Economic models of intraregional and interregional migration in the Northern Great Plains region
by David Matthew OMeara
A thesis submitted in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE
in Applied Economics
Montana State University
© Copyright by David Matthew OMeara (1977)
Abstract:
This study, as one part of an extensive analysis involving rural community development in the
Northern Great Plains Region, analyzes the availability of labor in these rural communities by means of
population migration. The availability of labor is considered important because human capital will be
needed to develop the vast coal reserves in the region as the Nation becomes increasingly dependent
upon coal as a source for energy.
The review of previous work dealing with migration establishes a basis for the methodology, used in
the study. The elasticity between particular economic conditions in the Plains Region (especially the
change in average wage) and migration are of paramount concern. A model for intraregional migration
is treated separate from interregional migration in order to find differentials in responses to conditions
in the Region. In addition to the economic conditions, certain State Economic Area characteristics are
introduced as controls for other causes to migrate. The capability of analyzing each factor of mobility
separately for origin and destination sites is accomplished by studying relevant gross migration rather
than net migration. After adjusting for simultaneity between migration and wages (where migration is
considered a determining factor for the change in average wage in a Plains' State Economic Area), the
values of elasticities between wages and the migration flows showed interregional inmigration to be the
most responsive. This implies that those living within the Plains in 1965 received more nonpecuniary
benefit from site amenities than did those living outside the region. The elasticity between the wage
and interregional inmigration is 3.13. The stream responding the least to the wage is intraregional
outmigration with an elasticity of -.35. However, this value proved to be insignificant. The elasticity
values adjusted for simultaneous inconsistency bias relating the wages to intraregional inmigration and
interregional outmigration are 1.68 and -.87 respectively. Besides the wage, other economic conditions
also proved to be important in determining the migration flows. STATEMENT OF PERMISSION TO COPY
In presenting th is thesis in p a rtia l fu lfillm e n t o f the req u ire­
ments fo r an advanced degree a t Montana State U n iv e rs ity , I agree
th a t the Library shall make i t fr e e ly a v a ila b le fo r inspection.
I
fu rth e r agree th a t permission fo r extensive copying o f th is thesis
fo r scholarly purposes may "be granted by my major professor, o r, in
his absence, by the D irecto r of L ib ra rie s .
I t is understood that
any copying or pu blication o f th is thesis fo r fin a n c ia l gain shall
not be allowed without my w ritte n permission.
Signature
Date
X -
/6 -7 7
ECONOMIC MODELS OF INTRAREGIONAL AND INTERREGIONAL MIGRATION
IN THE NORTHERN GREAT PLAINS REGION
by
DAVID MATTHEW O'MEARA
A thesis submitted in p a rtia l fu lfillm e n t
of the requirements fo r the degree
Of
MASTER OF SCIENCE
in
Applied Economics
Approved:
Med^y Major Department
Graduate Be
3ean
MONTANA STATE UNIVERSITY.
. Bozeman, Montana
A p r il, 1977
iii
ACKNOWLEDGMENTS
I would lik e to extend my g ra titu d e to the Chairman o f my
Graduate Committee, Dr. Lloyd Bender, fo r his immeasurable guidance
in the conception and form ulation o f th is thesis.
I am also indebted
to Dr. Richard Stroup, Dr. Verne House, Dr. John M. Marsh, and George
Temple fo r th e ir invaluable suggestions.
This study is one p art o f a la rg e r p ro jec t to determine methods
o f estim ating local impacts which is sponsored by the Economic Devel­
opment D ivision o f the Economic Research Service, United States De­
partment o f A gricu ltu re using interagency funds from the Environmental
Protection Agency.
Special thanks are extended to Mrs. Peggy Humphrey who expertly
typed the fin a l d r a f t , and to Mrs. M arjo rie Powers fo r her unlim ited
patience in typing the i n i t i a l copies.
F in a lly , thanks go to Gary Orser. whose computer knowledge and
s k ills were essential fo r the completion o f the study.
The re su lts and conclusions are the sole re s p o n s ib ility of the
author and do not necessarily represent the viewpoint o f others who
contrib uted, or the p o lic ie s o f the Montana State U n iv e rs ity , the
Economic Research Service, or the Environmental Protection Agency.
TABLE OF CONTENTS
Page
VITA.....................
ACKNOWLEDGMENTS
LIST OF TABLES.
LIST OF FIGURES
ABSTRACT. . . .
CHAPTER I :
ii
iii
vi
v iii "
ix
I
J u s tific a tio n of P ro je c t.........................
The Problem and Approach.........................
O b jectives.................• .............................. ....
Description of Study Area and Data. .
Population in Northern Great P l a i n s ................................. .... . . .
Economic C o n d itio n s .....................................
Trade A r e a s ...................................................................................................
Data and Sources...............................................................................................
14
14
16
16
CHAPTER I I :
REVIEW OF LITERATURE
M o b ility Theories .......................................................... .................
M igration as Simultaneously Determined With Local Economic
Conditions...................................................... . . .....................
Income Growth and M igration ......................................................
Labor Force P a rtic ip a tio n and M igration .............................
Income and Labor P a rtic ip a tio n Rates.....................................
Outmigration and Inm igration. . . ...........................................
Employment Changes and M ig ratio n ..............................................
Summary ...................................................................................................
CHAPTER I I I :
METHODOLOGY.........................................................
cxi Ln co CO
INTRODUCTION...................................................................................
19
19
23
24
27
28
29
31
33
34
Disaggregation o f the M igration Data................................. ....................
36
In trareg io n al Model ...................................... ■ ...................................... ....
36
In terreg io n al Models..................... . . ..................................................
37
In terreg io n al Inm igration .................
38
In terreg io n al Outmigration..................................................................
38
Variables Influencing M igration ..........................................
39
Local Economic Conditions ......................................................................
41
Control Variables .......................................................................................
45
The Simultaneous Aspects of Migration . . . ......................................
51
The Wage Model........................................................................................................ 53
Interdependency Between Wages and Migration . . . . . . . .
56
V
TABLE OF CONTENTS (c o n t'd .)
Page
In tra re g io n a l System.
In terre g io n a l System...................................................................................
CHAPTER IV:
58
59
INITIAL MIGRATION MODELS ......................................................
62
C o llin e a rity Among Variables Considered in the M igration
Models............................................................................................................
Ordinary Least Squares Estim ates..............................................................
64
68
CHAPTER V:
ESTIMATION OF THE MIGRATION MODELS.......................................
75
Two-Stage Estimation o f the In tra re g io n a l Model ................................
In tra re g io n a l Wage Model...........................................................................
Economic Conditions ...................................................................................
C ontrols............................................................................................................
Two-Stage Estimation o f the In terre g io n a l Model ................................
In te rre g io n a l Wage Model..........................................................................
Economic Conditions and Controls. .
..............................................
Three-Stage Regression R esults.....................................
Two-Stage Results ...........................................................................................
75
78
78
79
80
82
83
83
85
CHAPTER V I:
SUMMARY, ILLUSTRATION, ANDCONCLUSIONS ...........................
89
Illu s t r a tio n of R e s u lt s ......................................................................
Lim itations of the Models, Data andE s tim a te s ................................
C onclu sio n s......................... ! ..........................................................................
Recommendations ...............................................................................................
90
93
95
97
APPENDICES......................................................................... ......................................
99
Appendix A:
Numbered SEA's and States Used in Study.....................
100
Appendix B: Estimates fo r Aggregated M igration Models and
Wage M o d e ls ...................................................................
Appendix C: S ta tis tic a l Inference on Competing M igration
Models............................................................................................................
LITERATURE CITED.......................................................................................................
108
112
Vl
LIST OF TABLES
Table
I-I
Page
MIGRATION WITHIN AND AMONG NORTHERN GREAT PLAINS
STATES, 1965-1970...............................................................................
10
OUTMIGRATION AND INMIGRATION, PLAINS STATES AND ALL
OTHER CONTIGUOUS UNITEDSTATES,1965-1970..................................
12
STATE OF RESIDENCE IN 1970 BY STATE OF BIRTH, NORTHERN
GREAT PLAINS STATES...........................................................................
13
J II-I
VARIABLE DESCRIPTIONS ANDVARIABLES............................................
42
III-2
VARIABLES, RELATIONSHIPS, AND FULL MODELS FOR INTRAREGIONAL MIGRATION............................. .............................................
60
VARIABLES, RELATIONSHIPS, AND FULL MODELS FOR INTER­
REGIONAL MIGRATION...........................................................................
61
MULTICOLLINEARITY AMONG INDEPENDENT VARIABLES, INTRAREGIONAL MIGRATION MODELS, NORTHERN GREAT PLAINS, 19651970 ........................................................................................................
65
MULTICOLLINEARITY AMONG INDEPENDENT VARIABLES OF INTER­
REGIONAL OUTMIGRATION MODELS, NORTHERN GREAT PLAINS,
1965-1970..........................................................................
66
MULTICOLLINEARITY AMONG INDEPENDENT VARIABLES OF INTER­
REGIONAL INMIGRATION MODELS, NORTHERN GREAT PLAINS,
1965-1970...............................................................................................
67
OLS INTRAREGIONAL MIGRATION MODELS, NORTHERN GREAT
PLAINS, 1965-1970...............................................................................
70
OLS INTERREGIONAL OUTMIGRATION MODELS, NORTHERN GREAT
PLAINS, 1965-1970...............................................................................
71
OLS INTERREGIONAL EMIGRATION MODELS, NORTHERN GREAT
PLAINS, 1965-1970...............................................................................
72
TWO-STAGE LEAST SQUARES, INTRAREGIONAL MIGRATION
MODELS, NORTHERN GREAT PLAINS, 1965-1970 . . . . . . . .
76
1-2
1-3
III-3
IV - I
IV -2
IV -3
IV -4
IV -5
IV - 6
V -I
vi i
LIST OF TABLES (c o n t'd .)
Table
V-2
V-3
Page
TWO-STAGE LEAST SQUARES, INTERREGIONAL MIGRATION
MODELS, NORTHERN GREAT PLAINS, 1965-1970 ..............................
81
STANDARD DEVIATIONS OF COEFFICIENTS BETWEEN TWO- AND
THREE-STAGE REGRESSIONS.....................................
84
v iii
LIST OF FIGURES
Figure
Page
I-I
Northern Great Plains Study Area..................................................
18
III-I
Schematic Diagram o f the In te rre la tio n s h ip between
M igration and Wages Tested in th is S t u d y ..............................
57
ix
ABSTRACT
This study, as one part o f an extensive analysis involving rural
community development in the Northern Great Plains Region, analyzes
the a v a ila b ilit y o f labor in these ru ra l communities by means of pop­
u la tio n m igration. The a v a ila b ilit y o f labor is considered important
because human c a p ita l w ill be needed to develop the vast coal reserves
in the region as the Nation becomes increasingly dependent upon coal
as a source fo r energy.
The review of previous work dealing with m igration establishes a
basis fo r the methodology, used in the study. The e la s t ic it y between
p a rtic u la r economic conditions in the Plains Region (e s p e c ia lly the
change in average wage) and m igration are of paramount concern. A model
fo r in tra re g io n a l m igration is treated separate from in terreg io n al
m igration in order to fin d d if fe r e n t ia ls in responses to conditions in
the Region. In addition to the economic conditions, c e rta in State
Economic Area c h a ra c te ris tic s are introduced as controls fo r other
causes to m igrate. The c a p a b ility o f analyzing each fa c to r of m o b ility
separately fo r o rig in and d estinatio n s ite s is accomplished by studying
re le va n t gross m igration ra th e r than net m igration. A fte r adjusting
fo r sim ultaneity between m igration and wages (where m igration is con­
sidered a determining fa c to r fo r the change in average wage in a
P la in s ' State Economic A rea), the values o f e la s t ic it ie s between wages
and the m igration flows showed in terre g io n a l inm igration to be the
most responsive. This im plies th a t those liv in g w ith in the Plains in
1965 received more nonpecuniary b e n e fit from s ite amenities than did
those liv in g outside the region. The e la s t ic it y between the wage
and in terre g io n a l inm igration is 3 .1 3 . The stream responding the
le a s t to the wage is in trareg io n al outm igration with an e la s t ic it y of
-.3 5 . However, th is value proved to be in s ig n ific a n t. The e la s t ic it y
values adjusted fo r simultaneous inconsistency bias re la tin g the wages
to in tra re g io n a l inm igration and in terre g io n a l outm igration are 1.68
and -.8 7 re s p e c tiv e ly . Besides,the wage, other economic conditions
also proved to be important in determining the m igration flows.
CHAPTER I
INTRODUCTION
Recent s c a rc ity of energy derived from o il has awakened a quest fo r
new o il reserves, and more im po rtantly, a search fo r o i l substitu tes.
Coal is a major source fo r the production o f e le c tric a l energy as well
as a te c h n ic a lly p o ten tial source fo r gas used in re s id e n tia l heating.
The Fort Union coal formation in Montana and several coterminous states
is not only ric h in th is fo s s il f u e l , but the reserves also can be
recovered with r e la tiv e ease.
This abundant supply, the f e a s ib ilit y of
mining, and the q u a lity o f energy release and chemical composition of
the coal give the Fort Union coal an important part in the Nation's
energy program.
Large flows o f c a p ita l and human resources w ill be needed to mine
and process coal in the region.
This study is one p a rt o f a research
e f f o r t dealing with the problem o f form ulating a methodology to estimate
the economic impacts of coal development on rural communities in the
Northern Great P lain s.
The general problem of th is study is to re la te
local economic conditions to the m igration of people.
The area of
special in te re s t is w ithin f iv e states in the Northern Great Plains where
strip p a b le subbituminous coal deposits are located.
M igration is an
important aspect o f local impacts o f development fo r two reasons.
F ir s t ,
m igration is the single most important source of population change as
labor demands increase.
Second, the wage increases required to a t tr a c t
and hold ad d itio n al labor to an area w ill increase the c a p ita l and
-2 -
operating costs o f community f a c i l i t i e s .
Thus, local government costs
w ill be dependent upon the responsiveness o f m igration to changes in
local economic conditions, e s p e cia lly wage le v e ls .
J u s tific a tio n of Project
The expansion o f coal mining and conversion throughout the la t t e r
h a lf o f the 1970-1980 decade and into the f i r s t h a lf of the 1980's w ill
add to the economic base o f the study area.
The M itre Corporation
reported th a t the known coal resources in the Rocky Mountain region are
309,135 m illio n tons.
Recoverable reserves are estimated a t 33,345
m illio n tons (Huffman 1972).
Recoverable reserves w ill change over time
as new technology is developed and as the demand fo r coal increases.
In
1972, 31.8 m illio n tons o f coal were mined in the study area, while
a n tic ip a te d 1985 production is 306 m illio n tons.
The depletion o f 2,265
m illio n tons from current to ta l reserves o f 33,345 m illio n tons is pro­
jected during the 1972-1985 period (Federal Energy Adm inistration 1974).
Existing mines cannot d e liv e r the large q u a n titie s of coal expected
to be demanded in the near fu tu re .
Expansion of curren t mines and pro­
jected new mines by 1985 number 75 in the study area.
Average employ­
ment fo r a 1 0 -m illio n ton mine is estimated a t 220 employees ( Polzin
1974).
Because o f these 220 basic jo b s , other in d ire c t and induced job
opportunities w ill emerge.
The local population may increase by even
la rg e r amounts since each worker may have several dependents.
-3 -
Another source o f the increased labor demand could be from new coal
g a s ific a tio n and power generation plants w ithin the region.
Eighteen
new coal g a s ific a tio n plants are in the proposal stage (Federation of
Rocky Mountain States 1975).
Although few of these plants are expected
to be completed (Stroup 1976), th e ir impacts are c u rre n tly being evalu­
ated (U. S. Environmental Protection Agency 1976).
The peak construc­
tio n employment fo r building a plant capable o f producing 250 m illio n
standard cubic fe e t of gas per day approaches 3,000 employees while the
average operating employment w ill be about 625 ( Polzin 1974).
ating plants in the area are projected a t 68 (Smith 1975).
New gener­
A plant pro­
ducing 500 megawatts annually w ill need a peak construction employment
o f 625 and w ill require an average permanent work force o f 43 employees
.(Polzin 1974).
Since the study area is sparsely populated and has e s s e n tia lly an
a g ric u ltu ra l base, there may not be enough local labor to sustain
development.
Therefore, the size and c h a ra c te ris tic s of the labor force
lik e ly w ill change in the next decade, p rim a rily through changes in
m igration flows re la te d to the Northern Great P lains.
The expansion of the economic base, th a t is , "the change in those
a c t iv it ie s bringing export revenue or tra n s fe rrin g income into the
community from outside" w ill create an increase in the demand fo r labor
(Bender 1975, p. I ) .
demands.
Higher wages w ill be re fle c te d in these new labor
Thus, the lo c a lit y w ill experience increases in inmigration
-4 -
and decreases in outm igration.
The cause o f the change in m igration
rates p rim a rily is through events in the market place which r e fle c t the
economic conditions o f a lo c a lit y .
Also important are nonmarket aspects
such as amenities and social t ie s , which may change as the community
expands over tim e, and these also may a ffe c t changes in m igration.
M igration in to and from any community has consequences th a t are
sometimes overlooked when decisions concerning local growth are made.
When a community experiences an in flu x of people, the community may pro­
vide new f a c i l i t i e s such as s tre e ts and highways, sewage treatment
p lan ts, storm and sa n ita ry sewers, and schools.
There is also the burden
of increased f i r e , p o lic e , w e lfa re , and recreation services.
These
expenditures put stress on the revenue stru ctu re o f the community.
It
is thus important to id e n tify and qu an tify the migrants entering and
leaving developing areas so th a t r e lia b le decisions can be rendered.
Policy decisions also w ill depend upon the background conditions
of the community.
Each community is d iffe r e n t in terms o f average
income of re sid e n ts , d is trib u tio n o f income among re sid e n ts , ethnic
composition, social group s tru c tu re , c u ltu ra l values, tr a d itio n s ,
topography, settlem ent p a ttern s, in h e rite d physical s tru c tu re s , re la tio n
o f major transport f a c i l i t i e s , land values, commercial composition,
in d u s tria l mix, labor s k i l l s , governmental s tru c tu re , local government
e ffic ie n c y , in te re s t c o n flic ts , and many other facto rs ( Isard 1957).
These differen ces may appeal to d iffe r e n t types of migrants such as aged
versus young workers, s k ille d as opposed to the u n s k ille d , or profes­
sional against the white c o lla r worker.
Depending on these differences
as well as local economic conditions, p o licy decisions may need to be
d iffe r e n t from one community to the next.
A d d itio n a lly , planners may consider the e x te r n a litie s which are
imposed on a community when it s population changes.
For some m igrants,
the social cost to the community may be greater or sm aller than the
in d iv id u a l's p riv a te cost to relo cate in the community.
This may r e s u lt,
as f a r as the community is concerned, e ith e r in too many migrants
entering the community when the social cost is greater than the p riv a te
cost, or too few when p riv a te costs are greater than the benefits
received.
The o rig in a l population may wish to estimate the type and
extent o f development and it s impact on then, and then e x tra c t revenues
and make expenditure changes in the d ire c tio n s and amounts th a t minimize
fis c a l e x te r n a litie s .
The Problem and Approach
The general research ob jective is to r e la te m igration responses in
the Northern Great Plains communities to local economic conditions.
Three major concerns w ill be re fle c te d in the s p e c ific objectives and
methodology:
(a ) a generalized m igration model, (b) nonmarket considera­
tio n s , and (c ) the in terrelated ness of cause and e ffe c ts .
The generalized model to be constructed must be consistent in the
explanation o f both inmTgration and outm igration.
T h e o re tic a lly , market
-6 -
conditions in both destinatio n and o rig in areas influence the m igration
flow s.
Data are not a v a ila b le to r e la te an in divid ual m igrant's u t i l i t y
function to the conditions a ffe c tin g a decision to m igrate.
Rather
than th a t, conditions a t sending and receiving areas are linked to migra­
tio n flow s.
The observed m igration flows consist of people who were a t
the margin when conditions changed.
to three m igration streams:
The generalized model is applied
(a) in t r a regional m igratio n , (b) in t e r ­
regional in m igration, and (c) in terre g io n a l outm igration.
The ju s t if ic a t io n fo r three m igration streams is th a t nonmarket
considerations may influence the responsiveness of labor to changes in
economic conditions.
I
or social s tru c tu re .
People may have a psychic investment in a community
Family and social tie s may be valuable to them.
Development in any one community w ill a t t r a c t labor from nearby lo c a li­
tie s and from d is ta n t labor pools.
The magnitude and response of those
p o ten tial migrants in nearby areas may d if f e r from reactions of migrants
in more d is ta n t areas.
The reason th a t there are d ifferences is th a t
both market and nonmarket costs ris e as distance increases.
Wage ra te
changes w ill a t t r a c t more people from surrounding areas than from f a r o f f areas.
These are d ire c t pecuniary costs.
There may also be non-
pecuniary or psychic costs th a t d e lin e a te migrants as long- or shortdistance m igrants.
Short-distance migrants may fin d th a t there is
^Nonmarket considerations w ill influence the choice of explanatory
variab les in the m igration models as discussed in Chapter IV .
-7 -
little
loss in l i f e styles and environmental conditions when moving to
nearby areas.
However, over a more d is ta n t move, the thought of being
removed from one's place o f o rig in may play a ro le in the decision
process.
Therefore, th is study disaggregates the m igration streams
into in tra re g io n a l and in terre g io n a l moves and allows the c o e ffic ie n ts
associated with the economic variables to d if f e r fo r each segment.
The th ird major concern o f th is study is th a t some o f the local
economic conditions th a t influence m igration flows may be influenced in
turn by m igration.
Economic theory states th a t as wages in an area
change, p o te n tia l migrants are induced to stay or leave depending on
th is wage change.
However as migrants leave the supply o f labor changes,
thus influencing wage le v e ls in a lo c a lit y .
The interdependency of
m igration and local economic conditions means th a t an ordinary le a s t
squares estim ation technique w ill have a simultaneous inconsistency
bias.
In order to compensate fo r th is bias, addition al equations besides
the m igration equation must be introduced fo r each addition al endogenous
v a ria b le .
In th is case, only one a d d itio n al equation w ill be brought
in to the system, an equation fo r the change in wages over the m igration
period.
By doing so, the c o e ffic ie n t fo r the e ffe c t of wages on migra­
tio n can be solved as an in teg ra l p art of a system of equations.
This
is accomplished through the two-stage le a s t squares regression technique.
-8 -
Objectives
The objectives o f the study are:
1)
To estim ate the e ffe c ts o f selected economic variables
a t o rig in s and destinations and among gross m igration
flows fo r S tate Economic Areas in the Northern Great
P lains.
2)
To adjust fo r nonmarket influences in an e f f o r t to
distingu ish between the responsiveness o f separate
m igration flows.
3)
To adjust fo r the sim ultaneity between local wage
changes and m igration.
The re su lts of th is analysis can be used to a n tic ip a te how respon­
sive m igration is to local economic conditions; th a t is , where people
w ill move as conditions change.
Although beyond the scope o f th is
study, the re su lts can be used to estim ate the e ffe c t o f local economic
conditions on tax revenues and expenditures of developing communities.
I
Description o f Study Area and Data
p
The S tate Economic Areas considered to be the most representative
of the Northern Great Plains are located in Eastern Montana, North and
2A S tate Economic Area (SEA) is defined by the Bureau of Census as
a subdivision o f a s ta te consisting of a single county, or groups of
counties, th a t have s im ila r social and economic c h a r a c te r is tic s .,
-9 -
South Dakota, Eastern Wyoming, and Eastern Colorado.
This Region is
somewhat extended geographically throughout the Great Plains where
,
population den sity, employment o p p o rtu n itie s , and income le v e ls are
g en erally divergent.
been a g ric u ltu re .
The common bond among these areas h is to r ic a lly has
The Plains area is b a s ic a lly rural in the sense th a t
there are very few c it ie s with a population of 50,000 or more.
Areas in the Northern Great Plains have experienced inm igration and
outm igration even though t h e ir economic bases have been r e la tiv e ly
s ta b le .
Part o f the reason may be a normal turnover o f labor as the
market and it s p a rtic ip a n ts adjusted over time.
Area is ru ra l southeastern Montana (SEA4).
A ty p ic a l S tate Economic
The 1970 population was
33,310 and gross outm igration and inm igration from 1965-1970 was 8,746
and 5,649 persons re s p e c tiv e ly .
Thus, 15 percent of the 1970 popula­
tio n in th is SEA had moved in to the SEA w ith in the past 5 years.
The data presented in Table I - I are the m igration flows w ithin and
among states in the study area.
These numbers represent the magnitude
of aggregated in tra re g io n a l m igration; th a t is , these fig u res are given
by s ta te ra th e r than by a single SEA.
liv in g in Colorado SEA's in 1970.
For example, 130,035 people were
The number of people moving from
Colorado to an adjacent s ta te (Wyoming) is 6,950.
The number of people
moving to each o f the other states (nonadjacent) in the Region from
Colorado is 1,950, 916, and 3,338. The greatest number o f people in
)
any one m igration flow is w ith in s ta te s . The next la rg e s t flow is from
a s ta te to adjacent s ta te s .
- I O-
TABLE I - L
MIGRATION WITHIN AND AMONG NORTHERN GREAT PLAINS STATES,
1965-1970*
State of
Residence
1970
State c)f Residence 1965
South
North
Dakota
Dakota
Wyoming
Montana
Montana
Colorado
35,855
5,753
2,329
North Dakota
3,415
33,957
4,260
South Dakota
1,506
7,355
33,815
1,626
1,950
Wyoming
3,429
1 ,380
2,901
13,902
6,960
Colorado
4,918
3,656
5,341
11,914
130,035
*Source:
,
4,000
918.
3,338
916
U. S. Department o f Commerce, Bureau o f the Census, U. S.
Census o f Population, M igration Between S tate Economic Areas,
Washington, D. C ., I 97DT
i
I
-ITThe number o f people in each stream moving between states in the
study area and nonadjacent states is r e la t iv e ly small (Table 1 -2 ).
However, the aggregate o f the flows to and from states outside the study
area is a large part of to ta l m igration.
Data in Table 1-2 are the
number o f people moving to and from states in the study area and a ll
other s ta te s .
The number moving w ith in North Dakota is 33,957 people.
Those moving to the adjacent s ta te of Montana are 5,753 (Table I - I ) .
The m a jo rity of separate flows to nonadjacent states and to North Dakota
are sm aller (Table 1 -2 ).
However, C a lifo rn ia and Washington are
exceptions to the magnitude o f in te ra c tio n fo r most states in the study
area.
R ed istribution o f population re s u lts in an informal network fo r
communicating job inform ation among people throughout the Nation.
Those
people who have moved into the Northern Great Plains from elsewhere
lik e ly m aintain communications with people in other locations where
they have liv e d .
The re d is trib u tio n o f population is indicated by con­
tra s tin g s ta te o f b irth with place o f residence a t any point in tim e.
These data are reported fo r states in the study area in Table 1-3.
Approximately on e-h alf of the population reside in th e ir s ta te of b irth
in these s ta te s .
The fa c t th a t large numbers were born elsewhere
indicates th a t inform ation about other locations, is a v a ila b le .
-f
A 2 r-
TABLE 1 -2 .
OUTMIGRATION AND INMIGRATION, PLAINS STATES AND ALL OTHER CONTIGUOUS
UNITED STATES, 1965-1970*
O th er
C ontiguous
S ta te s
Montana
In
Out
Alabama
V ir g in ia
Arkansas
C o n n e c tic u t
Delaware
F lo r id a
G eorgia
Illin o is
In d ia n a
Kentucky
L o u is ia n a
Maine
M aryland
M assachusetts
M ich ig a n
M is s is s ip p i
M is s o u ri
New Hampshire
New J e rs e y
New York
N o rth C a ro lin a
Ohio
Oklahoma
P e n nsylvania
Rhode Is la n d
South C a ro lin a
Tennessee
Vermont
West V ir g in ia
W iscon sin
Texas ■
New Mexico
A riz o n a
Iowa
C a lif o r n ia
Nevada
Oregon
Utah
W ashington
403
257
665
1,263
554
293
376
128
53
62
1,495 1,097
740
301
1,845 1,940
809
578
304
403
506
348
180
363
552
419
451
710
1,349. 1,164
417
377
1,306 1,142
135
227
591
403
1,101
1,342
515
438
1,257 1,027
1,215
699
996
850
80
72
332
257
182
414
35
70
545
167
1,367 1,748
3,186 2,047
1,112
924
3,404 1,021
1,109
909
14,469 10,629
1,155
924
5,754 2,9 40
1,988 1,876
21,335 7,766
♦Source
N o rth
Dakota
In
Out
393
1,026
679
270
165
1,887
629
2,942
905
336
537
104
767
618
2,164
133
1,541
41
513
1,024
634
1,263
762
800
48
338
408
17
197
2,773
3,264
635
.2,119
2,181
10,798
418
2,646
502
5,284
N o rth e rn G reat P la in s S ta te s
South
Dakota
Wyoming
Out
In
Out
In
250
291
429
290
139
1,150
396
1,837
886
354
308
182
367
500
1,052
381
1,136
193
346
942
464
1,013
720
628
27
315
356
591
127
1,3 99
1,979
408
625
1,570
4,708
94
596
268
1,736
317
1,007
328
263
55
1,857
696
3,649
1,019
209
570
138
803
453
1,5 59
420
2,124
75
393
831
367
819
1,380
538
no
403
440
135
41
3,487
2,780
663
2,126
6,448
10,315
822
1,985
384
5,436
182
590
230
124
100
706
462
1,851
642
293
304
145
244
402
809
174
1,195
28
462
804
301
891
653
592
35
77
264
155
112
1,381
1,477
397
455
6,320
4,087
229
543
143
1,049
325
634
415
176
95
1,130
364
1,520
676
152
745
52
423
224
836
290
990
24
205
621
348
453
1,455
576
43
304
335
13
138
631
3,230
1,456
2,598
783
8,394
905
1,685
3,262
3,752
264
487
204
99
100
558
163
1,183
520
159
318
114
240
158
1,264
365
907
33
411
810
241
931
1,087
430
13
180
85
144
106
340
2,242
1,637
1,0 10
922
4,4 5 8
568
801
2,925
908
C olorado
Out
In
1,990
5,589
2,643
1,392
230
6,771
3,137
11,345
3,1 80
1 ,680
2,254
489
3,798
2,390
5,713
996
10,100
583
2,944
6,207
1,828
5,631
7,461
3,8 90
523
1,417
2,232
307
372
3,767
22,186
9,806
13,663
5,733
54,050
3,300
5,545
6,6 50
11,099
2,712
5,279
2,380
2,413
465
8,096
3,413
21,244
5,767
2,567
3,298
837
4,7 19
4,291
9,131
1,502
11,481
893
5,741
14,294
2,665
9,957
9,613
6,786
510
1,730
2,198
524
797
6,767
26,599
13,433
8,601
11,260
45,281
2,400
3,837
8,059
6,525
U. S. Departm ent o f Commerce, Bureau o f th e Census, U. S. Census o f P o p u la tio n
M ig ra tio n Between S ta te Economic A re a s, 1970, W ashington, D. C.
-1 3-
TABLE 1-3.
STATE OF RESIDENCE IN 1970 BY STATE OF BIRTH, NORTHERN
GREAT PLAINS STATES*
S tate o f B irth
Montana
S tate o f Residence, 1970
South
North
Dakota
Dakota
Wyoming
Colorado
400,119
5,180
3,506
8,602
10,491
North Dakota
38,750
452,112
16,211
3,605
10,651
South Dakota
12,680
14,800
472,797
85,98
18,413
Wyoming
10,419
1 ,089
2,886
146,877
23,072
Colorado
6,735
1,927
2,645
16,340
992,219
Montana
*Source:
LI. S. Department o f Commerce, Bureau of the Census, LI. S.
Census of Population, State o f B irth , Washington, D. C .,
1970.
-1 4~
Population in Northern Great M a in s .
The combined 1970 population in the State Economic Areas included
in the study area was 3,222,362 persons.
The la rg e s t c it y is B illin g s ,
Montana with a 1970 population of 87,350.
Other major c it ie s are
Bismarck (3 4 ,6 7 0 ), Mandan (1 1 ,0 9 3 ), Dickinson (1 2 ,4 0 5 ), W illis to n
(11,280)
in North Dakota; Miles C ity (9 ,0 2 3) in Montana; and Casper
and Sheridan, Wyoming (39,361 and 10,856) re s p e c tiv e ly .
The main flow
o f population throughout the 1960-1970 decade was from the ru ral areas
to these leading trade centers.
D alsted, L e is t r it z , and Hertsgaard
(1974) m aintain th a t there was a high ra te o f net outm igration from
each o f the areas.
Since a l l the areas had more outm igration than in -
m igration , i t is assumed th a t labor supplies in the past have been
reduced through labor m o b ility .
They concluded th a t the age group
below 25 (th a t might be prone to m igrate so as to increase th e ir
expected lif e t im e earnings) were a la rg e r percentage o f the population
than is tru e o f the Nation as a whole.
Therefore, age d is trib u tio n
may be important in determining m igration flows re la te d to the Northern
Great P lains.
Economic Conditions.
The tra d itio n a l economic base in the Northern Great Plains has
been a g ric u ltu re .
In the recent past, however, there was marked a
decrease in job opportunities in th is sector of the economy.
In
-1 5 -
Montana from 1960-1970, there was a decline o f 22.1 percent in a g r i­
c u ltu ra l employment, and the la rg e s t percentage gain in employment was
in the service in d u s trie s .
gained 7.4 percent.
Employment in a ll sectors w ith in Montana
North Dakota and South Dakota had the la rg e s t
declines in a g ric u ltu ra l employment (38.6 p ercen t). , Also, the to ta l
labor force p a rtic ip a n ts in each o f these states declined by 2.5 percent.
Wyoming's percentage decline in a g ric u ltu re was 26.7 percent, where the
service in dustries had the g reatest increase (29.3 p e rc en t), and to ta l
employment in a ll sectors increased by 3 .9 percent.
For the Northern
Great. Plains as a whole, there was a 33.9 percent decrease in a g r i­
c u ltu ra l employment during the 1960-1970 period; mining employment
increased by 8 .8 percent, manufacturing by 33.1 percent, and to ta l
employment by 6.1 percent.
Therefore, d iv e r s ific a tio n in the in d u s tria l
sector is evident throughout the Northern Great P lain s.
Furthermore,
declines in a g ric u ltu ra l employment have resulted from farm and ranch
consolidation, but the acreage devoted to th is sector remained about
constant (D alSted, L e is t r it z , and Hertsgaard 1974).
As fo r the economic w ell-being o f the labor fo rc e , in the Region,
incomes o f the residents in the Northern Great Plains usually f a l l
short o f the national average.
The 1970 per capita money income in
North Dakota was the lowest a t 72.2 percent o f the national average,
w hile Wyoming's northern area had 100.8 percent of the national average.
- 1 6-
Trade Areas.
C itie s w ith in the study area th a t are considered complete shopping
centers are:
Bismarck, Mandan, Minot, Dickinson, W illis to n in North
Dakota; M iles C ity in Montana; Sheridan and Casper in Wyoming; and Rapid
C ity in South Dakota.
In addition to the complete shopping centers,
the area is serviced by the primary w h o le s a le -re ta il center of B illin g s ,
Montana.
Bismarck, Mandan, Minot, Casper, and Rapid C ity also perform
c e rta in wholesaling functions and can be described as secondary
w h o le s a le -re ta il centers (Hoover 1971).
On the frin g es o f the study area there are four major m etropolitan
centers; Minneapolis -S t. Paul, Omaha, Denver, and Spokane.
G enerally,
the Dakotas r e la te economically to Minneapolis -S t. P aul, Wyoming to
Denver, and western Montana to Spokane.
However, these la rg e m etropoli­
tan areas do have some overlapping o f influence w ith in the study area.
Data and Sources
The data used in the study are p rim a rily from the 1970 LI. S. Census
of Population.
This census is the most recent documentation o f cross-
sectional m igration flows a v a ila b le fo r SEA's in the Northern Great
P lain s.
This secondary source is not as accurate or applicab le to the
1970 decade as would be more recent primary data.
However fo r purposes
o f th is study, the secondary data were a v a ila b le re a d ily and less co stly
than th a t from any other source.
-1 7 -
The m igration data are calculated from a 15 percent sampling o f
the population.
The independent variab les used in the study are calcu­
la te d on the basis of 5 percent to 20 percent samples.
Therefore, these
O
data may be subject to both sampling and response e rro rs .
People liv in g
in a Plains SEA in 1970 who liv e d elsewhere in 1965, and those who liv e d
elsewhere in 1965 but were liv in g in a Plains SEA in 1970 are observed
as the m igration flow s.
The c h a ra c te ris tic s of the o rig in and destina­
tio n areas th a t are re le va n t in the explanation o f m igration are based
on the 1970 sampling of the population.
c h a ra c te ris tic s are 1970 observations.
U nfortunately, most of these
I t would have been b e tte r to
have obtained both the 1965 and T970 observations.
The re su lts of the
study thus are dependent on data th a t r e ly heavily on secondary sources,
and data observed a t the end o f the m igration period.
The primary area under observation is demonstrated in Figure I - I .
For the s p e c ific observation SEA's and states see Appendix A .
Thus
Nebraska SEA's are excluded from the in tra re g io n a l model, because they
are not considered p art of the Plains Region fo r these flow s, but.are
included as o rig in s and destinations fo r the in terre g io n a l models.
The
adjacent states and portions of Plains states are excluded from a ll
models.
This was a re s u lt of the data and no economic ju s t if ic a tio n is
given fo r t h e ir exclusion.
O
See U. S. Department of Commerce, U. S. Census o f Population,
1970 fo r sampling procedures.
Area encompassing Northern
/ \S>y Great Plains SEA's fo r in t r a regional population movements
/N S y Area encompassing Northern Great
Plains SEA's fo r in terreg io n al
population movements
/ ___/ Area encompassing nonadjacent states
outside the Region fo r in terregion al
population movements
/LU/
Figure I - I .
Northern Great Plains Study Area.
Area encompassing adjacent states not
included in the study
CHAPTER I I
REVIEW OF LITERATURE
In attempts to apply c la s s ic al m igration theory, two major
approaches have been implemented.
These are (a) the behavior of in d i­
viduals according to t h e ir personal preferences, and (b) the adaptation
o f the theory to regional adjustment.
In divid uals can be studied in the
context o f human investment concepts where migrants maximize net benefits
expected over th e ir remaining productive life tim e ( Sjaastad 1962).
In
a regional system, people respond to differences among lo c a lit ie s .
M o b ility Theories
For an in d iv id u a l, a geographic move is a calculated decision in
response to changes in his preferences, or to possible net increases in
his op po rtu n ities.
The individual is assumed to take account of the
c e rta in ty o f his expectations and inform ation on income p o s s ib ilitie s ,
the value of nonmarket events and conditions, differences in price
le v e ls , and differences in r e la tiv e prices in each p o te n tia l lo ca tio n .
These expectations are evaluated with respect to corresponding costs
over the life tim e .
The p re d ic tiv e power of th is theory is dependent upon the ra tio n a l
decision processes of in d iv id u a ls .
I f the market is the sole source of
income, and i f information is c e rta in and costless, moving costs minor,
and labor s k ills uniform, then people w ill adjust u n til wages are
exactly equal in every labor market as described by Isard (1975).
-2 0 -
Differences in wage le v e ls can be a ttrib u te d only to a temporary d is ­
eq u ilib riu m , or to a v io la tio n of an assumption.
The value of the
c la s s ic al theory is th a t it s assumptions can be relaxed to is o la te
p a rtic u la r determinants of m igration.
Thus, abstracting from classical
assumptions, mere d ifferences in r e la tiv e prices of goods in two regions
due to tran sportation costs could re s u lt in wage d iffe re n c e s .
The c la s s ic al theory has been adapted to regional microcosms fo r
empirical v a lid a tio n .
In th is case, events in a regional economic
system are taken as indications o f opportunities fo r people.
I f the
same assumptions are made, then m o b ility w ill lead to conclusions s im ila r
to those o f the c la s s ic al model.
Empirical in vestig ations of m igration
responses of in d iv id u a ls using the c la s s ic al theory are extremely
expensive.
Empirical work on regional labor m o b ility has been more
popular, probably because data are a v a ila b le .
For the most p a rt, regional m o b ility models can be c la s s ifie d as
g ra v ity , push, or pull models.
R e illy (1929) was one of the f i r s t to
r e la te the Newtonian Theory o f mass and distance to the movement of
people between two places.
His model can be stated as:
M1-J = f0Vj(dH ) r 2
o>
This G ravity Model says th a t movement between two centers ( i ) and ( j )
is d ir e c tly proportional to the product of th e ir populations and
in versely proportional to the square of the distance separating them.
-2 1 -
G ravity models locate people in proportion to some e x is tin g population
d is tr ib u tio n , or to some preconceived d is trib u tio n .
Thus, distance and
population sizes are important variab les in the g ra v ity form ulations.
But g ra v ity models do not specify the behavorial content associated with
it s determinants.
When g ra v ity variab les are used, they are usually
surrogates fo r the flow and cost of information to market p a rtic ip a n ts ,
or fo r the lik e lih o o d o f obtaining a given employment or wage.
The abundance of net m igration data has spawned the group of migra­
tio n models characterized as push or pull models.
Gross migration data
can be reported as the sum of a ll gross inm igration (outm igration) to
a sing le place from (to ) a ll other places.
Total outm igration can be
expressed only as a function o f economic conditions a t the place of
o r ig in , hence, the term "push" model, such th a t
n
I Mij- = f (Pi ,Ui ) ,
j = I ............n.
(2)
M igration from i is a function of it s population a t r is k , (Pi ) , and it s
local economic conditions, (Ui ) .
Total inm igration to place i is
expressed as
n
I Mj . = T(P11U1) ,
j = I,
...,
n;
(3)
th a t is , population s ize and local economic conditions a t place i .
These p a rtia l models postulate (a) th a t the same economic conditions a t
-2 2 -
place i "p u ll" as well as "push" people.
When used w ith net m igration
(the ty p ic a l use) the models become
I
- Z (Mj-V = f ( P v ui > ’
j = 1......... n.
(4)
Net m igration is the d iffe re n c e between gross inm igration and outmigra­
tio n , and is reported fo r a s p e c ific place.
No flows from point to
point can be is o lated from th is form ulation.
truncates the m igration flow s.
This model a r b it r a r ily
Not only are the determinants lim ite d to
the p a rtic u la r study area but the separate flows are masked by aggrega­
tio n .
Recent census data are reported as flows of people from one point
to another.
These data allow economic conditions a t both o rig in and
d estinatio n to be included as determinants o f separate m igration flows.
One of the f i r s t o f these models was developed by Lowry (1966).
Data
were fo r flows between selected SMSA1s in the United States.
Isard (1960) and others have modified the basic in te ra c tio n model
to incorporate the r e la tiv e a ttra ctiven ess between two lo c a lit ie s .
Conditions in o rig in s and destinations in an attempt to trace the flows
o f people from less a ttr a c tiv e to more a ttr a c tiv e areas.
These models
are usually specified in the follow ing manner:
f f D1 j 'Ui >•
J
I I •••I n
(5)
-2 3 -
where m igration is d ir e c tly proportional to the conditions of a ttr a c t iv e ­
ness in the destin atio n (Uj.) and in versely proportional to the conditions
of a ttractiven ess in the o rig in (Ui ) .
But the fundamental problem with
such form ulations is th a t they are p a r tia l models.
People are assumed
to be absorbed in to the sending area without any e ffe c t upon the economic
system (Alonso 1974).
S im ila rly , when people leave an area no impact
from a decreased labor supply is a part of the model.
Recent research has attempted to conceptualize m igration as a part
o f the economic system o f an area.
The approach requires a simultaneous
determination o f m igration , wages, and other possible v a ria b le s .
The
remainder o f the chapter w ill review in greater d e ta il th is research and
it s im plications fo r th is study of m igration to and from the Northern
Great Plains and other locations throughout the United States.
M igration as Simultaneously Determined With
Local Economic Conditions
In order to understand f u l ly the ra tio n a le underlying m igration as
an in teg ra l part o f a system, the remainder of fh is chapter describes
and analyzes several models which emplpy simultaneous estim ation tech­
niques.
Knowledge may be gained as to how migration is in te rre la te d
with p a rtic u la r economic conditions in a given area.
With the exception
of Petto (1972), these models are applied to r e la t iv e ly large urban
areas.
Also, with the exception o f Greenwood (1 9 75 ), these models use
net m igration ra th e r than gross outm igration and gross inm igration as
-24-
the dependent v a ria b le .
The discussion ou tlines the endogenous variab les
included in each model and reasons fo r th e ir in clusion .
Income Growth and M ig ra tio n .
Petto (1972) formulated a model consisting of net m igration, income
growth, and labor force p a rtic ip a tio n functions.
His main ob jective was
to obtain evidence th a t m igration influences the growth o f per capita
income in the Ozarks.
He also found th a t among c e rta in age groups per
capita income also was s ig n ific a n t in explaining net m igration flows.
S p e c ifica tio n of the per capita income growth function presupposes
th a t m igration might e ith e r increase of decrease per capita income in a
p a rtic u la r s it e .
I f net outm igration is s e le c tiv e o f younger or more
educated workers, then the people remaining a t the s ite are the less
productive members o f the labor force and m igration may impede the
growth of per capita income.
A large proportion of the net outmigration
flows may very well be constituted by these c h a ra c te ris tic s .
However,
i t can also be argued th a t net outm igration also tends to reduce the
supply o f labor services, thereby exerting an upward pressure on wage
le v e ls and per capita income.
I f th is increase in per capita income
re fle c ts favorable economic opportunity, then there also may be a
reduction in net outm igration, hence a c irc u la r interdependence.
Greenwood (1975) also posits th a t, a p r i o r i , one cannot th e o re ti­
c a lly determine the d ire c tio n a l e ffe c ts o f m igration on income changes
-2 5 -
in an area.
He states th a t m igration a ffe c ts both the supply o f and
demand fo r la b o r.
The e ffe c ts on labor supply are obvious.
However,
he continues by arguing th a t increased inm igration w ill increase labor
demand as w e ll.
Inm igration may have a p o sitiv e in flu e n c e , because of
the m u ltip lie r e f f e c t , on lo c a lly produced goods and services, thus
increasing income.
Also i f inm igration induces investment in a lo c a lit y ,
then la b o r's marginal product may be expected to increase in the
receiving community.
Therefore, income may have e ith e r a p o sitiv e or
negative re la tio n s h ip with m igration depending on the magnitudes of
the s h ifts in supply and demand.
Greenwood's em pirical re su lts found
th a t, a t le a s t fo r very large m etropolitan areas, m igration does not ,
appear to have appreciably a ffected income growth.
Okun (1968) developed two simultaneous equation models including
income growth functions expressed as service income per c a p ita .
He
found th a t in both models net m igration was s ig n ific a n tly and p o s itiv e ly
re la te d to the absolute growth in service per capita income.
His
findings conclude th a t in te rs ta te m igration in the 1940's tended to
widen the in e q u a lity in the d is trib u tio n o f service income per capita
among the sta te s .
Conversely, m igration may be caused by income changes.
Petto (1972)
included per capita income growth in the m igration function as a
measure of income expectations.
Thus as economic opportunities
increased in an area, net outmigration should be reduced; however,
-2 6 -
income change over the period o f m igration was not s t a t is t ic a lly s ig n i- '
fic a n t.
He postulates th a t perhaps income growth was the re s u lt ra th e r
than a cause o f m igration.
Income increases may not have been a good
in d ic a to r o f economic opportunity e s p e cia lly i f large increases in
income were in counties w ith i n i t i a l low le vels of income where net
outm igration was high.
Greenwood (1975) also adds th a t, inasmuch as
income changes should in d ic a te economic expectations,
given the curren t income le v e l, the greater the expected
fu tu re increase in income, fo r which the current actual
increase may be a good proxy, the poorer is the current
level as a proxy fo r the discounted value o f the stream
o f expected fu tu re returns (p. 7 ).
Of the 12 gross inm igration and outm igration equations estimated by
Greenwood the c o e ffic ie n ts fo r income change was s ig n ific a n t only tw ice.
Apparently migrants moved to areas where th e ir expectations were' the
highest which may be d iffe r e n t from the areas c u rre n tly experiencing
high income growth ra te s , or expectations were not met because the
aggregate m igration response was so great th a t an increased labor supply
depressed wages.
Results obtained by Okun (1968) also in dicate th a t the absolute
change in service income per capita is not a good measure o f economic
opportunity.
He concluded th a t migrants were a ttra c te d to states with
r e la t iv e ly high le v e ls o f per capita income, and th a t these states also
had concurrent rapid increases in service per capita income.
Outmigra­
tio n from states in the 19401s tended to be from states w ith low
-2 7 -
absolute income le v e ls .
But these states tended to have large increases
in income so th a t the income increase appeared to have been.caused by a
decreased labor supply.
Labor Force P a rtic ip a tio n and M ig ra tio n .
P e tto '(1 97 2 ) considered the possible in te rre la tio n s h ip between
m igration and labor force p a rtic ip a tio n ra te s .
I f m igration is a func­
tio n of the labor force p a rtic ip a tio n r a te , then those counties experi­
encing rapid declines in p a rtic ip a tio n rates (which hopefully measures
employment op po rtu n ities) should, have greater net outm igration.
The
reason is th a t when job opportunities are not a v a ila b le workers do not
expect to gain employment and they then become discouraged and eventually
may drop from the labor fo rc e .
But Petto abstracted from inm igration
because his data were net m igration.
He th erefo re could not determine
the e ffe c t o f inm igration on p a rtic ip a tio n rates which in turn might
a ffe c t outm igration.
(
Okun (1968) treated the change in a g ric u ltu ra l labor p a rtic ip a tio n
rates (percent o f the labor force in a g ric u ltu re in 1940 subtracted
from the p a rtic ip a tio n ra te in 1950) as exogenous in his i n i t i a l model,
but as endogenous in his second model.
In both cases, the sign on the
c o e ffic ie n t was negative and s ig n ific a n t.
This means th a t as the labor
p a rtic ip a tio n rates in a g ric u ltu re decrease, net m igration to th a t area
increases.
A possible explanation o f th is is th a t states may experience
-2 8 -
declines in the share o f the labor force engaged in a g ric u ltu re but not
in the a g ric u ltu ra l service sector.
I f th is is tru e , then labor demand
may increase in the farm service secto r, depending on to ta l a g ric u ltu ra l
output and the s u b s titu tio n o f purchased inputs fo r la b o r.
Thus, net
inm igration may accompany a decline in the r e la tiv e labor contribution
to farming output.
Income and Labor P a rtic ip a tio n Rates.
Changes in per capita income have been postulated to have an e ffe c t
on labor force p a rtic ip a tio n ra te s .
The reason fo r th is is th a t inas­
much as "measures in per capita income r e fle c t b e tte r job op po rtu n ities,
then areas having r e la t iv e ly greater increases in per capita income
should have sm aller declines in p a rtic ip a tio n rates" (Petto 1972, p. 1 4 ).
However, in the Ozarks i t was observed th a t as incomes increase a t a
rapid r a te , labor p a rtic ip a tio n declined.
Possibly as incomes increased,
the underemployed (such as housewives and high school students) dropped
from the labor force as the counties became more a fflu e n t.
Okun (1 968) treated the percentage of people employed in a g ri­
c u ltu re as endogenous but solely determined by predetermined v a riab le s .
He found th a t in the 19 4 0 's ,- states with comparatively high levels o f
service income per capita had a f a i r l y small decline in the percentage
o f t h e ir labor force in a g ric u ltu re .
Okun also explored the hypothesis th a t numerically la rg e decreases
in the percentage o f the labor force in a g ric u ltu re would tend to
-2 9 -
induce large increases in service income per c a p ita .
The reason fo r
th is is th a t, per se, there may be a b e tte r a llo c a tio n o f resources and
a large increase in per capita service income as the percentage of the
labor force in a g ric u ltu re declines.
This is based on the assumption
th a t service income per worker was lower in the a g ric u ltu ra l sector
than in. the nonagric u ltu ra l sector during the 1940's.
,
Also i f nonparticipants do not contribute to the area's economic
a c t iv it y , then a decline in the number o f people working in an area
would tend to prolong poverty.
This.hypothesis has been substantiated
by both Petto (1 972) and Okun (1968).
Outmigration and In m ig ratio n .
' Greenwood (1 975) disaggregated outm igration to flows from one SMSA
to another, and to nonmetropolitan areas.
in a s im ila r fashion.
Inm igration was disaggregated
I t was hypothesized th a t not only is the base
c iv ilia n labor force in m etropolitan areas d iffe r e n t from those in non­
m etropolitan areas, but also i t is lik e ly th a t the various factors
influencing the m igration streams may have d iffe r in g impacts depending,
on orig ins and d e s tin a tio n s .
A d d itio n a lly , each m igration equation
specified the jo in t dependency between outmigration and inm igration.
The underlying reason is th a t recent migrants may be more m igration
prone than the indigenous population.
Therefore, areas having a large
proportion o f recent migrants may tend to have high m igration flows in
the opposite d ire c tio n ( M ille r 1967).
Greenwood's re s u lts seem to
-3 0 -
j u s tify th is s p e c ific a tio n in a ll o f the 12 m igration equations
estimated.
At f i r s t glance one would expect th a t those who have migrated fo r
the f i r s t time would not b e n e fit g re a tly by a second move because the
f i r s t move should be considered the best a lte rn a tiv e among a ll other
choices.
Migrants are not lik e ly to reside in low income areas a fte r
th e ir i n i t i a l move and any subsequent moves would not o rd in a rily gain
the migrant large incremental b e n e fits .
Under these circumstances the
two m igration streams may be strongly dependent on one another only i f
the migrant is a temporary mover, the extreme case being a seasonal
migrant worker.
Studies by Goldstein (1973) and M i l l e r . (1 967) concluded th a t in
fa c t a disproportionate p art of the migrant flow from 1955-1960 had
previously moved e ith e r p rio r to the period or during the 1955-1960
period.
M ille r (1 973) argues th a t those who have moved a t le a s t once
show a w illingness and a b il it y to move.
A d d itio n a lly a prospective
migrant loses any investment in human c a p it a l, in terms o f knowledge
o f his place of o rig in once he moves from an area.
little
Since he has very
investment in the area of destin atio n he may s t i l l
propensity to m igrate again.
have a high
F in a lly , a recent migrant has a greater
a v a ila b ilit y of inform ation from his friends and re la tiv e s in his
place o f o rig in .
This reduces his costs and increases his benefits
i f the decision is made to re tu rn .
-3 1 -
Employment Changes and M ig ra tio n .
In several studies, employment changes are used as an in d icatio n of
labor demand and job opportunities in an area.
expected s h ifts in labor demand.
I t could also denote
Therefore, outmigration may be lower
and inm igration higher the greater the employment change.
Greenwood
(1975) disaggregates the change in employment into changes in manufac­
tu rin g , other nonmanufacturing, and government.
However, th e ir aggre­
gate change is the v a ria b le jo in t ly determined with m igration.
The
reason th a t to ta l change in employment is disaggregated is th a t manu­
fa c tu rin g is considered to be the economic base Of a m etropolitan area
and is exogenously determined, and nonmanufacturing employment is con­
sidered endogenous.
Basic in dustries are influenced by events outside
the area and nonbasic in dustries area function o f local conditions
including changes in the basic in dustry.
A ll three are influenced by
gross inm igration and outmigration by the extent of the re su lta n t
s h ifts in the supply and demand of la b o r.
This argument is supported
by Borts and Stein (1964) who postulate th a t through induced investment,
expansion in job opportunities w ill re s u lt from high rates o f inmigra­
tio n .
Muth (1971) also concludes th a t m igration not only a ffe c ts but is
a ffected by employment growth.
I f the assumption is made th a t both the
i
supply and demand schedules have f i n i t e e la s t ic it ie s , the amount of
m igration into an area w ill be greater than t:he increase in employment.
-3 2 -
I f the labor supply were in f in i t e l y e la s tic as Borts and S tein (1964)
%
suggest, then employment would increase by the same fra c tio n a l amount
as m igration in absolute terms.
Under th is assumption m igration could
not explain any changes in wages or per capita income.
This is supported
by Mazek (1966) who argues th a t money wages are completely r ig id and
th a t an excess supply o f labor exists in a ll local labor markets. I f
,
th is is tru e , employment grows only when demand fo r labor increases.
■
Thus i f m igration indeed influences the demand fo r la b o r, then i t is
plau sible to include m igration as an explanatory v a ria b le a ffe c tin g ;
employment change.
The Muth model tre a ts m igration and to ta l c iv ilia n
employment as simultaneously determined.
His re s u lts tend to support
Mazek's model fo r small c it ie s , but found th a t fo r large m etropolitan
areas the e la s t ic it y of employment change with respect to migration is
e s s e n tia lly u n ity .
Due to the sim ultaneity between employment change
and m igration under the above assumptions, most areas experience
m u ltip le e ffe c ts as they expand or c o n tract.
Perhaps the best speci­
fic a tio n fo r testin g the sim ultaneity between employment growth and
m igration is to use the change in service oriented employment growth as
opposed to to ta l growth since one would not expect m igration to a ffe c t
d ir e c tly the growth, in basic in d u s trie s .
As mentioned above, Okun did
fin d th a t m igration s ig n ific a n tly influenced service income per c a p ita .
A lte rn a tiv e ly , increases in population through m igration may influence
service employment.
-3 3 -
Summary
In the past, there has been some success in tre a tin g m igration as
endogenous w ith in a system of behavioral re la tio n s h ip s .
For purposes
o f th is study, m igration and labor p a rtic ip a tio n rates are not jo in t ly
dependent because i t is assumed th a t these rates are a ffec te d by demand
conditions which are a cause o f m igration not caused by m igration.
Employment growth is considered exogenous again due to changes in condi
tions outside the study area.
Inm igration and outm igration are not
interdependent because the m igration data accounts fo r only successful
5-year movements from ah area to an area.
Those migrants who did not
remain in an area between 1965 and 1970 are not included in the migra­
tio n data.
The only simultaneous re la tio n s h ip postulated here is the
interdependency between m igration and the change in wage per employee
over the m igration period;
The ra tio n a le fo r th is s p e c ific a tio n w ill
be developed in the next chapter.
Gross outm igration and inm igration data are used ra th e r than net
m igration data.
I t is d i f f i c u l t to in te r p r e t e m p iric a lly the e ffec ts
of net m igration on economic conditions and vice versa.
Also, by
separating out the gross flows the e ffe c ts o f economic changes and of
\
separate m igration flows are c le a rly exposed.
In essence, the reasons
fo r receiving migrants in the Northern Great Plains may be d iffe r e n t
than fo r migrants leaving the Northern Great Plains.
CHAPTER I I I
METHODOLOGY
The major concerns o f th is chapter are the methodological approaches
used in studying Northern Great Plains m igration.
Therefore, follow ing
a b r ie f introduction to each methodological approach, the remainder of
th is chapter w ill discuss f i r s t the ra tio n a le fo r disaggregating the
data in to three d iffe r e n t models.
Second, variab les considered fo r the
m igration models w ill be introduced.
the endogenous and exogenous elements.
w ill be presented.
These variables w ill include both
T h ird , the technique o f analysis
F in a lly , an i n i t i a l s p e c ific a tio n o f the m igration
models w ill be completely defined.
A m igration study can analyze the re d is trib u tio n o f people in the
context o f e ith e r m o b ility facto rs a t one lo c a lity or in a framework
involving conditions th a t influence m igration in both the destination
and o rig in lo c a lit ie s .
The use of gross migration data from each o rig in
lo c a lit y to a ll destinations allows conditions a t both lo c a lit ie s to be
measured.
Furthermore, gross m igration flows can be separated to take
in to account movement in opposite d ire c tio n s .
Therefore, th is study
u t iliz e s gross m igration flows due to the above considerations.
w ill be discussed below, three models are used:
As
(a) ih t r a regional
I
m igration, (b) in terre g io n a l in m igration, and (c) in te rre g io n a l out­
m igration.
The variab les th a t are hypothesized to influence m igration f a l l
into three separate groups:
(a) control v a ria b le s , (b) economic condi­
-3 5 -
tions a t the o r ig in , and (c) economic conditions a t the d e s tin a tio n .
The variab les in each group were suggested by a review o f previous
studies and through the a p p licatio n o f economic theory.
The data were
i n i t i a l l y compiled on the basis th a t each v a riab le was a measure of
an unique s ite condition.
S pecific procedures w ill be implemented
la t e r in the study as an in d ic a tio n o f the actual uniqueness o f each
v a ria b le .
The m igration data are disaggregated into the three streams
id e n tifie d above.
The disaggregation allows the estimated c o e ffic ie n ts
as well as thfe variab les included to be d iffe r e n t fo r each stream.
The
basic reason fo r considering three m igration models and fo r expecting
differences among them is the assumption th a t amenities are valued
d iffe r e n t ly .
I f s ite amenities (whether scenery or social tie s ) are
important to people, then a d is ta n t move may require greater fin a n c ia l
rewards than moves w ith in the region.
This implies th a t in traregion al
and in terre g io n a l flows of people should be analyzed separately.
Further
more i f the c h a ra c te ris tic s o f people in in terre g io n a l flows (th a t is ,
d is ta n t m igrants) are d iffe r e n t fo r inmigrants and outmigrants, then
these flows should be analyzed separately.
More inform ation can then
be gained about the behavioral c o e ffic ie n ts by separating the two migra­
tio n flow s.
F in a lly , an o b je ctiv e of the study is to take account of the in te r ­
relatedness o f events associated with m igration.
The general approach
-3 6 -
required is a simultaneous estim ation of the parameters associated with
the endogenous and exogenous variables in a system o f equations.
The
endogenous variab les are those in te rre la te d variables whose values are
explained by the model.
The exogenous variables are those independently
determined variables th a t contribute to providing an explanation of
the s ize o f m igration flows.
Disaggregation of the M igration Data
The m igration data are separated into two sets on the basis of
distance of move.
Moves w ithin the region are considered to be o f a
r e la t iv e ly short distance and are between State Economic Areas.
In te r ­
regional moves are from (to ) S tate Economic Areas w ith in the region to
(from) states which are not contiguous to the region.
The data on
in terre g io n a l m igration flows are segmented fu rth e r into inm igration
and outm igration observations.
In tra re g io n a l Model.
The in t r a regional model is an analysis of short distance m igration
streams where i t is presumed th a t s ite amenities are not given up a t
the o rig in as people move to a nearby place j ; th a t is , the cost o f the
move is low because distance is short and because amenity values are
retain ed .
The peripheral boundary of th is m igration data set is p o litic a l
and geographic in nature.
The m igratory f ie ld is delim ited by North
z
-37-
and South Dakota, and the portions of Montana, Wyoming, and Colorado
east of the Continental D ivide.
The data subset is fu rth e r defined as '
the number o f people (older than four years o f age) moving from a State
Economic Area in the region to another S tate Economic Area in the region
w ith in the same s ta te or an adjacent s ta te in the region between 19651970.
Variables measuring the local economic conditions described e ls e ­
where necessarily r e la te to the o rig in and destinatio n S tate Economic
Areas.
In s h o rt, th is model defines the short distance migrant as a
Northern Great Plains resident whose move has been w ith in the region.
In terre g io n a l Models.
The in terre g io n a l m igration streams consist of inm igration into
the region and outm igration to d is ta n t lo catio n s.
The observations
include those not included in the in tra re g io n a l s e t, except th a t moves
to (from) adjacent states were omitted (see Figure I - I ) .
Migrants in
these streams are presumed to have high moving costs because of the
physical distance and the assumption th a t o rig in amenities are lo s t in
the move.
Furthermore, these streams are presumed to be s e le c tiv e ; th a t
is , the c h a ra c te ris tic s and behaviors o f the migrants are d iffe r e n t.
I f so, then the c o e ffic ie n ts associated with the variab les measuring
economic conditions w ill be d iffe r e n t.
The in terre g io n a l models hypothesize th a t wage d if fe r e n t ia ls between
o rig in and destin atio n are necessary in order to compensate people fo r
moving.
Large proportionate changes in wages a t the margin are
-38-
ex pec ted to be needed in order to draw people into the region.
Further­
more, the in te rre la tio n s h ip between wages and m igration (p a rt of which
is the supply of labo r) is expected to be d iffe r e n t from th a t in the
in tra re g io n a l model.
This is because the number o f people in each of
the in terreg io n al m igration streams are small in comparison to the
number o f people moving between S tate Economic Areas w ith in the Northern
Great P lain s.
(However, cumulative in terre g io n a l m igration may be
la rg e r fo r any one s i t e . )
In terre g io n a l In m igration. — In terreg io n al inm igration is defined
as the number of people moving from each of 39 contiguous states outside
the Northern Great Plains (and adjacent areas) to a S tate Economic Area
w ith in the Northern Great P lain s.
streams.
For each Plains SEA, there are 39
The observations are those people (older than four years of
age) liv in g in one o f 39 o rig in states in 1965 but in a Plains SEA in
1970.
In terre g io n a l Outm igration.--T h e in terreg io n al outmigration stream
is the opposite o f the stream mentioned above.
Therefore, the observa­
tions are those, people (plder than fo u r years of age) liv in g in an
o rig in SEA in the Plains in 1965 but in one of the 39 states outside the
Plains in 1970.
Depending on the in terre g io n a l m igration flow being
observed, the level o f aggregation a t o rig in and d e s tin a tio n is e ith e r
a s ta te or an SEA.
For the in tra re g io n a l model the level o f aggregation
a t both the o rig in and d estinatio n are SEA's.
-39Variables Influencing Migration
The m igration models in th is study are b a s ic a lly a modified in t e r ­
action model taking the general functional form:
SE.
SET ’
where:
3
(DISTi k )(RTSi k )(RTSj 1 )
(M E i )(FARi )
RU.
NI.. - is the observed number of people who moved from o rig in
1J
i to destin atio n j ;
Uj - are measures o f the attra c tiv e n e s s of the Jt *1 d e s tin atio n ;
Ui - are measures of the a ttractiven ess o f the i**1 o rig in ;
SEi - measures the unattractiveness o f the i**1 o rig in ;
SEj - measures the unattractiveness o f the j* * 1 d e s tin atio n ;
DIST-- and RTSik - measures i ' s r e la tiv e importance and
J
position in economic space;
RTSi j - measures i ' s r e la tiv e importance to destinatio n j ;
AGE- - is a c h a ra c te ris tic measure o f the p o ten tial fo r
1
m igration in i
PM. - measures the inform ation flow a v a ila b le a t i from j ;
J
'
Since th is functional form is m u ltip lic a tiv e the data are tran s­
formed in to common logs. This double log form ulation assumes th a t the
e rro r term is specified as TOe . Under th is s p e c ific a tio n the r e la tio n ­
ship between m igration and it s determinants are lin e a r .
^Associated w ith each v a ria b le is a regression c o e ffic ie n t. Only
the s p e c ific re la tio n s h ip s between the m igration flow and it s determi­
nants are hypothesized in th is functional form.
-40-
FAR. - is a c h a ra c te ris tic measure o f the p o ten tial fo r
m igration in i ;
RU- - is a c h a ra c te ris tic measure o f the p o ten tial fo r
m igration in i ;
P1- - measures the population a t r is k in i ; and
D IS jj
-
is the distance between i and j .
This general functional form describes the organization o f the
independent variab les in the m igration equations.
Variables measuring
local economic conditions are those postulated as measures o f a ttr a c tiv e
L
ness or unattractiveness o f a place.
th is study.
These are of primary in te re s t in
A ll other variab les are control v a ria b le s .
This c la s s if i­
cation system w ill be used throughout the remainder of the study fo r
presentation o f concepts and re s u lts .
The functional form also implies an expected sign fo r the c o e f f i­
c ie n ts .
Consider the m igration stream from place i to place j .
The
conditions which a t t r a c t people to a place should have a negative sign
fo r these conditions a t the o rig in and a p o sitiv e sign fo r these condi­
tions a t the d e s tin a tio n .
Thus, a la rg e change in wages a t i would tend
to reduce the stream from i to j , and increase the stream from j to i .
Those economic conditions tending to be u n a ttra c tiv e to labor m o b ility
would e x h ib it signs which are the opposite.
The functional form also implies signs fo r the control v a riab le s .
Thus, distance from one place to another, ( D j j ) , is expected to be
(
negatively associated with the stream of migrants from i to j .
Age
-41also is expected to be negatively re la te d to movement from point i
(although i t is not expected to influence the selection o f point j ) .
The precise d e fin itio n s of the variab les to be considered in each
model are presented in Table I I I - I . These are organized in the manner
in d icated , and are referenced throughout the remaining sections of the
study.
Local Economic Conditions.
Migration is a temporal geographic a llo c a tio n of people.
Thus,
migrants are expected to make decisions based upon expectations of
changes as well as to actual conditions over the period.
Four variables
are hypothesized to r e fle c t local ecqnomic conditions a t the o rig in and
d estinatio n in each model:
(a) beginning period wage (Wgg), (b) change
in wages over the m igration period (CW), (c) change in nonagricultural
employment (CEMP), and (d) a measure o f subemployment (SE).
Each o f the
f i r s t three is expected to be negatively re la ted to outm igration, where
the variab les are measures of conditions a t the o rig in .
Therefore,
high wages a t the beginning of the period should retard outm igration.
The measures of the same economic conditions a t the destin atio n imply a
p o sitiv e sign.
Subemployment in an area is expected to promote out­
m igration from the area and decrease inm igration.
\
-42table
III-I.
VARIABLE DESCRIPTIONS AND VARIABLES
V a ria b le D e s c rip tio n
V a ria b le
M ig ra tio n V a r ia b le s :
Number o f m ig ra n ts from an S E A (I) to an S E A (j) between
1965-1970.
'
'
M ..
Number o f o u tm ig ra n ts fro m an S E A (i) to a s ta te ( j ) ,
1965-1970.
OM..
Number o f in m ig ra n ts fro m a s t a t e ( j )
between 1965-1970.
IMi J
to an S E A (i)
C o n tro l V a r ia b le s :
P o p u la tio n :
Number o f p e op le r e s id in g in an S E A (i), 1970.
D is ta n c e : Road m ile a g e between th e la r g e s t tra d e c e n te r
in an o r i g in SEA to th e la r g e s t tra d e c e n te r in a d e s tin a ­
t io n SEA. For lo n g d is ta n c e moves, th e m ile a g e is between
s t a t e c a p it a ls .
D is ta n c e : Road m ile a g e between th e la r g e s t town in an
S E A (i) to th e la r g e s t town in any one o f th e a d ja c e n t
S EA's, S E A (k).
/
POP^
DIS j J
DISTi l ,
R ural P o p u la tio n : Number o f pe op le c la s s if ie d as r u r a l
in an S E A (i), 1970.
RUi
Farm W orkers: Number o f people 16 y e a rs o f age and o ld e r
employed on farm s in an S E A (i), 1970.
FAR.
E d u c a tio n : Number o f pe op le having 16 o r more ye a rs o f
s c h o o lin g in an S E A (i), 1970.
EdUi
P ast M ig r a tio n : Number o f pe op le born in one s t a t e , b u t
l i v i n g in a n o th e r s ta te in 1970.
PM.
Town S iz e : R a tio o f th e p o p u la tio n o f th e la r g e s t town
in an SEA o r a s ta te to th e p o p u la tio n o f th e la r g e s t
town in an S E A (i), 1970.
Town S iz e : R a tio o f th e p o p u la tio n o f th e la r g e s t town
in any one o f th e a d ja c e n t (k ) SEA's a d ja c e n t to an
S E A (i), to th e p o p u la tio n o f th e la r g e s t town in th e
S E A (i).
•
*
RTSi i
RTSk i
Economic C o n d itio n s a t O rig in s and D e s tin a tio n s ; 1965,
1965-1970, 1976:
Employment Change: Change in q u a r t e r ly n o n a g r ic u ltu r a l
employment in an SEA,or a s t a t e , 19 6 5 -1 9 7 0 .a
CEMP- i
’
Wage Change: Change in q u a r t e r ly p e r employee s a la r ie d
income in an SfA o r a s t a t e , 1965-1970.
CWi .
J’
I n i t i a l Wage: B e g in n in g o f th e p e r io d , q u a r t e r ly p e r
employee s a la r ie d income in an SEA o r a s t a t e , 1965.
W ,-
Subemployment: P e rce n t o f p o s s ib le weeks n o t worked
in an SEA o r a s t a t e , 1970.
SEi •
J’
t J, i
aSee County B usiness P a tte rn s f o r d e f i n i t i o n o f employment. The
sam pling acco un ts f o r 66 p e rc e n t o f th e c i v i l i a n la b o r fo r c e .
-43-
The change in per employee wage ,(CW1-) is considered to be endogenous
in any p a rtic u la r S E A (i).5
This v a ria b le is calculated by dividing
to ta l wage and salary revenue from nonagric u ltu ra l employment in a
given year by the to ta l number o f .nonagricultural employees in the year.
Wages are expected to influence m igration through labor demand.
Labor
tends to migrate to areas where the present value o f th e ir fu tu re
expected income streams are the g re a te s t.
As an SEA(i) experiences a
sharp increase in demand, the wage increases.
M igration from th is SEA
to other lo c a lit ie s decreases since the p o ten tial migrants in th is area
expect to gain more personal b e n e fit by remaining where they are than
by moving elsewhere.
The greater th is change in the wage ra te the
greater is the attra c tiv e n e s s of the area.
At the same tim e, the
migrant from outside the area w ill consider relocating in areas experi­
encing rapid wage increases.
The change in nonagricultural employment (CEMP) is a measure of
job opportunity in lo c a lit ie s .
Where jobs are expanding i t is assumed
th a t job vacancies are also expanding and are being f i l l e d by outside
la b o r, as well as by e x is tin g local la b o r.
5
Changes in demand fo r labor
The primary means by which an area increases or decreases its
labor supply is through labor m o b ility . I f there are large amounts o f
m igration from an SEA(i). over a period, then the wage ra te is expected
to increase ( c e te ris paribus) . An increase in inm igration (IMj1
O w ill
make a v a ila b le to an area a greater number of people w illin g to supply
lab o r. This increase in labor supply w ill tend to decrease wages.
-4 4 -
in destinatio n lo c a lit ie s are due to exogenous changes in demand fo r
products produced in these lo c a lit ie s .
Therefore, any community
experiencing an expansion in job opportunities w ill have a v a ila b le
fo r employment people who otherwise would have searched fo r jobs in
other lo c a lit ie s , and people who fin d th a t th is lo c a lit y provides fo r
them benefits th a t they could not obtain elsewhere.
The i n i t i a l wage in o rig in and d estinatio n areas, (W65) , is also
included as an economic condition v a ria b le .
This v a riab le controls
fo r the possible expectations of p o ten tial migrants in fu tu re changes.
I t is presumed th a t areas which have high i n i t i a l wage le v e ls a ttr a c t
people who believe th a t wage le v e ls w ill remain high.
I t is possible
but not probable th a t migrants may move to low wage areas in a n tic ip a ­
tio n o f large wage changes over the next 5 years.
Subemployment measures the u n d e s ira b ility of an SEA.
The precise
d e fin itio n of th is v a ria b le is the percent of weeks not worked by those
a c tiv e ly in the labor fo r c e .6
I t is calculated by subtracting the
actual number of weeks worked in an SEA (s ta te ) by the labor force from
the to ta l possible weeks th a t could be worked to obtain the number of
weeks not worked.
This fig u re is then divided by the to ta l number of
possible weeks th a t could be worked in 1970.
The g re a ter the percentage
6This v a ria b le includes, a ls o , the unemployed; these people are
in the labor force because they were a c tiv e ly looking fo r jobs.
-45-
of weeks not worked by the labor force in a given SEA ( s t a t e ) , the
g reater is the p ro b a b ility o f in divid u als becoming discontent with an
area and moving elsewhere.
However, they are not li k e l y to choose a
d estinatio n th a t also has a r e la t iv e ly high subemployment ra te .
Control V a ria b le s .
The control variab les are c h a ra c te ris tic measures o f Northern
Great Plains SEA's.
Furthermore, only a c e rta in combination of these
control variab les are hypothesized to influence a given m igration stream
Presented in th is section are ju s tific a tio n s fo r these Plains control
variab les as possible influences on Northern Great Plains m igration.
The v a ria b le (RTS^ ) is the r a tio of the population o f the la rg e s t
town in any one SEA adjacent to an S E A (i), to the population of the
la rg e s t town in a sending SEA(i) fo r the in trareg io n al and in terregion al
outm igration models.
For the in terre g io n a l inm igration model the SEA(i)
is denoted as a receiving SEA.^
Also included is the road mileage
between these two trading centers (DIST1-
.
The town r a tio is expected
to be negative in each model, while the distance is hypothesized to be
p o s itiv e .
As the town population in an adjoining SEA is la r g e .r e la tiv e to
the la rg e s t town in a sending SEA, in te ra c tio n between the o rig in SEA
and possible destinatio n SEA's (s ta te s ) may decrease.
A d d itio n a lly ,
^These towns are g enerally designated as Rand McNally trade centers
-46-
the remoteness o f a lo c a lit y increases as the distance between these
two towns is g re a t, thus increasing the p ro b a b ility o f in te ra c tio n
between an o rig in SEA and possible destinatio n areas other than large
adjacent SEA's.
The reason fo r the inclusion o f (R T S ^ ) in the inm igration model
is th a t as the town population in any adjacent SEA is greater r e la tiv e
to the la rg e s t town in an S E A (i), the adjoining SEA may compete b e tter
fo r inmigrants because through growth i t may have a la rg e r number of
job opportunities a v a ila b le .
Furthermore, given a fe a s ib le distance
between these two towns, people may be w illin g to relo cate in the adja­
cent town and commute to th e ir p a rtic u la r town of employment which may
be located in another SEA.
That is , these two variab les are hypothesized
to influence m igration because o f the importance of s p a tia l economic
influences in regional economics.
The hypothesis assumes th a t nearby
towns have some impact on how another town appears to po ten tial
migrants.
This impact is the re s u lt of spheres o f influence in economic
space theorized in regional economics.
The n e x t'p o s itiv e ly re la te d v a ria b le to in t r a regional and in t e r ­
regional outm igration is the ra tio o f the la rg e s t town in the destina­
tio n s ite to the largest-tow n in the o rig in s ite (RTSj1- ) .
This v a riab le
denotes the r e la tiv e importance of an SEA(i) to any destinatio n ( j ) .
As the town size in an SEA(j) increases, it s importance in terms of
opportunity increases r e la tiv e to the opportunity in S E A (i).
Also the
-4 7 4*
ra tio o f the c it y size in the j
•
destin atio n s ta te to the o rig in SEA(i)
is expected to be negatively re la te d to inm igration since migrants from
s ta te ( j ) are lik e ly to remain there i f there are good job opportuni­
tie s .
I t is assumed th a t town sizes are surrogates fo r opportunity and
amenities in a c e rta in area.
Thus, la rg e r towns have fa s te r growing
labor forces and probably la rg e r absolute numbers o f people employed.
A d d itio n a lly , the la rg e r the town the greater the p ro b a b ility th a t th is
town w ill be able to provide amenities not obtainable in sm aller towns.
Median age (AGE1- ) in an SEA is also a v a ria b le included in the
m igration equation fo r movements from i to j .
th is flow is expected to be negative.
It s re la tio n s h ip with
People in those lo c a lit ie s with
high median age populations are considered to have r e la t iv e ly more time
and c a p ita l invested in th e ir p a rtic u la r lo c a lit ie s .
The incentive to
move would then have to be greater fo r the long-tim e resident than fo r
those who have rece n tly entered the labor fo rce.
Hoover (1971, p. 175)
demonstrates by using census data from 1960 th a t rates of m igration by
sex by sin g le years o f age peaked between 18 years and 20 years of age
fo r both males and females.
The ra te o f m igration decreased to the age
o f 60 and then roughly leveled out.
The farm v a ria b le (FARi ) is also negatively re la te d to the m igration
flow from i to j .
This v a ria b le denotes th a t those liv in g on farms have
a personal preference fo r doing so and are not lik e ly to sever th e ir
roots, g enerally b u ilt up over a lif e t im e , given th a t a ll other condi­
-4 8 -
tions do not change.
As mentioned e a r lie r , (AGE1- ) and (FAR1- ) do not
influence the m igration flow from j to i (in m ig ra tio n ).
The ru ral population v a ria b le (RlK) is included in the in tr a regional model and the in terre g io n a l inm igration model.
Demographers
have stressed th a t h is to r ic a lly (u n til possibly the recent decade) the
predominate m igration flow is rural to urban (Hoover 1971), (Greenwood
1974), (Isard= 1975), ( Balstad 1974).
Lewis (1974) found th a t out-
migrants from the Northern Great Plains between 1965-1970 had the
tendency to move to more urbanized areas, and was th e re fo re p o s itiv e ly
re la te d to outm igration.
However, (RU1- ) is a proxy fo r social tie s in
the region and th erefo re may tend to impede m igration out o f a Plains
s it e .
Furthermore, inmigrants from other states may not fin d i t
b e n e fic ial to relo cate in less populated areas.
Therefore, (RU.) is
negatively re la te d to both the in t r a regional flow and the in terregion al
inm igration flow .
Migrants have h is to r ic a lly been observed to fo llo w established
m igration flows.
Therefore, the measure of th is p rio r m igration flow
(PM.- -) is p o s itiv e ly re la te d to each m igration stream.
*
For in tr a -
90
regional and in terre g io n a l outm igration, (PMj) is the number o f people
who were born a t the o rig in i but reside in destination j in 1970.
For
the in terre g io n a l inm igration flo w , (PMj) is defined as the number of
people liv in g in a destinatio n in 1970 who were born in an o rig in area
-4 9 -
fo r th is m igration stream.
The la te n t migrant considering lo c a litie s
where old frien d s and re la tiv e s have already moved may fin d i t less
c o s tly to fo llo w .
A d d itio n a lly , th is migrant may be compelled to fo llo w
p rio r m igration streams because g re a ter knowledge and more information
is provided him through d ire c t contact with former m igrants.
M ille r
(1973) also stresses th a t, when using regression analysis to explain
m igration , a v a ria b le should be entered into the equation to control fo r
the e ffe c t of differences in the propensity to m igrate so th a t the
economic variab les are not biased.
Distance (DIS1-J-) is calculated as the road mileage from the la rg e s t
town in the i tfl SEA to the la rg e s t town in the Jtfl SEA fo r the in tr a regional model.
Vinning (1955) and H uff (1960) have stressed th a t the
m igrant's view of remoteness and distance may not be uniform throughout
the m igration f i e l d .
Nearby areas, in f a c t , may appear d iffe r e n t in~
economic, s o c ia l, and geographical aspects from one another.
lo c a litie s may appear to the migrant as more or less uniform.
D istant
Informa­
tio n and less uncertainty about an area increases as the distance between
possible a lte r n a tiv e destinations decrease.
A d d itio n a lly , the fa rth e r
a lo c a lit y is from a p o te n tia l migrant the greater the p ro b a b ility th a t
some intervening lo c a lit y w ill enter in to the m igrant's selection among
possible d e s tin a tio n s .
These intervening a lte rn a tiv e s may not necessarily
be a one-dimensional concept but could very well be c ir c u la r .
The
migrant may consider a ll possible lo c a lit ie s w ithin a c e rta in radius of
-50-
his o r ig in , thereby decreasing the p ro b a b ility of m igration to a d is ta n t
lo catio n even fu rth e r.
Therefore, fo r the in tra re g io n a l flo w , which is
a r e la t iv e ly short distance haul, distance is expected to be highly
s ig n ific a n t and negative.
Distance (DISij.) fo r the in terre g io n a l flows is calculated as the
road mileage from the c a p ita l c it y o f the s ta te in which the SEA in
question is located to the s ta te c a p ita l o f the destination s ta te .
Following the Vinning (1955) and Huff (I9 6 0 ) hypothesis, distance should
not fig u re as an important determinant of in terreg io n al flow s.
However,
i f distance is a measure o f costs and i f information is scarce over long
distances where these distances c o n s titu te many intervening opportuni­
t ie s , then distance would also be a fa c to r in explaining these flow s.
The re la tio n s h ip is also negative to these m igration streams.
Population (POPi ) , on the other hand, is expected to have a p o sitiv e
sign in a ll three models.
This v a ria b le is a measure of the population
a t ris k in i , fo r the in tra re g io n a l and in terreg io n al outmigration
models.
That is , the la rg e r the population in any o rig in ( i ) , the
greater the number of po ten tial migrants in th at area.
The variab le
is introduced into the in terreg io n al inm igration model to control fo r
size in destinatio n areas.
The la rg e r the population in any destination
(in th is case P la in 's SEA's) the g reater the p ro b a b ility th a t la rg er
numbers o f migrants w ill relo cate in th a t area from places outside the
-51-
region.
The next two sections il lu s t r a t e the models to be considered
in the remaining chapters o f the study.
The Simultaneous Aspects o f Migration
Regional impact analysis postulates th a t a demand fo r labor is a
function o f basic economic a c t iv it y .
A basic economic a c t iv it y is one
which responds to demand and supply conditions outside the region in
question; th a t is , i t is not influenced by consumption decisions w ithin
the region.
A derived demand fo r labor is generated fo r service a c t i v i ­
tie s associated w ith the new a c t iv it y and with the added population.
Thus, labor demand is determined outside the regional system.
Labor supply in the short run consists of the indigenous residents
and migrants.
As local economic conditions change, labor responds.
Therefore, given a new or old level o f exogenously determined demand,
the adjusted level o f wages w ill depend on the adjustment o f labor
supply.
in e la s tic
The aggregate supply o f labor in a community may be r e la tiv e ly
depending on the speed o f adjustment through changes in migra­
tio n or changes in labor force p a rtic ip a tio n .
For a single a c t iv it y or
occupation w ithin the labor market area, the supply is more e la s tic
because i t can be a ffec te d by not only m igration and labor p a rtic ip a tio n
rates but by the in te rc h a n g e a b ility o f occupations and s k ills among the
e x istin g labor fo rc e .
The labor supply of any one in divid ual employer
may closely approximate a p ric e -ta ke rs market the la rg e r the local labor
market.
U ltim a te ly , the average wage in any one community w ill change
-5 2 -
depending on the response o f the aggregate supply o f la b o r, which is
p rim a rily influenced through the d iffe r e n tia ls in m igration rates into
or from th a t lo c a lit y .
The a p p lic a tio n of the ordinary le a s t squares technique when the
system is simultaneous leads to what is re ferred to as a simultaneous
inconsistency bias.
This means th a t given a c e rta in level of the
endogenous v a ria b le which is included as an explanatory v a ria b le , there
w ill be predicted a c e rta in level on the dependent v a ria b le th a t w ill
not in turn produce the o rig in a l level given to the explanatory
g
v a ria b le .
To specify a method so th a t the estimates are consistent,
two-stage analysis must be incorporated.
Furthermore to incorporate the
consistent and most e f f ic ie n t method, three-stage analysis is .im p le ­
mented.
The two-stage le a s t squares technique involves f i r s t , the in d ire c t
estim ation of the endogenous v a ria b le s .
Second, the reduced form e s t i­
mates (estim ates determined by a ll predetermined v a ria b le s ) are used fo r
sing le le a s t squares estim ation o f each equation in the system.
In th is
study, the change in wages (CW1- ) is an endogenous element specified by
an equation which includes the le v e ls o f inm igration and outmigration
O
By correctin g fo r simultaneous inconsistency bias in a given a re a ,
i f weekly wages are $150, m igration might be 250 people. Simultaneously,
i f m igration is 250 then wages should be 150, i f a ll other influences
are constant.
-5 3 -
(m igration a ffe c ts the supply o f la b o r, hence wages,
IM j1-) -
Each m igration equation in turn contains the v a ria b le change in wages
(higher wages a t t r a c t and hold m ig ra n ts ).
The two-stage method allows
fo r th is sim ultaneity and th erefo re produces consistent estim ates.
This study also u t iliz e s three-stage analysis.
This method may
prove applicable i f the residuals across equations are c o rre la te d .
If
the residuals are c o rre la te d , three-stage analysis w ill produce e f f ic ie n t
estimates o f the c o e ffic ie n ts (estim ators having sm aller va rian c es ).
Improvement of the models by th is method is a p o s s ib ility dependent upon
the s p e c ific a tio n o f the complete system.
However, the fin a l choice
between two- and three-stage re su lts depends upon changes in e ffic ie n c y
o f the estim ator.
I f no change is recorded then the two-stage findings
should be chosen.
The reason fo r th is is th a t s p e c ific a tio n bias is
magnified by the dynamic properties of the three-stage methodology.
The Wage Model.
A wage model must be developed to compensate fo r the inconsistency
bias.
The dependent v a ria b le in th is model (CW1-) is expressed as a
function of supply and demand conditions, social c h a ra c te ris tic s , and
the position and importance of an SEA(i) in economic space.
Classical labor m o b ility theory im plies the assumption th a t there
is a p erfect m o b ility between two locations and th a t there is r e la t iv e ly
little
unemployment.
This study assumes th a t there is not perfect
-54-
m o b ility between two lo c a lit ie s .
Because o f differences o f tastes and
preferences among in d iv id u a ls , the supply does not s h if t s u ffic ie n tly
to equate wages a t p rio r le v e ls , and s h ifts in demand in one lo c a lity
,
I
are not accompanied by meaningful decreases in demand in other lo c a li­
tie s .
Therefore, the re la tio n s h ip between the change in wages and the
change in nonagricultural employment (CEMP1.) is expected to be p o s itiv e .
Depending on the marketplace events taking place between 1965-1970,
low i n i t i a l wage areas may be re la ted to large changes in wages over
the period or vice versa.
Industry may fin d i t economically fe a s ib le
to relo cate or expand in areas th a t have r e la t iv e ly cheap lab o r.
It
w ill be a m atter of em pirical in te rp re ta tio n whether the demand fo r
labor increased in low income areas r e la tiv e to high income areas when
holding labor m o b ility constant.
Subemployment (SE^) is used as a surrogate measure o f the economic
condition in an S E A (i).
I t is expected to be negatively re la te d to the
change in wages in an SEA(I)'.
Those areas experiencing high subemploy­
ment rates are thought to have an excess labor supply a t the end o f the
period and e v id e n tly did not expand s u ffic ie n tly over the period to
adequately accommodate the e x istin g labor fo rc e .
The e ffe c ts of migra­
tio n on the change in wages have already been discussed above.
Several social variab les are entered in to th is equation to control
fo r s p e c ific SEA c h a ra c te ris tic s of the population th a t may influence
the change in wages.
Education is a measure of the investment in human
-55-
c a p ita l and is calculated as the absolute number of people in an SEA(i)
having 16 or more years o f schooling.
Becker (T97T, p. 173) s ta te s ,
"theory im plies th a t unpleasant occupations are forced to pay higher
wage rates to compensate entrants fo r the unpleasantness.11 However,
he proves th a t there is a p o sitiv e re la tio n s h ip between investment in
human c a p ita l and earnings when costs in tra in in g and differences in
opportunities and a b il it ie s are not held constant.
The age v ariab le
is p o s itiv e ly re la ted to the change in wages because SEA's with higher
median age (assuming a normal d is trib u tio n of age among SEA's) since a
la rg e r proportion o f the population have had a v a ila b le to them more
experience and opportunity than younger members of the labor force.
The
fin a l nonmarket c h a ra c te ris tic is the number o f people in an SEA c la s s i­
fie d as r u r a l.
The re la tio n s h ip is expected to be negative since
h is to r ic a lly these populations have been c h ro n ically stagnant.
There is probably l i t t l e doubt th a t the average wage change across
'
SEA's w ith in the Northern Great Plains does not equal the average fo r
other SEA's in other regions in the United States.
However, in order to
p a r t ia lly determine the v a ria tio n of each SEA w ith in the Northern Great
Plains from the average wage change, there is included in th is equation
the p a rtic u la r position and importance of an SEA w ith in it s own economic
domain.
The v a riab les (R TS^) and (B IS T .^) may have e ith e r p o sitive or
negative signs depending on the Northern Great P la in s ' actual demography.
I f a ll town sizes adjacent to an SEA(I) are large r e la tiv e to the town
-5 6 -
s ize in an S E A (i), then the SEA in question may b e n e fit from an adjacent
area 's growth.
Since the v a riab le is expressed as a r a t io , then as the
town size in the adjacent area increases the ra tio w ill increase, thus
becoming p o s itiv e ly re la te d to the change in wages in the SE A (i).
Also,
as the distance between these two town sizes increase the benefits
from an adjacent large town decreases, thus dampening the change in
wages in th a t SE A (i).
A lte rn a tiv e ly , i f the town size in an SEA(i) is la rg e r than any
adjoining SEA's town s iz e , then th is SEA is considered the dominate
in d u s tria l center.
Therefore, any growth in the la rg e s t town in an
adjacent SEA w ill take away what presumedly would have taken place in
the la rg e s t town in the area, thus decreasing possible wage changes in
the S E A (i).
However, the ra tio as defined above would be negatively
re la te d to the change in wages under these circumstances.
Interdependency Between Wages and M ig ra tio n . — Economic theory con­
veys the concept th a t d iffe r e n tia l wage rates imply d iffe r e n tia l migra­
tio n ra te s .
Simultaneously, d if fe r e n t ia ls in m igration rates cause
d iffe r e n t ia ls in wage ra te s .
The schematic diagram in Figure I I I - I
demonstrates th is hypothesis.
Given an exogenous change in employment ( l e t 's assume th a t the
national demand fo r coal increases), the demand fo r labor in the area
containing th is natural resource w ill increase,
On one side, th is
increase in demand w ill a ffe c t m igration because there are now job
V
-57M o b ility
Factors
In tr a regional
Migration
/
In t e r - \
regional
Inm igration
/
In te r -X
regional \
Outmigration
Indigenous
Population
Labor
Demand
Labor
Supply
Employment
Status of
Labor Force
Basic
Employment
Wage
Change
Locational
C haracteristics
Endogenous elements, those elements determined by the system
Exogenous fa c to rs , those facto rs determined independent of
the system
Figure I I I - I . Schematic Diagram of the In te rre la tio n s h ip between
M igration and Wages Tested in th is Study.
(
-58-
vacancies in the area and m igration is assumed to respond to these
op po rtu n ities.
Furthermore, any increase in demand fo r labor w ill
influence the wage r a te .
As wages change, economic theory says th at
people w ill move to those areas with r e la t iv e ly higher ra te s .
However,
m o b ility is a component of the supply o f T a b o r.. Therefore, those areas
experiencing heavy m igration w ill experience s h ifts in th e ir labor supply
which in turn a ffe c ts local wage ra te s .
A d d itio n a lly , each community
has d iffe r e n tia ls in other facto rs th a t influence m igration.
An example
may be th a t some areas are more densely populated than others or th a t
some are more a g r ic u ltu r a lly o rien ted .
Therefore, there are exogenous
variab les included in the system to control fo r d iffe r e n tia ls in the
in fra s tru c tu re between communities.
In order to implement the method­
ology described above, the remaining chapters of the study w ill estim ate,
te s t, and compare the follow ing models.
^
In trareg io n al System.
The d e fin itio n o f the in t r a regional model was given in an e a r lie r
section o f th is chapter.
Since the observation m atrix in th is model
contains both the inm igration and outm igration figures fo r all_ lo c a li­
t ie s , only one equation is needed to explain both flow s.
The c o e ffi­
cients o f the variab les th a t are measures o f conditions in both destina­
tio n and o rig in areas can be obtained in one equation.
Therefore, i f
the c o e ffic ie n t associated with CWi i s .-B1 and the c o e ffic ie n t associ­
ated with CWi is oti , the introduction o f an inm igration model would
-59-
obtain estimates such th a t a-j would be associated with
would be associated with
CWj- ,
CW1- and -B-j
The inm igration in th is model is there­
fo re only explained by those variables in the m igration equation th a t
are measures of conditions in both o rig in s and destin atio n s.
The
s p e c ific a tio n o f the f u l l in t r a regional model is shown in Table I I I - 2 .
9
In terreg io n al System.
The in terre g io n a l inm igration and outm igration, and wage models
were also previously defined in th is chapter.
Because the origins are
not destinations in the observed m igration flow s, separate m igration
models are developed fo r in terre g io n a l inm igration and outm igration.
The fu ll, m igration models are specified in Table I I I - 3 .
Gross outmigra- •
tio n and a gross immigration is explained by economic conditions in
o rig in and destin atio n areas and s p e c ific control variab les related to
the Plains region.
The wage model is explained by both the gross
inm igration and outm igration flqws re la te d to each Northern Great Plains
SEA.
A d d itio n a lly , the wage model is also explained by c e rta in economic
conditions and SEA c h a ra c te ris tic s .
9A more complete system may be to t r e a t CWj also as endogenous as
is done w ith CW1- since the la rg e s t proportion of migrants to any SEA(j)
are from SE A (i).
-60TABLE I I I - 2 .
VARIABLES, RELATIONSHIPS, AND FULL MODELS FOR INTRAREGIONAL MIGRATION
In trareg io n al
Model
Items and Independent
Variabl es
Controls:
Population, 1970:
o rig in to destinatio n
POP.
+
DtSj j .
-
Town Size:
d e s tin a tio n /o rig in
RTSj j
+
Town Size:
Largest a d ja c e n t/o rig in
--I
-u.CO(
TT
Distance:
a t o rig in or destinatio n
Wage
Model
CW1
-
+ or -
DISTik
+
+ or -
FAR.
-
RU1
+
-
EdU.
+
+
AGE1
-
-
PMj
+
Economic Conditions a t D estination:
Employment change, 1965-1970
CEMP.
+
Wage Change, 1965-1970
CUj
+
Distance:
Largest a d ja c e n t/o rig in
Farm Workers, 1970:
o rig in
Rural Population, 1970:
Education, 1970:
Median Age, 1970:
o rig in
o rig in
o rig in
Past M ig ratio n, 1970:
destination
+
Wage, 1965
Subemployment, 1970
s^ j
-
Economic Conditions gt O rigin:
Employment change, 1965-1970
CEMP1
-
Wage change, T965-1970
CW1
-
Wage, 1965
“ 65.
Subemployment, 1970
M ig ratio n , 1965-1970
SE1
" ij
+
-
+
+
+
-61I
TABLE I I I - 3 .
VARIABLES, RELATIONSHIPS, AND FULL MODELS FOR INTER­
REGIONAL MIGRATION
C ontrols:
Population, 1970:
d estinatio n
At o rig in or
O rigin to destinatio n
Town Size: O rigin (d e s t.)/d e s tin a tio n (o rig in )
Wage
Model
-GW,
i
In terreg io n al ModeIs
IM
OM
ij
ij
Items and Independent
Variabl es
Distance:
i
POP.
+
+
D IS ij
-
-
+
+
<
'
1"
Town Size: Largest a d ja c e n t/d e s titio n (o rig in )
RTS1k
-
-
+ or -
Distance: Largest a d ja c e n t/d e s titio n (o rig in )
D I 5Tlk
+
+
+ or -
Farm Workers, T970:
FARi ,
O rigin
Rural Population, 1970:
tio n (o rig in )
Education, 1970:
Destina­
O rigin
Median Age, 1970:
O rigin
Past M ig ratio n , 1970:
Destination
RU. '
1
-
-
EdU1
+
+
AGE.
-
—
PMU
+
+
Economic Conditions a t D es tin a tio n :
Employment change, 1965-1970
CEMP
-
+
Wage change, 1965-1970
CW
-
+
Wage, 1965
Subemployment, 1970
“ 65
SE
+‘
-
Economic Conditions a t O rig in :
Employment change, 1965-1970 ,
CEMP
+1
-
Wage change, 1965-1970
CW
+
Wage, 1965
+
Subemployment, 1970
W/r
65
SE
Inm igration
IM ji
Outmigration
< •
+
-
•
+
+
+
I
CHAPTER IV
INITIAL MIGRATION MODELS
In th is chapter the two basic problems of concern are the fin a l
s p e c ific a tio n or selection o f variables fo r each m igration model and
special problems re la te d to study data.
Thus, the m igration models are
developed such th a t they are consistent with a p r io ri evidence, arid
where each explanatory v a ria b le to the. extent possible is uncorrelated
w ith the remaining variab les in the model.
Two major procedures are follow ed.
F ir s t , c o e ffic ie n ts o f determ i­
nation are calculated by regressing each explanatory v a ria b le on the
other independent variab les in the model.
This demonstrates the degree
o f m u ltic o llin e a r ity present in the data.
This data c h a ra c te ris tic is
a primary concern because the study is in terested in sin g le estimates o f
p a rtic u la r conditions influencing m igration.
The to ta l p re d ic tiv e power
o f the m igration models is not a paramount concern.
Second, a fte r
variab les are selected , ordinary le a s t squares is applied to the models.
These are reported in succeeding sections as i n i t i a l models.
The fin a l
ordinary le a s t squares models presented in th is chapter are consistent
with the two-stage and three-stage le a s t squares findings in tfie next
chapter.
The ra tio n a le fo r c e rta in variab les hypothesized to influence
population m o b ility was discussed in Chapter I I I .
These variables were
considered on the basis of hypotheses developed in other studies and
-63from established economic theory.
Thus, each v a ria b le was ju s t if ie d
as an influencing fa c to r on a p a rtic u la r migration stream, and it s .
expected re la tio n s h ip to the m igration stream.
This chapter w ill explain
fu rth e r the selection o f variables from those hypothesized by the f u l l
models in Tables I I I - I and I I I - 2 .
A ll o f the variab les considered in Chapter I I I cannot be used in
each o f the m igration models fo r several reasons.
I t is evident fo r
instance th a t more than one v a ria b le could to a great degree measure
or explain s im ila r c h a ra c te ris tic s .
E ith er rural or farm populations
are included in any one model because fo r the Northern Great Plains
e ith e r v a ria b le is in d ic a tiv e of strong social tie s which decrease
m o b ility .
Median age of the population, on the other hand,is a control
v a ria b le re la tin g to the m o b ility c h a ra c te ris tic o f people a t the
o r ig in , but not lo g ic a lly an influence in people's selection of a
d e s tin a tio n , th e re fo re , i t is not considered appropriate fo r an inmigra­
tio n model.
The problem of col lin e a r it y perhaps is another major reason th at
a ll variab les considered in the f u l l model should not be included in
the m igration models.
Population and economic a c t iv it y are not spread
evenly throughout the region.
Places which include large towns and
c itie s lik e ly w ill e x h ib it large absolute changes in population,
employment, income, and other fa c to rs , a ll o f which are c o rrelate d .
-64The major purpose o f the m igration models is to r e la te local
economic conditions to m igration streams; th a t is , the c o e ffic ie n ts are
to be in terp re te d in d iv id u a lly .
Therefore, m u ltic o lli nearit y among
variab les is a serious consideration.
C o llin e a r ity Among Variables Considered
in the M igration Models
Based on the sample data used in the study, ordinary le a s t squares
is used to regress each explanatory v a ria b le of m igration on the remain­
ing variab les in each model.
The degree o f m ulti col lin e a r it y among the
variab les can be measured by examining the c o e ffic ie n ts o f determination
re s u ltin g from each regression.
These c o e ffic ie n ts are d e ta ile d in
Tables I V - I , IV -2 , and IV -3 fo r each m igration stream.
This procedure
does not single out which variables are in te rre la te d with other s p e c ific
v a ria b le s .
However, the 0.98 R opposite ru ral population in Table IV -I
could a ris e whenever ru ra l population is correlated with, only one other
explanatory v a ria b le or with a lin e a r combination o f other explanatory
v a ria b le s .
Displayed in Table IV -I are the variab les dropped i n i t i a l l y from the
in tra re g io n a l model because of suspected m u ltic o llin e a r ity .
The
variab les having zero re s tric tio n s imposed on th e ir c o e ffic ie n ts are
Farm Workers (FARi ) , Rural Population (RUi ) , Median Age (AGEi ) , Change
in Employment (CEMPi ) , and Wage in 1965 (Wg^ ) , where th e ir respective
R^'s are 0 .9 4 , 0 .9 8 , 0 .6 3 , 0 .8 6 , and 0 .9 3 .
The re s u lt o f dropping
these variab les from the hypothesized model leaves the highest degree
I
TABLE IV - I.
MULTICOLLINEARITY AMONG INDEPENDENT VARIABLES, INTRAREGIONAL MIGRATION MODELS,
NORTHERN GREAT PLAINS, 1965-1970a
Model
Items and Variables
In itia l
Full
Final
_______ r 2...............
Controls:
Population a t O rig in , 1970
Distance: o rig in to d estinatio n
Town Sizes: d e s tin a tio n /o rig in
Farm Workers a t O rig in , 1970
Rural Population a t O rig in , 1970
Median Age a t O rig in , 1970
Distance: la rg e s t a d j. town to o rig in
Town Sizes: la rg e s t a d j./o r ig in
Past Migrants a t D estination, 1970
.95
.46
.52
.94
.98
.63
.73
.67
.49
Economic Conditions a t D estination:
Employment change, 1965-1970
Wage change, 1965-1970
Wage in 1965
Subemployment, 1970
.86
.56
.84
.23
Economic Conditions a t O rigin:
Employment change, 1965-1970
Wage change, 1965-1970
Wage in 1965
Subemployment, 1970
.91
.83
.93
.61
.
.53
.44
.52
.33
.43
.51
.58
.57
.45
.18
.44
.41
.43
.42
.14
.43
.42
.13
.77
.81
. .26
.33
aFor each model, each variable is regressed on a ll other variables in the model.
TABLE IV-2.
MULTICOLLINEARITY AMONG INDEPENDENT VARIABLES OF INTERREGIONAL OUTMIGRATION
MODELS, NORTHERN GREAT PLAINS, 1965-1970a
Model
ItaTis and Variables
Full
In itia l
Final
Controls:
Population a t O rig in , 1970
Distance: o rig in to destin atio n
Town Sizes: d e s tin a tio n /o rig in
Farm Workers a t O rig in , 1970
Median Age a t O rig in , 1970
Distance: la rg e s t a d j. town to o rig in
Town Sizes: la rg e s t a d j./o r ig in
Past Migrants a t D estination, 1970
.76
.40
.65
.69
.75
.68
.35
.75
.76
.38
.63
.64
.75
.61
.32
.63
-.5 4
Economic Conditions a t D estination:
Employment change, 1965-1970
Wage change, 1965-1970
Wage in 1965
Subemployment, 1970
.72
.52
.41
.06
.55
.32
Economic Conditions a t O rigin:
Employment change, 1965-1970
Wage change, 1965-1970
Wage in 1965
Subemployment, 1970
.84
.68
.82
.39
.61
.68
.24
.35
.35
.32
aFor each model, each variable is regressed on a ll other variables in the model.
.51
.34
TABLE IV-3.
MULTICOLLINEARITY AMONG INDEPENDENT VARIABLES OF INTERREGIONAL INMIGRATION
MODELS, NORTHERN GREAT PLAINS, 1965-19703
Model
Items and Variables
Full
In itia l
Final
R2
C o n tro ls :"
Population- a t D estin atio n , 1970
Distance: o rig in to destinatio n
Town Sizes: d e s tin a tio n /o rig in
Rural Population a t D es tin atio n , 1970
Distance: la rg e s t a d j. town to destinatio n
Town Sizes: la rg e s t a d j./d e s tin a tio n
Past Migrants a t -D estination, 1970
.81
. 56
.64
.76
.67
.48
.72
Economic Conditions a t D estination:
Employment change,, 1965-1970
Wage change, 1965-1970
Wage in 1965
Subemployment, 1970
.84
.65
.84
.30
Economic Conditions a t O rigin:
Employment change, 1965-1970
Wage change, 1965-1970
Wage in 1965
Subemployment, 1970
.68
.37
.24
.04
.38
.48
.40
.40
.21
.28
.31
.27
.28
.31
.38
.26
.49
.38
.27
.09
.09
.03
.03
-
aFor each model, each variable is regressed on a ll other variables in the model.
-68o f c o rre la tio n a t 0.81.
I t is also in te re s tin g to note th a t by dropping
these v a ria b le s , the c o e ffic ie n t o f determination fo r population a t the
o rig in (POP1. ) decreases from 0.95 to 0.53.
In reviewing the remaining two muI t i c o l l in e a ri t y ta b le s , one fo r
in terre g io n a l outm igration and the other fo r the in terreg io n al inmigra­
tio n model, several control and economic variables also are considered
inappropriate because o f the re s u lta n t R 's (m u lt ic o llin e a r it y ) .
The
control variab les remaining in the outm igration model th a t are considered
s u ffic ie n t in c o n tro llin g fo r SEA c h a ra c te ris tic s are Population (POP1- ) ,
Median Age (AGE1-) , and Past Migrants a t Destination (P M j).
For the
inm igration model only Distance from O rigin to Destination (D IS^j) and
Rural Population (RU1-) are control variab les excluded from the fin a l
model.
Because the economic variab les are o f primary in te re s t fu rth e r
consideration w ill be given to the in divid ual c o e ffic ie n ts associated
with each economic v a ria b le .
Although examination of the m u ltic o llin e a r­
i t y tables reveals the fin a l models fo r each m igration stream, the
next section w ill in d ic a te the reasons fo r imposing zero re s tric tio n s on
some of the economic conditions in the fin a l models.
Ordinary Least Squares Estimates
Economic theory has established th a t people seek occupations,
lo ca tio n s , and jobs where most lik e ly net benefits are obtained by an
in d iv id u a l.
Therefore, wages, employment, i n i t i a l wage, and subemploy­
-69ment are a ll economic conditions th a t should influence population
m o b ility .
However, these conditions are not always easy to measure and
many times proper surrogates do not e x is t.
Therefore, several of the
economic variab les are dropped because, f i r s t , there may be the possi­
b i l i t y o f m u ltic o llin e a r it y which has been already mentioned.
Second,
the a v a ila b le data may not measure the intended hypotheses and therefo re
are not appropriate fo r analysis.
be th e o r e tic a lly c o rre la te d .
T h ird , the variab les themselves may
For example, i t could be argued th at
communities experiencing high p a rtic ip a tio n rates in employment also
are lik e ly to have low subemployment ra te s .
Thus the exclusion of
several economic variables does not imply th a t fo r some m igration
streams economic conditions are not re le v a n t.
I t only implies th a t the
fin a l models are the best models obtainable on the basis o f the r e s t r ic ­
t iv e data a v a ila b le and because of e x p lic it economic circumstances in
the Northern Great P lains.
The zero re s tric tio n s imposed on the control variab les resulted in
a refined s p e c ific a tio n of the m igration models.
The models are
fu rth e r refined in th is section by follow ing the c r it e r ia outlined above.
Thus, economic variables are dropped from the models when th e ir c o e f f i­
cients do not e x h ib it the correct sign or when a v a ria b le can s t i l l be
expressed as a lin e a r combination of one or more variab les in the model.
The ordinary le a s t squares estimates fo r each model are presented
in Tables IV -4 , IV -5 , and IV -6 .
The remaining economic variables in
-70TABLE IV-4.
OLS INTRAREGIONAL MIGRATION MODELS, NORTHERN GREAT
PLAINS, 1965-19703
Items and
Independent Variables
C ontrols:
Population a t O rig in , 1970
Distance: o rig in to destination
Town Sizes: d e s tin a tio n /o rig in
Farm Workers a t O rig in , 1970
Rural Population a t O rig in , 1970
Median Age a t O rig in , 1970
Distance: la rg e s t a d j. town
to o rig in
Town Sizes: la rg e s t a d j./o r ig in
Past Migrants a t D estin atio n , 1970
Exp.
Sign
+
_________ Model
In itia l
________ Final
B
t
B
t
C oeff. St a t .
C oeff. St a t .
1.14
- 1 .46
0.20
5.03.
6.57
1.97
1.21
-1 .4 6
0.20
6.43
6.66
1.90
+
0.37
-0 .2 3
0.40
1.45
1.79
6.09
0.26
-0 .2 6
0.41
1.45
2.32
6.42
Economic Conditions a t D es tin a tio n :
+
Employment change, 1965-1970
+
Wage change, 1965-1970
+
Wage in 1965
Subemployment, 1970
1 .67
3.10
-2 .0 3
4.71
4.93
3.03
1 .68
■ 3.12
-2 .0 3
4.79
5.00
3.07
-
0.06
0.06
0.63
0.00
-0 .2 9
0.95
+
-0 .0 2
0.00
Economic Conditions a t O rig in :
Employment change, 1965-1970
Wage change, 1965-1970
Wage in 1965
Subemployment, 1970
-
+
+
-
In terc e p t
-17.04
-16.26
S ta tis tic s :
?
~
F S ta tis tic
Degrees of Freedom Residual
Standard Error
aData are in double logarithmic form.
0.63
24.00
171.00
0.52
0.62
29.00
173.00
0.52
-71TABLE IV-5.
OLS INTERREGIONAL OUTMIGRATION MODELS, NORTHERN GREAT
PLAINS, 1965-1970a
Items and
Exp.
Independent Variables
Sign
_______________________________
C ontrols:
Population a t O rig in , 1970
Distance: o rig in to destination
Town Sizes: d e s tin a tio n /o rig in
Farm Workers a t O rig in , 1970
Median Age a t O rig in , 1970
Distance: la rg e s t a d j. town
to o rig in
Town Sizes: la rg e s t a d j./o r ig in
Past Migrants a t D estination, 1970
+
1.62. 17.05
1.94
-0 .4 3
-0 .0 3
-0 .3 9
-0 .9 5
15.39
3.30
0.79
5.84
1.52
-3.87
8.39
0.42
-0.11
0.71
4.95
3.07
19.03
0.76
26.14
0.24
4.52
0.17
3.74
-
-0.11
-0.67
3.54
3.55
-0.13
-0 .7 5
5.52
5.42
+
1.02
3,79
0.84
3.07
-
+
-
+
-
+
Economic Conditions a t D estin atio n :
+
Employment change, 1965-1970
+
Wage change, 1965-1970
+
Wage in 1965
Subemployment, 1970
Economic Conditions a t O rig in :
Employment change, T965-1970
Wage change, 1965-1970
Wage in 1965 .
Subemployment, 1970
__ .________ Model____________
In itia l
______ Final
B
t
B
t
C oeff. St a t .
C oeff. S t a t .
In terc e p t
-5 .6 9
- 1 .51
S t a t is t ic s :
R2
F S ta tis t ic
Degrees of Freedom Residual
Standard Error
aData are in double logarithmic form.
0,70
182.00
923.00
0.471
0.68
284.03
928.00
• 0.486
-72TABLE TV-6.
OLS.INTERREGIONAL INMIGRATION MODELS, NORTHERN GREAT
PLAINS, 1965-1970a
Items and
Independent Variables
Exp.
Sign
C ontrols:
Population a t D estin atio n , 1970
Distance: o rig in to destinatio n
Town Sizes: d e s tin a tio n /o rig in
Rural Population a t D estination,
1970
Distance: la rg e s t a d j. town
to destinatio n
Town Sizes: la rg e s t a d j . /
destinatio n
Past Migrants a t D estin atio n , 1970
+
1.63
16.47
1.67
15.51
+
0.20
4.71
0.20
4.67
+
0.64
8.55
0.62
7.98
-0 .2 6
0.71
5.88
20.25
-0 .2 6
0.71
5.77
20.22
1.08
2.00
6.32
6.53
1.01
1.87
0.30
5.35
5.61
0.94
-0 .8 6
2.50
-0 .8 5
2.49
0.27
1.75
0.27
1.75
_
+
Economic Conditions a t D es tin a tio n :
+
Employment change, 1965-1970
+
Wage change, 1965-1970
+
Wage in 1965
Subemployment, 1970
Economic Conditions a t O rig in :
Employment change, 1965-1970
Wage change, 1965-1970
Wage in 1965
Subemployment, 1970
____________ Model__________
I n i t i a l ___________ Final
B
t
B
t
C oeff. S.tat.
C oeff. ' St a t .
+
-16.49
In terc e p t
-15.92
I
S t a t is t ic s :
0.60
152.40
926.00
0.59
R2
F S ta tis tic
Degrees of Freedom Residual
Standard Error
aData are in double logarithmic form.
N
)
0.60
137.20
925.00
0.59
-73-
the fin a l in tra re g io n a l model are Wage change (CWj. ) , Wage in 19 6 5 .(W65 ) ,
and Subemployment (SEj.) a t destinatio n lo c a lit ie s and Wage change (CW1-I
a t o rig in lo c a lit ie s .
The in terre g io n a l models also are refin ed
because o f unexpected signs.
by deletin g variables
The economic variab les th a t remain in
the fin a l outm igration model are Employment change (CEMPj.) a t destina­
tio n and Employment change (CEMP1- ) , Wage change (CW1- ) , and Subemployment
(SEi ) a t o r ig in .
In Table IV -6 , the economic variab les th a t remain fo r
addition al analysis in the inm igration model are Wage change and I n i t i a l
wage a t the destin atio n SEA, and Wage change and Subemployment a t the
o rig in s ta te .
I t should be emphasized again th a t due to lim ita tio n s of
the data used, the fin a l models do not contain a ll the measures of
economic conditions hypothesized to influence m igration.
is influenced by the same economic conditions.
A ll migration
The important d i f f e r ­
ence is the d if fe r e n t ia l e ffe c t of each condition on m igration.
To the
extent possible (those in the fin a l analysis) a major concern in the
remaining chapters is to is o la te the d if fe r e n t ia l e ffe c ts o f economic
conditions on separate m igration streams.
O
The R 's (c o e ffic ie n ts of determ ination) fo r the fin a l in tr a ­
regional and in terre g io n a l outm igration and inm igration models are
0 .6 2 , 0 .6 8 , and 0.60 re sp e c tiv e ly .
In Appendix C i t was concluded th a t
on the basis o f the F s t a t is t ic s , the fin a l models in each case explains
s ig n ific a n tly as much v a ria tio n in m igration as did the f u l l and i n i t i a l
-74re s tric te d models.
The re s u lta n t models developed in th is chapter are
assumed to be void o f any fu rth e r v io la tio n in the assumptions fo r
c la s s ic al normal regression a n a ly s is , except fo r the stochastic proper­
tie s of change in wage (CW1-) used as an explanatory v a ria b le fo r migra­
tio n .
As already mentioned, economic theory implies th a t th is v io la ­
tio n is present when analyzing m igration .
Therefore, by implementing
the fin a l models from th is chapter the interdependency between migra­
tio n and wages w ill be considered in Chapter V.
CHAPTER V
ESTIMATION OF THE MIGRATION MODELS
The re su lts of th is chapter are based on the fin a l migration models
a rrived a t in Chapter IV .
There are three major goals to be considered:
(a) to estim ate the m igration models while tre a tin g the change in
average wage endogenous to a P la in s ' SEA, (b) to te s t fo r the possi­
b i l i t y o f improving the e ffic ie n c y o f the estim ates, and (c) to d e te r­
mine the impact of the remaining m igration fa c to rs .
The primary objec­
tiv e in the two-stage analysis is to obtain a consistent estimate of the
endogenous explanatory v a ria b le in each model th a t s ig n ific a n tly in f lu ­
ences the m igration stream.
Therefore, both the c o e ffic ie n t and its
estimated variance are observed.
In the analysis o f the three-stage
method, comparison o f the standard deviations (square root o f the
variances) o f a ll c o e ffic ie n ts in each model are compared with those o f
the two-stage regression.
I f e ffic ie n c y is not gained then the consis­
te n t two-stage re su lts w ill be re tain ed .
Two-Stage Estimation of the In trareg io n al Model
The two-stage estimates o f the in tra re g io n a l m igration and wage
models are presented in Table V - I .
In th is section, special in te re s t
is given to the estimated c o e ffic ie n t and t-v a lu e associated with the
wage change (CW^) .
The consistent estim ate, as was the ordinary le a s t
squares estimate (Table IV - 2 ) , is not s t a t is t ic a lly s ig n ific a n tJ 0
^0The s t a tis t ic a l te s t associated with each v a ria b le used in the
study is a general te s t concerning the a p p licatio n of a p r io ri economic
-76TABLE V -I.
TWO-STAGE LEAST SQUARES, INTRAREGIONAL MIGRATION MODELS,
NORTHERN GREAT PLAINS, 1965-1970a -
Model
Items and
Independent Variables
Controls:
Population a t O rig in , 1970
Distance, o rig in to destination
Town Sizes, d e s tin a tio n /o rig in
Rural Population a t O rig in , 1970
Median Age a t O rig in , 1970
Education a t O rig in , 1970
Distance, large a d ja c e n t/o rig in
Town Sizes, la rg e s t adjacent/
o rig in
Past Migrants a t D estination,
1970 '
Wage, CW^
M ig ratio n, M ..
B c o e ff.
t - s t a t ,.c
1 .22
-1 .4 6
0.19
B c o e ff.
6.39
6.66
1.90
0.25
1.37
0.19
-0 .3 3
0.03
-0 .3 3
-0 .2 6
2.34
-0 .0 9
0.41
6.42
Economic Conditions a t Destination :
1.68
Wage Change, 1965-1970
3.13
Wage in 1965
-2 .0 3
Subemployment, 1970
4.78
5.01
3.06
Economic Conditions a t O rigin:
Employment Change, 1965-1970
Wage Change, 1965-197Qb
Wage in 1965
Subemployment, 1970 K
M ig ratio n, 1965-1970°
In terc e p t
S ta tis tic s :
Degrees of Freedom Residual
Standard Error
-0 .3 5
2.37
1.62
0.63
10.84
6,84
-0 .1 2
8.12
0.02
0.12
0.02
0.11
0.99
2.96
0.99
16.17
,
t-s ta t.c
2.53
173..00
o..52
174.00
aData in double logarithm ic form.
t3These are endogenous explanatory v a ria b le s .
cThe t - s t a t i s t i c associated with a .05 sign ificance level is 1.645.
-77-
Econpmic theory implies th a t wage d is p a ritie s are a nominal measure
o f obtainable b e n e fit among lo c a lit ie s , th erefo re an in centive fo r poten­
t i a l migrants to move, w hile c o n tro llin g fo r other aspects o f m igration.
The change in average wage o f an o rig in SEA does not in d ic a te th a t groups
of p o te n tia l migrants remain a t the o rig in as average wages increase.
There is the p o s s ib ility th a t SEA's have both a disproportionate number
of high paying jobs as well as a disproportionate number o f low paying
jobs.
Thus, the p o ten tial migrants may have remained in lo c a litie s to
increase th e ir fu tu re income without i t being re fle c te d in the average
wage o f the community.
Though data are lim ite d and expensive to o b ta in ,
a more accurate measure o f wages may be the in divid ual wage o f each
migrant before m igration is considered.
Furthermore, average wage may
not measure an economic in centive to m igrate i f there is l i t t l e v a ria ­
tio n across the observed SEA's, where some SEA's have large m igration
flows r e la t iv e to other SEA's.
Under these circumstances the average
wage w ill not s ig n ific a n tly influence m igration.
The change in average
wage (CW^) does, however, influence the selection of a d e s tin a tio n .
Those SEA's having a large change in average wages also have large
inm igration flows from w ithin the region.
theory. Therefore, the null hypothesis is Hq IB1- = O and the a lte rn a ­
tiv e hypothesis is Ha=Bi f 0. The sign ificance level of the te s t is
.05. That is , an in f in i t e number of samples from a population with
B1- = 0 w i l l , 5 percent of the tim e, get a value of B- which w ill lead
to an in co rre c t re je c tio n of the null hypothesis.
-78In tra re g io n a l Wage Model.
The wage model is specified to include an SEA's c h a ra c te ris tic s and
economic conditions a t the beginning, end, and during the m igration
period.
Also included are variables measuring the economic opportunity
of an SEA's la rg e s t trade center r e la tiv e to nearby areas, and the
remoteness of these trading centers.
M igration is th e o re tic a lly
included as a market function v a ria b le a ffe c tin g the labor supply in an
area.
Economic Conditions.
M igration (M-jj) s ig n ific a n tly influences the change in average wages.
The re la tio n s h ip is p o s itiv e as expected, but the c o e ffic ie n t is not
la rg e .
T h e o re tic a lly , the m o b ility of labor influences wages by
increasing or decreasing the labor supply.
In the present s itu a tio n ,
m igration decreases the supply of la b o r, thereby increasing wages.
The
small c o e ffic ie n t may be the re s u lt o f an e la s tic demand fo r labo r.
The change in employment (CEMPi ) does not e x h ib it the correct sign
but strongly influences the wage change.
I t should be remembered th a t
the calculated wage is an average wage, where to ta l nonagricultural
salary revenue is divided by to ta l nonagricultural employment.
There­
fo re , those areas experiencing large increases in new employment, a t
the same tim e, did not experience an equal or greater increase in salary
income per person.
-7 9 -
The other two v a ria b le s , wage level in 1965 (Wg5 .) and subemployment (SE1-) did not s ig n ific a n tly a ffe c t the change in wages over the
period.
C ontrols.
Education (EdU1- ) and median age (AGE1- ) also did not s ig n ific a n tly
a ffe c t the change in average wages.
The ru ra l population v a riab le (RU1- )
i n i t i a l l y was thought to be negatively re la te d to wage changes.
Rural
areas are not expected to develop p a ra lle l with densely populated areas.
However, the p o s itiv e sign is in d ic a tiv e o f a predominately rural region,
and those SEA's having large ru ral populations have a g reater change
in wages.
There are two ad d itio n al variab les th a t influence the wage change
in an S E A (i).
These are the lo catio n of the la rg e s t town in an SEA
experiencing the wage change, r e la tiv e to the la rg e s t adjacent town in
any one o f the adjoining SEA's (DIST1-^ ), and the r a tio o f these two
town sizes (RTS1-^ ).
The tr ic k le - o u t e ffe c t of growth centers is the ra tio n a le fo r the
inclusion of these variab les (Hoover 1971).
However, th e ,re la tio n s h ip ,
of these variab les to the wage change may be the r e s u lt o f the demog­
raphy of the study area.
The negative sign apparently suggests th a t
in most cases the la rg e s t adjoining town has more economic opportunity
than the la rg e s t town in the SE A (i).
This re la tio n s h ip can be explained
by looking a t the numerator of the r a tio ; as the la rg e s t town in the
-80adjacent SEA is large r e la tiv e to the la rg e s t SEA(i) town, the ra tio
becomes la rg e .
The r e la t iv e ly greater size in the adjacent SEA is
thought to dominate surrounding areas by a ttra c tin g new industry: or
expanding e x is tin g industry.
Thus, i f an adjacent area captures much
o f the economic a c t iv it y , wages in i are depressed.
F in a lly , as an SEA(i) is located more remotely in terms of distance
from the la rg e s t adjacent town, the tr ic k le -o u t e ffe c t is not as g re a t.
Therefore, wage changes are not as g re a t.
The c o e ffic ie n t o f determina­
tio n fo r the ordinary le a s t squares o f th is model is .8 0 , and the in t e r cept is 2.53 (Appendix B).
11
Two-Stage Estimation o f the In terreg io n al Model
The two-stage estimates of the in terre g io n a l model are displayed
in Table V-2.
In the two m igration equations, the endogenous explanatory
v a ria b le is the wage change between 1965-1970 (CW1- ) .
Both outmigration
and inm igration are endogenous explanatory variab les (OM^.,. IM j1O in
the wage model.
n
p
The R is not calculated fo r the two- and three-stage re s u lts .
The reason fo r th is is th a t only ordinary le a s t squares analysis
c o rre c tly decomposes the sum o f the to ta l v a ria tio n in the dependent
v a ria b le in to what is explained by the independent va riab le s |and,what
is random. Therefore, an R2 , i f derived fo r two- and three-stage analy­
s is , might measure a comparable goodness o f f i t , but cannot be in te r ­
preted as the proportion o f the v a ria tio n in the dependent v ariab le
explained by the right-hand side v a ria b le s .
i
TABLE V -2.
TWO-STAGE LEAST SQUARES, INTERREGIONAL MIGRATION MODELS, NORTHERN GREAT PLAINS, 1965-1970*
Model
M ig ra tio n
Item s and
Independent V a ria b le s
0Mi. i
B c o e ff.
C o n tro ls :
P o p u la tio n a t O r ig in o r D e s t., 19/0
Town S iz e s , d e s t in a t io n / o r ig in
R ural P o p u la tio n a t O r ig in , 1 970
Median Age a t O r ig in , 1970
E du ca tio n a t O r ig in , 1970
D is ta n c e , la r g e s t a d j. tow n/
o r i g in ( d e s t. )
Town S iz e s , la r g e s t a d ja c e n t/
o r i g in ( d e s t. )
Past M ig r a tio n a t D e s tin a tio n , T 9 7 0
Economic C o n d itio n s a t D e s tin a tio n :
Employment Change, 1965-1970
Wage Change, 1965-1970
Wage, 1965
Subemployment, 1970
Economic C o n d itio n s a t O r ig in :
Employment Change, 1965-1970
Wage Change, 1 965-1970b
Wage, 1965
Subemployment, 1970
k
O u tm ig ra tio n , 1965-1970°
In m ig r a tio n , 1965-1970b ,
In t e r c e p t
Wage, CW1
1Mi i
t-s ta t.c
1.6 6
16.26
-3 .8 9
8 .4 2
0 .7 6
25.96
0 .1 6
3.64
B c o e f f.
t-s ta t.c
1.05
0.32
7.8 3
6 .7 0
0.01
-0 .4 4
0.24
0.07
4 .3 0
11.35
1 0 .4 6
-0 .2 6
18.04
0.32
0.66
6 .5 9
17.53
0.02
3.17
3.13
3.7 0
-1 .2 2
10.11
9.0 2
3 .1 8
-0 .1 2
15.95
0.03
0.29
0.01
-0 .0 2
0.27
6.21
0.71
-0 .8 4
5.47
4.5 9
-1.11
3 .0 5
0 .8 9
3.1 9
0.23
1.4 2
S ta tis tic s :
Degress o f Freedom R esidual
S tandard E r r o r
t-s ta t.c
0.94
-0 .1 3
-0 .8 7
-2 1 .0 6
-1 .3 1
B c o e f f.
928.00
0 .4 9 -
aData a re In d o ub le lo g a r ith m ic fo rm .
^These a re endogenous e x p la n a to ry v a r ia b le s .
cThe t - s t a t i s t i c a s s o c ia te d w ith a .05 s ig n if ic a n c e le v e l is 1.6 45.
3.63
926.00
0.3 7
925.00
-82The consistent estim ate o f the wage change a t o rig in is s ig n ific a n t
and ex h ib its the expected negative sign.
The average wage change a t
o rig in SEA's fo r in terre g io n a l outm igration is concluded to be an
in centive fo r p o te n tia l migrants a t the margin of p ro b a b ility to e ith e r
remain in or move from a community in the Northern Great P lains.
The endogenous explanatory v a ria b le (wage change a t d e s tin atio n )
in the in terre g io n a l inm igration model is also s ig n ific a n t and exhibits
the expected p o s itiv e sign.
That is , those SEA's experiencing a large
wage change between 1965-T970 had an in flu x of migrants from states ou t­
side the region.
The remaining control variables and economic conditions
in both the outm igration and inm igration models are s ig n ific a n t.
For
the inm igration model subemployment a t destinatio n (SE1-) is retained
in the two-stage regression because i t e xh ib its the c o rrect sign.
The
two-stage outm igration model is specified in the same manner as was the
ordinary le a s t squares model in Chapter IV .
In terreg io n al Wage Model.
As described e a r lie r in th is chapter, the wage v a ria b le is a func­
tio n o f c h a ra c te ris tic s and economic conditions of a P la in s ' SEA.
The
m igration variab les ex ert d ire c t pressure on the change in wages o f an
SEA through labor supply conditions.
-8 3 —
Economic Conditions and C ontrols.
Outmigration and inm igration do not s ig n ific a n tly a f f e c t wage
changes under the consistent estim ation technique.
the expected signs hypothesized by economic theory.
However, they display
In theory, out­
m igration exerts upward pressure on wages by decreasing labor supply.
A lte rn a tiv e ly , inm igration increases labor supply forcing wages down,
everything else being constant.
The other variab les th a t do not s ig n ific a n tly influence the wage
change are ru ra l population (RU1. ) and wage income a t the beginning of
the period (W6 5 ) .
The s ig n ific a n t variab les th a t did not have the
expected signs are employment change (CEMP1) and subemployment (SE1) .
Three-Stage Regression Results
Three-stage regression is performed on the m ig ra tio n models in an
attempt to increase the e ffic ie n c y of the estim atesi
Depending upon
the decrease in the estimated standard deviations o f the c o e ffic ie n ts
in each m igration model, the selection between two- and three-stage
analysis is made.
The two-stage method (consistent estim ation) may be
selected because s p e c ific a tio n bias is not a major concern in single
equation estim ation.
Shown in Table V-3 are the estimated standard deviations of each
c o e ffic ie n t in both the two- and three-stage regressions.
In each
instance the decrease in the standard deviation is less than I percent
TABLE V-3.
STANDARD DEVIATIONS OF COEFFICIENTS BETWEEN TWO- AND THREE-STAGE REGRESSIONS.
______ M igration Models9_________________
Tntraregional
In terregion al
M igration
Outmigration
Inm igration
2SLS
3SLS
2SLS
3SLS
2SLS
3SLS
C ontrols: ,
Population0
Distance, o rig in to destin atio n
Town Sizes, d e s tin a tio n /o rig in
Median Age a t O rig in , 1970
Distance, la rg e s t a d j. town/observed0
Town Sizes, la rg e s t adj./observed0
Past Migrants a t D estin atio n , 1970
.185
.111
.064
.185
.110
.063
Economic Conditions a t D e s tin a tio n :
Employment Change, 1965-1970
Wage Change, 1965-1970
Wage, 1965
Subemployment, 1970
.352
.624
.661
.349
.619
.655
.356
.356
Economic Conditions a t O rig in :
Employment Change, 1965-1970
Wage Change, 1965-1970
Subemployment, 1970
.191
.220
.102
.191
.218
.101
.102
.101
.462
.452
.029
.028
.044
.042
.
.024
.189
.279
.024
.188
.277
.134
.128
.047
.042
.089
.048
.038
.087
.046
.036
.310
.410
.382
.292
.391
.373
.366
.163
.320
.134
aModels correspond to those in Tables V -I and V-2.
^Reported as a t o rig in fo r in terre g io n a l outm igration and in t r a regional m igration, and a t
destinatio n fo r the in te rre g io n a l inm igration model.
cReported as r a tio o f o rig in town fo r in terre g io n a l outm igration and in traregion al migra­
tio n , and a t destinatio n town fo r in terre g io n a l inm igration model.
-85-
when estim ating the in tra re g io n a l model under the three-stage method.
Thus, the increase in e ffic ie n c y is not s u ffic ie n t to warrant threestage a n a ly s is .
That is , the p o s s ib ility of s p e c ific a tio n bias may give
r is e to a d is to rtio n o f the c o e ffic ie n ts not ju s t if ie d by the small
change in the standard deviatio n s.
Displayed also in Table V-3 are the standard deviations between
the two regression methods fo r the in terre g io n a l models.
The estimated
standard deviations o f the c o e ffic ie n ts also decreased by less than I
percent in each case fo r the outm igration model.
For the inm igration
model the m a jo rity of deviations remained about the same or decreased by
less than I percent.
Only wage change and subemployment a t o r ig in , and
wage change and wages in 1965 a t destinatio n have estimated standard
deviations th a t decreased by more than I percent.
Thus i t is concluded
th a t fo r both the in tra re g io n a l model and the in terre g io n a l models
sing le equation estim ation is approp riate.
The re s u lts presented in
the remaining sections o f the study are th erefo re based on the consistent
two-stage estim ates.
Two-Stage Results
From Tables V -I and V -2, population (POP..) a t o rig in fo r both
in tra re g io n a l and in terre g io n a l outm igration controls fo r the popula­
tio n a t r is k .
Population (POPi ) a t destinatio n indicates the greatest
movement o f in terre g io n a l inmigrants to the more populated SEA's in
the Northern Great P lain s.
Distance (DIS. •) between o rig in and destin aIJ
-86tio n is a fa c to r entering the in tra re g io n a l model, but does not
influence in terre g io n a l m igration.
Since the distance o f the movement
is short, costs both pecuniary and psychic are sm all.
Thus, migrants
a t le a s t moving w ith in the Plains region are able to distingu ish between
d e s tin atio n s.
They a re , th e re fo re , aware o f , and susceptible to ,
intervening op po rtu n ities.
The town population r a tio (RTSj1-) a t destination to o rig in displays
the expected sign and s ig n ific a n tly influences both in tra re g io n a l and
in terre g io n a l in m igration.
This v a ria b le is intended to control fo r
amenities and economic opportunity a t destinatio n r e la t iv e to o rig in .
An o rig in or d e s tin atio n SEA's major trading center measured by
it s population (thus the trading c e n te r's status in the c e n tra l-p la c e
hierarchy) r e la t iv e to the la rg e s t trading center in any one adjacent
SEA's (RTSjc1-) and the distance between these two trading centers (DIST^^)
are also included in the two models mentioned in the above paragraph.
However, the distance v a ria b le in the in trareg io n al model is not s ig n i­
fic a n t a t the .05 le v e l.
Past m igration a t destinatio n (PMij-) p a r t ia lly
explains a ll three streams, and th erefo re is an important fa c to r re la te d
to both Northern Great Plains in tra re g io n a l and in terre g io n a l migra­
tio n .
In reviewing Tables V -I and V -2, the m igration streams are not
influenced by a l l o f the economic v a ria b le s .
In some cases the
surrogates used as measures of economic conditions a t o rig in and destin a­
i
-87“
tio n are in ap propriate.
In tra re g io n a l migrants do however respond to
changes in average wage (CWj) and i n i t i a l average wage (Wg5- ) a t
d e s tin a tio n .
Furthermore, SEA's experiencing high subemployment rates
(SE-) are in d ic a tiv e o f undesirable destinations fo r in trareg io n al
migrants.
A fte r c o n tro llin g fo r s p e c ific Plains c h a ra c te ris tic s , economic
conditions in both o rig in and d estinatio n determine in terre g io n a l out­
m igration flows.
The change in nonagricultural employment (CEMPj) in
the 39 contiguous states outside the region are inducements fo r people
to move.
Since a large number of migrants remained in these states
over the m igration period i t is apparent they e ith e r had a job before
making the
move or were able to obtain a job a f t e r a r riv in g . ,
Increases in both employment (CEMPj) and wages (CWj) in the Plains
reduce the movement o f people from the region.
The wage change in th is
i.
model is treated as endogenous and is th erefo re p a r t ia lly dependent
upon the actual amount o f outm igration.
F in a lly , those Plains areas
experiencing high subemployment rates (SEj) had a d d itio n al people5on
I
the margin o f m o b ility moving in te rre g io n a lly . The study did not
p a r titio n the migrant streams by employment status.
Therefore, no
inference can be made about the actual c h a ra c te ris tic o f the migrant.
The observed increase in outmigration was lik e ly a r e s u lt o f e ith e r a
depression in the economy of an SEA and th erefo re a m igration stream
-8 8 -
la rg e ly comprised o f those who were subemployed, or a stream comprised
of employed people who were able to fin d employment elsewhere.
Change in average wage (CWi ) and the beginning o f the period wage
(Wg5 ^) a t destinatio n are incentives fo r migrants to relocate in the
Plains region.
Since i t was found th a t inm igration does not s ig n if i­
c a n tly influence the wage change a t destinatio n (Table V -2 ), wages
between the region and the remaining 39 contiguous states were not
a ffec te d by m igration.
However, subemployment (SEi ) a t destination
in h ib its the in flu x o f inm igrants.
F in a lly , p o sitiv e changes in wages
a t o rig in states (CWj) is an in centive fo r people not to relocate in
the P la in s , while subemployment (SEj) a ffe c ts the in terreg io n al inmigra­
tio n flow in the opposite d ire c tio n .
I t is evident th a t measures of only several o f the economic condi­
tions used in the study are appropriate in explaining Northern Great
Plains re la te d m igration.
Because the study is only one part of an
extensive p ro ject on the impact of population changes accompanying
coal development in ru ra l communities of the Northern Great P lains, the
economic variab les of primary in te re s t are the change in average wages
across SEA's in the region.
Through the methodology follow ed, estimates
are obtained lin k in g these wages to the separate m igration flows in
a n tic ip a tio n o f regional development.
CHAPTER VI
SUMMARY, ILLUSTRATION, AND CONCLUSIONS
The purpose o f th is study is to lin k local economic conditions to
population m igration.
The population m igration is expressed as groups
or streams o f migrants moving from an o rig in to a destinatio n between
1965 and 1970.
Because the models c o n s titu te gross flows between two
places, facto rs th a t influence m igration are computed fo r o rig in and
d e s tin a tio n .
The f i r s t o b jective is to lin k economic conditions a t
these lo c a lit ie s to the movement o f people between them a f t e r c o n tro llin g
fo r other s ite c h a ra c te ris tic s .
The methodology r e lie s on the assump­
tio n th a t people a t the margin o f m o b ility respond to conditions in a
set o f places.
These conditions e ith e r a t t r a c t or repel people by
in dicating an increase or decrease in economic opportunity.
The second o b je ctiv e is to disaggregate the data in to an in tr a regional model, and in terreg io n al outm igration and inm igration models.
This method o f analysis is used on the assumption th a t d is ta n t popula­
tions react to conditions in the Plains d if fe r e n t ly .
That is , migrants
who move w ith in a c e rta in proxim ity o f an o rig in are not as rig id when
weighing costs of the move in re la tio n to it s rewards i f nearly the
same s ite values are received in both lo c a lit ie s .
Additional incentives
or la rg e r increments must be a v a ila b le fo r people who move over a
greater distance, assuming geographic preferences are important to
people.
Once the data are disaggregated, estimates o f the e ffe c t of
-90local economic conditions are made.
The estimates in each model are
compared to those in the aggregated models (Appendix B).
A model is postulated fo r each m igration stream from existing
hypotheses in the lit e r a t u r e .
exists among these d a ta .
But a high degree of m u ltic o llin e a r it y
Thus these o rig in a l models are modified by
excluding those variab les explained by the remaining variab les in each
model.
Although these constrained models are lim ite d in meaning and
use, they are the basis o f the remaining analysis o f the study.
The th ird and fin a l o b jective is to adjust fo r the sim ultaneity
between m igration and the wage change in a P lain s' SEA.
This is
accomplished by introducing two-stage le a s t squares regression analysis.
This method is implemented because economic theory im plies th a t the
explanatory v a ria b le , wage change, is endogenous; th a t i s , i t is a
function of m igration.
Furthermore, three-stage le a s t squares also is
i
applied to the m igration models in an attempt to gain more information
about the estim ates.
Illu s t r a tio n o f Results
Each m igration model is expressed as both m u ltip lic a tiv e and m u lti­
v a ria te .
Because th is form ulation forces the c o e ffic ie n ts to be
exponents, the data are transformed in to double logarithm ic form, making
the model conform to normal lin e a r regression a n alysis.
Thus, the
c o e ffic ie n t is defined as the e la s t ic it y o f m igration with respect to
I
f
^
-91it s determinant.
Because the determ ination o f m igration is m u ltiv a ria te ,
these are defined as p a rtia l e la s t ic it ie s .
V e rb a lly , th is means th a t,
fo r example, the p a rtia l e la s t ic it y o f m igration with respect to the
wage change in a P la in s ' SEA is approximately the percent increase (or
decrease) in the amount of m igration i f the wage changes by I percent
\
while the remaining determinates are held a t a constant le v e l.
Given the d e fin itio n o f the estimated c o e ffic ie n ts , they now can
be applied m eaningfully to the m igration streams.
purposes the wage change v a riab le (CW1. ) is used.
For illu s t r a t iv e
The dependent v a riab le
fo r the in tra re g io n a l m igration model is calculated as the number of
people from an SEA(i) to an SEA(j) in the Plains region between 19651970 (M ^ j).
In order to simulate the impact of a change in wages on
m igration , the estimated c o e ffic ie n t must be applied to each m igration
stream, and f i n a l l y summed over these streams.
Thus, the to ta l change
in a m igration stream can be expressed as follow s:
ACW.
bI t CWT
1, 2, . . . , n
n = number of observed
migration streams
That is , each m igration stream from i to j increases or decreases by the
same percent dependent upon the e la s t ic it y between m igration and wages,
and the percent change in wages.
Thus, the absolute change in m igration
is conditional on the m igration streams between an SEA(i) and a ll other
S E A (j)'s .
The same formula is applied to the in terre g io n a l models.
-92The estimates associated with the wage change v a ria b le are given
in Tables V -I and V-2.
The estimated e la s t ic it ie s are -0 .3 5 and 1.68
fo r in t r a regional outm igration and inm igration re s p e c tiv e ly .
The
e la s t ic it ie s associated with in terre g io n a l outm igration and inmigra­
tio n are -0 .8 7 and 3 .1 3 .
The to ta l change in population due to migra­
tio n from a I percent increase in the wage change a t an SEA(i) can be
estimated by the follow ing re la tio n s h ip s :
= -.0035%M.j
j = I , 2, . . . , n
%AMj. = . 0168% .
j = I , 2, . . . , n
JaOM1J = -.OOSTjOMi j
j = 1, 2 .......... m
JAlMji = .0313JIMj.
J
J
j = I , 2, . . . . m
'
Therefore, from a I percent increase in the wage change each in tr a regional outm igration stream w ill decrease by 0.^35 percent, and each
inm igration stream w ill increase by 1.68 percent.
A decrease of 0.87
percent and an increase o f 3.13 percent is also expected fo r each in t e r ­
regional outm igration and inm igration stream in response to the increase
in wages.
However, the magnitude of the change in m igration is condi­
tio n a l
on the number o f people
in
each stream between
SEA(j)
or a s t a t e ( j) from 1965
to
1970.
anSEA(i) and an
-9 3 -
Lim itations o f the Models, Data and Estim ates.
There are several lim ita tio n s a ttrib u te d to the models and data
(th e re fo re the estim ates) which should be discussed before any conclu­
sions are made.
The major problems are:
(a) prediction .of to ta l migra­
tio n , (b) a conclusive generalized m igration model, (c ) disaggregation
o f the m igration data, and (d) v a ria b le measurement, e s p e c ia lly the
wage v a ria b le .
The study re la te s c e rta in economic conditions to Northern Great
Plains m igration.
However, because the models were not able to explain
much o f the 1965-1970 m igration , p redictio n of to ta l fu tu re migration
is lim ite d .
During the conception o f the p ro je c t, the influence of
the m ilit a r y complex in the region was considered m inim al.
A control
fo r these complexes may increase the p re d ic tiv e power o f the models,
and should be considered in any fu rth e r research.
Furthermore, the
attempt to r e la te d iffe r e n t m igration e la s t ic it ie s with respect to
(
economic conditions fo r in t r a regional and in terreg io n al m igration by
disaggregating the data may have also decreased the to ta l p re d ic tiv e
power o f the models.
F in a lly , i t should be noted th a t estimates a rrived
a t by normal le a s t squares regression analysis lose confidence in pre­
d ic tin g m igration as the v a ria b le changes beyond it s mean value.
There­
fo re , caution should be taken when projecting the impact of any one
explanatory v a ria b le on m igration.
-9 4 -
Another problem o rig in ated from the data used in the study.
The
models had to be modified because of apparent m ulticol lin e a r it y in the
-I
data.
This was d e a lt with extensively in Chapter IV .
-
The re s u lt is
th a t these constrained models may have increased s p e c ific a tio n e rro rs .
H yp o th e tic a lly, i t was considered important to have each m igration
stream explained by a ll economic conditions a t o rig in and d e s tin a tio n ,
which was not the case in the fin a l a n alysis.
Also, the d ire c t measure­
ment o f the economic variab les in T965, 1970, or 1965-1970 may lim it
the in te rp re ta tio n and use o f the e la s t ic it ie s fo r explaining migra­
tio n from 1965 to 1970.
Conditions may have been much d iffe r e n t fo r
interm ediate m igrants, thus the reasons fo r t h e ir movement may not have
been measured.
Testing the gain in inform ation by disaggregating the data (s ig n i­
fic a n t d ifferences in the estimates of each model) should also be con­
sidered.
The m igration m atrix was not p a rtitio n e d in th is study in such
a way as to allow a te s t o f disaggregated models.
Thus, the study
cannot conclude th a t separate analysis o f in tra re g io n a l and in t e r ­
regional m igration fo r the Northern Great Plains is appropriate.
F in a lly , wages are not adjusted to in dicate the real wage o f a
lo c a lit y , only a nominal wage is used.
When adjustments are not,made
fo r the cost o f liv in g and in fla tio n among lo c a lit ie s , the wage as
measured may not r e f le c t the incentive fo r a migrant to move.
D isparities
in these rates across lo c a lit ie s could r e s u lt in l i t t l e m igration in to
-95absolute high average wage areas having a t the same time high rates
of in fla t io n .
Thus, the re la tio n s h ip between m igration and wages
would tend to be negative or in s ig n ific a n t.
Conclusions
Based on the data and a n a ly tic a l technique used in the study, i f
po licy makers are concerned with the a v a ila b ilit y of labor in the
Northern Great P la in s , there are various avenues th a t can be taken.
People can e ith e r be induced to remain in the region or a ttra c te d to
i t from the contiguous states outside the region.
In order to r e d is t r i­
bute the population w ith in the Plains ( in t r a regional m ig ra tio n ), wages
in those SEA's in need o f human c a p ita l must increase.
A d d itio n a lly ,
those SEA's experiencing a decrease in subemployment rates w ill a t tr a c t
addition al people from the P lains.
To induce migrants to remain in the
region, employment and wages can be increased.
The e ffe c t of job oppor­
tu n itie s outside the region may decrease the a v a ila b ilit y o f labor by
increasing outm igration, but the impact is sm all.
People w ill enter the
Plains region as wages increase, but w ill consider otherwise i f high
subemployment rates p e rs is t, holding a ll else constant.
Furthermore,
p o ten tial migrants w ill m igrate to the region as economic opportunity
declines in the contiguous states outside the region.
The fa c t th a t fo r in trareg io n al m igration wages a t the o rig in do
not a ffe c t the outm igration flow is only a s t a tis t ic a l r e s u lt.
In
-9 6 -
theory, th is and other in s ig n ific a n t facto rs do influence m o b ility .
E ith er a d iffe r e n t methodology or an, improved surrogate in dicating
expected increase in fu tu re income may prove to be s t a t i s t ic a l ly s ig n i­
fic a n t.
However, the conditions above do extend some inference about
th e ir influence on the m igration streams.
In examining Tables V -I and V-2, the c o e ffic ie n ts associated with
wage change in the Plains in d icate th a t the separate m igration flows
respond d if fe r e n t ly .
To increase the in terre g io n a l outm igration flow
by the same percent as other flo w s, wages outside the region would have
to increase by a greater percentage.
Wages would not have to increase
as much to induce migrants to the Plains region.
This indicates th a t
the Plains population may have a regional preference and receive addi­
tio n a l benefits in the form o f geographic ren t by liv in g in the Northern
Great P lains. ,
As fo r distance between o rig in and d e s tin a tio n , the re s u lts tend
to support the contention th a t .in te rn a l migrants distin g u ish between
a lte r n a tiv e locations and p re fe r lo c a lit ie s close to th e ir o rig in s .
For long distance flow s, the migrant does not distin g u ish among a lte rn a ­
tiv e destinations by distance.
Long distance outm igration is more
responsive to population size than the other two streams.
In traregion al
and in terreg io n al inm igration respond in r e la tiv e ly the same magnitude.
In comparing past m igration a t d e s tin a tio n , the long distance moves
respond more, as expected.
Since past m igration is the main information
-97source fo r current migrants over long distances, the e ffe c t of past
m igration should be g re a te r.
Over short distances, inform ation about
a lte r n a tiv e destinations is e a s ily obtainable through conditions in
the market place.
Recommendations
Economic theory has established th a t c e rta in conditions influence
the m o b ility of people.
To analyze th is m o b ility a wide v a rie ty of
methods have been proposed and implemented.
In the c la s s ic a l context,
wage d if fe r e n t ia ls among lo c a lit ie s is the basis fo r the geographic movement of people.
In a more contemporary framework addition al incen­
tiv e s have been introduced.
In e ith e r case the methodology must
incorporate, as much as possible, the behavioral patterns and preference
functions of those migrants being studied.
The ideal approach would be
the d ire c t measurement of each in divid ual m igrant's cost and b en efit
from making a move.
In th is manner not only would the determinants
o f m o b ility be improved, but addition al information about the character­
is tic s of the migrant streams could be gained.
Another methodological problem is the use of e ith e r net m igration
or gross m igration as the m igration flow to be studied.
The importance
of lin k in g p a rtic u la r economic conditions to separate flows necessitated
the use o f gross m igration i,n th is study.
However, i f the accuracy of
projecting the to ta l net flow a t a lo c a lit y is the major concern then
net m igration may prove b e n e fic ia l.
-98FinalTy, in th is study the supply and demand fo r labor (people) in
an area is analyzed through m igration.
The re la tio n s h ip between wages
and market conditions' in an area could be linked without considering
the o rig in or destinatio n o f the labor supply.
That is , the analysis
may be improved by exclusively incorporating local area supply and
demand conditions a t given points in tim e.
APPENDICES
-100-
APPENDIX A
NUMBERED SEA'S AND STATES USED IN STUDY
L
-101State Economic Area
Number
N
Montana (2)
1
Montana (3)
2
Montana (4)
3
North Dakota ( I )
4
North Dakota (2)
5
North Dakota (3)
6
North Dakota (4)
7
North Dakota (5)
8
South Dakota ( ! )
9
South Dakota (2)
10
South Dakota (3)
11
South Dakota (4)
12
South Dakota (5)
13
Wyoming (2)
14
Colorado (3)
15
Colorado (4)
16
Colorado (5)
17
Colorado (B)
18
Colorado (D)
19
Nebraska ( I )
20
Nebraska (2)
21
Nebraska (3)
22
Nebraska (4)
23
Nebraska (5)
24
Origins and
►Destinations
In traregion al
Model
Origins (desF t i nation)
In terregion al
& Aggregated
Models
-102-
S ta te s
Alabama
A riz o n a
Arkansas
Cal i fo rm " a
C o n n e c tic u t
Delaware
F lo r id a
G eorgia
I llin o is
In d ia n a
Iowa
West V ir g in ia
Kentucky
L o u is ia n a
Ma i ne
M aryland
M assachusetts
M ich iga n
W ashington
M is s is s ip p i
M is s o u ri
W iscon sin
Nevada
New Hampshire
New J e rs e y
New Mexico
New York
N o rth C a ro lin a
Ohio
Oklahoma
Oregon
P e n nsylvania
Rhode Is la n d
South C a ro lin a
Tennessee
Texas
Utah
Vermont
V ir g in ia
Montana
N o rth Dakota
South Dakota
Wyoming
Colorado
Ida ho
Nebraska
Kansas
M innesota
Number
1
■x
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
O rig in s
r ( d e s tin a tio n s )
I n t e r r e g io n a l
Model
r
O rig in s
( d e s tin a tio n s )
A ggregated Model
-103-
APPENDIX B
ESTIMATES FOR AGGREGATED MIGRATION
MODELS AND WAGE MODELS
I
i
I
APPENDIX TABLE B - I .
OUTMIGRATION MODELS, AGGREGATED DATA 1965-1970, NORTHERN GREAT PLAINSa
Items and
Independent Variables
In itia l
R estricted Model
B
tC oeff.
value
1.63
-0 .0 5
0.01
-0 .1 9
-0 .5 2
0.35
-0 .1 3
0.82
12.74
- 1 .06
0.36
-2 .9 0
-0 .9 7
3.71
-3 .7 8
21.78
1.75
-0 .0 9
-0 .0 4
-0 .3 2
-0 .6 8
0.52
-0 .1 4
0.83
13.23
-2.12
-1 .0 4
-5.07
-1.21
5.78
-4.13
28., I 5
Economic Conditions a t D estination:
Employment Change, 1965-1970
0.20
Wage Change, 1965-1970
-0 .8 2
Income in 1965
2.50
Subemployment, 1970
0.16
3.52
-2.71
5.41
2.24
0.09
Economic Conditions a t O rigin:
Employment Change, I 965-T970
Wage Change, 1965-1970
Income in 1965
Subemployment, 1970
-4 .8 6
-1 .1 5
-4 .8 8
1.62
S ta tis tic s :
In te rc e p t
Degrees o f Freedom Residual
Standard Error
-0 .2 3
-0 .3 0
-1 .5 0
0.41
-8 .1 7
135.00
I,
0.2057
Final
Restricted Model
B
tC oeff.
val ue
1.71
18.32
-
-3.73
-8 .9 3
0.86
41.95
2.06
0.02
0.65
-0 .0 7
-0 .4 7
-1 .9 6
-1.71
-0.17
- 1 .37
-7 .7 2
-7 .6 5
0.89
3.45
1.09
-6 .8 8
319.00
I,
0.2213
aTwo-Stage le a s t squares with data in double logarithm ic form.
nearest fra c tio n of hundreds.
^The t-v a lu e associated with a .05 sign ificance level is 1.645.
•
4.31
0.04
I , 144.00
0.2383
C oefficients rounded to
-104-
Nonmarket Controls:
Population a t O rig in , 1970
Distance, O rigin to Destination
Town Sizes: D e s tin a tio n /o rig in
Farm Occupants, L970
Median Age a t O rig in , 1970
Distance la rg e s t a d j. tow n/origin
Town Sizes: Largest a d j./o r ig in
Past Migrants a t D estination, 1970
Full Model
B
t- k
C oeff.
value0
-105APPENDIX TABLE B-2.
INMIGRATION MODELS, AGGREGATED DATA 1965-1970,
NORTHERN GREAT PLAINS3
Items and
Independent Variables
Full Model
B
t- .
C o e ff.
value0
Nonmarket Controls:
Population a t D es tin atio n , 1970
1.77
-0.21
Distance, O rigin to Destination
Town Sizes: O rig in /d e s tin a tio n
0.07
Rural Population a t D estination,
1970
-0.61
Distance la rg e s t a d j. town/
destin atio n
0.75
Town Sizes: la rg e s t a d j . /
destin atio n
-0 .1 7
Past Migrants a t D estin atio n , 1970 0.59
10,57
-3 .8 7
1 .45
S ta tis tic s :
In terc e p t
Degrees of Freedom Residual
Standard Error
1.19
12.28
0.24
5.99
6.75
0.82
11.85
-3 .9 7
15.54
-0 .3 2
'0 .7 4
-8 .0 4
28.32
-0 .3 0
4.65
2.99
- L 91
2.23
2.60
9.87
8.84
-1 .2 9
-5 .2 3
0.31
3.25
-3 .9 4
Economic Conditions a t D estination:
-0 .0 2
Employment Change, 1965-1970
1.37
Wage Change, 1965-1970
1.66
Income in 1965
Subemployment, 1970
-0 .5 5
Economic Conditions a t O rigin:
Employment Change, 1965-1970
Wage Change, 1965-1970
Income in 1965
Subemployment in 1970
Final
R estricted Model
B
tC o e ff.
value
•
0.31
-2 .2 7
1.77
0.26
4.64
-7 .2 9
5,24
2.89
- 16.73
I ,136.00
0.3074
- 18.15
1,142.00
0.3453
aTwo-Stage le a s t squares with data in double logarithm ic form.
C o e ffic ie n ts rounded to nearest fra c tio n of hundreds.
h"he t-v a lu e associated with a .05 sig n ifican ce le v e l is 1.645.
-1 0 6 -
APPENDIX TABLE B-3.
CHANGE IN WAGE MODEL, NORTHERN GREAT PLAINS,
1965-1970
Items and
Independent Variables
Education in O rig in , 1970
Rural Population in O rig in , 1970
Median Age in O rig in , 1970
Income in 1965
Employment Change in O rig in , 1965-1970
Subemployment in O rig in , 1970
Distance la rg e s t a d j. tow n/origin
Town Sizes: O rig in /la rg e s t a d j.
Outmigration, 1965-1970
In m igration, 1965-1970
S ta tis tic s :
In terc e p t
Degrees o f Freedom Residual
Standard Error
R2
F S ta tis t ic
OLS
B
F .
S
ta
t.
C o e ff.
Two-Stagea
B
t- c
C o eff. Valuev
0.58
0.23
-0 .9 9
-0 .0 0
-0 .1 2
-0 .6 4
-0 .3 2
-0.01
-0 .0 0
0.00
0.24
0.00
-0 .4 2
0.03
-0 .1 2
0.29
-0 .2 6
0.02
0.01
0.01
258.02
70.13
125.95
0.00
385.242
123.570
578.50
3.85
1.54
0.06
5.76
1,141.00
0.08003
0.69
258.20
13.25
0.17
-4 .7 2
0.29
-17.58
6.97
-19.77
3.39
0.56
-0 .6 4
3.13
1,141.00
0.0069
•
aWage equation associated with Final Aggregated Outmigration and
Immigration Models. Data is double logarithm ic form, and c o e ffic ie n ts
rounded to nearest fra c tio n o f hundreds.
^The F s t a t is t ic associated with a .05 sig n ifican ce level is 3.84.
cThe t-v a lu e associated with a .05 sig n ifican ce level is 1.645.
-107APPENDIX TABLE B-4.
ORDINARY LEAST SQUARES FOR CHANGE IN WAGE MODEL,
NORTHERN GREAT PLAINS, 1965-1970
Items and
Independent Variables
Education a t O rig in , 1970
Rural Population a t O rig in , 1970
Median Age a t O rig in , 1970
Income in 1965
Employment Change a t O rig in , 1965-1970
Subemployment a t O rig in , 1970
Distance la rg e s t a d j. tow n/origin
Town Size: la rg e s t a d j. tow n/origin
Outmigration, 1965-1970
In m igration, 1965-1970
M ig ratio n, 1965-1970
S ta tis tic s :
In terc e p t
Degrees of Freedom Residual
Standard Error
F S ta tis tic
OLSa
B
C o eff. S t a t .c
OLSb
B
F
b
C o eff. S ta t.
0.24
0.00
-0 .4 2
0.02
-0 .1 2
0.28
-0 .2 6
0.01
-0 .0 0
-0 .0 0
0.02
0.21
-0 .3 6
0.04
-0 .1 2
0.12
-0 .3 3
-0 .0 9
0.32
6.70
3.28
0.03
67.47
1.03
120.70
-46.08
0.01
4.53
171.76
0.07
19.49
0.04
262.34
39.76
333.38
9.33
1.55
0.19
3.13
925.00
0.08254 ■
0.67
191.82
2.83
174.00
0.06482
0.80
79.98
aWage equation associated with Final Long Distance Inm igration and Out­
m igration Models. Data is in double logarithm ic form, and c o e ffic ie n ts
rounded to nearest fra c tio n o f hundreds.
bWage equation associated with Final Short Distance M igration Model.
Data is in double logarithm ic form, and c o e ffic ie n ts rounded to nearest
fra c tio n o f hundreds.
cThe F s t a t is t ic associated with a .05 sign ificance level is 3.84.
- 1 08-
APPENDIX C
STATISTICAL INFERENCE ON
COMPETING MIGRATION MODELS
-109Each o f the m igration models were broken down from the f u ll model
to models having fewer v a ria b le s .
This appendix in fe rs a s t a tis t ic a l
F te s t on the three separate models.
We w ill work back from the fin a l
models to the f u l l model 'to s t a t is t ic a lly te s t the influence o f addi­
tio n a l explanatory variab les on the^m igration mean.
The F s t a tis t ic
to be calculated is :
where:
N - Q
O - k
I_____
JZD
OO
CO
I----------
m
SSRg - SSRk
r Q -M -Q
Q = number o f independent variab les in the la rg e s t model;
k = number of independent variab les in the sm allest model; and
n = number o f observations.
The null hypothesis is then:
Ho :
bM
Bp - O
= Bk+2
The hypothesis is rejected when the calculated F s t a t is t ic is
g reater than the F d is trib u tio n value given by Fq_k n_g.
In tra re g io n a l M igration Models
Test between Final R estricted Model and I n i t i a l R estricted Model
1OU
I
-
'd CO
- .62"
12 - 10
.6 :
T
.01' [1721 _ .86
2
.37
.37
2.32
/
-110F2 iy i a t the I percent level of s ig n ifican ce = 4.61 using Vg = °°
Therefore the hypothesis th a t
Hq :
ai 2 := a14 = 0 is not re je c te d *
1
Test between I n i t i a l R estricted Model and Full Model:
F = 2.94
,
F5 ,l 66' = 3 - 02
Therefore:
H0 :
aIQri a l6 =: a4 = a13 = a17 = 0 1S not re je c te d .
Thus, the fin a l re s tric te d model does not s ig n ific a n tly reduce the
explanation o f v a ria tio n in m igration , but the s lig h t d iffe re n c e is due
to randomness.
Outmigration, In terre g io n a l Moves
Test between Final R estricted Model:
F = .17
F5 , 923 "
3.02
Therefore:
Hg:
a i = an
= aIO = a3 = a6 ^ 0 15 not rejected
*The variab les associated with these parameters are given in Tables
I V - I , IV -2 , and IV -3 .
-111Test between I n i t i a l R estricted Model and Full Model:
F = .31
F4,919 = 3,32
'Therefore:
Hq:
a8 = a7 = a-j5 = a-|g = 0 is not re je c te d .
Thus, the same is concluded fo r th is model.
In m igration, In terre g io n a l Moves
Test between Final R estricted Model and Full Model:
F = .37
F5,920 = 3 ' 02
Therefore:
bI * ba = b12 = a6 = bI 2 = a15 = 0 is not re je cte d .
LITERATURE CITED
/
113
LITERATURE CITED
Adams, Russel B ., and Barchent, John R. Trade Centers and Trade Centers
o f Upper Midwest, Upper Midwest Economic Study. Urban Report
No. 3. Minneapolis : U niversity o f Minnesota, Sept. 1963.
Alonso, W illiam . P olicy-O riented In terreg io n al Demographic Accounting
and a G eneralization o f Population Flow Models. Working Paper
No. 247, In s titu te o f Urban and Regional Development. Berkeley:
U niversity o f C a lifo r n ia , June-July 1974.
j
Anderson, T. W., and Rukin, Herman. "Estimation of the Parameters of
a Single Equation in a Complete System of Stochastic Equations,"
Annals Mathematical S ta tis tic s 20 (September 1949): 46-63.
Bechdolt , B. V ., and Niedereorn, J . H. "An Economic D erivation of the
G ravity Law of Spatial In te ra c tio n ," Journal o f Regional Science 9
(August 1969): 273-282.
Becker, G. S. "Investment in Human C a p ita l: A Theoretical A nalysis,"
Journal o f P o litic a l Economy, Suppl. Part 2 (October 1962): 9-49.
,______ .
Economic Theory.
New York:
A lfred A. Knopf, In c ., 1971.
Bender, Lloyd. Predicting Employment in Four Regions o f the Western
S ta te s . Tech. B ul. No. 1:529. U. S. Department o f A g ric u ltu re ,
Economic Research Service, Economic Development D iv is io n .
Washington, D. C ., Nov. 1975.
Bennion, E. G. The Cowles Commission, "Simultaneous Equation Approach,
A S im p lified Explanation," Review o f Economics and S ta tis tic s 34
(March 1952): 49-56.
Berry, R. A ., and S olig o 9 R. "Some Welfare Aspects o f In tern atio n al
M ig ratio n ," Journal o f P o litic a l Economy 77 (September 1969):
778-794.
Blanco, C icely. "The Determinants of In te rs ta te Population Movements,"
Journal of Regional Science (Summer 1963): 77-84.
Bluestone, Abraham. "Major Studies of Workers Reasons fo r Job Choice,"
Monthly Labor Review 78 (March 1955): 301-306.
Borts, George H ., and S te in , Jerome L. Economic Growth in a Free Market.
New York: Columbia U niversity Press, 1964.
~
114
Bowles, Samuel.
M o b ility ."
356-362.
"Migration as Investment Approach to Geographic
Review o f Economics and S ta tis tic s 52 (November 1970):
Dalsted, Norman L ., L e is t r it z , Larry F ., and Hertsgaard, Thor A. Energy
Resources Development in Northern Great Plains: A Summary of
Economic Impacts. V o l. 21. Fargo: North Dakota State U n iv e rs ity ,
June 1974.
Fabricant, Ruth A. "An Expectational Model o f M ig ratio n ," Journal of
Regional Science 10 (A p ril 1970): 13-34.
Federal Energy A dm inistration. P roject Independence. Stock No. 411800029. Washington, D. C.: U. S. Government P rin tin g O ffic e , 1974.
Federation o f Rocky Mountain S tates, In c. Energy Development in the
Rocky Mountain Region: Goals and Concerns. Feb. 1975.
Goldstein, G. S ., and Moses, L. W. "A Survey of Urban Economics,"
Journal o f Economic L ite ra tu re 162 (June 1973): 471-515.
Greenwood, Michael J . "Research on In tern a l M igration in the United
States: A Survey," Journal o f Economic L ite ra tu re (July 1975):
397-433.
______ . "A Simultaneous-Equations Model of Urban Growth and Migra­
tio n ," Journal o f the American S ta tis tic a l Association (Forthcoming
A r t i c le ji
Haggett, P eter. Locational Analysis in Human Geography.
Edward Arnold, 1965.
London:
Hamilton, Horace C. "Population Pressure and Other Factors A ffecting
Net Rural-Urban M ig ratio n ," Social Forces 30 (December 1951):
209-2T5.
Hansen, Nales M. Location Preferences, Migration and Regional Growth,
A Study of the South and Southwest United S ta te s . New York:
Praeger Publishers, 1973.
I
Hoover, E. M. "The Concept o f a System o f C itie s ," Economic Development
and Cultural Change (T955): 196-198.
115
_________ . An Introduction to Regional Economics.
Knopf, In c ., 1971.
New York:
A lfred A.
H uff, D. L. "A Topographic Model of Consumer Space and Preferences,"
Papers and Proceedings. Regional Science Association (1960): 159173.
Huffman, L. "Survey of Coal A v a ila b ilitie s by S ulfur Content," Report
to the Environmental Protection Agency. The M itre Corporation,
MTR-6068, May 1972.
Is a rd , W alter. Introduction to Regional Science.
N .J .: P re n tic e -H a ll, In c ., 1975.
Englewood C lif f s ,
_________ , and Coughlin, Robert. Municipal Costs and Revenues Resulting
from Community Growth. American In s titu te o f Planners. W elIesly:
Chandler-Davis Publishing C o., 1957.
Kementa, Jon.
1971.
Elements of Econometrics.
New York:
The Macmillan C o.,
Lewis, Eugene P. "Empirical Least Squares Regression Models fo r Employ­
ment In - and O ut-m igration, and Income D is trib u tio n in the Northern
Great Plains o f the United S tates," M. S. th e s is . Montana State
U n iv e rs ity , 1974.
Lowry, Ira S. M igration and M etropolitan Growth, Two A nalytical Models.
San Francisco: Chandler Publishing C o., 1966.
Mazek, Warren. The E ffic ie n c y of Labor M igration w ith Special Emphasis
on Depressed Areas. Working Paper, Cur. 2. In s t it u t e fo r Urban
and Regional Studies. Washington U n iv e rs ity , June 1966.
M ille r , Ann R. "The M igration o f Employed Persons to and from Metro­
p o lita n Areas o f the United S ta te s," Journal of American S ta tis ­
tic a l Association (December 1967).
M ille r , Edward. "Is Out-M igration Affected by Economic Considerations?,"
Southern Economic Journal 39 (January 1973): 396-405.
Muth, R. F. "M igration: Chicken or Egg?," Southern Economic Journal
37 (January 1971): 295-306.
Okun, B. "Population M igration and S tate In e q u a lity : A Simultaneous
Equation Approach," Economic Development and C ultural Change
(January 1968).
116
P a rr, Joftn Bi; '!'Outmigration and the Depressed Area Problem," Land
Ecoriomics 26 (May 1966): 149-159.
-------T
;
" s '
P etto , Anthony.' "An Economic Analysis of M igration fo r the Ozarks
, Region, 1950-1970." Ph.D. d is s e rta tio n . U n iversity o f Missouri
a t Columbia, 1972.
,
P o lzin , Paul E. "Montanans on the Move," Montana Business Q uarterly 13
(Summer 1975).
______ . Water Use and Coal Development in Eastern Montana. Bureau
o f Business and Economic Research, Rpt. No. 57. Missoula:
U n iversity o f Montana, Nov. 1974.
Raimon, Robert L. " In te rs ta te M igration and Wage Theory," Review of
Economics and S ta tis tic s 44 (November 1962): 428-438.
Rand McNally Road A tla s , T970.
Rogers, A. B.. "A Regression Analysis o f In terreg io n al M igration in
C a lifo rn ia ," Review o f Economics and S ta tis tic s 53 (June 1971):
262-267.
;
Romans, J. I .
"Benefits and Burdens o f M ig ratio n ," Southern Economic
Journal 50 (January 1974).
Sjaastad, Larry.A . "The Costs and Returns o f Human M ig ra tio n ," Journal
o f P o litic a l Economy 70 Suppl. (October 1962): 80-93.
Smith, Joseph B. A L istin g of Proposed, Planned or Under Construction
Energy Projects in Federal Region V I I I . U. S. Bureau o f Mines,
The Subcommittee to Expedite Energy Development, August 1975.
S ta tis tic a l Package fo r the Social Sciences. SPSS Manual.
Version 4 .0 1 . New York: McGraw-Hill, In c ., 1975.
Xerox
Stroup, Richard, and Thurman, W alter. "W ill Coal G a s ific a tio n Come to
the Northern Great P lains?," Montana Business Q uarterly 14
(W inter 1976).
Tarver, James S.
207-213.
"Predicting M ig ra tio n ," Social Forces 39 (March 1961):
U. S. Department o f Commerce. Bureau of the Census. County Business
P atterns. Washington, D. C.: U. S. Government P rin tin g O ffic e ,
7965:
117
. County Business Patterns. Washington, D. C.:
Government P rin tin g O ffic e , 1970.
U. S.
____ . U. S. Census o f Population: M igration Between State
Economic Areas. Washington, D. C.: U. S. Government P rinting
O ffic e , 1970.
_________ . U. S. Census o f Population: State Economic Areas.
Washington, D. C.: U. S. Government P rin tin g O ffic e , 1970.
_________ . LI. S. Census of Population: S tate of B ir t h .
D. C.: U. S. Government P rin tin g O ffic e , 1970.
Washington,
. U, S. Census o f Population: LI. $. Summary. Part I ,
S e c tio n 'll Washington, D. C .: U. S. Government P rin tin g O ffic e ,
1970.
LI. S. Environmental Protection Agency. O ffic e of Research and Develop­
ment. D ra ft F ir s t Year, Progress Report o f Technology Assessment
of Western Energy Resource Development, V o l. I I - B , Supporting
A nalysis. Washington, D. C.: LI. S. Government P rin tin g O ffic e ,
1976.
Vinning, R. "A D escription of C ertain Special Aspects o f an Economic
System," Economic Development and C ultural Change 3 (1955): 147195.
-Vv .
■■ ' ;■
' ■
.
i' I:
%
Zr
f
MONTANASTATEUNIVERSITYLIBRARIES
N3T8
Om25
c o p .2
DATE
O'Meara, David M
Economic m odels o f
in t r a r e g io n a l and i n t e r ­
r e g io n a l m ig ra tio n in
th e N orthern Great P la in s
r e g io n
ISSUED
TO
"ZWEfeKt
i
Z
Z 7 /
Z
A /S '
XL. J