages: An
analysis of energy impacts." Paper presented at the Western
Agricultural Economics Association Annual Meeting, Bozeman, MT.
July 23-25.
Cooper, J.P.
1972 "Asymptotic covariance matrix of producers for linear regres-
sion in the presence of first-order autoregressive disturbances."
Econometric 40(March):305-310.
Durbin, J.
1970 "Testing for serial correlation in least-squares regression
when some of the regressions are lagged dependent variables."
Econometric 38(May):410-421.
100
-------
Fair, R.C.
1970 "The estimation of simultaneous equation models with lagged
endogenous variables and first order serially correlated
errors." Econometrica 38(May):507-516.
Griliches, Zvi
1961 "A note of serial correlation bias in estimates of distributed
lags." Econometrica 21(January):65-73.
1967 "Distributed lags: A survey." Econometrica 35(January):16-49.
Hammonds, T.M., R. Yadav, and C. Vathama
1973 "The elasticity of demand for hired farm labor." American
Journal of Agricultural Economics 55(May):242-245.
Johnson, S.S.
1961 An econometric analysis of the demand for and supply of farm
labor. Unpublished Ph.D. thesis. Iowa State University.
Kmenta, J.
1971 Elements of Econometrics. New York: The Macmillan Company.
Leistritz, F. Larry
1973 "Coal Development in North Dakota." North Dakota Farm Research
30(7).
Magee, Stephen P.
1976 International Trade and Distortions in Factor Markets. New York:
Marcel Dekker, Inc.
Northern Great Plains Resources Program
1975 Effects of Coal Development in the Northern Great Plains.
Denver: Northern Great Plains Resources Program.
Schuh, G.E.
1962 "An econometric investigation of the market for hired labor in
agriculture." Journal of Farm Economics 46(May):307-321.
Schuh, G.E., and J.R. Leeds
1963 "A regional analysis of the demand for hired agricultural labor."
Papers and Proceedings of the Regional Science Association
11:295-308.
Synergy, Inc.
n.d. Econometric Software Package User's Manual. Washington, D.C.:
Synergy, Inc.
101
-------
Tyrchniewicz, Edward W., and G. Edward Schuh
1966 "Regional supply of hired labor to agriculture." Journal of
Farm Economics 68(August):537.
U.S. Department of Agriculture
n.d. Agricultural Statistics. Washington, B.C.: U.S. Department
of Agriculture. Selected years.
U.S. Department of Agriculture
1977 Changes in Farm Production and Efficiency, 1977. Statistical
Bulletin No. 581. Washington, D.C.: U.S. Department of
Agriculture, Economic Research Service.
U.S. Department of Commerce
n.d. Statistical Abstracts of the United States. Washington, D.C.:
U.S. Department of Commerce, Bureau of the Census. Selected
years.
U.S. Department of Commerce
1979 State Quarterly Economic Developments. Washington, D.C. : U.S.
Department of Commerce, Economic Development Administration.
U.S. Department of Labor
1979 Employment and Earnings 26(2). Washington, D.C.: U.S. Depart-
ment of Labor, Bureau of Labor Statistics.
U.S. Department of Labor
1977 Employment and Earnings, States and Areas, 1939-75. Bulletin
1370-12. Washington, B.C.: U.S. Department of Labor, Bureau
of Labor Statistics.
102
-------
THE GRASP SOFTWARE/DATA SYSTEM
AND ITS APPLICATIONS TO SOCIAL RESEARCH
Celia A. Allard I/
INTRODUCTION
We live in an age of data accumulation made possible by the
computer. Government bureaus, private agencies, and businesses gather
data and maintain files of information relevant to their particular
needs. The sheer bulk of data accumulated on the scale that has
occurred over the last three decades would have been dismissed as
impossible 50 years ago, yet is taken for granted today. The real
challenge, of course, lies not in accumulating data, but in processing
it once it has been gathered.
The ability to analyze data via computer, once confined to computer
specialists, became the province of a much larger data-oriented audience
with the introduction of statistical packages such as SPSS, BMD, OSIRIS,
and SAS. Valuable though the available canned systems may be, all were
designed to operate efficiently on small or moderately sized data files.
The version of SPSS at Montana State University (MSU) limits the user
to 5,000 variables (Nie and others, 1975, p. 37). Although there is in
theory no limit to the number of cases that an SPSS file may contain, the
amount of available computer storage effectively limits the size of the
file at many computer installations..
The MSU version of SPSS will accept a file of 10,000 cases and
5,000 variables. Analysis of an SPSS system file created from this
data would be extremely unwieldy, since SPSS stores the value of each
variable for each case in one word of computer storage. The data for a
file of 10,000 cases with 5,000 variables would occupy 50,000,000 words
of computer storage. The amount of computer time required to perform
even simple statistical analyses on a file of this size is quite high,
making SPSS analyses of large amounts of data cost-prohibitive.
JV Celia Allard is Service Manager at the Center for Social Data
Analysis, Montana State University.
103
-------
The Center for Social Data Analysis (CSDA) at MSU routinely analyzes
a data file considerably larger than the one described above: a version
of the Continuous Work History Sample (CWHS) (U.S. Department of Commerce,
1976). The CWHS file contains 1,317,484 cases and 1,030 variables; it is
a one percent sample of the labor force employed in social security covered
occupations. CWHS data are compiled by the Social Security Administration
from employers' First Quarter report forms and selected other sources to
provide infomation on the worker's sex; race; date of birth; annual state,
county, and industry of employment; and an estimate of annual FICA income
from first quarter earnings. The CSDA version of the CWHS contains these
data for the years 1960 and 1965 to 1975. Work histories for individuals
employed through any number of successive years may be constructed, since
case identification numbers are included in the sample for each period.
The longitudinal nature of the CWHS allows a researcher to study
worker profiles for different industries during different years as well
as follow the movement of workers from place to place and industry to
industry. Migration researchers in particular have taken advantage of
the CSDA's ability to analyze the CWHS quickly and cheaply; their version
of the CWHS has been the focus of a number of demographic studies under-
taken during the last two years.
GRASP: A PACKAGE FOR LARGE DATA FILES
Analysis of the CWHS and other large files at MSU is made possible
by GRASP (Generalized Rapid Access Software Package), a statistical
package developed at the CSDA. GRASP is similar to other statistical
packages from the user's standpoint, but vastly different from other
packages in data base orientation. GRASP was specifically designed to
provide rapid, economical access and analysis of large data files
(Gilchrist and Wardwell, 1978; and Gilchrist and Allard, 1980).
The most fundamental difference between GRASP and SPSS is the
arrangement of cases in the data files on which they operate. SPSS
data files are arranged in the traditional case orientation, with all
variables attached to each case record, and cases arranged sequentially
in the file (Figure 1). This type of data file organization is by far
the most common; it reflects the fact that data are usually gathered
case-by-case. A typical survey, for example, involves questioning
participants on a number of topics. Each participant is interviewed
individually, so the results of a single interview are stored quite
naturally as one case in a data file.
Data are usually analyzed by variable although accumulated and
stored by case. What is of interest to the researcher are the responses
104
-------
of all participants to given questions rather than the particular
responses of any one individual. Consequently, it makes sense to store
data by variable rather than by case for purposes of data analysis. Much
of the efficiency of GRASP can be attributed to the variable orientation
of the data files, in which all values for each case form a single long
record (Figure 1).
Case oriented data files usually are arranged sequentially. In
GRASP variable-oriented data files, the records are ordered by a data
directory rather than physically arranged in a particular order. File
structure is entirely imposed by the data directory. Once a variable
name has been entered in the directory, data for that variable can be
separated physically from the data for any of the other variables re-
corded there (Figure 2). The data directory enables GRASP to quickly
locate the data for a particular variable rather than search for it
through a sequential file of variables.
The variable oriented data file structured by the data directory
partially explains the CSDA's ability to analyze large data files with
the limited resources available at MSU. A GRASP data analysis can be
performed using only those records which contain variables actually
being analyzed. For example, an analysis designed to explore the rela-
tionships between sex, race, and income from the data file of Figure 2
would require only the data directory and File 1. All the data encom-
passed by the directory need not be physically present in any one
location at the time of analysis. Consequently, variables from the
files can be stored on magnetic tapes when they are not needed for
analysis and transferred to disk packs as required.
The data in GRASP files in addition to being variable oriented are
word packed to optimize computer storage space. As mentioned earlier,
SPSS stores each data item in a word of computer storage. GRASP, by
contrast, packs data into words on the basis of maximum number of codes
per variable. For example, a dichotomous variable has a miximum of two
codes "yes" and "no". For any dichotomous variable, GRASP will pack
the values for 32 cases into a single word of storage. This means that
all values of a single dichotomous variable for all 1,317,484 cases in
the CWHS file can be stored in 1/32 of the space that would be required
if each data item occupied a full word of computer storage (Figure 3).
Similarly, variables having higher numbers of codes can be packed 16
variables to a word, 8 variables to a word and so on.
Dichotomous variables are not created merely to save storage space
however. GRASP is programmed to define samples extremely rapidly using
logical combinations of dichotomous variables. In practice, dichotomous
variables are used to define a basic sample; that sample can be refined
as necessary using variables that are not dichotomous (and thus require
105
-------
FIGURE 1. EXAMPLES OF CASE-ORIENTED AND VARIABLE-ORIENTED DATA FILES
Case Orientation
SEX RACE
Case 1
Case 2
Case 3
Case 4
Case 5
Case 6
1 1
2 1
1 1
1 2
1 1
2 3
Variable Orientation
AGE
32
17
25
27
46
34
INCOME
10000
5750
16000
24000
30000
17000
Case 1 Case 2 Case 3 Case 4 Case 5 Case 6
Record 1:
Record 2:
Record 3:
Record 4:
SEX
RACE
AGE
INCOME
1
1
32
10000
2
1
17
5750
1
1
25
16000
1
2
27
24000
1
1
46
30000
2
3
34
17000
FIGURE 2. RELATIONSHIP OF DATA DIRECTORY TO PHYSICAL STORAGE OF DATA
Directory
SEX
RACE
AGE
INCOM
OCC
YRSEXP
RANK-
EDUC
PAEDUC
PAOCC
UNION
HRSWORK
BOR:
REGIO
TAX/
Data
SEX, RACE, INCOME
OCC, EDUC, PAOCC, PAEDUC
TAX, UNION
YRSEXP, RANK, HRSWORK
AGE, BORN, REGION
106
-------
FIGURE 3. ILLUSTRATION OF PACK VARIABLE STORAGE
one value in one word of computer storage
32 dichotomous values in one word of computer storage
more computer time to process). For example, the CWHS file could be
used to define a sample of miners in the Western region in 1960 with
the following group of dichotomous variables:
SELECT: EMP60 AND WR60 AND MINING60
For comparative purposes, this sample could be redefined for a number
of different years or repeated for other regions of the country. The
definition of each sample using dichotomous variables would require
just a fraction of a minute of computer time.
Perhaps the most important implication of the speed of this
sampling technique is the resulting potential for analyzing large
data files using the same stepwise strategy that characterizes the
typical analysis of small data files. Taking full advantage of GRASP1s
sampling capability, a researcher might plan her first computer run
to consist entirely of sample definitions that would enable her to
explore the size implications of the various samples. She could choose
the samples best suited to her purposes for further analysis and con-
tinue the analysis as another series of steps, each designed in accor-
dance with information obtained in previous computer runs.
A stepwise analysis of this type is common with small data files,
but the sheer size of large data files stored in the traditional manner
makes stepwise analysis of these files prohibitively expensive. The
expense is rooted in the case-oriented structure of the file, which
dictates that all records must be read by the computer to extract the
values of the variable being analyzed. For a computer, reading data
records is a much more time consuming process than calculating results.
Since GRASP reads only the records containing the variables being
analyzed, the time and expense associated with reading data records is
kept to a minimum. Both approaches require approximately the same
amount of computer time for calculations, but GRASP spends much less
time reading data records.
107
-------
THE NATIONAL COUNTY DATA BASE
Not only do the data directory and variable orientation of the
files make possible analysis of large data files with limited resources,
they also make possible the creation of large data files from smaller
ones. The National County Data Base (NCDB) is a result of GRASP's unique
merging capability. When GRASP merges data files, it does not physically
combine their records; instead, it combines their variable names in a
single data directory. The NCDB is actually a huge number of variable-
oriented records sharing a common data directory. All files merged in
the NCDB contain county level data, and it is a fairly simple matter for
GRASP to add new county level files to the NCDB. With each file merger,
the NCDB becomes a more versatile analytic tool, since any variable in
the file can be analyzed in terms of any other.
The National County Data Base contains demographic, economic,
environmental, and social information for all counties in the United
States. Most of the data are longitudinal, permitting comparative
studies over time. The NCDB is currently comprised of the following
seven files, which are described briefly below:
(1) BEA Employment and Income Series
(2) County Portion of the 1947-1977 Consolidated County and
City Data Book
(3) Area Measurement File
(4) Area Resource File
(5) Fourth Count County Summary Tape (Human Resources Profile)
(6) Population and Net Migration Estimates
(7) Bowles 1950-1960, 1960-1970 Net Migration Rates
BEA Employment and Income Series 2j
The BEA employment and income data provide information concerning
the number of workers within industries and the aggregate income of wage
earners within industries. These county-level data are available annu-
ally for the period from 1969 through 1977. Used independently or in
conjunction with the CWHS, the BEA series provide the information neces-
sary for analysis of changes in county-level work force and industrial
structures. They are also used in conjunction with sociodemographic
characteristics from the Fourth Count County Summary Tape to identify
what types of counties are experiencing what types of change in employ-
ment structure and earnings.
27 Restricted data; permission to analyze must be obtained from
Dick Stuby, EDD,USDA.
108
-------
County and City Data Book
The 1914-1977 consolidated county information consists of a
longitudinal series of basic social, economic, and demographic data
including items drawn from a variety of governmental and private
agencies. Detailed data are provided for population, employment, vital
statistics, school enrollment, health, income, public assistance, social
security, banking, housing, government employment and finance, elections,
crime, manufacturing, retail and wholesale trade, selected services,
mineral industries, agriculture, and weather.
The county portion of the County and City Data Book is incorporated
into the NCDB; the CSDA may eventually nest the cities within their coun-
ties and aggregate across units to obtain additional county level charac-
teristics.
Area Measurement File
The Area Measurement File is provided by the Bureau of the Census
and is used to append the coordinates of county population centroids
to the county characteristics data system. With these coordinates,
distances between counties can be calculated to provide a measurement
of distance in migration studies. The land and water area of counties
are also obtained from these data and are used to provide a rough
indication of the amenities available for development of recreation
within the county.
Area Resource File
The Area Resource File is a county-based compendium of health-
related data developed in 1971 by the Health Resources Administration.
It contains detailed information on health manpower, facilities, and
training; population characteristics and economic data; hospital
utilization and expenditure data; and environmental data pertaining to
health, such as large animal population, elevation, mean temperature
and humidity, precipitation, and water hardness index. Various data
elements are available for different years; the ARF is updated and
expanded periodically. The version of the ARF available in the NCDB
is that of August, 1978.
Fourth Count County Summary Tape
The Fourth Count County Summary Tape, or Human Resources Profile
(Hines, et al., 1975) is a selection of demographic, social, and econ-
omic characteristics of counties drawn from the 1970 and 1960 censuses
of population. The 1970 data are drawn from 15 percent and 20 percent
samples, and the 1960 data from the 25 percent sample. It is the major
data set available for contrasting characteristics of metropolitan and
109
-------
nonmetropolitan counties. When variables from this source are merged
into the CWHS file, the CWHS can be used to construct gross flows of
worker movement between counties grouped on the basis of detailed
characteristics.
The major characteristics included in these data deal with popula-
tion growth and distribution; population composition with respect to
age, sex, and family structure; labor force participation and occupa-
tional structure; education; income; and incidence of poverty.
Population and Net Migration Estimates
The Bureau of the Census makes available a tape including the
1973-1977 population and components of change estimates for each county.
These estimates are prepared under the Federal-State Cooperative Program
for Local Population Estimates. The data consist of provisional and
final point estimates of population for each year, births and deaths
from April 1, 1970 to the end of the respective calendar year (i.e.,
1970 to 1974, 1970 to 1975, etc.), and residual net migration (i.e.,
the difference between net change and natural increase).
These data can be used in conjunction with county characteristics
data to identify variations in net migration experience by type of
county. They can be used in conjunction with CWHS data to compare
patterns of net employment movement with net residential migration.
Bowles Net Migration Rates
Estimates of net migration by age, sex, and color have been pre-
pared for each county (and larger aggregations states, divisions,
and regions). Bowles and Tarver (1965) prepared these data for the
1950-1960 decade; Bowles, Beale, and Lee (1975) have extended the
earlier work to the 1960-1970 decade. These data provide mobility
trends affecting rural and nonmetropolitan counties and regions in the
decades prior to the onset of the nonmetropolitan migration turnaround.
Used in conjunction with the later data (i.e., 1973-1977), they aid in
defining counties experiencing varying degrees and forms of change.
These counties are then analyzed in terms of their structural, func-
tional, and geographical characteristics to provide the bases for
inferring causes for and consequences of demographic change.
Possible Additions
Other data sets under consideration for addition to the National
County Data Base include the County Business Patterns to provide FICA-
covered employment information by size of firm to the 4-digit SIC
(Standard Industrial Classification) level; the Census of Governments
110
-------
data on county-level revenues and expenditures to examine fiscal and
service impacts of growth; and ultimately, the 1980 census county
characteristics data as they become available.
THE NCDB COUNTY STANDARD
The greatest difficulty the CSDA encountered in collating these
data sets was not a function of software or hardware; it was the purely
mechanical problem of coordinating the county codes in the various data
sets. A county code standard which would overcome the discrepancies
among the individual source data sets had to be created in order to
merge these data files into a single National County Data Base. Such
resolution was necessary to provide the basis for a single file in which
county data were comparable regardless of the data set of origin. The
county standard allows an analysis of any variable in the National County
Data Base in terms of any other variables. The standard is the 1970 FIPS
county code standard with extra codes added to resolve specific problems
(Figure 4).
The most troublesome county coding discrepancy encountered con-
cerned the treatment of counties in Virginia and Alaska. Some data
sets contained aggregated data for Alaska and Virginia, while others
contained data for the individual units (the new Bureau of the Census
units for Alaska, the independent cities and county remnants of Virginia).
FIGURE 4. NCDB-FIPS COUNTY STANDARD DISCREPANCIES
Total counties and county equivalents in FIPS standard: 3142
Units added to NCDB standard + 96
United States code 1
Statewide codes 50
Alaska aggregate codes 3
Virginia Aggregate codes 32
Wisconsin aggregate code 1
Outdated/new Virginia FIPS codes 9
Units removed from NCDB standard: - 2
Kalawao, HI 1
Yellowstone Park, part, MT 1
Total counties and county equivalents in NCDB standard: 3236
111
-------
Data that had already been aggregated could not be disaggregated. The
alternative seemed to be to aggregate the data for the individual units
and retain only aggregated data in the NCDB. There was not, however,
sufficient information to aggregate some of the variables present
median values, ranking values, and some percentage change values, for
example.
This problem was resolved by retaining the county codes for the
individual units and adding a special series of codes for aggregated
units of Alaska and Virginia. This solution has the advantage of pre-
serving both the individual unit data and the aggregated data; it has
the disadvantage of requiring that some values be omitted during every
data analysis to avoid duplication of county units.
The NCDB contains data for the individual units from all files
except those files offering only aggregated data. It contains aggre-
gated data from the aggregated files and also aggregated variables from
other data sets where the aggregated values were simple calculations
such as sums and means. For the sake of completeness, the CSDA plans
to aggregate as much of the individual Alaska and Virginia data as
possible. Some of the calculations required are rather elaborate, how-
ever, the aggregation will proceed slowly and steadily over time.
APPLICATIONS OF THE SOFTWARE/DATA SYSTEM TO SOCIAL RESEARCH
The GRASP software/data system offers many advantages to those
engaged in social research. The National County Data Base is unique
in the nation in its array of social and economic indicators, and the
number of indicators available will increase each time a new data set
is added. The addition of the 1980 census county characteristics data
will broaden significantly the scope of the NCDB and make it a more
valuable tool for social research. The versatility of the NCDB is
further heightened by its relationship via GRASP to the CWHS file.
Worker-oriented data from the CWHS can be computed for county units and
attached as new variables to the NCDB. Conversely, county level NCDB
data can be attached as contextual properties to cases in the CWHS file.
A researcher possessing a county level data set can submit that file
for inclusion in the NCDB to considerably extend her base of variables
available for analysis.
The Center for Social Data Analysis at Montana State University
provides access via GRASP to the Continuous Work History Sample and
the National County Data Base for researchers in government agencies,
private corporations, and universities. The GRASP software itself is
not available for transfer to other computers at this time. GRASP was
developed on the Sigma 7 computer at MSU, and some of its programming
112
-------
code is machine specific. Montana State University is currently
acquiring a new computer, however, and GRASP must be converted to run
on the new system. The new version of GRASP will be written in FORTRAN
77, a language which operates on many machines. Meanwhile, the CSDA
will provide both data and data analyses for interested researchers.
The CSDA will provide magnetic tapes containing those cases and
variables specified for those researchers wishing to acquire data for
analysis on their own computer systems. Since the seven data sets
discussed earlier effectively have been merged as the National County
Data Base and have a common county standard, variables from any of the
seven original data sets can be treated as though they came from the
same data set (as in fact they now do). The sample of cases specified
may be a few selected counties, a state, a region, or the nation as a
whole.
The data from these tapes can be transferred to the computer at
the researcher's institution, and can be analyzed using methods avail-
able. This service enables researchers to acquire data of a breadth
that they might have neither the time nor the resources to acquire
in any other way, since duplicating our collation is rather complicated
and collecting the data available in libraries is extremely time con-
suming. The disadvantage of acquiring data tapes for computer transfer
is that the efficiency of GRASP for data analysis is lost when this
option is taken, however.
The researcher can specify the analysis and have it executed at
the Center for Social Data Analysis to take full advantage of the data
retrieval and analytic capabilities of GRASP. The researcher is allowed
considerable flexibility in her analytic specifications. The traditional
approach to the analysis of a large data file has been to acquire the
maximum amount of information possible during a single computer pass
through the data file. The rationale for this originates in the case-
oriented structure of most data files. When each record in a file must
be read by the computer to extract information, as much information as
possible must be gleaned from a single computer run. There are two
disadvantages to this approach. First, if the researcher specifies an
elaborate analysis and has incorrectly estimated the samples or mis-
specified the variables on which that analysis is based, much of the
data she received from the computer may be virtually worthless for her
purposes. Second, much of the data generated by a one-pass analysis
may prove unnecessary or irrelevant for the project. Unfortunately,
the cost of examining large data files by computer has always been ex-
pensive. Sample counts and test runs simply have not been possible
except in instances where funding is not an issue.
This picture changes dramatically with GRASP. GRASP's variable
oriented data files and rapid sampling techniques make a stepwise
113
-------
analysis of large data files both logistically and financially feasible.
The CSDA encourages researchers to explore the size implications of all
samples under consideration before the full data analysis is run. Many
of the analyses performed consist of several computer runs, each building
on the results obtained from previous runs. A stepwise approach to data
analysis is encouraged for researchers who are unfamiliar with the data
file being analyzed, and for those who wish to probe the data file to
see what information is available.
This does not mean that the CSDA never performs a traditional one-
pass data analysis. Such analyses are frequently performed for
researchers who know exactly what they want from a data file and are
familiar enough with the data that the chances of satisfaction with
the resultant analyses are high. All is not lost even in those cases
where the one-pass analyses are disappointing. GRASP data analyses
are cost effective, so all or part of a wayward analysis can be re-
peated without impoverishing the researcher.
Finally, the researcher can benefit from the GRASP software/data
system by taking advantage of the analytic routines available. GRASP
is still young compared to more widely available statistical packages;
while all the common statistical routines are available in GRASP, some
of those less frequently used are not. However, GRASP has a few ex-
tremely useful procedures that are unavailable elsewhere, such as the
nonrecursive path analysis routine and the comparative case analysis
procedure.
A researcher wishing to undertake a nonrecursive path analysis
must first develop a model consisting of a number of related variables
arid the postulated causal relationships between them (Figure 5). Input
for a path analysis of the above model would be a list of the variables
involved and a list of the paths between them (e.g., SEX to DEGREE).
Given this input, GRASP identifies the path coefficients for all postu-
lated paths in the model. GRASP also decomposes the zero order correla-
tion coefficients between the variables into causal paths (both direct
and indirect) and noncausal paths and then identifies the relative pro-
portion and unique contributions of each of these components to the
original zero-order correlation. GRASP illustrates these paths graphi-
cally and gives the products of the path coefficients involved. Paths
can be added and deleted at the discretion of the researcher during the
analysis.
FIGURE 5. CAUSAL MODEL FOR PATH ANALYSIS
DEGREE-
f RANK ) SALARY
'EXPERIENCE-
114
-------
GRASP's comparative case analysis routine allows analysis of
relationships between individual cases. Any statistical package can
determine the number of persons in a sample within a given salary range,
but GRASP can also determine how many are within X dollars of another
and, for those who differ by X dollars, the average difference. The
matching capability of GRASP is not limited to case by case comparisons
of one characteristic; ranges may be specified for several characteris-
tics and the comparison made using all of them. The following GRASP
commands would define a comparison of all cases in the MSU Sociology
Department; the output lists the cases matched and the difference in
values for each of the variables involved in the comparison.
FIGURE 6. AN EXAMPLE OF THE GRASP MATCHING COMMAND
SAMPLE: (COLLEGE=6) AND (DEPT=6158)
MATCH: SEX (DIFF>0) AND DEGREE (DIFF=0) AND RANK (DIFF=0)
AND SALARY (DIFF<>50000)
The above analysis matches all pairs of cases differing in sex and
salary while having the same degree and rank.
GRASP will also limit the matching to particular specified cases.
For example, a researcher might choose a particular county of refer-
ence which experienced a decline in population over a given period.
GRASP could determine which other counties in the country have exper-
ienced a similar population decline. The comparison could be extended
to changes in industrial composition or income levels.
Another unique feature of GRASP is its series of linkages designed
to accomodate user-defined procedures. These linkages offer the re-
searcher the opportunity to truly custom design her analysis. If she
is conversant in standard FORTRAN, she can write her own subroutine to
be linked to GRASP. Alternatively, she can specify the type of analysis
desired and leave the actual programming to CSDA staff. The most elab-
orate routine of this type developed to date takes as input crosstabula-
tion data from three time periods and uses these data to calculate a
series of migration-related ratios.
During the past three years, a number of researchers have taken
advantage of the GRASP software/data system to further their investi-
gations. Much of the early research using GRASP focused on the non-
metropolitan turnaround phenomenon through analyses of the CWHS. The
National County Data Base has existed for less than a year, so research
utilizing that file has been less extensive. A partial list of research
topics appears below:
115
-------
(1) Industrial deconcentration trends within states of interest
compared to national trends;
(2) Patterns over time in nonmetropolitan retention of recent
metropolitan-to-nonmetropolitan migrants;
(3) Variations by region of the nation in sequence and extent
of metropolitan employment deconcentration;
(4) Separation of intraregional and interregional metropolitan-
to-nonmetropolitan migration flows;
(5) Separation of intrasystemic and intersystemic movement in
deconcentration processes;
(6) In-depth descriptive profiles of changing socioeconomic
trends in individual metropolitan and nonmetropolitan
counties and groups of counties;
(7) Patterns of change in workforce structure and industrial
distribution at the state level or by other groupings of
county units;
(8) Income inequality by region and race, focusing on counties
in the South; and
(9) Impacts of mining developments on population and economic
structure of rural states in the Northern Great Plains.
Some of these analyses utilize the CWHS, some the NCDB, and some use
both of these files. One of the more comprehensive analyses executed
to date required the creation of a data file incorporating variables
from both the NCDB and the CWHS. The records for this file were
counties comprising selected SMSA's, and the variables were drawn from
the NCDB and computed for each county of interest from the CWHS. Vari-
ables also were computed from combinations of the variables already in
the file. This custom-designed file contained several hundred variables
from a variety of sources when it was complete; it represents a truly
creative use of the GRASP software/data system.
An impressive amount of research has already been completed using
GRASP, but the investigations undertaken thus far only begin to tap
the potential uses of the system. With the existing data bases,
longitudinal and cross-sectional analyses of social, economic, and
demographic phenomena can be obtained with a breadth that was previ-
ously impossible at such speed and low cost. GRASPfs unique merging
116
-------
capabilities have dissolved the boundaries between discrete county-
level data sets with the creation of the National County Data Base,
a file which will become increasingly versatile as new data sets are
incorporated. The possibilities for exploratory analyses represented
by the GRASP software/data system are limited only by the imagination
and creativity of the researchers who take advantage of the opportuni-
ties offered by the system.
117
-------
References
Bowles, Gladys K., and James D. Tarver
1965 Net Migration of the Population, 1950-60, by Age, Sex and
Color. Washington, D.C.: U.S. Department of Agriculture,
Economics Research Service.
Bowles, Gladys K., Calvin L. Beale, and Everett S. Lee
1975 New Migration of the Population, 1960-70 by Age, Sex and
Color. Washington, D.C.: U.S. Department of Agriculture,
Economic Research Service.
Gilchrist, C. Jack, and John M. Wardwell
1978 "GRASP: A rapid access to the continuous work history sample."
in Policy Analysis with Social Security Research Files. Pub-
lication No. (SSA) 79-11808. Washington, D.C.: U.S. Depart-
ment of Health, Education, and Welfare.
Gilchrist, C. Jack, and Celia A. Allard
1980 "New Strategies for Processing Large Data Files in Migration
Research." In David L. Brown and John M. Wardwell (eds.),
New Directions in Urban-Rural Migration. New York: Academic
Press.
Hines, Fred K., David L. Brown, and John M. Zimmer
1975 Social and Economic Characteristics of the Population in
Metropolitan and Nonmetropolitan Counties, 1970. Agricultural
Economics Report No. 272. Washington, D.C.: U.S. Department
of Agriculture, Economic Research Service.
Nie, Norman H., C. Hadlai Hull, Jean G. Jenkins, Karin Steinbrenner,
and Dale H. Bent
1975 SPSS: Statistical Package for the Social Sciences. Second
edition. New York: The McGraw-Hill Company.
U.S. Department of Commerce
1976 Regional Work Force Characteristics and Migration Data.
Washington, D.C.: U.S. Department of Commerce, Bureau of
Economic Analysis.
U.S. Department of Commerce
n.d. Population and Net Migration Estimates (1973-1977): Current
Population Reports, Federal-State Cooperative Program for
Population Estimates. Series P-25, P-26. Washington, D.C.:
U.S. Department of Commerce, Bureau of the Census.
118
-------
U.S. Department of Commerce
1977 County and City Data Book: 1977. Washington, D.C.: U.S.
Department of Commerce, Bureau of the Census.
U.S. Department of Health, Education and Welfare
1978 The Area Resource File: A Manpower Planning and Research Tool.
Publication No. (HRA) 78-69. Washington, D.C.: U.S. Depart-
ment of Health, Education, and Welfare.
Wardwell, John M., C. Jack Gilchrist, and Celia A. Allard
1980 "A national county data base for rural research." Newsline
8(July):22-32.
119
-------
MODELING DYNAMICS AND DISEQUILIBRIUM IN LOCAL IMPACT
ANALYSIS: A LITERATURE ASSESSMENT
George S. Temple If
INTRODUCTION
Existing models of rapid growth are equilibrium models. But rapid
growth, illustrated by boomtowns associated with western energy devel-
opment, probably is as clear an example of a disequilibrium process or
dynamic adjustment process as one is likely to find. Current work on
dynamics and disequilibrium would seem to have a ready application.
This paper examines that work in the context of rapid regional growth,
and discusses its applicability.
Descriptions of the rapid growth process are familiar. Western
energy development mostly occurs in sparsely settled rural areas. The
indigenous population supplies enough labor to satisfy the requirements
of the local, uncomplicated economies. Energy projects are of large
scale compared to these economies. Therefore they are often disruptive
of local labor markets and public fiscal institutions. Impacts are
quite visible, and take time to work themselves out.
Changes in population summarize the timing and magnitude of impacts.
Many factors influence local population changes in response to employ-
ment changes. Foremost among these are the local multiplier effects,
labor force participation rates, the demographic composition of the
population, and the extent of commuting.
Local county employment multipliers are small due to the simple
economies which service agriculture and provide basic consumer services.
There is little manufacturing. Most agricultural inputs and consumer
goods are imported and agricultural products are exported in raw form
from the region. The same tendencies will hold true for coal mining
activities in the near future. Also excess capacity in the private
service sectors is another reason that incremental multipliers tend to
be small and nonlinear.
_!/ Department of Agricultural and Applied Economics, University of
Minnesota.
120
-------
Participation rates and the demographic composition of the labor
force change as employment increases rapidly in a rural area. During
construction periods, for instance, much of the new labor is composed
of single workers with extremely high participation rates. The partic-
ipation rate of residents increases as well during periods of high
labor demands. In each case, the effect is that population increases
more slowly than employment, especially in more densely settled areas.
Finally, workers may commute long distances to jobs if local
housing is unavailable or expensive, especially when construction crews
are large relative to the resident work force. Population impacts in
these cases are distributed over a wide geographic area along major
commuting routes.
However, these are initial effects. Once the construction phase
is finished, jobs associated with the energy project appear more stable
and of long term. Workers settle in the region with their families.
Rather than commute long distances, workers buy local housing, often
new construction. The service sector expands. Average participation
rates may return to "normal" levels. All of these imply migration into
the region as the local economy adjusts to its new equilibrium position.
There is no reason to suppose that this adjustment takes place
instantaneously. Since it requires migration as part of the equili-
brating response, the adjustment is likely to be rather slow. Indeed,
the stereotype of a boomtown is a position seemingly far from equilib-
rium. Part of the problem of analyzing impacts, then, is the issue of
the duration of the disequilibrium situation. This is disequilibrium
dynamics.
LITERATURE ON DISEQUILIBRIUM MODELING
There is a recent and growing literature on disequilibrium
modeling. In this section, that material is considered from the view-
point of rapid growth analysis. None of the papers are oriented toward
regional analysis. The question is whether any of the approaches can
be adapted to this kind of analysis. The critical concern is whether
the disequilibrium models incorporate adequate dynamic features.
The available literature can be divided into three groups: the
disequilibrium macroeconomic models, which take a general equilibrium
(general disequilibrium) approach; microeconomic models, which are
usually concerned with single markets and deal in partial rather than
general analysis; and a somewhat miscellaneous category of work invest-
igating the theoretical foundations of disequilibrium and dynamics.
121
-------
The groups are not equally relevant to regional modeling. Most important
are the microeconomics, econometrically oriented papers. The others turn
out to be peripheral, but are addressed first.
Theoretical macroeconomic models are presented in papers by Barro
and Grossman (1971), Buiter (1975), and Fortes (1977). These analyses
are not useful for rapid growth applications because the models are not
dynamic, but rather comparative static disequilibrium models. The
essence of the rapid growth problem is that the disequilibrium associ-
ated with it is a consequence of the significant dynamic changes coming
about at bottom it is an adjustment problem. Comparative statics are
insufficient to deal with this.
What the models do taking Barro and Grossman as an example
is to consider two sectors, a labor market and a goods market. The
existence of excess demand or supply in one sector is assumed, and the
implications for excess demand or supply in the other sector are examined.
Little attention is paid to any economic process which could eliminate
the disequilibrium. Rather the approach is used to shed light on tradi-
tional macro concerns static questions. A paper by Gourieroux,
Laffont and Monfort (1980) is the only macroeconomic paper that is empir-
ically oriented. Its contribution is to consider methods of estimation
of disequilibrium models in a simultaneous equations context. The only
other paper to consider this issue is one by Quandt (1976). Further
discussion is deferred to the empirical section.
Another category consists of papers which deal with the meaning of
disequilibrium, and with the economic theory deriving from it. A paper
by Kornai and Weibull (1977) is an example.
The paper discusses a queuing model applied to a shortage economy.
The model is meant to apply to economies such as those in eastern Europe
where queuing for consumer goods is a normal aspect of the economy. The
price of a good includes both a money price and waiting time in a queue.
Alternatives to buying the good are buying something else (with no wait-
ing time), postponing a decision, or forced saving. There is no uncer-
tainty in the model; the consumer knows just how long he must wait if
he joins the queue. The waiting time is not constant; the length of
the queue can vary, but at any time it is certain how much waiting time
is associated with a given queue. The model is used to predict queue
length. This depends on such factors as the strength of preferences
of consumers for the good in excess demand compared to other goods,
willingness to queue, and how rapidly a unit of the good is used up.
The paper is useful because of the way it integrates queuing into
what is really equilibrium analysis. One usually thinks of queuing as
a manifestation of excess demand and hence of disequilibrium. But in
122
-------
this model there are no forces set in motion to eliminate the queue.
Everyone who wants the good enough to join the queue eventually is
able to buy it. All demand is eventually satisfied. Queues do not
lengthen indefinitely. The price of the good is the sum of the dollar
price and the waiting time. This produces an equilibrium queue length.
Markets don't "clear" at each point in time, but this is because paying
the price for the good requires time.
This suggests that there can be "equilibrium excess demand" and
"disequilibrium excess demand". The latter obtains when forces are
set in motion which change the behavior of demanders or suppliers or
both; that is, when something happens to eliminate queues.
For rapid growth analysis queuing probably is to be regarded as a
disequilibrium phenomenon. The Kornai-Weibull model might be applic-
able during the disequilibrium phase, but would be incomplete in that
it would not explain the changes moving the rapid growth economy to-
wards equilibrium.
The final group of papers considered are microeconomic in approach.
The papers in this section form a homogeneous group in terms of subject
matter. All are primarily about the econometrics of disequilibrium
models. All consider variations of the same basic model, and ask simi-
lar questions. There are signs that this work is becoming the standard
approach to disequilibrium analysis. Structural aspects common to all
the papers are considered first.
Probably the most accessible papers of the group are by Fair and
Jaffee (1972) and by Rosen and Quandt (1978). Other representative
papers are by Quandt (1976), by Maddala (1979), and by Gourieroux,
Laffont and Monfort (1980). The models in these papers are not meant
to be definitive dynamic disequilibrium models. The dynamic structure
is primitive or sometimes nonexistent. In some cases the models make
no economic sense. They are intended only to illustrate the variations
in estimation methods as one moves to alternative specifications. How-
ever, the disequilibrium structure, which is common to all models, is
well specified. It is the disequilibrium structure which leads to the
interesting econometrics.
The key criticism is that the well-developed disequilibrium struc-
ture, combined with an incomplete dynamic structure, implies peculiar
dynamic behavior for some variables. This is most easily seen using
one of the more fully developed models such as the Rosen-Quandt model
or Fair and Jaffee's Quantitative Method. Such a model consists of
four equations:
123
-------
Q = f(pt, x)
(2) Qj - g(pt, X*)
(3) Qt - min (Q^, Q*)
(4) pt = k(q£ - q^), 0 < k < l
Equations (1) to (3) comprise the static structure of the model. Equa-
tion (4) supplies the dynamics. The first two equations are ordinary
demand and supply curves. Equation (3) specifies the disequilibrium
structure. The plausible reason behind it is that if price is at any
disequilibrium level, restriction on quantity actually traded should
come from the function which implies the smaller quantity. So if
price is below the equilibrium price, quantity traded will equal quan-
tity supplied and some demand will be unsatisfied. The converse holds
if price is initially too high. That is, given voluntary markets, sup-
ply and demand curves represent the maximum quantities that will be
traded at any price.
An implication of the model is that 6 is never negative, quantity
can never decline. This is easily seen by writing:
^t _ fc d s ^t
Now t is just the slope of the demand curve or the supple curve
(whichever is operative according to equation (3)). Given normally
sloped curves, the result is that Q >^ 0. This is fine as long as
the source of the disequilibrium is expansion. The reader can easily
verify that if either the demand curve or the supply curve shifts out-
ward, q grows and p moves to new equilibrium values at rates which
depend on k and on the slopes of the curves.
But this plausible behavior breaks down if either curve shifts
inward. It is equally easy for the reader to verify that in this
case, p moves at a finite rate to its new equilibrium value. The
behavior of q, on the other hand, exhibits two kinds of movement.
Initially it declines instantaneously. This is required by the min
condition. Then it grows at a finite rate until it reaches its new
equilibrium level. Since the initial decline is instantaneous, it is
"outside" the model in a sense, and does not contradict the conclusions
about 6 .
t
124
-------
The salient feature about this is how peculiar it is. When a
curve shifts inward, it is normally associated with a decline in
quantity. To model this by a large, instantaneous drop followed
by growth at a finite rate is not plausible. Moreover, it is the
disequilibrium specification that is really responsible for the
situation.
One could certainly object that this criticism is artificial,
and occurs only because the dynamic structure is not fully specified.
The assumption of instantaneous shifts in demand or supply curve is
equivalent to^assuming that X is infinitely large. If X is really
finite, then Q would decline at a finite rate.
But this raises as many questions as it answers. Most important
is whether such a specification is consistent with disequilibrium. If
demand and supply curves shift at finite rates, price and quantity may
change just as fast, implying continual equilibrium. One might rescue
disequilibrium by supposing that prices adjust more slowly, but this
is merely an assertion. Indeed, the common assumption is that prices
change faster than other things; otherwise there is no point in de-
scribing slides along, as opposed to shifts in, curves.
Even if one assumes disequilibrium, how to model it is unclear.
Equation (3) might not be satisfied. If P and Q adjust more slowly
than X , Q may be outside the demand and supply curves. This contra-
dicts the understanding of those curves as representing maximum quanti-
ties people wish to exchange. Some other explanation of disequilibrium
would be needed, and there is no obvious candidate.
The lesson to be drawn is that a proper dynamic specification is
not only necessary but is a prerequisite to modeling disequilibrium.
Only when the dynamics are in place can any disequilibrium structure
be imposed on the model.
This suggests that the four equations above are a prototype rather
than a finished product. This is admittedly true in terms of the
dynamics. But the arguments above suggest it may also be true in terms
of the disequilibrium structure. The next step would be to discuss
some of the ways in which the model might evolve. But before doing
this it is advisable to examine some of the econometric issues posed
by the prototypical model, while recognizing that these may or may not
be relevant to later versions.
The basic econometric problem seems to be partitioning the obser-
vations between the demand curve and the supply curve, assuming that
125
-------
observations lie on one or the other (but not both). How this is best
done statistically depends on the information available to use in allo-
cating the observations. The model with the least information consists
of equations (1) to (3). Price is exogenous. One way of proceeding is
to derive the probability that any observation lies in a particular
curve, conditional on the values of the exogenous variables including
price. From this a likelihood function can be found and maximized.
This is the method used in all the papers mentioned above. Another way
is to use the segmented regression technique developed by Cook, Lubov
and Stinson at the University of Minnesota. This technique allows one
to test for the optimal division of observations into one, two, or three
regression lines. In this case, we know there are two lines and the
technique would optimally assign the observations to each line.
Models with additional information are those in which price plays
some extra role. The role can vary. One possibility is that the direc-
tion of the price change signals whether excess demand is positive or
negative, and hence which curve the observation applies to. Another
possibility is that price is an endogenous variable through some rela-
tionship such as equation (4). Each type of model leads to a different
likelihood function, and it is the derivation of these that occupies
most of the material in the papers named above.
There are a number of subsidiary points which round out the econ-
ometric literature. Quandt (1976) points out that if there is insuf-
ficient information in the model the likelihood function is unbounded.
One example is if the model consists of equations (1) to (3). Unbound-
edness implies that iterative maximization processes do not converge.
In general, however, this problem disappears if one is able to supply
more information, such as equation (4). Another problem pointed out by
Quandt is that it may not be easy to test the hypothesis of an equilib-
rium structure against a disequilibrium model. This is especially true
if the parameter set under the equilibrium hypothesis is not a subset
of the parameter set for the disequilibrium model. Technically, the
hypotheses are not nested. Maddala (1979) points out that the source
of this problem is model structure. One must ask what in the model
produces the (possible) disequilibrium, and then try to design a test.
This can be done in some cases.
Finally, it is interesting to note the changes that occur in esti-
mation methods as very small changes occur in dynamic specification.
The alternatives considered in these papers amount only to current or
lagged values of prices in the supply and demand curves. To begin with,
Maddala points out that the correct specification depends on the inter-
pretation of the model. The direction of causality determines the
specification: if prices change in response to excess demand, the
126
-------
lagged formulation is right. If excess demand exists because prices
can't adjust immediately, the current specification is preferred. How-
ever, these alternatives don't affect the likelihood functions much.
But this conclusion changes when more than one market is contained in
the model. Both Quandt and Gourieroux, Laffont and Monfort (1980)
consider multimarket models. Quandt?s is specified using current
prices. The other paper uses lagged endogenous prices. This model
turns out to have a likelihood function which is more easily maximized.
LITERATURE ON DYNAMICS
Truly dynamic models of the labor market are even less well devel-
oped than disequilibrium models. At best there are some indications
of the right direction to take. This section will summarize a model
adapted from Sargent (1979). By way of introduction, models presented
by Nadiri and Rosen (1974) and Lucas and Rapping (1969) briefly are
considered.
The Lucas and Rapping model is a macroeconomic and the Nadiri and
Rosen model a microeconomic model of labor markets. They are similar
in that each model has a short run and a long run aspect. What moti-
vates the Lucas-Rapping model is a desire that labor supply functions
be homogeneous of degree zero in all variables in the long run (as
classical optimizing behavior requires) but allow for responses to
money wage changes in the short run (to model the business cycle).
This is accomplished by deriving an aggregate supply function which is
a difference equation. In the short run, aggregate supply responds to
price changes. In the long run, if prices are stable for a long period
the supply function becomes perfectly inelastic. What is suggestive
about this model is that it is dynamic in that a difference equation is
used. However, the temporal aspect is rudimentary because the equation
is derived from a two-period model. This point will be developed more
fully later.
The Nadiri-Rosen paper can be thought of as a bridge between the
formal dynamic model to follow and the disequilibrium models of the
last section. Like the Lucas-Rapping paper, there is an explicit com-
parison of short run versus long run properties. Lucas-Rapping generate
the long run from the short run by assuming regularity in certain inde-
pendent variables, such as the price level. They impose regularities
such as constant prices as part of the definition of the long run. Once
this is done, the properties of the supply function are deduced. Nadiri
and Rosen, in contrast, define the long run as the period when labor
demand satisfies a particular functional form. The short run is a
127
-------
period of adjustment to this equilibrium. Thus the Nadiri-Rosen model
can be out of equilibrium in the short run. This is not allowed in the
Lucas-Rapping model, which is explicitly in equilibrium continuously.
The Nadiri-Rosen model assumes a Cobb-Douglas production function.
Factors independent variables include both stocks of inputs and
utilization rates. The first problem considered is to derive long run
factor demand functions. This is done in standard fashion by minimizing
input costs subject to the production function. Several interesting
results emerge. First, long run scale effects occur in factor stocks
but not in utilization rates. This coincides with a. priori expectations,
and results from the assumption of unitary elasticity of substitution
between any factor stock and its corresponding rate of utilization.
Second is the effect of factor price changes. Factor prices are defined
to include a number of components. For example, labor costs include
items such as wage rates, which imply that costs depend on utilization
rates; labor costs also include some user costs such as personnel depart-
ments which correlate with labor stocks. The effect of a change in
factor prices depends on what part of the overall factor price changes.
As an example, an increase in wage rates increases demand for labor
stocks but decreases utilization rates, with the latter effect dominating
total labor hours. The reason that stocks increase is that there is a
substitution effect between stocks and utilization rates.
Short run disequilibrium is allowed for by a type of partial adjust-
ment mechanism. Specifically, the actual change in the quantity of an
input in a period is a linear function of the difference between long
run equilibrium levels and last period's levels of all inputs (not solely
the input of interest). This provides feedback and interaction among
inputs; for example, a large gap between desired and actual stocks may
affect utilization rates. These functional forms can be translated into
equations identical in form to standard distributed lab models, implying
that such models can be interpreted in terms of feedbacks and inter-
actions among independent variables. Finally, conditions for dynamic
stability and speed of response are derived. One interesting result is
that all inputs have the same adjustment coefficients. This is a con-
sequence of the reduced form of a set of difference equations. So the
lag structure will be equivalent across inputs. The economic meaning
of this is that disequilibrium with respect to one input necessarily
implies disequilibrium elsewhere in the set of factors, as long as the
firm operates on its production function frontier.
The Lucas-Rapping model is attractive in that it derives a (dynamic)
difference equation from optimization by economic agents. The Nadiri-
Rosen model is useful in showing that continuous equilibrium is not
necessary for a well-founded dynamic model of economic optimization.
128
-------
However, both models suffer from the use of short run versus long run
thinking. This distinction becomes unnecessary when a complete dynamic
structure is specified.
The following model from Sargent (1979) is an example of such a
structure. It is a simple model of labor demand only. To be complete
it needs to be combined with a labor supply model and a statement of
equilibrium or disequilibrium market behavior. But it illustrates the
methodology dynamic models require.
Assume that a firm uses a single factor, labor. It faces a linear
production function and also finds it costly to adjust its labor force
from one period to the next. It is a perfect competitor in all markets.
The firm desires to maximize V ,
- f
where b is a discount rate, n is employment, f is value of output,
{a ..} is a sequence of shocks to productivity, and (w ,.} is a sequence
of wage rates.
In this formulation, the firm has an infinite time horizon, but
this is not crucial. In words, the firm maximizes the discounted sum
of production, net of factor costs and adjustment costs. Necessary
conditions include the Euler equations. In this problem, each Euler
equation is a second-order difference equation in employment, one for
each j , which has the form:
bdnt+j+l - d<1+b>nt+j + dnt+j-l = Wt+j - f - at+j
(For details of the mathematics producing this and the following results,
see Sargent (1979)). Solution of this difference equation yields the
following demand curve for employment.
= nt+j - I f bl(w - f -
This states that this period's employment will differ from last period's
by an amount that depends negatively on future wages, positively on
future productivity, and negatively on costs of adjusting the work force.
A model like this can be extended in several directions. More com-
plicated objective functions are easy to imagine, although one can very
quickly introduce terms that make solution of the difference equation
129
-------
difficult. Introducing uncertainty can be done readily, either by the
addition of random errors directly or by defining sequences such as
{a ..} to be stochastic processes. As long as V contains only linear
or quadratic terms it is easy to handle uncertainty.
As indicated above, the real bottleneck with this approach is that
making the model more realistic also makes it harder to solve. The dif-
ficulty of solution increases rapidly, so that progress is slow. Never-
theless, even simple models such as the above may produce valuable
insights. When combined with the other components needed to complete
a full labor market model, the results may be useful for analyzing local
impacts.
-------
References
Barro R., and H. Grossman
1971 "A general disequilibrium model of income and employment."
The American Economic Review 61(l):82-93.
Buiter, William H.
1975 Economic Policy in Short-Run Models and in Long-Run Equilib-
rium: A Theoretical Framework and Some Applications. Research
Memorandum 192. Princeton, N.J.: Princeton University,
Econometric Research Program.
Fair, R., and D.M. Jaffee
1972 "Methods of estimation for markets in disequilibrium."
Econometrica 40(3):497-514.
Gourieroux, C., J.J. Laffont, and K. Monfort
1980 "Disequilibrium econometrics in simultaneous equations systems."
Econometrica (January):75-96.
Kornai, J., and Weibull, K.
1977 The Normal State of the Market in a Shortage Economy: A Queue
Model. Seminar Paper 91. Stockholm: University of Stockholm,
Institute for International Economic Studies.
Lucas, R., and L. Rapping
1969 "Real wages, employment and inflation." Journal of Political
Economy (September-October):721-754.
Maddala, G.S.
1979 Disequilibrium Econometrics. Paper presented at the meetings
of the American Agriucltural Economics Association, Pullman,
WA, July 30.
Nadiri, M., and H. Rosen
1974 A Disequilibrium Model of Demand for Factors of Production.
New York, NY: National Bureau of Economic Research.
Fortes, Richard
1977 Effective Demand and Spillovers in Empirical Two-Market Dis-
equilibrium Models. Discussion Paper 595. Cambridge:
Harvard Institute of Econometric Research.
Quandt, Richard
1976 Maximum Likelihood Estimation of Disequilibrium Models.
Research Memorandium 198. Princeton, N.J.: Princeton
University, Econometric Research Program.
-------
Rosen, H., and Richard Quandt
1978 "Estimation of a disequilibrium aggretate labor market."
Review of Economics and Statistics (August):371-379.
Sargent, T.J.
1979 Macroeconomic Theory. New York: Academic Press.
132
-------
ASSESSING COMMUNITY IMPACTS THROUGH THE BLM PERMITTING PROCESS
David C. Williams I/
INTRODUCTION
The primary objective of the Administration's energy policy for
the early 1980's is to reduce America's dependence on imported oil.
The cheapest and most immediate way to do that is to conserve
first by cutting back consumption, but more importantly by making
more effective use of what we do consume. In the long run, we have
to concentrate on using renewable energy resources such as solar.
For the middle run, however, our energy policy calls for new
or more intensive development of existing oil and gas production,
and for significantly increased use of our massive coal resources.
Most of that development will come in the West. And within the West,
much of that development will occur on the public lands. Public
lands are those Federal lands managed by the Bureau of Land Manage-
ment. They total over 417 million acres over 20 percent of the
land surface of the United States. Since they are concentrated in
the eleven Western States and Alaska, they are a higher portion of
Western lands roughly three-fourths of Nevada, and one-half of
Alaska, for example.
More than one-half of the coal reserves of the country are on (or
under) public lands. BLM now is developing a comprehensive coal
management program which will lead to a regular, long-range leasing
of public lands for coal mining, starting in 1981. Much of that
coal will be used for generation of electric power, or for conversion
into synthetic fuels. The remainder will be shipped out of t;he area,
mostly by rail, but increasingly by coal slurry pipelines. In a
sense, the coal "converted" in the region will be shipped out too,
by electric power transmission lines, or by pipelines for synthetic
gas. Given the amount, and the scattered pattern of, the public
lands, it is going to be quite unusual to develop major energy
projects in the West without involving the Bureau of Land Manage-
ment (BLM) .
\J Chief, Office of Special Projects, Bureau of Land Management.
133
-------
Energy projects are going to need either: 1) a right-of-way
(R/W) grant for pipeline(s), such as BLM is issuing for the Alaska
Natural Gas Line, the Northern Tier oil line, and the ETSI (Energy
Transportation System, Inc.) coal slurry pipeline; or 2) a sale of
public lands for a power plant (or synthetic fuel plant), as BLM did
to the Intermountain Power Plant (IPP) , and is considering for the
Allen-Warner Valley Power Complex, the Deseret power plant, and
proposals at Bisti, New Mexico and White Pine, Nevada. These power
plants generally also need a transmission line R/W over the public
lands, and other Federal lands (especially National Forests). BLM
will also be considering coal gasification proposals, such as the
Mountain Fuel plant in Utah, and oil shale facilities in Colorado
and Utah.
Each of these projects involves a "major Federal action,"
the R/W grant or sale/use of land which requires an Environmental
Impact Statement. BLM is, except in rare cases, the lead agency on
these EIS's, BLM grants, each year, thousands of R/W's for small oil
and gas pipelines.
Over the past few years, the Bureau has been heavily involved
in major energy projects of national significance including the
Trans-Alaska Pipeline (TAPS)and the SOHIO Pipeline from California
to Texas. The Bureau has experienced the strain of dealing with
projects which cover several States, and are large, complex, highly
visible, and politically sensitive. That description is now
applying to more and more projects requesting grants, permits, or
sale of land from BLM. In the summer of 1979, BLM established the
Office of Special Projects (OSP) to handle major Bureau (and Depart-
ment of the Interior) projects that require special attention and
coordination.
The Office of Special Projects has two major elements, as shown
in Figure 1, the OSP Organization Table. The Washington office
manages the projects and provides coordination with headquarters
offices of BLM, DOI, and other Federal agencies. Of more interest to
those involved in the EIS process is the Special Projects Environ-
mental Impact Team (SPEIT), located here in Denver. It should be
noted, however, this is a Washington Office located in the field,
the only one of its kind in the Bureau. SPEIT is responsible for
the entire EIS process, including application (and pre-application),
scoping, preparing the draft and final environmental impact state-
ments, and drafting the decision document. The 31 persons on the
Special Projects Impact Team include the Team Manager and an
Environmental Coordinator, who is responsible for the quality control
134
-------
FIGURE 1. OFFICE OF SPECIAL PROJECTS ORGANIZATION TABLE
jjects Project Coordination
ipact team Core Group
ffice) (Washington Off ice)
Chief
/ Williams
I
/ Secretary
Walker
/ 'f
Project Administrators
Gurr Montross Patton
I ' 1
L_ , I
r : . :
Project Leaders ^^^
Writers-Editors/Clerks ^ ~^
S S
I
vi c.
o
0)
Technical Team
Engineer Hydrologist
Wildlife Biologist Botanist
Economist Air Quality
Geologist Specialist
Landscape Community
Architect Planner
Team Manager
Tulloss
Environmental
Coordinator
Public Affairs
Officer
Administrative
Officer
Support Team
Supervisory Printing Specialist
Senior Cartographer
Cartographic Technician
Illustrator
Word Processors (3)
135
-------
on all EIS's. A project leader is assigned to each project full-
time for the entire EIS process, and has the ability to call on the
11 technical specialists, including an economist and a community
planner. To make sure the EIS gets out on time, the team has a
report production capability.
Responsibility for issuing the right-of-way grant, permit, or
selling of land still rests with the appropriate BLM State Director.
For pipelines crossing several States, one State Director has
authority. The Washington project administrator for OSP stays
involved in the project until after the permits are issued, to assure
that the schedule is met, that necessary coordination takes place,
and of great importance to us that the findings and recommenda-
tions of the EIS are reflected in the terms and conditions of the
grant.
Four major projects have been assigned to the Office of Special
Projects: The ETSI coal slurry pipeline, the Northern Tier Pipeline,
the Alaska Natural Gas Transportation System (ANGTS), and our only
non-energy project the MX Missile System proposed by the Air
Force for the desert of Nevada and Utah. These projects are described
in Figure 2. OSP also monitors the Trans-Alaska Oil Pipeline,
coordinates the Bureau's activities relating to the Bisti power plant
proposal in northwest New Mexico, and is about to accept the
application for the Pacific Coast Pipeline which will ship oil from
the central valley of California to Midland, Texas (along the pro-
posed SOHIO route).
The "Special Project Process" is described in more detail in
Figure 3. This paper relates the major steps in the process to
needs for information about community impacts.
THE EIS PROCESS
The concept of the fully integrated ElS/decision process is one
of the strong points of the Special Project approach to energy project
management. There is a continuous flow, under sustained direction,
from before the application is filed until after the grant is issued.
The project is responsible for the pre-application/application,
scoping, the EIS preparation plan, the draft and final EIS documents,
the decision document, and the R/W grant of sale documents, including
terms and conditions.
136
-------
FIGURE 2. CURRENT OSP PROJECTS-FEBRUARY 1, 1980
The Office of Special Projects' current assignments are listed and
briefly described below.
Energy Transportation Systems, Inc. (ETSI) - This project proposes a
1700-mile long subsurface coal slurry pipeline that will transport
37.5 million tons of coal per year from the Powder River Basin in
Wyoming to electrical generating plants and markets in the South.
The Wyoming State Office has been assigned the field responsibility.
OSP is responsible for scoping, preparing the environmental impact
statement (which is under contract), and drafting the decision
document. George Gurr is the assigned Project Administrator, and
Dick Traylor is the assigned Project Leader. Presently, the project
is in the environmental impact statement preparation phase.
Northern Tier (NT) - On January 17, 1980, the President chose the
Northern Tier pipeline for expedited permit treatment. This project
proposes an approximately 1600-mile long pipeline to transport up to
933,000 barrels per day of crude oil from the west coast to Minnesota.
The Montana State Office has the responsibility for issuing the right-
of-way grants. OSP is providing the Washington Office coordination.
Dan Patton is the assigned Project Administrator. Presently, the
project is in the "expedited" permit granting phase.
Mobile Intercontinental Ballistic Missile System (MX) - The Air Force
is proposing to base 200 MX missiles on public lands in Nevada and
Utah by the year 1986. The Air Force is the lead agency responsible
for the environmental impact statement. The Nevada State Office has
the field lead (working with the Utah State Office). The OSP has been
assigned the responsibility for coordinating with the Air Force in its
EIS preparation process to insure that effects on public lands and
lightly populated regions are addressed. David Williams is the assigned
Project Administrator and Bob Pizel is the assigned Project Leader.
Presently, the project is in the scoping phase.
Alaska Natural Gas Transportation System (ANGTS) - This project proposes
a 48-inch diameter pipeline to transport natural gas from the Prudhoe
Bay oilfields to markets in Southern California and the Midwest. The
proposed route parallels the Trans Alaska Pipeline route through Alaska
and the Alcan Highway through Canada. Near the U.S. - Canadian Border,
the route splits. The western leg follows an existing pipeline system
137
-------
FIGURE 2. ('CONTINUED)
to San Francisco. The northern border leg travels through Montana,
North and South Dakota, and terminates in Iowa where it connects with
existing pipelines leading to the Midwest. Field responsibility is
assigned to the Alaska, California, and Montana State Offices. OSP
is proving the Washington office coordination for BLM. Larry Montross
is the assigned Project Administrator. Presently, the project is in
the permit granting coordination phase for the western leg and northern
border. The Alaska Natural Gas Transportation System Coordinator, in
the Office of the Assistant Secretary, Land and Water Resources, has
overall Departmental responsibility for ANGTS. OSP also serves as
BLM's Washington Office monitor of the Trans-Alaska Pipeline.
138
-------
FIGURE 3. SPECIAL PROJECT PROCESS
Project proposal is Identified bv
Headquarters throupl Congressional
or Secri'ary contacts with applicant
, Prefect proposalIB identified byI
1 Statt Office througt contact with j
apr ' leant i
EARU
Office of Special Projects (140) IB notified
and confirms Information and significance of
project from various field offices, OSP
notifies W.O Divisions
DESIGNATION
OSP (140) prepares option paper and forwards
to Directorate.
Director designates a special project and
assigns responsibility to OSP * Project Is
added to the Bureau's Issues Management
System 01 Secretary's Critical Issue Manage-
ment System The Director Issues an instruc-
tion memorandum outlining the responsibilities
of OSP and involved State Office and setting
j out expected termination of OSP role
Chief, OSP. assigns Project Administrator and
Team Manager assigns Project Leader
A Project Committee convenes, consisting of
Project Administrator; Team Manager; Project
Leader; State Director or designated authority;
and OEPR staff member.
COORDINATION
AND PROJECT
PREPARATION
(Project Committee prepares scoping, scheduling,
| and Implementation plans.
Project Leader and State Director or designated
authority hold Federal agency scoping meeting to
identify involved, local, State and Federal
agencies and to establish coordination and
agreement requirements.
Project Leader manages the project including the
draft and final EIS's and the decision document.
OSP with Bureau Executive Secretariat provides
final review of assessment and decision document,
briefs declsionmakers.
Decisions are made; applicant notified.
PROJECT
DECISION
State Office prepares right-of-way grant or Bale
document, with guidance from Lands and Righta-of-Way;
OSP ma> continue as W.O. coordinator on project.
OSP effort Is terminated; development phase
monitoring Is assigned to appropriate field office
and W.O. Assistant Director.
*lf not designated, project Is assigned to appropriate
State Office and W.O. Dlviilon
139
-------
Pre-Application/Application
One of the biggest problems BLM and other permitting agencies
have had is getting a complete and adequate application. To get the
clock running, applicants often file before they have all the data
needed. Sometimes they don't want to spend too much time or money
before applying, and getting a sense of the BLM reaction. Sometimes
they don't really have the project in final form; the project which
changes its dimension while the application is being considered is
all-too-familiar.
A more complete application is needed to do a good job of
evaluating project impacts. That means a clear and detailed descrip-
tion of the project. For consideration of community impacts, it
is necessary to know at a minimum the best projections of
employment by number, by year, by skill, and by income. It is
necessary to see the pattern of employment over the life of the
project, and if possible the projections of local vs. outside hiring.
It's helpful, too, to have the projected spending in the community
for materials and services.
For the ETSI project, the project office developed a listing of
the information needed at the beginning of the process; now that
list is being tried on the next project the Pacific Coast Pipeline.
The application will be filed with the California BLM State Office.
Previous personnel are working with the applicant, the State Office,
and the State government to obtain that quality application needed.
The State government is involved because under BLM and CEQ guide-
lines, the office has the authority to prepare joint EIS's with State
and local governments. California will require an EIS on the
Pacific Coast project, and the OSP is negotiating to make it a joint
one prepared by BLM and fully acceptable to the State.
At the pre-application stage, the OSP also makes the determina-
tion of who will do the EIS, and how. Under the Office of Special
Projects, EIS's may be prepared in one of three ways: by the
Special Projects Impact Team in Denver (in-house), as will be done
on the Pacific Coast project EIS; by a third-party contract, as on
the ETSI pipeline; or by the Federal applicant, as on the MX missile
project. Under third-party contracts, the applicant identifies
consultants (not associated with the applicant) who are qualified;
BLM makes the selection; and a contract is signed between the
applicant and the contractor. Payment is made direct, and no Federal
funds are involved. BLM sets out the scope of work, manages the
project closely, and prepares the final EIS and decision document.
-------
A mix of the three techniques likely will be used for EIS preparation
on BLM projects.
Scoping
The requirement of "scoping" as the first step in preparing an
EIS was one of the significant changes in the CEQ (Council of Envir-
onmental Quality) regulations made in 1979. The scope of an EIS
is the range of actions, alternatives, and impacts to be included in
the document. Scoping determines the significant issues to be con-
sidered in the EIS; and identifies the insignificant issues which
will not be considered.
In the ETSI project, the first go through the scoping process
under the new office, the Federal agencies involved were identified:
BLM, Forest Service, Fish and Wildlife Service, Corps of Engineers,
and U.S. Geological Survey. Two Federal scoping meetings with
these agencies (in Denver and Washington), and nine public scoping
meetings along the length of the project (Wyoming to Mississippi)
were called by BLM to identify and rank issues. At the public
meetings, 469 people attended for presentations of the proposal;
they then broke into work groups to assign priorities to the issues
identified.
For the ETSI project, 249 people "voted" on the issues. More
than half, 142, identified "subsurface water" as the most significant
issue. Local socioeconomic impacts was second, with 51 votes and
49 were concerned about employment effects of the project. Concern
was especially expressed about the construction phase impacts.
Based on these meetings, the Special Projects Impact Team pre-
pared ETSI Coal Slurry Pipeline Proposal: A Report on Public Involve-
ment in Identification of the Issues. A copy of this report was
sent to every person who attended the scoping meetings, to State
governments, and to interested Federal agencies. It played a key
role in developing the preparation plan.
Preparation Plan
Following the scoping meetings, the SPEIT Project Leader called
a meeting of the OSP staff (project administrator and project
leader), the Department's Office of Environmental Project Review, the
Fish and Wildlife Service (the EIS cooperating agency), the applicant
(ETSI), and the contractor (Woodward-Clyde) to establish the prepar-
ation plan. The groups worked out how to do the project: what
141
-------
agencies are involved, how to coordinate with States, what staff are
assigned, and what is to be the schedule. Primary attention was
given to the allocation of resources to the issues identified in
the scoping process.
Socioeconotnic impacts were the second issue. The major sub-
issues are long-term jobs and future water users, with secondary
concerns about construction impacts. About one-fourth of the effort
is devoted to socioeconomic impacts in the EIS, compared to about
one-half for the primary issue water. With completion of the
Preparation Plan, and its approval by the Director of BLM, the EIS
analysis is started.
EIS (Environmental Impact Statement)
The EIS addresses the issues identified in the scoping process,
with the emphasis assigned by the preparation plan. For the ETSI
project EIS, the chapter on "Affected Environment" will cover:
water resources, socioeconomic considerations, vegetation, wildlife,
aquatic biology, cultural resources, agriculture, air quality,
recreation resources, transportation networks, and wilderness.
Based on the scoping process, no sections will be prepared on
geologic setting, climate, visual resources, noise, forestry
resources, mineral resources, topography, or soils sections which
might apply to other EIS's, and which under the old guidelines would
have applied to every EIS. The time and effort saved by narrowing
the scope of the EIS can be applied, then, to the key issues, in
this case, water resources and socioeconomic impacts.
For community impacts (used interchangeably with socioeconomic) ,
we generally need to know: existing conditions (a baseline for the
future without the project) and the conditions after the project.
It will be the difference between the two which is defined as the
"impact" of the project. Impacts from the construction and long-
term operational phases of the project will need to be determined.
Further, the OSP will need to have information in enough detail to
assess what the problems are, and what can be done about them.
That is where models come in. It is my experience that the
patterns of rapid growth from energy development are similar, that
an experienced analyst can describe the population growth curve and
identify major problems based on a knowledge of the existing
community and the proposed project. Detailed information is needed
information oh who, when, and where. The who is numbers of people,
142
-------
by age, family size, and income. These characteristics will deter-
mine the impact on the existing community, and on the service needs
for the new residents. The when is the phasing the schedule for
impacts will do much to determine how severe they are. And finally,
the where defines the location of the need. It is no longer enough
to know the overall impact; it is necessary to know specifically
what communities and jurisdictions are affected.
The specific "Socioeconomic Considerations" to be addressed in
the chapter on "Environmental Consequences" in the ETSI EIS cover
both construction, and operation, maintenance, and abandonment, all
listed below:
Construction Operation, Maintenance, Abandonment
Employment Employment
Population Growth Population Growth
Housing Housing
Service Requirements Service Requirements
Local Tax Revenues Local Tax Revenues
Cost/Revenue Cost/Revenue
Gross Regional Product Gross Regional Product
Economic Related Activity Economic Related Activity
Economic Dislocations
Separate sections will address these impacts for the four
major elements of the project: preparation plants, water supply
system, slurry pipeline system, and dewatering plants. (This EIS
does not consider the supplying coal mines nor the using power
plants.) The service requirements are going to be more detailed;
the IPP EIS, for example, considered: water, sewage, solid waste,
education, law enforcement, fire protection and public health.
Though improvement is being made, EIS's have been slow to
address and especially to address with any devotion the "socio-"
portion of "socioeconomic" impacts. IPP discussed the "quality of
life" as: employment, shopping, inflation, age, homogeneity/values,
social and personal problems. The EIS, however, did not have any
measures recommended to deal with any of these problems.
It can be expected that EIS's prepared by BLM will give much
more consideration to social impacts. The Bureau is now embarking
on a two-year program to identify and deal with the "Social Effects
of the Federal Coal Management Program in the West." This cooperative
program with the State governments no doubt will provide aid to
143
-------
BLM's evaluation of major energy facilities as well. The project
briefing document states one reason for moving ahead now is:
4. Existence of economic, demographic (population), and
community service/facility models at Bozeman, Laramie,
Fargo, Ft. Collins, and other locations. These automated
approaches are operational and future work appears to be
largely refinement-oriented. Understanding and projecting
social effects remains a major scientific and administrative
deficit.
Decision Document
The final step in the EIS process is the document sent to the
President (as for Northern Tier), the Secretary of the Interior
(as for IPP), or the Director of BLM (as will be done on the ETSI
project). This document lays out for the decision-maker the options
available, and the information available on the effects, advantages,
disadvantages, costs and benefits, and mitigation measures of each
option. To a large extent, this information will come from the
EIS. In the Office of Special Projects, the same team that prepares
the EIS will prepare the draft decision document. The major issues
should flow right from the scoping process through the EIS to this
decision document.
Not everything from the EIS will be in this document; in fact,
the decision piece should be a compact distillation of the findings
and recommendations of the EIS. Further, there will be some informa-
tion in the document which came after the EIS, including the results
of any negotiations between the Bureau and the applicant on mitigation
measures.
The heavy reliance on the EIS, however, will be a significant
Improvement over previous practices, in which the connection between
the two documents was not clear. Certainly the community impact of
the project generally failed to make the decision document.
Summary
The place of community impact modeling in the BLM permitting
process is in the full extent of the EIS process, from scoping
through EIS preparation to the decision document. This is a major
expansion from the present role of modeling.
-------
COMMENTS AND SUGGESTIONS
The following points are made as helpful comments and sug-
gestions on the role of modeling community impacts, and should not
be considered as BLM positions related to its EIS process:
EIS's in the past have given little attention to socio-
economic impacts, but this is being corrected.
The major assessment of socioeconomic impacts is being
done by models; we may have come, however, from the point
of too few models to too many there must obviously be
duplication, and the decisionmaker is hard put to decide
which one to use.
The EIS's and the models within them have been designed
as "one-shot" affairs, not updated as information changes,
as they must be.
Some of the recent work on models has concentrated on the
beauty of the methodology or the numbers, and not on its
usefulness to evaluators, decisionmakers, or affected
communities.
EIS's have not developed mitigation measures for community
impacts, and even the impacts themselves have not been
carried forward to the decision documents.
The discussion of community impacts has been heavily
quantitative rather than qualitative. Social impacts
have been given little attention beyond some vague
reference to quality of life.
Model makers, therefore, are going to have to:
- focus on the user of the information concentrate
on what the decisionmaker needs to know, not what can
be produced;
- be wary of employment projections provided by the
applicant; they will invariably change (and usually for
the worse);
- build in a capability to adjust to changed schedules
(they will always change) and changed employment
projections;
145
-------
give a range of potential impact it is the scale,
direction, and timing that is important, not the
precise number; in fact, using very precise numbers
gives an impression of accuracy and validity that may
not be there;
use factors in the models which come from rapid growth
experience or apply to rapid growth situations; national
averages and "stable-state" characteristics are mis-
leading, to say the least;
address some of the qualitative questions through use
of numbers, e.g., the number of unemployed wives;
write clearly and briefly; and
know when to quit.
146
-------
PANEL COMMENT
LOCAL IMPACTS OF RESOURCE DEVELOPMENT
Jack D. Edwards I/
ACTIVITIES OF BLM IN IMPACT ASSESSMENT AND RELATED POLICY ISSUES
The Bureau of Land Management is involved in a broad range of
environmental impact statements since most major Federal land man-
agement actions have significant (and often controversial) effects
on the quality of the human environment. The social and economic
consequences on individuals, groups, communities, and regions are
often substantial. People are an important part of the human envi-
ronment and can express their concerns on how a specific project
would affect their pocketbook or way of life.
BLM has the lead in preparing:
-145 grazing management EISs covering 170 million acres of
Federal rangelands in the West, under a court mandate;
-12 timber management EISs on Federal timberlands, primarily in
the Pacific Northwest;
-9 regional coal statements, with 2 to be completed this year,
prior to additional leasing of Federal coal, and another
scheduled for completion in 1981;
-7 outer Continental Shelf Oil and Gas Lease Sale EISs, which
are targeted for completion in 1980-81.
In addition, BLM cooperates with other agencies in many other
EISsAir Force MX Missile Project underway in Nevada and Utah.
JL/ Project Leader, Office of Special Projects, Bureau of Land
Management, Denver, Colorado
147
-------
PRIORITY INFORMATION NEEDS FOR MORE EFFECTIVE
IMPACT ASSESSMENT AND POLICY ANALYSIS
Information needs are defined to include both data and analyt-
ical models. A first prerequisite is to have competent economists/
social scientists, preferrably on interdisciplinary teams near the
action. Comprehensive automated data bases and sophisticated impact
assessment models will be of limited value without a professional
social scientist to apply to the study area and translate results
for decisionmakers.
Summary Data Assessment
Economic and demographic data are generally adequate at the
county level, although data are not always utilized correctly.
There are continuing problems keeping data bases current and going
below county level to deal with communities or individual users/
operators. Social data, especially values and attitudes, are ex-
tremely limited and normally have to be acquired first hand in the
area.
BLM has three projects underway to improve/expand our social
and economic data bases:
1. Ranch Budget project - This is multi-year cooperative
effort with ESCS, FS, and BLM. It is geared toward
improving ranch operator cost and return data throughout the
West in order to strengthen user and community impact
analysis in Grazing Management Statements.
2. Census Users Group, Bureau of Census - has just completed
a two-year project for BLM for an automated economic/
demographic data (county) base with some data synthesis/
analysis. Pilot applications have been made in five states.
There is a need for continuing work to acquire sub-county
data.
3. Social Effects of the Federal Coal Management Program in
the West - Although this project will focus on social
effects of coal development, it will benefit social
analysis of other resource development programs in the
region. A handout on its current status is available.
1-48
-------
Analytical Models
Since models are abstractions from reality, they can be useful
in giving indications of direction, but should not be expected to
furnish precise numbers for exact future dates. As has been pointed
out by Leistritz and others:
-Models are tools for decisionmakers and don11 replace cieci-
sionmakers and are not an end in themselves;
-Models should be tailored or selected to fit situation,
pocketbook, and timeframes;
-There is a need for validity assessment of models.
David Betters, Colorado State University, conducted an abbre-
viated comparative analysis of two models that have been used
frequently in BLM environmental statements. This included the
Harris Model, used exclusively on the OCS oil and gas sale EIS; and
DYRAM, applied mostly to BLM grazing statements.
Like much other good research, the comparison confirmed what
was already suspected. The more sophisticated, expensive Harris
Model was more appropriate for analyzing impacts on large complex
regional economies (OSC, Coal Leasing), whereas DYRAM was more
applicable for assessment of range and timber management proposals
of a smaller magnitude in essentially rural economies.
SPECIFIC REACTIONS TO IDEAS PRESENTED IN PAPERS
The papers seemed to be for three audiences, or combinations
thereofmodel builders, model users, and decisionmakers. After
elimination of two papers (Temple & Mankhaus-Adams) geared strictly
for model builders, my brief critique was limited to the remaining
three.
Bender-Parcels pointed up the need to be aware of structural
changes in rural economies; and to document and incorporate changes
in economic baseline conditions into analytical models, for more
precise measurement for project-associated impacts. More specific
guidance on how to go about incorporating structural changes into
models would be helpful.
Murdock-Leistritz made three useful observations based on their
H9
-------
assessment of selected impact models:
1. Recognize limitations of impact assessment models. Most
of the current crop of impact models are limited to econ-
omic and demographic factors and do not cover social
and environmental aspects.
2. Need to assess validity of models.
3. Need for selectivity in choosing (or tailoring) model to
meet decisionmakers' needs and constraints.
Shriver's - Construction Labor Demand System (CLDS), coupled
with compatible regional impact models, could provide improved
estimates of direct-secondary jobs resulting from energy projects.
Although this system has to be taken on faith until applications are
completed, it would be useful in mitigating boom town impacts.
CONCLUDING REMARKS
So long as model-builders (economists) talk only with one
another and confine their discussions to matters of technique
rather than substance, there is a tendency to be clever rather than
useful, to substitute assumptions for facts, and to be more con-
cerned with the model (form) than with results. At the same time,
decisionmakers have been remiss in failing to specify their modeling
needs in reasonable timeframes and in providing necessary funding,
while user translations of model results have often missed the mark.
This workshop reflects a positive effort in reversing the situation
by bringing model builders, users, and decisionmakers together for a
constructive interchange of ideas and needs.
150
-------
PANEL COMMENT
LOCAL IMPACTS OF RESOURCE DEVELOPMENT
Judith A. Davenport I/
The Wyoming Human Services Project, begun in 1975, trains
multidisciplinary teams to go into energy-impacted communitites and
assist those communities to develop mitigation strategies to deal
with their human services problems resulting from rapid growth.
There are two major components to the Wyoming Human Services Project:
a university-based training program (funded by the National Institute
of Mental Health, DHEW) and a community-based service program.
The university-based component, housed at the University of
Wyoming, includes a Project Director, Field Director, multi-
disciplinary teaching faculty and support staff. Students are re-
cruited into a two-semester course aimed at developing skills in
community development and human service delivery in impacted areas.
These students, who must be seniors or in their last year of graduate
school, are recruited from a variety of "people-oriented" disciplines,
such as adult education, anthropology, communications, law, nursing,
psychology, recreation and social work. The curriculum includes
content on such topics: the nature and problems of impact communities,
team building, establishing helping relationships, community organiza-
tion, program planning and use of community resources. Upon gradua-
tion, teams of approximately five members are selected for one year
appointments to the community-based program. These teams then
receive additional intensive training for one month on campus arid
in the community.
In the community-based service program, team members, called
Project Associates, go into impacted communitites which have requested
assistance. Teams are financed by the local community from numerous
sources, including the Economic Development Administration, municipal
government, state government and energy companies. Team members are
placed, at no direct salary cost to the agency, in settings related
to their educational preparation. For example, a nurse might be
_!_/ Associate Professor and Director of the Wyoming Human Services
Project, Department of Social Work, University of Wyoming.
151
-------
placed in a mental health center.
Agency service personnel tend to be so overwhelmed with
increased caseloads and new assignments that they welcome any form
of assistance. Assistance is given through team members, who spend
twenty hours per week providing direct services. This tangible help
not only provides a needed service, but aids in securing community
sanction for the team. "Project" team activities consume the
remaining twenty or more work hours of each week. Team members,
usually from five or so different disciplines, work together to
help the community define impact-specific problems of service
delivery and to design and implement measures intended to alleviate
those problems.
Engaging in such activities and endeavors can be extremely
difficult for existing agency personnel. Attempting to cope with
the influx of people and problems may make them so "crisis-oriented"
that they have virtually no time for planning and preventive measures.
Such measures become more feasible with the assistance of Project
Associates, who work closely with a community advisory board and
local agency personnel to ensure local input and involvement in
problem definition and solution. In effect, the community-based
service component functions partly as a rural adaptation of a
traditional social work program the health and welfare planning
council, found in most urban areas and called by such names as Council
of Community Agencies or Council of Community Services. Clearly, the
thrust of the project is toward enhancing the community's ability
and capacity to solve its own problems. Therefore, the team does
not attempt to impose its own definitions of the problems. Problems
are identified and addressed by the community with team members
serving as catalysts, enablers and facilitators.
The Wyoming Human Services Project is in its fifth and final
year of operation and is in the process of analyzing training and
operations components. Much information related to working with
human services agencies and providers has been gleaned through project
experiences.
Providing technical assistance and facilitating planning efforts
to these communitites has been both challenging and rewarding.
Specific areas of concern are the lack of financial and other
resources, lack of planning information, and lack of planning time.
Tools for planning relating to human services and rapid growth are
almost non-existent. Information now contained in the Environmental
Impact Statements prove to be almost useless in assisting human
services personnel to surmount the many "people" problems related
152
-------
to an expanding stress-ridden population.
Several suggestions related to improving the utility of EIS's
for decisionmakers in human services could be made. The first, and
most obvious, would be the inclusion of material related to human
services in all EIS's. This information should include possible
riitigation strategies and possible funding sources, if needed, Dave
Williams' paper discloses that the Bureau of Land Management is
moving toward that direction in its increased concern with socio-
t-conomic impacts.
Some way of monitoring the degree of social change experienced
in the community from the beginning of impacting would be extremely
helpful. This would provide the community planners with information
on an on-going basis so that they could adjust their planning and
services for the changes occurring.
It would be extremely important for those persons who are
hired to conduct the EIS's in the Rocky Mountain/Plains states to
be from the West, or at least to have a thorough understanding of the
values and culture of this part of the country. This is more signi-
ficant in analyzing the social impacts than in projecting physical
impacts. Human services delivery and the changes expected to occur
due to rapid growth should be analyzed by someone from a human
services background. Many social scientists are employed to conduct
socioeconomic assessments but basically their assessments tend to be
grounded solely in theory and have little relevance to practicality.
Suggested mitigation strategies are either so global as to be
meaningless to decisionmakers or else contain suggestions which
are obvious but which do not contain problem-solving techniques.
If EIS's are to be relevant to human services providers and
decisionmakers, then some of the above suggestions should be
implemented. Communities experiencing problems of increasing
populations need all the assistance they can receive in order to
design well-planned communities.
153
-------
PANEL COMMENT
LOCAL IMPACTS OF RESOURCE DEVELOPMENT:
INTERIM REPORT ON THE HCRS/MCRO ENERGY-IMPACT PROJECT
Elise McNutt
Gage Skinner
Shell! Bischoff I/
The HCRS Mid-Continent Regional Office began investigating
energy impact issues in FY 80 owing principally to the expanding
role of energy as a theme of national and regional signficance.
Energy development, with all of its attendant social and economic
consequences, particularly affects several western states in the
Mid-Continent administrative area notably Utah, Wyoming, Colorado,
Montana, and the Dakotas.
Coal production is the cornerstone of the Administration's
accelerated energy program. Contending that delays in fully exploit-
ing America's coal resources can no longer be tolerated, a balance
is being sought between often conflicting environmental and energy
goals. An Interior Department report released in 1980 indicates
that coal production on Federal lands was 100 million tons in 1979,
a 25 percent increase from the 1978 level. By 1985, the report
predicted that 200 million tons would be mined on Federal and Indian
lands in the West, still a small percentage of what is expected to
be 1.03 billion tons mined overall in 1985. Coal now accounts for
more than 80 percent of United States' fossil fuel reserves but
supplies only 18 percent of the country's energy needs.
In addition to rising demand for development of coal and for
development of uranium resources, a tremendous effort is underway to
advance a synthetic fuels program. Legislative initiatives in
Congress are afoot to combat the energy crisis by creating a govern-
ment/ industry partnership that would initially authorize spending
$20 billion over 5 years to develop oil and gas from coal, shale
rock, tar sands, and other non-traditional sources.
J7 Heritage Conservation and Recreation Service, Mid-Continent
Region, P.O. Box 25387, Denver Federal Center, Denver, Colorado 80225
154
-------
While energy and mineral development boom and bust cycles are
not new to many western states, the coming decades promise to con-
front energy-impacted communities with disrupted local economies and
altered social traditions on a scale perhaps not heretofore imagined.
The entire spectrum of quality-of-life issues is surfacing in the
context of the 1980's. The type and quality of public services that
will be available to rapid growth communities is of increasing concern
to both government and the private sector.
The role of the HCRS Mid-Continent Region should be clear: its
business centers on recreation planning and development and on
natural and cultural resource protection and preservation. HCRS
conducts this business in a manner that is conducive to government/
private sector partnership, and develops service programs with the
conviction that public recreation and heritage values play a key
role in sustaining the health, well-being, anri social fabric of any
community.
The challenge in the coming months is to ensure that these
values are incorporated in the most effective way possible in
response to requests for planning assistance from impacted communities.
The initial strategy has led HCRS to become an active partner
in the broad-based energy development community. During FY 80, the
objective has been to systematically identify those individuals,
agencies, localities, and corporations which have a clear stake in the
Region's energy development.
Field contacts by HCRS/MCRO energy team members have resulted
in our own enlightenment about energy issues; in our acceptance and
recognition as partners in addressing energy impact needs; in the
identification of key pilot communities and areas toward which we will
focus our assistance. The following is only a partial list of those
communities HCRS now is serving on a pilot basis:
1. Ticaboo, Utah
2. Colstrip/Forsyth, Montana
3. Wright City andHanna/Elmo Area, Wyoming
4. Craig/Meeker/Canon City/Paonia, Colorado
Further, HCRS is focusing attention on financial assistance,
especially for front-end planning, citizen involvement, corporate
involvement, and basic community self-help or planning strategies such
155
-------
as the park/school concept, fundraising, volunteerism, advisory
councils, and so forth.
Two objectives designed to give practical application (trans-
ferability) of these strategies based on real, pilot project communi-
ties are: 1) a government/private sector-funded publication dealing
with case studies on how impacted communities are solving problems
and providing services in recreation and heritage preservation, and
2) an integrated regional workshop designed to address those issues,
engaging the participation of all segments of the energy development
community, with a lead role played by HCRS in formulating the agenda.
A more specific description of the HCRS/MCRO energy-impact
team's activities to date is outlined below:
Collected a List of Selected Contacts Dealing with Energy-
Impacted Areas, i.e., Public, State, Private, and Federal,
- Ticaboo, Utah. Coordinating role with the Utah Outdoor
Recreation Agency and HCRS to ensure that recreation and
cultural resource planning are fully examined and incor-
porated in joint site development formulated by Garfield
County School District, Plateau Resources mining firm,
Ticaboo Development Corporation, and architectural
consultants. Special focus on the park/school recreation
planning concept. Desired results: A planned unit
development that does not take on the character of a
"company town," that serves and stimulates the varied
recreational interests and diversity of the development's
residents and visitors, and that utilizes private sector
commitments in the most efficient manner possible. Greater
application of the HCRS "tool kit" strategies is expected
as the Ticaboo development progresses, as well as a
heightened awareness and action statement of the Utah
Outdoor Recreation Agency to deal with Utah's impacted
communities.
- Colstrip, Montana. In cooperation with the Western Energy
Company, Wirth Associates, Inc., the Montana Department of
Community Affairs, and other appropriate parties, HCRS
will gather documentation relative to the history and
development of Colstrip and Colstrip's park and recreation
system. Of particular interest is an examination of labor
productivity, turnover rates, attitudinal and social
benefits, and facility design and placement. The objective
156
-------
is to determine the impacts of a park recreation
program on an energy-development community with a
track record of advance planning and citizen, private
sector, and government involvement. What is the
state-of-the-art at Colstrlp? Are there common problens
and common solutions transferable to nearby Forsyth and
to energy-impacted settlements elsewhere?
Craig/Meeker/Canon City/Paonia, Colorado. Provision
of general and specific front-end planning assistance,
including but not limited to formation of impact-oriented
citizens' committees, formulation of a local citizen
planning process, development of a student intern and
adequate local master planning program, land acquisition
strategies, park/school concepts, and other applicable
"tool box" strategies in the HCRS Technical Assistance
program.
- Wright City/Hanna-Elmo, Wvoming. HCRS proposes inves-
tigation into Wright City (ARCO), a planned unit develop-
ment , also with a track record in front-end planning for
recreation. An analysis will be done of the role of
the recreation program in serving Wright City residents
and addressing issues of labor productivity, social
stability, citizen involvement, and other factors germane
to Wright City's unique development. There will also be
an examination of at least one other heavily impacted
Wyoming area.
- Finally, additional selected examples of land acquisition
or private sector involvement strategies in the HCRS,
Mid-Continent Region, will be cited in the publication,
again in the context of energy-impacted communities as
they cope with this ever-expanding problem affecting the
Nation's growth.
The Technical Assistance Division of HCRS is involved in energy-
impacted areas for the purpose of understanding the recreation needs
in a rapid growth area and of providing assistance in increasing
recreation and preservation opportunities.
Special attention to the recreation needs of impacted areas
will spur much needed analysis. More students are needed to provide
rapid growth towns with the solid data to defend recreation planning
and implementation. Recreation planning is needed early in the
development stages to assure quality-of-life and deter social problems.
157
-------
PANEL COMMENT
LOCAL IMPACTS OF RESOURCE DEVELOPMENT
Jean Ackerman I/
I work with a firm that is involved in several energy impact
situations; in each case the focus of our job is to provide
technical assistance to the local government to handle the impacts,
as opposed to designing impact measurement techniques.
I'm one of those people out in the trenches using the tools
that you have developed and I'm here today to report on how well your
tools are working for us, and people like us. Since I am an
economist, I will primarily be addressing economic impact issues.
We seem to be in a second generation of impact assessment
technology, where we have just begun to measure the effectiveness
of our tools. Some preliminary test results are in and they are
not encouraging.
Since there are no single accepted tools with which to measure
employment or population, we tool users are using multiple techniques
simultaneously. We cannot know absolutely whether one technique
will be more accurate than another, prior to the impact; all we can
know is whether the different techniques are producing radically
different results. If they are, we know to investigate and explain
the differences as best as we can.
There are a host of reasons why impact measurement techniques
have failed to anticipate impacts accurately that are totally
unrelated to the technique itself. The impactor may have needed
federal legislation to provide development incentives or financing
that never materialized; a labor strike or labor union problem may
cause a total shift in the number and residence of the construction
labor force.
I/ Briscoe, Maphis, Murray, and Lament, Inc., Boulder, Colorado.
158
-------
Rifle, Colorado is one community that is tired and callous from
hearing about the potentially severe population impact that may
accompany oil shale development. The pounds of impact studies for
the Rifle area address what will happen when the impact comes not
how to prepare now for future development which may occur at some
indefinite time.
Perhaps we in the energy impact field should focus more of our
attention on developing ways to help impacted communities be flexible
and responsive to energy impact when and if it occurs, rather than on
ways to measure the impact more specifically. Housekeeping efforts
such as insuring that local land development policies make new develop-
ment support itself, structuring a tax system to capture the new
impact, and improving the budget-making process are a few ways.
The policies available to the impactor with respect to housing,
transportation, work week schedules, education, community facilities,
etc. are powerful mitigation tools themselves. Rather than drama-
tizing the potential impact through a detailed numbers game, and
developing grand, expensive capital improvements projects and
programs to mitigate the impact, time may be more productively
spent working with the impactor to understand how they might
mitigate future potential impact before it occurs, thereby minimizing
the potential impact in the first place.
I get the impression that many energy impact researchers, like
true scientists, seek a controlled laboratory environment where the
community stays constant, and impacts can be imposed and measured
distinctly. We technical assistance people mess up the laboratory
by trying our best to mitigate as we identify potential impacts, so
that the eventual impact is imposed on a community that is as
prepared fiscally, economically, and socially as possible.
At least in the area of economic impact, we have some identifi-
able tools to use and examine. Usable social impact techniques are
hard to find. In absence of any defined social measurement technique,
we are using a social impact panel of experts to review the economic
impact results of one of our projects and to suggest the social
impacts these economic effects may cause through a modified delphi
technique. Our social "tool" is the composite grey matter of our
panel of sociologists with prior experience in energy impact.
This is a pragmatic way to examine social impacts very specific-
ally and is probably an improvement over a case study approach.
Although we hope this approach is useful, we know our "tool" is not
159
-------
founded on social theory which can be tested in any way.
This technology-transfer workshop has given me some specific
help, and I probably learned more than I offered. You may want to
consider broadening the participants to include impacted community
representatives in your future work sessions.
» U.S. OOVfKNMtNT HUNTING OFFICC 1M1 -757-009/8014
160
-------