APPLICATION OF AN ANALYTIC MIGRATION

Transcription

1 WHITE FLIGHT AND CENTRAL CITY LOSS: APPLICATION OF AN ANALYTIC MIGRATION William H. Frey, Center for Demography and Eology Department of Soiology, University of Wisonsin - Madison This paper2 utilizes an analyti migration framework to assess the aggregate impat of seleted ommunity -level fators on white population losses experiened in entral ities of large metropolitan areas. The framework parameterizes analytially distint omponents of loal and long distane migration streams whih ontribute diretly to entral ity population hange. Eah omponent an be speified as a funtion of ommunity -level attributes whih are relevant to the explanation of speifi in- and out -migration streams. In this appliation, previously advaned raial and nonraial attributes of entral ities and their surrounding suburbs are used to estimate framework omponents based on 1970 ensus data for white movement streams assoiated with the entral ities of large SMSAs. These estimates are then used to asertain the impat that the entral ity raial omposition exerts on net white out -migration from seleted ities. The data demonstrate that the aggregate impat of raially linked "white flight" has been minimal. I. Analyti Migration Framework The framework was developed in order to analyze population hange in both the ity and suburbs of a metropolitan area through ommunity determinants of movement streams that ontribute diretly to suh hange (see Frey, 1977a). Beause eah ontributing stream responds to different sets of ommunity attributes, the framework an be used to assess the net -migration onsequenes of ity, suburb, and metropolitan attributes whih influene movement levels in one or more streams. The ore of the framework onsists of a series of stream -speifi parameters whih an be linked to a demographi aounting equation, Through this linkage, relationships an be speified between ommunity attributes, stream movement levels and aggregate population hange in ities and suburbs. The Framework Parameters Eah of the framework parameters are assoiated with one of the following movement streams: I. Intrametropolitan City -to- Suburb or Suburb -to-city Mobility Streams II. In- migration Streams to Cities or suburbs from outside the SMSA III. Out- migration Streams from Cities or Suburbs to plaes outside the SMSA The framework assumes that ity and suburban population hange are linked to population hange at the metropolitan level and that the streams listed above represent all avenues whereby the ity or suburb population is affeted by movement within and from outside the metropolitan area. With one exeption, the framework parameters assoiated with eah stream represent rates whih are applied to various "at risk" populations of residents and movers. These are listed in Figure A. Beginning with the intrametropolitan ity- to- suburb stream (stream IA), the rate at whih a ity resident will move to the suburbs during an interval is defined as the produt of the parameters and p. This separation of parameters is promptedly empirial studies whih show that residential mobility results from two major stages of deision - making -- the deision to move (made by a resident) and the hoie of destination (made by the mover), and that eah stage is influened by different ausal fators (Butler et al., 1969; Speare, Goldstein and Frey, 1975). Therefore, the i parameter denotes the rate at whih a ityresident will move anywhere within the SMSA, and the p parameter denotes the rate at whih a i ÿ origin mover will reloate in the suburbs. As will be demonstrated below, this distintion permits the analyst to ausally relate different sets of ommunity attributes to eah stage of the mobility proess. In a similar manner, the rate at whih a suburban resident will move to the ity (stream IB) is defined as the produt of framework parameters is and In- migration to the entral ity or suburbs from outside the SMSA (streams IIA and IIB) is also seen to be the produt of two framework parameters. For eah stream, the number of in- migrants rather than the rate of in- migration is speified. In- migrants to the entral ity are defined as the produt of parameters M and p M denotes the number of -migrants to the S SA as a whole, and p denotes the rate at whih SMSA in- migrants loate in the entral ity. This separation of parameters is justified on the basis of findings that long - distane migrants are initially attrated to metropolitan -wide eonomi or labor market attributes (Lansing and Mueller, 1967). The ity or suburb residential loation within the metropolitan area then beomes a seondary deision for SMSA in- migrants whih is made on the basis of different sets of fators. Finally, only one framework parameter is assoiated with out -migration streams from metropolitan ities and suburbs (streams IIIA and IIIB). The Demographi Aounting Equation The framework parameters are linked to a demographi aounting equation whih allows their effets to be translated into aggregate hanges in ity and suburb population sizes during an interval. If one begs s with P, the ity population at time t, and P, the suburb population at time t, it is posssible to ompute the ity and suburb populations of age n and over at time t+n using the relationships in Figure B. By employing these relationships, the migration framework an be used to relate ommunity attributes to aggregate population hange in entral ities and suburbs. The key mehanisms for the analysis are the framework parameters whih are assumed to be ausally related to various attributes. More speifially, eah 454

2 i Mo p Figure A: IA - INTRAMETROPOLITAN CITY -TO- SUBURB MOBILITY MOBILITY INCIDENCE RATE OF CITY RESIDENTS The rate at whih ity residents* move anywhere within the SMSA between t, t+n SUBURB DESTINATION PROPENSITY RATE OF CITY MOVERS The rate at whih ity -origin movers reloate to a suburb destination between t, t+n IIA - IN- MIGRATION TO THE CITY FROM OUTSIDE THE SMSA MIGRATION INTO THE SMSA Total number of migrants into the SMSA between t, t+n CITY DESTINATION PROPENSITY RATE OF IN- MIGRANTS The rate at whih SMSA In- Migrants reloate to a ity destination between t, t+n Movement Streams and - OUT -MIGRATION FROM THE CITY TO OUTSIDE THE SMSA OUT- MIGRATION INCIDENCE RATE OF CITY RESIDENTS The rate at whih ity residents migrate out of the SMSA between t, *residents who Assoiated Framework Parameters is ps+ do not out - migrate between t, t+n IB - INTRAMETROPOLITAN SUBURB -TO -CITY MOBILITY MOBILITY INCIDENCE RATE OF SUBURB RESIDENTS The rate at whih suburb residents* move anywhere within the SMSA between t, t+n CITY DESTINATION PROPENSITY RATE OF SUBURB MOVERS The rate at whih suburb -origin movers reloate to a ity destination between t, t+n IIB - IN- MIGRATION TO THE SUBURBS FROM OUTSIDE THE SMSA MIGRATION INTO THE SMSA Total number of migrants into the SMSA between t, t+n SUBURB DESTINATION PROPENSITY RATE OF IN- MIGRANTS The rate at whih SMSA In- Migrants reloate to a suburb destination between t, t+n IIIB - OUT -MIGRATION FROM THE SUBURBS TO OUTSIDE THE SMSA OUT -MIGRATION INCIDENCE RATE OF SUBURB RESIDENTS The rate at whih suburb residents migrate out of the SMSA between t, t+n Figure B: Demographi Aounting Equations (1) P* - spm - s(p - Pm)ip+8 + s(ps - + s (2) P8* - spemb - s(p8 - PSms)i8p8 + s(pt - Pm)ip + smopo-s where: = ity population age n and over at time t+n Pt = ity population at time t s* suburb population age n and over at time t+n = suburb population at time t s = survival rate for movers, migrants, or nonmovers

3 framework parameter an be expressed as a funtion of a number of ommunity attributes whih serve as independent variables. For example: i = f (X) wherejx. denotes one of k ommunity attributes whih are related to the residential mobility inidene rate of ity residents. The other framework parameters an be speified as funtions of the same or different attributes. After the parameters have been speified as funtions of relevant ommunity attributes, the demographi aounting equations an be used to assess the aggregate impat of an attribute (or ombination of attributes) on population hange in an individual ity or suburb during an interval t, t+n. - II. Appliation to Central City "White Flight" In this appliation of the analyti framework, we are interested in asertaining.the extent to whih the size of the ity's Blak population influenes aggregate white loss due to the seletive suburban reloation of residential (intrametropolitan) movers, and the suburban destination hoies of in- migrants to the metropolitan area. The motivation for this investigation draws from an earlier study we had undertaken to assess the relative importane of both raial and nonraial influenes on reent white ity -to- suburb movement in large SMSAs (Frey, 1977b). Based on a ross -setional analysis of movement streams in 39 SMSAs during the period, our findings indiated that raial influenes did not predominate. Signifiant raial desegregation in entral ity shools and the ourrene of raial disturbanes during the period ontributed little to the explanation of ity- to -suburb white flight, while eologial features of the SMSA and ity- suburb fisal disparities proved to be important determinants. One raial fator -- the perent of the entral ity population whih was Blak -- did influene white out -movement, partiularly in non -Southern ities, and prevented us from dismissing raial fators ompletely as flight determinants. The present analysis represents a somewhat restrited appliation of the framework in the sense that ommunity attributes will only be assessed as determinants of the destination propensity parameters p, psi, and This fous on the destination propensity parameters only an be justified on the basis of our earlier finding that the raial fator, perent ity Blak, influenes white ity -to- suburb movement primarily through the ity -suburb destination hoies of ity -origin movers, and only minimally through the mobility inidene of ity residents (denoted by framework parameter i ) (Frey, 1977b). It is also onsistent studies of residential mobility motivations whih indiate that the deision to move is affeted less by "white flight" onsiderations than by the family's need to make housing adjustments oinident with hanges in its size and omposition (Rossi, 1955; Speare, Goldstein and Frey, 1975). One further restrition will be the fous only on movement -indued hanges to the size of the white ity population, thus disregarding the effets of fertility and mortality on aggregate hange. The Data The data for the investigation are taken from the Census subjet report Mobility in Metropolitan Areas (U.S. Bureau of the Census, 1973) whih lassifies 1970 residents of ities and suburbs of the 65 largest SMSAs aording to their 1965 residene loations, and from whih it is possible to ompute white (nonblak) population and framework parameters for the interval that are neessary to pursue this analysis. These data will be used for two purposes: (a) to speify framework parameters p p, and p as funtions of ommunity a tributes; and alulate the inrement to white ity population loss in seleted SMSAs that an be attributed to the ommunity attribute, perent ity Blak. Speifiation of the destination propensity rates as funtions of ommunity attributes will be aomplished in ross -setional multiple regression analyses, using as ases, the 39 SMSAs whih were examined in the earlier study. In order to alulate inremental white population hange in seleted SMSAs that is assoiated with different values of p, p, and pow using equation (1) in FigureB it s neessary to obtain atual values for the remaining framework and population parameters in that equation. These an also be omputed from the 1970 Census subjet report, although for this purpose it is useful to rearrange the terms of that equation (see footnote to Table 1). Speifying Framework Parameters The ommunity attributes that are used to estimate destination propensity parameters p p, and p onstitute those raial and nonraial attributes whih proved to be the most important determinants of white ity -to- suburb movement in our earlier study. These attributes and their abbreviations are as follows. BLK -- Perent City Blak CIT -- City Share of SMSA Population EDX -- Suburb /City Eduational Expenditures Per Capita (x 100) TAX -- Suburb/City Tax Revenues Per Capita (x 100) CRM -- City Crime Rate PSD -- Postwar Suburban Development CMT -- City- Suburb Commuters CTA -- Central City Age: The number of years between the ensus year when the ity first attained a population of 50,000 and the year 1970 SRG -- Southern Region: (South =1, Other Regions =0) SxB -- Interation of SRG and BLK We now proeed to speify the framework parameters p p and p as funtions of the ommu attributes lust presented in regression analyses. Eah parameter is regressed on all of the attributes for the 39 SMSAs that form the basis of this investigation. The resulting equations appear as follows: p = BLK CIT EDX TAX CRM PSD CMT CTA SRG SxB R2 =.92 (3) 456

4 p R2 =.84 p R2 =.93 = BLK CIT EDX TAX CRM PSD CMT CTA SRG SxB = BLK CIT EDX TAX CRM PSD CMT CTA SRG SxB It is diffiult to evaluate the relative importane of eah attribute from the unstandardized oeffiients presented here. It is, nevertheless, apparent that the perent ity Blak inreases the suburb propensity of ity movers and dereases the ity propensity of suburb movers and SMSA'in-migrants. Eah of these effets is greatly moderated in Southern ities. The Aggregate Impat on White City Loss We move on to the major aim of this analysis: to asertain the aggregate impat on white ity loss whih an be attributed to the ity's Blak population size as it affets the destination hoies of white residential movers and SMSA inmigrants. This aggregate impat will be assessed in three SMSAs: Cleveland, Dayton, and Dallas. Eah of these had a fairly sizeable perentage of Blaks in the entral ity at the beginning of the migration interval: 33% for Cleveland, 26% for Dayton, and 22% for Dallas. Presented in Table 1 are the population and framework parameters for Cleveland, Dayton, and Dallas whih are neessary to es- timate P1-27 for eah ity. The values for pa- rameters p, p, and p are estimated from equations t, (e), and (53ased on atual values for the ommunity attributes shown in Table 2. The values for the remaining framework and population parameters were omputed from atual mobility and population data for the SMSAs reported in the 1970 ensus. To assess the aggregate impat of BLK, the following strategy will be taken: First, we assume various atual and hypothetial numbers of Blaks in eah ity for Seond, we translate these atual and assumed numbers into values of Perent City Blak (BLK). Third, we ompute parameters p p, from the atual and hypothetial. values of BLK using e- quations (3), (4), and (5). Fourth, we ompute 1970 white ity population figures (P1270) based on atual and hypothetial values of p, p, and p using the demographi aounting equá tion The latter figures will allow us to ompare the aggregate hanges to eah ity's white population whih would have resulted from different raial mixes in the ity at the beginning of the movement interval. The results of this analysis appear in Table 3. For eah of the three SMSAs, the following series of assumptions is made about the number of entral ity Blaks in 1965: (A) the atual number of Blaks, (B) a 50 perent inrease in the atual number, (C) a 25 perent inrease in the atual number, (D) a 25 perent derease in (4) (5) the atual number, and (E) a 50% derease in the atual number. Shown in olumn (1) are the orresponding values of BLK whih are used to estimate the destination propensity parameters in olumns (2) through (4). The final three olumns display results of the omputations using the demographi aounting equation (1): the white ity population age 5 and over (olumn 5), the differene from the atual total (olumn 6), and the perent differene from the atual total (olumn 7). As our review of equations (3), (4), and (5) suggested, an inrease in the Perent City Blak is assoiated with a net derease in the white population. Yet the level of impat resulting from the drasti differenes in the number of ity Blaks is not substantial in any of the three ities. This effet is extremely small in Dallas -- resulting in part from the lesser influene of Perent City Blak in Southern SMSAs. Clearly, the aggregate "flight" impat of the entral ity raial omposition -- as transmitted through the destination hoies of loal movers and in- migrants -- is slight, over a five -year migration interval. III. Use of the Framework in "White Flight" Researh The investigation undertaken here represents an initial step toward a ausal analysis of white entral ity population hange utilizing the analyti migration framework. This framework, whih we have desribed in more detail elsewhere (Frey, 1977a), allows the researher to identify ity, suburb, and metropolitan determinants of movement streams whih ontribute diretly to population hange in the entral ity. Using this framework in onjuntion with readily available ensus data, it is possible to alulate inremental hanges in a ity's population assoiated with speifi ommunity attributes that serve as determinants of one or more movement streams. In this manner, the framework an be employed to establish ausal relationships between ommunity attributes, stream movement levels, and aggregate population hange in the entral ity, over the ourse of a migration interval. In the present appliation, we foused our attention on one ausal attribute -- ity raial omposition -- as it affets white entral ity hange through the seletive destination hoies of white intrametropolitan movers, and white in- migrants to the metropolitan area. Based on aggregate movement data from seleted large SMSAs, our findings indiate that suh effets were minimal over the interval. Hene, not only does the ity's raial omposition play a relatively minor role in explaining white movement from the ity to the suburbs (Frey, 1977b), but the total impat of its influene on aggregate white ity loss seems also to be exeedingly small, at least in the short - run. Although restrited in its fous to one ausal attribute and three framework parameters, this appliation of the analyti framework serves to illustrate its utility in an investigation of entral ity "white flight" determinants. In future reports, we plan to extend our ausal analysis of white population loss beyond this re-

5 stritive fous in order to inorporate a greater number of ommunity attributes as ausal fators, and to provide a more refined assessment of "flight" onsequenes for entral ity hange. FOOTNOTES 1 This researh is supported by grant No. 1 RO1 HD , "Migration and Redistribution: SMSA Determinants," from the Center for Population Researh of the National Institute of Child Health and Human Development. 2A more extented treatment appears in Center for Demography and Eology Working Paper University of Wisonsin -Madison. 3Fuller definitions and rationale for these fators appear in Frey (1977b). REFERENCES Brown, Lawrene A., and Eri G. Moore The Intra -Urban Migration Proess: A Perspetive. Geografiska Annaler 52B: Butler, Edgar W., et al Moving Behavior. and Residential Choie -- A National Survey. National Cooperative Highway Researh Program Report No. 81. Washington, D.C.: Highway Researh Board, National Aademy of Sienes. Frey, William H. 1977a. "Population Movement and City- Suburb Redistribution: An Analyti Framework." Presented at the 1977 Meetings of the Population Assoiation of Ameria, St. Louis, Missouri. (Working Paper 77-15, Center for Demography and Eology, University of Wisonsin- Madison.). 1977b. "Central City White Flight: Raial and NonRaial Causes." For presentation at the 1977 Meetings of the Amerian Soiologial Assoiation, Chiago, Illinois. (Disussion Paper No , Institute for Researh on Poverty, University of Wisonsin - Madison.) Lansing, John B., and Eva Mueller The Geographi Mobility of Labor. Ann Arbor: Institute for Soial Researh. Rossi, Peter H Why Families Move. New York: The Free Press. Shryok, Henry S. Jr., and Jaob S. Siegel The Methods and Materials of Demography. Washington, D.C.: U.S. Bureau of the Census. Simmons, James W Changing Residene in the City: A Review of Intraurban Mobility. Geographi Review 58: Speare, Alden Jr., Sidney Goldstein, and William H. Frey Residential Mobility, Migration and Metropolitan Change. Cambridge, Massahusetts: Ballinger Publishing Company. U.S. Bureau of the Census Census of Population: Subjet Reports Final Report PC(2) -2C. Mobility for Metropolitan Areas. Washington, D.C.: U.S. Government Printing Offie. Table 1: Population and Framework Parameters for the intervals used as inputs to Equation (1)b SNUB s(p p1965m s(p P1965m )i s(p P1965m )i p o Cleveland Dayton Dallas aframework parameters and p are estimated from equations (3), (4), and (5) in the text based on atual ommunity attributes (see Table 2). The other population and framework parameters are omputed from the 1970 Census subjet report Mobility in Metropolitan Areas (U.S. Bureau of the Census, 1973). bequation (1) an be rewritten as: P* s(pt - Ptm ) - s(pt - Ptm )i p + s(pt - Ptm )i p + sm p * s s s o where t 1965, n 5, and s represents the appropriate survival rate for eah mover, migrant, or nonmover group. 458

6 Table 2: Community Attributes used to Estimate Framework Parameters and for Interval in Cleveland, Dayton, and Dallas SMSAs Community Attributesa Cleveland Dayton Dallas BLK CIT EDX TAX CRM PSD CMT CTA SRG SxB Table 3: The Effets of Atual and Hypothetial Numbers of City Blaks in 1965 on Migration Framework Parameters p-s' and por during the Interval, and on the 1970 City White Population Age 5 and over, in Cleveland, Dayton, and Dallas SMSAs Assumed Number of City Blaks in 1965: Cleveland SMSA A. Atual Number B. Inrease by 100% C. Inrease by 50% D. Derease by 50% E. No Blaks. Dayton SMSA A. Atual Number B. Inrease by 100% C. Inrease by 50% D. Derease by 50% E. No Blaks Parameter Valuesb 1970 City White Population Age 5 and Over BLK Valuea Population Size Differene from (A) Pt Differene from (A) (1) (2) (3) (4) (5) (6) (7) Dallas SMSA A. Atual Number Inrease by 100% C. Inrease by 50% D. Derease by 50% E. No Blaks ablk is omputed for eah assumed number of ity Blaks in 1965 as. (assumed number of 1965 ity Blaks) 100 (assumed number of 1965 ity Blaks + atual number of 1965 ity whites) bcomputed from equations (3 ), (4 ), and (5 ) based on olumn (1) value of BLK and the atual values of CIT, EDX, TAX, CRM, PSD, CMT, CTA, SRG, and SxB whih appear in Table 2. Computed from equation (1) [see footnote to Table 1], based on values of and in olumns p$ si (2), (3), and (4) and on atual values for the other framework parameters whih appear in Table