to nonmetropolitan areas typically leaves workers farther from their jobs or closer. ... tion out of metropolitan areas yields a more energy-intensive configuration .... of the pairwise differences is statistically significant): They wind up commut-.
Population Research and Poficy Reviev~; 2 (1983) 189-206
189
Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
Is Population Decentralization Lengthening Commuting Distances? ' PETER A. MORRISON and ALLAN F. ABRAHAMSE
The Rand CmToration, 1700 Main Street, Santa Monica, California 90406, U.S.A.
Abstract This study questions the conventional wisdom about how commuting distances change when workers migrate from metropolitan to nonmetropolitan areas. It does not appear that decentralization yields a more energy-intensive configuration of residences and job locations: we find no indication that this migration lengthens the aggregate distance that workers commute. Some migrants do commute long distances, but their numbers are offset by many more who end up closer to their jobs. Our findings relate to two contrasting views ("sprawl" and "nucleation") of how workers are becoming repositioned in relation to their jobs as settlement patterns change; the latter view appears more realistic. We briefly discuss policy implications pertaining to alternative transportation modes for commuting, setting priority travel needs in an energy emergency, and telecommunications as a substitute for commuting.
I. Introduction
Over the last decade, national settlement patterns have changed considerably as people have moved out of large metropolitan centers. The dispersal of population to smaller urban centers and rural areas has generated an assumption that workers who move their households away from the m e t r o p olis typically engage in lengthy commutes to their jobs. This possibility raises a variety of social; economic, and political questions. For example, will this outward dispersal induce greater reliance on private automobiles and heighten consumption of energy for commuting? Will it leave commuters more vulnerable to any future fuel-supply "squeeze"? If, as some assume, many workers who move to nonmetropolitan areas retain metropolitan jobs, then continued outward dispersal may well mean (1) longer-distance commuting to work on average, and (2) greater reliance on private automobiles (since most nonmetropolitan areas lack transportation alternatives to the family car). If, as others claim, workers generally 0167-5923/83/503.00
9 1983 Elsevier Science Publishers B.V.
190 change jobs as well as residences when they leave the metropolitan areas, dispersal may reduce aggregate commuting distances if it promotes clustering nearer to jobs in satellite employment centers. The result will be "nucleation" rather than "sprawl." No one really knows whether contemporary migration from metropolitan to nonmetropolitan areas typically leaves workers farther from their jobs or closer. To find out, we analyzed data from a national sample of employed heads of families. These data showed the changes in the job-related commuting distances for people who made particular types of intercounty moves during the 1970s. These moves included migration from one metropolitan county to another, from one nonmetropolitan county to another, from a metropolitan to a nonmetropolitan county, and vice versa. Although our results have limitations, which we discuss later, they suggest a tentative answer: A small fraction of workers who migrate from metropolitan to nonmetropolitan areas do become long-distance commuters--if they settle in areas adjacent to the metropolis--but their numbers are offset by a larger fraction who end up closer to work. Overall, then, population dispersal is not lengthening the aggregate distance that workers commute. This finding is noteworthy because it casts doubt on at least one tenet of the conventional wisdom about settlement patterns--the assumption that migration out of metropolitan areas yields a more energy-intensive configuration of residences and job locations. If our findings are validated through independent replication, they may have a variety of policy implications (discussed below in Section V) pertaining, for example, to alternative transportation modes for commuting, telecommunications as a substitute for travel to work, and setting priority travel needs in an energy emergency. In the next section, we briefly review the findings from previous research about the link between settlement patterns and commuting distances and formulate our specific research questions. Section III describes the data set and our methodology. In Section IV, we present the findings of our study. Section V outlines the implications of those findings.
ii. Findings from Previous Research
Many demographic factors determine personal travel patterns, including where people live, where they work and how they get there, the household units into which they are grouped, and a household's stage in the family life cycle (Abrahamse and Morrison, 1981; Allen and Edmonds, 1980; Fritzsche, 1981; Garkovich, 1982; Hanson and Hanson, 1981; Keyes, 1976, 1980, 1982; Small, 1980; Sosslau, 1980; Spielberg and Andrle, 1982; Zelinsky and Sly,
191 1981; Zimmerman, 1980). However, the links between these demographic factors and personal travel patterns are poorly understood, and data for studying them are sparse (Hirst, 1980; Hoch, 1981; Saltzman and Newlin, 1981). In this study, we focus on the distance between where people live and where they work--and how that distance may change when they migrate out, or completely away, from metropolitan areas. Migration (which we define as a move from one county to another) is transforming U.S. settlement patterns in several ways (Fuguitt et al., 1981; Long, 1981; Long and DeAre, 1982): Net migration flows are causing large metropolitan areas to lose, and smaller metropolitan areas (particularly those of less than 500,000 residents) to gain, population. - - Decentralization is creating widening zones of growth in nonmetropolitan counties that adjoin metropolitan areas. Population is clustering farther out in small, freestanding nonmetropolitan cities and towns beyond commuting range of metropolitan areas.
-
-
-
-
These changes are repositioning many workers in relation to their jobs. However, as the Introduction noted, opinions differ on how these changes in settlement patterns may affect personal travel patterns, especially commuting distances (Bowles and Beale, 1980a, 1980b; Garkovich, 1982; Hoch, 1981; Keyes, 1980; Long, 1981; Rickaby, 1981; Small, 1980; Zelinsky and Sly, 1981). Beneath these differences, we can discern two implicit geometric images of settlement whose validity we shall partially test. One possibility is that when people migrate from metropolitan to nonmetropolitan counties, their distance to work lengthens considerably, either because they retain metropolitan employment links or because jobs and people may be more thinly spread out in many nonmetropolitan settings. This image of "metropolitan sprawl" envisions some of the migrants commuting long distances back to jobs in the metropolitan areas they have left [2], and envisions also long-diStance commuting becoming increasingly common in nonmetropolitan areas (Long, 1981). Another possibility is that the contemporary pattern of settlement in nonmetropolitan areas outside commuting range of metropolitan centers may result in shorter commuting distances (Zelinsky and Sly, 1981). In this image of "nucleation," settlement patterns are evolving toward a polycentric form, consisting of many autonomous "miniature cities" beyond the fringes of metropolitan areas. This settlement pattern enables people to commute from home to a job nearby. To date there is no conclusive evidence for either the sprawl or nucleation effects (either or both of which could operate). However, Bowles and Beale (1980a) show that more than four-fifths of workers who leave a metropolitan area for a nonmetropolitan one also have severed their former working ties
192
within five years and taken jobs in the nonmetropolitan sector, a finding that would argue more for nucleation [3]. Their studies also show that nonmetropolitan residents travel a shorter median distance to work than their metropolitan counterparts do (4.6 vs. 7.6 miles). However, the distributions around these medians are noticeably different. Figure 1 displays certain of their findings, which indicate that the issue is more complicated than the medians alone might suggest. Among workers who live outside metropolitan areas, a disproportionate fraction are concentrated in both the short-distance (under 5 miles to work) and long-distance (over 30 miles) categories, relative to workers living inside metropolitan areas. However, the few miles traveled by the many commuters who work close to home in nonmetropolitan areas evidently more than counterbalance the many miles traveled by the few who work far from home. Although this result foreshadows one of our central findings, the data used in the Bowles and Beale studies are inherently limited: they cannot reveal how a given individual's distance from work changes when he migrates, for example, from a metropolitan to nonmetropolitan area. Thus, these data cannot disclose whether metropolitan out-migration is raising or lowering the medians shown in Fig. 1. To extend this line of inquiry, we address the following research questions: 1. How do workers' commuting distances change when they migrate from metropolitan to nonmetropolitan counties? In particular, does that change
40
//~ Percent
30
of workers
Workers living outside
///~, //////
//Workers living inside ~
~\\\
20 10
/
0
, Under 1
Median Outside: Inside: 4,6 7.6 , I il , 1-4 e m_95 10-19
\ ~ ~" , 20-29
30+
Distance traveled to work (miles) SOURCE:Adaptedfrom Bowlesand Beale(1980a) Fig. 1. Distribution of commuting distances for workers living outside and inside metropolitan areas.
193
depend on whether or not the latter are adjacent--that is, within commuting distance--to a metropolitan county? 2. Do commuting distances after migration vary by the type of intercounty move made? 3. Do the commuting distances of metropolitan-to-nonmetropolitan migrants differ from those of the nonmigrant residents they have joined?
III. Data and Methodology DATA SOURCE
Our analysis is based on data from the University of Michigan Panel Study of Income Dynamics (PSID). The PSID is a longitudinal survey of a national sample of over 5000 U.S. families interviewed annually since 1968. We used data from the 12-year file of the PSID (extending from 1968 through 1979) and restricted our focus to workers who drove to work [4]. The PSID has two important advantages for this study: (1) its sample size is sufficient to generate adequate numbers of metropolitan-to-nonmetropolitan migrants; (2) it provides longitudinal coverage, allowing comparisons of individuals' distances to work before and after moves. From the original sample of 4802, we selected the 3013 family heads in the first wave who were members of an interviewed family in the twelfth wave also. This sample comprises 63 percent of 1968 family heads; it excludes dropouts from the PSID and newly-formed families that "split off" from original families. Based on year-to-year changes in the PSID county-of-residence code, we defined four types of intercounty migration: A move from one metropolitan county to another, including both intermetropolitan moves and those intrametropolitan moves that merely cross a single county line. -Metro-nonmetro: A move from a metropolitan to a nonmetropolitan county. -Nonmetro-metro: A move from a nonmetropolitan to a metropolitan county. --Nonmetro-nonmetro: A move from one nonmetropolitan county to another.
--
Metro-metro:
We further distinguished moves to nonmetropolitan counties according to whether or not the counties were adjacent to a metropolitan area. The PSID data set and our analytical procedures have several important limitations. First, because our sample contains only intercounty moves, we
194 cannot examine potential changes associated with intracounty decentralizat i o n - t h a t is, movement outward from urban centers to destinations within the same county. Second, we restricted our analysis to family heads who reported being employed and driving to work before and after a move. This leaves out heads who were temporarily unemployed or who commuted by some other mode [5]. Finally, the PSID is not a representative sample of all U.S. families. Rather, it is highly stratified and initially oversampled families with low incomes. Nevertheless, our sample represents a large, important group--stably employed workers who drive--somewhat overrepresenting the lower end of the income scale. We have no reason to suppose that our results would be drastically different if these limitations could be overcome.
IV. Findings The findings we present here derive from selected comparisons between (1) the commuting distances of metropolitan-to-nonmetropolitan migrants, before and after their moves, (2) commuting distances after migrations of different types, and (3) commuting distances of nonmetropolitan residents and of metropolitan-to-nonmetropolitan migrants joining those residents [6]. These comparisons will test the plausibility of the views that outward migration from metropofitan areas produces "sprawl" or "nucleation" patterns of settlement. CHANGES IN COMMUTING DISTANCES
How do workers' commuting distances change when they migrate from metropolitan to nonmetropolitan areas? Data on mean commuting distances suggest that they change very little. Within the pairs of pre- and post-migration distances shown in Fig. 2, the differences are modest, and most are not statistically significant [7]. The metropolitan-to-nonmetropolitan migrants end up driving about the same average distance after the move as before, with a faint indication that whatever change occurs may depend on whether or not the nonmetropolitan area chosen is adjacent to a metropolitan area. POST-MOVE COMMUTING DISTANCES BY TYPE OF MOVE
Do commuting distances after migration vary by the type of move made? The answer to this question will offer clues to how the common features of destinations shape commuting distances. Figure 3 shows the mean post-move commuting distances for several types of migrants. Migrants from metropolitan counties evidently end up closest to work (6 miles) when they move
195 0
5
10
15
10
15
D ESTI NATI ON 10 mi
Before:
All nonmetro counties
~
After:
9 mi
Before:
Adjacent county
10
After:
~ Nonadjacent county
0
6
I
mi
11 mi
Before: 8 mi
After:
~
~
I
mi ~-- I
5 Miles
SOURCE: Panel Study of Income Dynamics data
Fig. 2. Averagepre- and post-move commutingdistance for metropolitan-to-nonmetropolitan migrants.
to nonmetropolitan counties not adjacent to any metropolitan area. They end up farthest from work (14 miles) when they move to another metropolitan county (the difference between these two extremes is statistically significant). Nonmetropolitan migrants establish a similar pattern (although none of the pairwise differences is statistically significant): They wind up commuting 7 to 8 miles after moving to another nonmetropolitan area and 12 miles if they move to a metropolitan area. C O M M U T I N G DISTANCES OF MIGRANTS A N D RESIDENTS
Do the commuting distances of metropolitan-to-nonmetropolitan migrants differ from those of the residents they join [8]? In particular, are these migrants likelier than the residents to be long-distance commuters, suggesting that
196 10
15
M I G R A N T TYPE Metro-to9 mi All nonmetro 11 mi
9. . v
Adjacent 6 mi Nonadjacent
I I
-I
14 mi Other metro
Nonmetro-to7 mi All nonmetro 7 mi Adjacent
8 mi Nonadjacent
-I -I -I
12 mi Metro
10
15
Miles SOURCE: PanEl Study of Income Dynamics data
Fig. 3. Comparative commuting distance following move, by migrant type.
they have retained their metropolitan-area employment links? Figure 4 presents the relative commuting distances of residents and metropolitan-to-nonmetropolitan migrants for various types of counties. The key comparisons reveal no significant difference. Metropolitan migrants to nonmetropolitan counties are no farther from work, on average, than the nonmigrant residents they join--which they should have been if very many of them had actually retained jobs in their former metropolitan locations. As a more definitive test, we compared the distributions of distances to work for (1) the metropolitan migrants to adjacent nonmetropolitan counties and (2) the residents they join. Here again, evidence of long-distance commuting is absent: these distributions (shown in Fig. 5) are not significantly different from each other, and those workers who do commute long distances (in excess o f 15 miles) are not disproportionately newcomers.
197 10
5
I
DESTINATION
15
Residents: 10 mi All nonmetro counties
Migrants:
9 mi
Residents: Adjacent
-I
12 mi
Migrants: 11 mi
-
I
I
Residents: 9 mi
-I
Nonadjacent ~
0
Migrants: 6
miI I 5
10
15
Miles SOURCE: Panel Study of Income Dynamics data
Fig. 4. Comparativecommuting distance for residents and metropolitan migrants to other counties, by type. Thus far, the analysis has involved comparisons of means and distributions of distances to w o r k - - a n d as we have seen, they vary little for migrants and residents. However, when we compare the proportions of cases registering increases or dezreases, and the mean distances of those changes, we can find important differences between migrants and residents, as shown in Fig. 6. Among residents of all nonmetropolitan areas (see third bar from left), changes from year to year display a rough symmetry; there is about a 50-50 chance that residents will change their distance to work, and these changes average (plus or minus) 6 to 7 miles. Some of these changes may reflect year-to-year variations in respondents' reports, but the rest will be genuine differences resulting from changes in workplace or intracounty changes of residence. In contrast, migrants to all nonmetropolitan areas (see fourth bar
198 60
45
Percent
nigra;t;)
of
workers 30
Other residents / (n = 113)
15
~
/ ~
\X// I
I
I
Under 6
7-15
16-24
SOURCE:
/
25+
Distance traveled to work (miles) Panel Stud'/ of Income Dynamics data
Fig. 5. Distributions of commuting distance for metropolitan migrants to, and other residents of, adjacent nonmetropolitan counties.
from left) show a distinctly asymmetric pattern: They are far likelier to shorten than to lengthen their commuting distance (60 percent vs. 28 percent); but, when they do lengthen it, the increase is a hefty 13 miles--twice that of the residents. Because so few migrants end up 13 miles farther from work on average and so many end up 7 miles closer, the effects offset each other. This offsetting effect produced our earlier observations that metropolitan-to-nonmetropolitan migrants travel about the same average distance before and after their move (Fig. 2) and about the same as the residents they join (Fig. 4). The data in the right-hand panel of Fig. 6 also show an important difference between nonmetropolitan counties that are adjacent or nonadjacent to a metropolitan area. In both types of counties, the incoming migrants are far likelier to shorten than to lengthen their commute. However, the asymmetry of the change in commuting distance--larger increases than decreases--is confined to those migrants settling in adjacent counties. Their average increase is 17 miles (versus 10 miles for the residents) whereas their average decrease is only 8 miles (virtually identical to that of the residents). According to this analysis, then, metropolitan-to-nonmetropolitan migration typically leads to shorter commuting, consistent with the ~
199
Nonmetro destinations : Adjacent Nonadjacent counties counties
All destinations: Metro
Nonmetro
60% +12 mi
40% % Increasing commute 20%
0
+6 mi
+17 mi
+13 mi
l-ii
+Tmi
+
t-H
~iiiiiiJi
-9mi
~ -4 mi
-10 mi
'iiiiiiiiiiiii iiiiiiiiiiiiii
]
:::::::::::::
60%
~
i i i i!i !i ~
20% % Decreasing commute 40%
~
+10 mi
i]iii!ii
-8 mi -7 mi -5 m i - -
80% Residents Legend
+6mi I -3mi
Migrants
+4mi~ Average //mileage
~
U
/
change
-5 mi SOURCE: Panel Study of Income Dynamics data
Fig. 6. Changes in commutingdistance for metropolitanmigrants and other residents at destination.
200 image. However, a minority of the migration into counties adjacent to metropolitan areas does lead to much longer commuting, consistent with the "sprawl" image.
V. Implications of Findings Using a small but well-defined sample of intercounty migrants, we have examined how people's commuting distances change when they migrate from metropolitan to nonmetropolitan areas. We find no indication that such migration is lengthening the aggregate distance that workers commute. Although a minority of such migrants do become long-distance commuters, their numbers are offset by a considerably larger fraction who end up closer to their jobs. Our findings relate to the two contrasting views of what may be happening in the nation's nonmetropolitan areas. One is that when metropolitan residents disperse into the adjacent nonmetropolitan fringe, their jobs and homes become more separated, lengthening the journey to work. We find evidence of this "sprawl" effect, but it is atypical. Another view is that nonmetropolitan communities enable workers to live closer to their jobs in these satellite employment centers, thereby shortening distances to work. This "nucleation" image fits more closely with the realities of how metropolitan-to-nonmetropolitan migrants typically position themselves in relation to their jobs. Inherent limitations of our data make these results tentative and subject to important caveats. The broadest importance of our findings is that they cast doubt on one tenet of the conventional wisdom about the nation's changing settlement patterns--the assumption that migration out of metropolitan areas is yielding a more energy-intensive configuration of residences and job locations. They also cast a new light on other assumptions that pertain to specific areas of policy: The market for electric cars--electric vehicle use is limited by the distance such vehicles can drive before recharging. Until now, these "to-and-fromwork" cars have been considered suitable only for short-distance intracity commuting within large metropolitan centers. However, their range may be well suited to the daily driving requirements of nonmetropolitan households if most workers in nonmetropolitan areas live close to their jobs. It would behoove U.S. automakers and the Department of Energy's Office of Vehicle and Engine Research and Development to consider the electric car's potential for small cities and towns as well as for the large metropolitan areas. Setting priority travel needs--in planning for a possible disruption of oil supplies, federal emergency planners must develop fuel allocation criteria
201
based on some conception of priority travel needs. It is generally assumed that priority would be given to the travel needs of police, fire fighters, physicians, other medical personnel, and long-distance commuters who have no alternative transportation. Nonmetropolitan residents have usually been put in that latter category. It is true that they log more total mileage for trips of all purposes than their metropolitan counterparts do. However, our findings indicate that going to and from work is not the reason why, which implies that their commuting needs may merit no higher priority than the commuting needs of metropolitan workers. Substituting telecommunications for travel--the assumption that people commute farther as they move away from the cities has fostered a popular scenario: a future in which some fraction of the work force substitutes telecommunications for daily travel to work. In this scenario, the "sprawl" created by migration to the rural countryside is ameliorated by bringing the job to the worker's residence rather than the worker to the distant job. Telecommunications may benefit the decentralization of work in various ways, but our findings indicate that overcoming increased commuting distances is not among them.
Notes 1 Revision of a paper given at the 1982 Population Association of America meetings, San Diego, 1982. We acknowledge research support by Center Grant P50-HD-12639 from the Center for Population Research, National Institute of Child Health and Human DeveloW ment, U.S. Department of Health and Human Services. 2 In 1975, the mean one-way commuting distance of nonmetropohtan residents who worked in metropohtan suburbs was 25 miles, in contrast to the national average commuting distance of 8.5 miles (U.S. Bureau of the Census, 1979). 3 Sly (1982) provides related evidence and an insightful discussion of the metropolitan setting. His evidence indicates that the more decentralized U.S. metropolitan areas have lower per-capita consumption of gasoline than the less decentralized ones. The implication here is that dispersal does not compel longer commuting and, therefore, more gasoline consumption. 4 For further details on sample design, interviewing and follow-up procedures, and response rates, see Institute for Social Research (1972). Our choice to restrict the sample to drivers was dictated by technical considerations. The survey includes the distance to work for all family heads, regardless of travel mode, for waves 4 through 12. In wave 1, distance was not recorded at all; in waves 2 and 3, it was recorded only for those who drove to work. Thus, limiting the analysis to workers who drove greatly simplified the analysis. However, it probably excludes workers very close to their jobs, who might walk, and those very far from work, who might carpool. Mode of travel to work is coded differently on different PSID waves but can be recoded consistently to distinguish those employed family heads who (1) drove alone or with a family member, (2) carpooled, (3) took public transportation, or (4) walked.
202 5 Nationally, 85 percent of all U.S. workers drive to their jobs; 65 percent do so alone (U.S. Bureau of the Census, 1979). 6 We constructed two files, one describing actual moves (the "migration file"), the other describing comparable time segments without moves (the "nonmigration file"). For the migration file, we selected only the first identifiable migration of each type for each family head. We limited the file further to moves for which the prior and subsequent survey waves showed the family head driving to work and recorded the distance driven. Under this selection process, one family head could generate as many as four moves (one each of the four types shown below). (As an illustration, a family might have made two moves between metropolitan areas and two moves between nonmetropolitan areas, but only the first move of each type would be included in the migration f i l e - - a total Of two rather than four moves.) This procedure yielded a file consisting of 314 migrations:
MIGRATION FILE Type of Move
Number
Metro-metro Metro-nonmetro Nonmetro-metro Nonmetro-nonmetro
173 65 38 38 Total 314
We next created the nonmigration file to enable comparisons between migrants' and nonmigrants' commuting distances. For each family head, we selected a pair of successive years (waves) at random and put them in the nonmigration file if they met the following criteria: (1) county of residence was unchanged, (2) the family head drove to work, and (3) distance to work was recorded. The resulting file contained 1077 pairs of consecutive years without migration (one for each family head). For our analysis, this file represents residents who drove to work but remained in a given metropolitan or nonmetropolitan area for two consecutive years:
N O N M I G R A T I O N FILE Type Metropolitan Nonmetropolitan
Number 829 248 Total 1077
For each type in the migration and nonmigration files, we computed the following statistics: DISTI: Mean commuting distance before the move or nonmove DIST2: Mean commuting distance after the move or nonmove AVGINC: Mean of all before-after increases in distance AVGDEC:Mean of all before-after decreases in distance %INC: Percentage of cases with before-after increase %DEC: Percentage of cases with before-after decrease. Table I (Appendix) shows these summary statistics for all migrants by type of move and for nonmigrants by location.
203 7 Tests of statistical significance were based on rank sum tests. We note only those differences that could be detected with significance of 5 percent or less. 8 The migrants, of course, become residents, but we use "resident" here to refer to nonmetropolitan family heads in our nonmigrant file, i.e., heads who have lived in the area for two consecutive years.
References Abrahamse, Allan and Peter A. Morrison (1981). "Demographic influences on energy use: an analysis based on the panel study of income dynamics." Unpublished manuscript. Allen, Edward L. and James A. Edmonds (1980). "Energy demand and population changes," in Population and Energy: Implications for Policymakers. Hearings before the Subcommittee on Energy Research and Production of the Committee on Science and Technology, U.S. House of Representatives: 11-28. Bowles, Gladys K. and Calvin L. Beale (1980a). Commuting Patterns of Nonmetro Household Heads, 1975. Washington, D.C.: Economics, Statistics, and Cooperative Service, U.S. Department of Agriculture; and Athens, Ga.: Institute for Behavioral Research, University of Georgia, cooperating. Bowles, Gladys K. and Calvin L. Beale (1980b). "Commuting and migration status in nonmetro areas," Agricultural Economics Research 30: 15-22. Fritzsche, David J. (1981). "An analysis of energy consumption patterns by stage of family life cycle," Journal of Marketing Research 18: 227-232. Fuguitt, Glenn V., Daniel T. Lichter, and Calvin L. Beale (1981). Population Deconcentration in Metropolitan and Nonmetropolitan Areas of the United States, 1950-1975. Population Series 70-15, Applied Population Laboratory, University of Wisconsin-Madison. Garkovich, Lorraine (1982). "Variations in nonmetro patterns of commuting to work," Rural Sociology 47: 529-543. Hanson, Susan and Perry Hanson (1981). "The travel-activity patterns of urban residents: dimensions and relationships to sociodemographic characteristics," Economic Geography 57: 332-347. Hawley, Amos H. and Sara Mills Mazie (eds.) (1981). Nonmetropolitan America in Transition. Chapel Hill: Univ. of North Carolina Press. Hirst, Eric (1980). "Understanding residential/commercial energy conservation: the need for data," Journal of Energy and Development 6: 86-108. Hoch, Irving (1981). "Energy and location," in Amos H. Hawley and Sara Mills Mazie (eds,), Nonmetropolitan America in Transition. Chapel Hill: Univ. of North Carolina Press. Institute for Social Research (1972). A Panel Study of lncome Dynamics. Ann Arbor: The University of Michigan Survey Research Center. Keyes, Dale L. (1976). "Energy and land use," Energy Policy 4: 225-236. Keyes, Dale L. (1980). "The influence of energy on future patterns of urban development," in Arthur P. Solomon (ed.), The Prospective City. Cambridge, Mass.: MIT Press. Keyes, Dale L. (1982). "Energy for travel: the influence of urban development patterns," Transportation Research 16A: 65-70. Long, John F. (1981). Population Deconcentration in the United States. Special Demographic Analyses CDS-81-5, Bureau of the Census. Washington, D.C.: U.S. Government Printing Office. Long, Larry and Diana DeAre (1982). "Repopulating the countryside: a 1980 census trend," Science 217: 1111-1116.
204 Rickaby, P.A. (1981). "Six regional settlement patterns: alternative configurations for energy-efficient settlement," Environment and Planning B 8: 191-212. Saltzman, Arthur and Lawrence W. Newlin (1981). "The availability of passenger transportation," in Amos H. Hawley and Sara. Mills Mazie (eds.), Nonmetropolitan America in Transition. Chapel Hill: Univ. of North Carolina Press. Sly, David F. (1982). "The consequences of metropolitan decentralization for personal gasoline consumption," Population Research and Policy Review 1:183-195. Small, Kenneth A. (1980). "Energy scarcity and urban development patterns," International Regional Science Review 5: 97-117. Sosslau, Arthur B. (1980). Home-to-Work Trips and Travel. Final Report by COMSIS Corporation for Federal Highway Administration, Highway Statistics Division. F H W A / PL/81-002. Springfield, Va.: National Technical Information Service, microfiche. Spielberg, Frank and Stephen Andrle (1982). "The implications of demographic changes on transportation policy," Journal of the American Planning Association 48: 301-308. U.S. Bureau of the Census (1979). "The journey to work in the United States, 1975," Current Population Reports: Series P-23, No. 99. Zelinsky, Wilbur and David F. Sly (1981). U.S. Population Redistribution and Personal Energy Use. Final Report to the National Science Foundation. Population Issues Research Center, Pennsylvania State University (mimeo). Zimmerman, Carol A. (1980). "Energy for travel: household travel patterns by life cycle stage." Paper presented at the International Conference on Consumer Behavior and Energy Use, September 17-20, 1980, Banff, Alberta, Canada.
Destination
All counties Metro All nonmetro Adjacent nonmetro Nonadjacent nonmetro
All counties Metro All nonmetro Adjacent nonmetro Nonadjacent nonmetro
All counties Metro All nonmetro Adjacent nonmetro Nonadjacent nonmetro
All counties Metro All nonmetro
Origin
All counties
Metro
All nonmetro
Adjacent nonmetro
37 21 16
76 38 38 18 20
238 173 65 38 27
314 211 103 56 47
N
12.6 16.2 7.9
11.4 12.7 10,0 15.9 4.8
11.4 12.1 9.5 10.4 8,2
11.4 12.2 9.7 12.2 6.7
DIST1
8.4 10.9 5.2
9.5 11.5 7.4 7.3 7.5
12.9 14.3 9.1 11.4 5.9
12.1 13.8 8.5 10.1 6.6
DIST2
7.4 7.2 7.7
8.2 9.6 6,6 2.7 9.6
12.0 11.7 13.3 17.3 5.5
ll.0 11.3 10.1 11,9 7.9
AVGINC
17.7 25.3 11.1
16.0 20.8 12.7 17.3 5.0
9.0 10.3 6.8 8.2 5.3
10.4 11.8 8.5 11.1 5.2
AVGDEC
Summary Statistics for Migrants by Origin and Destination, and for Nonmigrants by Location
TABLE I
Appendix
40,5 42.9 37.5
46.1 50.0 42.1 38.9 45.0
45.8 52.6 27.7 31.6 22.2
45,9 52.1 33.0 33.9 31.9
%INC
40,5 33.3 50.0
35.5 28.9 42.1 55.6 30.0
44.1 38.2 60.0 55.3 66.7
42.0 36.5 53,4 55.4 51.1
%DEC
(Nonmigrants)
1077 829 248 113 135
N
10.2 10.1 10.5 11.4 9.7
DIST1
10.2 8.4 11.6 23.6 4.7
9.7 4.8
DIST1
10.5 10.6 10.3 12.0 8.9
DIST2
10.5 12.3 9.1 12.3 7.3
3.4 8.2
DIST2
6.2 6.0 7.0 9.7 3.8
AVGINC
8.8 11.8 5.9 1.3 7.9
3.8 15.5
AVGINC
5.5 5.3 6.0 9.3 4.2
AVGDEC
13.8 13.0 14.3 23.8 4.8
13.0 5.5
AVGDEC
29.3 31.2 23.0 27.4 19.3
%INC
51.3 58.8 45.5 37.5 50.0
40.0 33.3
%INC
27.0 26.4 29.0 22.1 34.8
%DEC
30.8 23.5 36.4 50.0 28.6
60.0 33.3
%DEC
Note: N = n u m b e r of pairs of successive years; DIST1 = m e a n c o m m u t i n g distance before the m o v e or n o n m o v e ; D I S T 2 = m e a n c o m m u t i n g distance after the m o v e or n o n m o v e ; A V G I N C = m e a n of all b e f o r e - a f t e r increases in distance; A V G D E C = m e a n of all b e f o r e - a f t e r decreases in distance; % I N C = p e r c e n t a g e of cases with b e f o r e - a f t e r increase; % D E C = percentage of cases with b e f o r e - a f t e r decrease. Source: Panel Study of I n c o m e D y n a m i c s data.
All counties Metro All n o n m e t r o Adjacent n o n m e t r o Nonadjacent nonmetro
Unchanged location
39 17 22 8 14
N o n a d j a c e n t n o n m e t r o All counties Metro All n o n m e t r o Adjacent nonmetro Nonadjacent nonmetro
N 10 6
Destination
Adjacent nonmetro Nonadjacent nonmetro
Origin
T A B L E I (continued) o~
to
