《The Effects of Recession on Household Composition“Doubling Up” and Economic WellBeing.doc》由会员分享，可在线阅读，更多相关《The Effects of Recession on Household Composition“Doubling Up” and Economic WellBeing.doc（51页珍藏版）》请在三一办公上搜索。
1、 The Effects of Recession on Household Composition:“Doubling Up” and Economic Well-BeingThis paper reports the results of research and analysis undertaken by U.S. Census Bureau staff. It has undergone more limited review than official publications and was released to inform interested parties of ong
2、oing research and to encourage discussion of work in progress. Any views expressed are those of the authors and not necessarily those of the U.S. Census Bureau. We thank Trudi J. Renwick, Rose Kreider, Charles Hokayem, David Hornick, Charles Nelson, and David Johnson for helpful comments and suggest
3、ions.ABSTRACTThe current recession presents the opportunity to study the effect of economic strain on contemporary family behavior and household structure. The extent to which young adults delay home-leaving, adults join households, or families combine households in response to economic stress is no
4、t well-documented. Using the US Census Bureaus Current Population Survey Annual Social and Economic Supplement for 2008 and 2010, we find an increase in the number and share of doubled-up households and adults residing in doubled-up households since the start of the recession in 2007. Results from l
5、ogistic regression models predicting doubled-up status suggest that younger adults and those who were not in the labor force were more likely to be doubled up in 2010 than in 2008. In general, doubled-up householders and adults were more disadvantaged and experienced a larger increase in poverty rat
6、es during the recession than their counterparts. One way people may cope with challenging economic circumstances is to combine households and household resources with other families or individuals. Preliminary evidence suggests that the number of persons and families sharing households in the U.S. h
7、as increased since the start of the recession in December 2007. For example, there were 1.6 million more multi-family households between 2008 and 2010, an increase of 12 percent (U.S. Census Bureau 2010). A recent study by The Pew Research Center also reports that 1 in 10 young adults aged 18 to 34
8、moved back with parents due to the recession (Pew Research Center 2009). However, the extent to which individuals and families actually share households and how household sharing has changed over time has not been well documented in the literature. The recent economic recession offers the opportunit
9、y to assess the effect of economic conditions on contemporary demographic behaviors. Although an extensive literature cites the effects of macroeconomic conditions on marriage (Liker & Elder, 1983, Conger & Elder 1994) and fertility (Billari & Kohler 2004; Goldstein et al. 2009; Kravdal 2002, Rindfu
10、ss et al. 1988), less attention has been given to how diminished economic resources affect family and household composition (see Wiemers 2010; Pew Research Center 2009 for exceptions). In this paper, we examine “doubled-up” households, which we define as households that include an adult who is not t
11、he householder, spouse or cohabiting partner of the householder. Cohabiting households (i.e. households having a household head cohabiting with another individual) may also be considered “doubled-up” households. However, in this paper, we do not define cohabiting households as doubled up if the only
12、 adults living in the household are the household head and his/her cohabiting partner (See Kreider 2010 for discussion of the recent increase in cohabiting households). Although recent research and media attention has focused on adult children moving back with parents or grandparents sharing a house
13、hold with their children and grandchildren, doubled-up households may also include nonfamily household arrangements as well. We use data from the U.S. Census Bureaus Current Population Surveys Annual Social and Economic Supplement (CPS ASEC) to explore both the socio-demographic characteristics and
14、the economic well-being of adults in doubled-up households. The estimates in this paper are based on responses from a sample of the population. As with all surveys, estimates may vary from the actual values because of sampling variation and other factors. All comparisons made in this paper have unde
15、rgone statistical testing and are significant at the 95-percent confidence level unless otherwise noted. For information on confidentiality protection, sampling error, non-sampling error, and definitions see and .As shown in Figure 1, the percent change in the number of doubled-up households has bee
16、n volatile over the past two decades ranging from -0.7 percent in 1998 to 6.4 percent in 2010. Sharp increases in doubling up occurred during or soon after economic recessions (1991-1992, 2000-2001, and 2008-2010). The most recent recession is marked by a disproportionate increase in the number of d
17、oubled-up households relative to total households, as shown in Figure 1. The pattern suggests a correlation between doubled-up households and economic recessions that warrants further investigation. In this paper, we focus on the effect of the most recent recession on doubling up. Specifically, we a
18、sk:(1) Has there been an increase in the number of “doubled-up” households or the number of persons in doubled-up households since the start of the recession at the end of 2007?(2) What are the socio-demographic characteristics of adults (including both householders and non-householders) who reside
19、in doubled-up households? Have these characteristics changed since the recession began in 2007? (3) Does doubling up improve the economic well-being of persons living in the household?BackgroundSharing a household with family members is a time-honored strategy for stretching thin resources (Hareven
20、1990; Ruggles 1987). Yet much of the literature that examines how macroeconomic factors influence household sharing has focused on housing affordability and availability (Mutchler & Krivo 1989; Whittington & Peters 1996; Kiernan 1989) or on the influence of policy changes (Bitler, Gelbach & Hoynes 2
21、006; Acs & Nelson 2004; Costa 1999; McGarry & Schoeni 2000; Ellwood & Bane 1985), and not on the effect of economic downturns.In an early study, Monahan (1956) used U.S. Census data to estimate the percentage of families who shared households with other families in the first half of the 20th century
22、. Monahan found that this type of doubling up increased between 1910 and 1955, peaking in 1947 when 9.9 percent of all families shared households with other families. Monahan attributed the increase in doubling up to macroeconomic factors i.e., the Great Depression, shortages of housing materials du
23、ring wartime, and limited postwar housing stock. Yet his descriptive analysis did not test these propositions. More recently, London and Fairlie (2006) examined the association between state unemployment rates and the living arrangements of minor children using data from the Current Population Surve
24、y (CPS) and the Survey of Income and Program Participation (SIPP). They concluded that unemployment was not significantly associated with children residing in a shared household in the CPS and found a weak relationship between unemployment and household sharing in the SIPP. Using the Panel Study of
25、Income Dynamics and the American Community Survey, Painter (2010) found that macroeconomic conditions, including unemployment and the housing market, were associated with household formation. Prior research has also found individual characteristics to be associated with coresidence of young adults w
26、ith parents. Researchers have noted an increase in the average age of home-leaving over the past few decades (Goldscheider and Goldscheider 1994). Previous studies also suggest that males were likely to stay in the parental home longer than females (Goldscheider and DaVanzo 1985; Mitchell et al. 198
27、9) and whites were more likely to leave home at younger ages than blacks or Hispanics (Aquillino 1991; Bianchi 1987; Painter 2010; and Sarksian, Gerena and Gerstel 2006). Marriage has a strong, negative effect on coresidence, whereas school enrollment increases the likelihood of coresidence (Aquilli
28、no 1991).Other studies identified a relationship between own employment status and living arrangements, although much of this literature focuses on home-leaving among young adults. Using data from the 1997 National Longitudinal Study of Youth, Kaplan (2009) concluded that moving from employment to u
29、nemployment increases the likelihood of moving back to the parental home by about 70 percent. Wiemers (2010) also found that adults transitioning to unemployment were twice as likely to join other households. Examining cohabiting couples, Kreider observed differences in employment status between exi
30、sting and newly formed cohabiting couples and concluded that economic pressures may have contributed to an increase in cohabitation between 2009 and 2010. Other research also suggests a positive association between unemployment and coresidence (Aassve et al. 2002; Avery, Goldscheider, and Speare 199
31、2; Ermisch and Di Salvo 1997; Aquillino 1991; Painter 2010).Data For this analysis, we use the U.S. Census Bureaus CPS ASEC for years 2008 and 2010 to isolate the changes occurring in household structure since the start of the recent recession. The CPS ASEC is well suited to examine the change in ho
32、usehold composition because it contains detailed demographic information on households and the relationship of individuals to the household head as well as other socioeconomic and demographic information. Collected annually between February and April, the CPS ASEC captures information on current hou
33、sehold composition and income for the prior year. The 2008 ASEC captures household composition in spring 2008 and income in calendar year 2007; the 2010 ASEC captures household composition in spring 2010 and income in calendar year 2009. Thus, the 2008 CPS ASEC reflects economic conditions prior to
34、the recession while the 2010 CPS ASEC reflects economic conditions during the recession. In this analysis, although we include households headed by persons under 18 years of age in the sample, our measure of doubling up focuses on “extra adults in the household” because minor children are typically
35、considered dependents. Although young adults ages 18 to 24 live with parents while attending school The CPS ASEC also classifies unmarried young adults temporarily residing in college dorms or away at school as living with their parents. We control for school enrollment in our models and also estima
36、te extra adult status for adults aged 25 years and older and for adults not enrolled in school., we include them in our sample but control for school enrollment in the models. The analytic sample consists of 75,659 households and 145,520 adults in 2008 and 76,067 households and 149,011 adults in 201
37、0. The weighted figures represent: in 2008, 116.8 million households and 224.5 million adults, and in 2010, 117.5 million households and 229.1 million adults. MethodsDefining Doubled-Up HouseholdsIn this analysis, we estimate the number and percent of households designated as doubled-up-that is, hou
38、seholds that include at least one extra adult. We define “extra adults” as persons aged 18 years and older who reside in a household and who are neither householders, nor the spouses or cohabiting partners of householders. Our definition of “extra adults” includes adults related to the householder a
39、nd thus captures adult children of householders who move in with parents as well as parents who move in with adult children. The CPS ASEC sample includes the civilian noninstitutionalized population. This universe includes civilians in households, people in noninstitutional group quarters (other tha
40、n military barracks) and military in households living off post or with their families on post (as long as at least one household member is a civilian adult). The universe excludes other military in households and in group quarters (barracks), and people living in institutions. As we are interested
41、in individuals residing in households, our estimates of “extra adults” in doubled-up households excludes the group quarters population. That is, estimates of “extra adults” include both relatives of the householder (except spouses) and non-relatives of the householder (except cohabiting partners). T
42、he “extra adults” concept also includes persons who are the roommates or housemates of the reference person. Predicting Doubled-Up Household Status Using a pooled sample of households in the 2008 and 2010 CPS ASEC, we estimate logistic regression models to predict the doubled-up status of households
43、. The dichotomous dependent variable in the household models is coded 1 if the household is doubled up and 0 if the household is not doubled up. Independent variables for the full model include survey year as well as the following household characteristics: type, income (in thousands), tenure and ge
44、ographic region. Other independent variables include householder characteristics such as age, sex, race-ethnicity, nativity, educational attainment, marital status, employment and personal poverty status. In this analysis, we estimate a set of nested models: Model 1 includes survey year as the only
45、covariate; Model 2 incorporates household characteristics and the householders socio-demographic characteristics; in Model 3, we test interactions between survey year and the covariates in order to assess whether determinants of doubled-up status changed for households over the course of the recessi
46、on.Predicting the Doubled-Up Status of AdultsUsing a pooled sample of households in the 2008 and 2010 CPS ASEC, we estimate two sets of logistic regression models to predict both the doubled-up status of adults and extra adult status. In models predicting doubled-up status, the dichotomous dependent
47、 variable is coded 1 if the adult resides in a doubled-up household, and coded 0 otherwise; in models predicting extra adult status, the dichotomous dependent variable is coded 1 if the adult is not the householder, or the spouse or cohabiting partner of the householder, and 0 otherwise. Independent
48、 variables in these analyses include the following demographic characteristics: age, sex, race-ethnicity, nativity, marital status, and school enrollment. The school enrollment variable is restricted to persons aged 18 to 24 years. We also incorporate covariates representing individual socioeconomic
49、 status, including educational attainment, employment and personal poverty status. For both analyses, we estimate a set of nested models: Model 1 includes year as the only covariate; Model 2 incorporates the demographic and socioeconomic characteristics of the householder. In Model 3, we test interactions between survey year and the covariates in order to assess whether the determinants
