U. S. Department of Health and Human Services.

Effects of Holding Assets on Social and Economic Outcomes of Families

Appendix:
Empirical Studies of Effects of Asset Holding

Appendix Table.
Empirical Studies of Effects of Asset Holding
Author Data Source Sample/Study
Population Method Outcomes
Analyzed Key Asset Explanatory
Variables Findings Author’s
Principal Conclusions Key
Weaknesses²

Economic Well-Being

Bynner & Despotidou (2001) National Child Development Study (U.K.). 12,000 British individuals born in 1988. OLS regression Labor market experience at age 23-33: years in full-time employment, years unemployed, self-employed, .… Savings and investments at age 23. Asset holding is associated with positive employment outcomes (p. 3). Labor market experience appears to be strongly affected by assets. The coefficient on asset measures is especially strong when the outcome is unemployment (p. 16). Endogeneity.

Carasso and McKernan (2005) Survey of Consumer Finances US households. Literature review. Assets, liabilities, and networth Homeownership. In 2001, homeowners had total asset holdings 16 times greater than renters ($210,400 vs $12,700); debt 12 times greater than renters (69,400 vs 6000) but were 4 times less likely to be delinquent on debt, and net worth 36 times greater than renters (171,700 vs 4,800). Homeownership status made a larger difference in net worth than income, education, single-family status, and minority status.

Cho (1999) NLSY. 363 women experiencing marital disruption between 1985 and 1989. Instrumental variables regression where assets are instrumented with their own lags. Seemingly unrelated regression. Log total family income, log per capita family income, log earned income, log welfare income, & log private transfers. Log financial assets, log home equity, log value of business or real property. Financial assets are positively associated with the economic well-being of women one year after marital disruption. Home equity and value of business/real property are not statistically, significantly related. Financial assets (as measured in the post-marital disruption year) have a positive effect on income/earnings and reduce welfare receipt. Asset accumulation could be a positive strategy for women and children at risk of marital disruption (abstract). Instrument for assets with poor instruments? Include other endogenous variables in model (human capital, social support). Could these bias instrumented variable? It is strange that they use per capita income and total family income as dependent variables. Presumably in many cases the husband was a contributor to these variables before divorce so it isn't surprising that they fall.

Goetzmann & Spiegel (2002) Office of Housing Enterprise Oversight housing price indices. Sales of homes in the United States from 1980-2000. Descriptive analysis comparing returns on financial investments. Returns on home sales & financial investments. Returns on housing, returns on financial investments such as S&P 500, Treasury bills, inflation, T-bills, etc. (1) Housing had a 4.2 percent nominal annual rate of return (appreciation) over the past 21 years; the consumer price index was 3.7 percent for the same time period (p. 259). It is dangerous for homeowners to devote too much of their wealth to an asset that has low historical return and a serious risk of loss over multiple-year horizons. If the government actively encourages homeownership, it has the responsibility to inform potential homeowners of the risks (p. 272). Doesn't consider implicit rent.

Jencks & Mayer (1989) Chicago survey. 1,617 non-hispanic blacks and whites pooled from surveys in 1983 and 1985. Regressions (likely OLS). Number of material hardships: couldn't afford food, no insurance, unmet medical care, housing problems, & cut-off utilities. Home ownership and access to credit as measured by ability to borrow $500 if needed. (1) Homeowners reported fewer hardships than tenants with the same income and needs. Homeownership may be a proxy for unmeasured economic resources and unmeasured forms of household efficiency (p. 108). (2) All else equal, a family's ability to borrow $500 in the event of an emergency did as much to reduce hardship as tripling family income (p. 109). Poverty statistics do not provide reliable information about material hardship. Family income is not the only critical determinant of material well-being--it explains only 14 percent of the variance in material hardship. Broader measures of economic resources explain only a little more. Better measures of both savings and credit would improve our ability to predict hardship, but probably not enough to alter basic findings (p. 111).

McCarthy, Van Zandt & Rohe (2001) Literature review. US households and housing markets. Literature review. Economic benefits and costs of homeownership on an individual and societal level. Homeownership. (1) House prices appreciated an annual average of 3.5% (city-based) and 4% (suburban) from 1987-1997, however this rate is very uneven across areas and over time (p. 23). (2) Homeowners hold a larger-than-optimal portfolio share in housing, exposing owners to higher overall portfolio risk (p. 43). Behavioral studies on tenure choice are much more readily available than research on the benefits and costs of homeownership. Although most households, including lower- and moderate-income households, benefit through home purchase, there has been little research that carefully measures and compares the relative benefits of ownership across socioeconomic groups (p. 44). Promoting homeownership as an across-the-board policy may not be the best route. Future research should focus on ways to preserve homeownership, develop a better understanding of the housing market, and examine how homeownership affects other economic variables.

Mills et al. (2006) Random assignment IDA survey data. 840 randomly assigned IDA program participants (treatment) and non-participants (control) individuals. Experimental program design with OLS and probit regressions. Homeownership and equity, non-retirement financial assets, business ownership and equity, enrolling in courses, computer purchase, retirement savings, home improvement, net worth. Eligibility to participate in the IDA program Eligibility for the IDA program: (1) Raised homeownership rates by almost 10 percentage points over 4 years for black renters, but reduced financial assets and business ownership. (2) Had no effects on homeownership for white renters, but their business equity rose. (3) Overall had no statistically significant effect on net worth, among other findings. The IDA program: (1) Raised homeownership rates by almost 10 percentage points over 4 years for black renters, but reduced financial assets and business ownership. (2) Had no effects on homeownership for white renters, but business equity rose. (3) Home owners use the IDA in different ways than renters (abstract).

Moore et al. (2001) IDA survey data. 318 current and former IDA participants. Descriptive. Economic effects as measured by (1) decreasing consumption, (2) purchasing a home or starting/expanding a business, and (3) increasing employment (p. 21). IDA Program Participation, which includes saving accounts, economic education, expectations for saving behavior, incentives to save, staff and peer support, etc (p. 46) (1) Thirty percent of participants said they had less money for leisure, eight percent said they had to give up food or other necessities (p. 20), 35 percent said they were less likely to save outside their IDAs. (2) Almost 75 percent said they were more likely to purchase or renovate a home and fifty-seven percent said they were more likely to start or expand a business. (3) Forty-one percent said they were more likely to increase work hours and 61 percent said they were more likely to increase their income in other ways (p. 47). Cross-sectional survey data with self-reported effects.

Nichols (2005) American Housing Survey (AHS). 9,711 U.S. households with fixed rate mortgages who have not refinanced or moved (from AHS 1985-2002). (1) Calculate compound annual rate of return that incorporates appreciation rate, implicit rent, mortgage contract, and mortgage income tax deduction (p. 23); (2) Logistic regressions of the probability of negative total dollar return to homeownership (p. 31); (3) OLS regressions of measures of housing return (conditional on positive return) and the appreciation rate (p. 31). Total returns on housing. Homeownership. (1) The annualized average return to homeownership is 1.9% based on appreciation alone, 9.8% based on appreciation and implicit rent; 8.5% based on appreciation, implicit rent, and mortgage contract, and 8.6% based on appreciation, implicit rent, mortgage contract, and mortgage income tax deduction (table 5.1, p. 23). (2) Single-family homes provide higher rates of return.

Raphael & Rice (2002) SIPP. 16-65 year old non-disabled individuals living in households with zero to three cars. OLS and Instrumental variables regressions (2SLS) where car ownership is instrumented by state insurance rates and gas taxes. OLS and 2SLS estimates are similar for employment and hours worked outcomes, but differ for the wage outcome. Employment, hours worked, log wages. Car ownership. Car ownership increases employment and hours worked, but not necessarily wages (p. 124). Large positive differences in employment rates and hours worked between those who do and do not own cars in both OLS and instrumented regressions. Instrumenting eliminates the positive impact car ownership has on wages in OLS regression (abstract).

Reid (2004) PSID and 55 low-income homeowners in Seattle, Washington. Low-income renters and homeowners. Descriptive comparisons of means before and after home purchase. Changes in home value and exit from homeownership to renting. Homeownership. For low-income homeowners, the financial returns to homeownership even after 10+ years averaged only 3 percent per year. For middle- and high-income whites, the returns averaged 5 percent per year (p. 29). Low-income homeowners do not see the same levels of house price appreciation as do higher income homeowners (p. 28). For low-income homeowners, the financial returns to homeownership even after 10+ years were lower than the return on Treasury bills.

Rossi & Weber (1996) GSS, NSFH, & ANES. 1,500 (GSS), 13,000 (NSFH), 13,000 (ANES) US Households. Descriptive OLS and logit regressions. 8 measures of household finances, including assets and liabilities. Homeownership. (1) Total household income of homeowners is more than $17,000 greater than renters (p. 12). (2) Homeowners have about $6,000 more in savings and about $5,000 more in mutual funds than renters (p. 12). (3) Homeowners are more likely to have credit card debt, installment debt, and personal bank loans. Homeowners are less likely to have education loans and to have bills more than 90 days overdue (p. 13). Claims for some social and individual benefits from homeownership are only weakly supported. Strong differences between owners and renters exist only in demographic characteristics (p. 1).

Sullivan (2005). PSID & SIPP. 9,400 (SIPP) and 15,666 (PSID) Working-age household heads with steady employment. OLS and Instrumental variables regressions where earnings are instrumented with unemployment spells (p. 6). Assets are not instrumented for but are measured at an initial period (p. 7). Changes in unsecured debt & changes in food and housing consumption. Unemployment induced earnings losses. Low-asset holdings as measured by non-positive financial assets, non-positive total gross assets, and total gross assets to earnings ratio of less than 0.12 (p. 7). Households with assets use unsecured debt to borrow when faced with temporary shortfalls in earnings. Households with low-assets do not borrow--likely because they do not have sufficient access to unsecured credit--and their consumption falls (abstract).

Social Well-Being

DiPasquale & Glaeser (1999) General Social Survey & German Socio-Economic Survey. 1,500 U.S. households and 13,000 German households OLS and instrumental variables regressions where homeownership is instrumented with the average homeownership rate by income quartile, race, and state (p. 356). Individual-level fixed effects. Social capital and good citizenship as measured by: number of nonprofessional organizations affiliated with, political knowledge, voting, participation in community planning, gardening, gun ownership, and church attendance. Homeownership. Both in the U.S. and in Germany, homeownership is strongly correlated with variables that attempt to measure good citizenship and social capital. A large portion of the effect of homeownership comes from lower mobility rates for homeowners. Standard economic incentives (both the effects of ownership and tenure) influence investment in social capital (p. 383).

Haurin, Dietz & Weinberg (2003) Literature review. US households. Literature review. The impact of neighborhood homeownership rates on residents. Homeownership. Theoretical discussions of neighborhood homeownership effects supports the hypothesis that homeownership rates can affect behavior. However, relatively few studies have rigorously empirically tested the theoretical hypothesis put forth. Attention should be given to determining causal and noncausal correlated effects and determining whether feedback effects are present. Further examination of neighborhood effects has relevance for social welfare policies.

Moore et al. (2001) IDA survey data. 318 current and former IDA participants. Descriptive. Perceived social and civic effects as measured by (1) more problems and more good relationships with family, (2) more problems with neighbors and more involvement in neighborhood, and (3) more respected in community (p. 21). IDA Program Participation, which includes saving accounts, economic education, expectations for saving behavior, incentives to save, staff and peer support, etc (p. 46) (1) About half of current participants said they were more likely to have good relationships with family members, (2) about one-third said they were more likely to be involved in their neighborhoods, and (3) about one-third said they were more likely to be respected in their communities (p. 47). Cross-sectional survey data with self-reported effects.

Reid (2004) PSID and 55 low-income homeowners in Seattle, Washington. Low-income renters and homeowners. Comparisons of means before and after home purchase. Changes in neighborhood characteristics Homeownership. Neighborhood benefits for shifting from renting to owning were minimal for low-income whites, but substantial for low-income minorities (p. 18).

Rossi & Weber (1996) GSS, NSFH, & ANES. 1,500 (GSS), 13,000 (NSFH), 13,000 (ANES) US Households. Descriptive OLS and logit regressions. (1) 22 measures of sociability (p. 16). (2) 64 measures of marriage and family issues (p.19). Homeownership. (1) Homeowners are not consistently more likely to be members of social networks (p. 17). (2) Homeowners-renter differences on marriage and family behavior and value issues are not very great (p. 20). Claims for some social and individual benefits from homeownership are only weakly supported. Strong differences between owners and renters exist only in demographic characteristics (p. 1).

Civic Engagement

Bynner & Despotidou (2001) National Child Development Study (U.K.). 12,000 British individuals born in 1988. OLS regression. Citizenship and values: voting, political interest, political cynicism, work ethic. Savings and investments at age 23. Although voting showed no association with assets, political interest is positively associated with asset holding (p. 16). Although voting showed no relation to assets, political interest is likely to be greater with asset holding (p. 16). Endogeneity.

DiPasquale & Glaeser (1999) General Social Survey & German Socio-Economic Survey. 1,500 U.S. households and 13,000 German households OLS and instrumental variables regressions where homeownership is instrumented with the average homeownership rate by income quartile, race, and state (p. 356). Individual-level fixed effects. Social capital and good citizenship as measured by: number of nonprofessional organizations affiliated with, political knowledge, voting, participation in community planning, gardening, gun ownership, and church attendance. Homeownership Both in the U.S. and in Germany, homeownership is strongly correlated with variables that attempt to measure good citizenship and social capital. A large portion of the effect of homeownership comes from lower mobility rates for homeowners. Standard economic incentives (both the effects of ownership and tenure) influence investment in social capital (p. 383).

Rossi & Weber (1996) GSS, NSFH, & ANES. 1,500 (GSS), 13,000 (NSFH), 13,000 (ANES) US Households. Descriptive OLS and logit regressions. 27 measures of political behavior (p. 24). Homeownership. (1) Homeowners and renters are not very different in general political interest (p. 23). (2) Homeowners are almost consistently more engaged in local politics than renters but the activism does not extend to all area (p. 23). Claims for some social and individual benefits from homeownership are only weakly supported. Strong differences between owners and renters exist only in demographic characteristics (p. 1). Endogeneity.

Child Well-Being

Aaronson (1999) PSID. 5,143 children reaching the age of 17 between 1975 and 1993. Probit regressions and instrumental variables regressions where homeownership is instrumented with the average homeownership rate by income quartile, race, and state (p. 366). High school graduate by age 19. Homeownership. Some of the homeownership effect found by Green & White (1997) is driven by family characteristics such as residential stability that are correlated with homeownership. Nevertheless, in the sense that homeownership increases residential stability, it appears to be correlated with education attainment (p. 356). Endogeneity of residential mobility and imperfect controls for the endogeneity of homeownership.

Bynner & Despotidou (2001) National Child Development Study (U.K.). 12,000 British individuals born in 1988. OLS regression Parenting at age 37: number of books child has, frequency of reading to child, frequency child reads. Savings and investments at age 23. Asset holding is not associated with parenting outcomes as measured by child's reading. Parenting outcomes did not appear to be improved by the presence of assets (p. 16). Endogeneity?

Dietz and Haurin (2003) Literature review. US households. Literature review. The impact of homeownership on private and social micro-level behavior. Homeownership. The literature contains solid studies that report positive direct and indirect effects of homeownership on child outcomes (p. 439). There is substantial evidence that homeownership has important effects on some household behaviors and outcomes. However, much of the past 30-year's literature on consequences of homeowning is deficient from a theoretical or econometric perspective (abstract).

Green & White (1997) PSID, 1980 PUMS, & High School and Beyond. 17 or 18 year old youths. Probit models and bivariate probit models that instrument for the endogeneity of homeownership with the relative cost of owning versus renting (p. 455). The correlation coefficients from the bivariate probit estimates suggest that homeownership is not an endogenous variable for the two outcomes considered (p. 456). Attending school at age 18 & having a child by age 18. Homeownership, housing tenure. Homeownership decreases the likelihood that children of homeowners have children and increases the likelihood that children of homeowners are in school at age 18. Findings are particularly important for low-income households (p. 441). Children of homeowners are less likely to have children and more likely to be in school at age 18. The dollar benefit per low-income household of parents being homeowners is at least $31,000 (p. 441).

Haurin, Parcel, & Haurin (2001) NLSY79 and NLSY Children. Mothers age 14-22 in 1979. Random effects regressions with an unknown instrument to control for the endogeneity of homeownership (p.3). (1) Child's home environment as measured by emotional and cognitive/physical scores. (2) A measure of child's behavior problems. (3) Child cognition as measured by mathematics and reading scores (p. 10). Homeownership. Homeownership is positively and statistically, significantly associated with an improved home environment, but only marginally statistically, significantly associated or not statistically, significantly associated with child cognition and child behavior (p. 13-14). Children of homeowners have better home environments, higher cognitive test scores, and fewer behavior problems than do children of renters. Owning a home leads to a 13 to 23 percent higher quality home environment, all else equal (abstract). Homeownership increases cognitive math test scores 9 percent and cognitive reading scores 7 percent relative to renters. Homeownership reduces the index of child behavior problems by about 3% (pp. 13-14). Not clear that controls for endogeneity. First stage regressions not shown and not clear what instrument is.

Henretta (1984) PSID. Individuals in the 1980 PSID who are heads of household and were sons or daughters in previous waves. Logit and OLS regressions. Child's homeownership and child's home value. Parent's income, homeownership, and home value for the last year the child was in the parent's home. (1) Parent's homeownership is positively and statistically, significantly associated with child's homeownership, for both whites and blacks (p. 134). (2) Parent's home value is positively and statistically associated with child's home value for whites, but not blacks (p. 135). (3) Parental gifts have no statistically significant association with child's equity or mortgage for whites, but do have a positive and statistically significant association for blacks (p. 137). Tentatively interprets the findings to suggest that the effect of parental home value reflects a socialization process in which the child's living situation affects the child's expectations concerning the proper or appropriate standard of living (p. 138). For blacks however, parental gifts (and not just child's expectations) appear to play a bigger role in determining a child's homeownership and value choices (p. 138). Endogeneity of parental income, homeownership, and home equity.

Kane (1994) CPS. Average sample of 500 dependent black and 3,250 white 18-19 year olds per year (p. 883). Full information maximum likelihood probit models to control for the endogeneity of parent's education (p. 887). Child's high school graduation and college entry. Parent's homeownership. Parent's homeownership is positively and significantly associated with child's high school graduation and college enrollment for blacks and whites (p. 888, 894, 896, 909). From 1973 to 1988, college costs drove enrollment rates down while increases in parental education levels of blacks exerted upward pressure on increased educational attainment by blacks (p. 878). Does not control for the endogeneity of family resources (income, homeownership, employment).

McGarry & Schoeni (1995) HRS. HRS respondents and their children who are not living at home. Logit, OLS, and fixed effect regressions. Whether parents transfer cash to their children and the amount. Whether children transfer cash and time to their parents and the amount of each. Child's total income and homeownership status, parent's financial situation and homeownership status. (1) Wealthy parents transfer larger sums of cash to their children, but the finding is not statistically significant in the fixed effects model (p. S206, S209). (2) Parents transfer larger sums of cash to lower income children (p. S208). (3) Wealthy children transfer larger sums of cash to their parents, but the finding is not statistically significant in the fixed effects model (p. S219). (4) Children transfer larger sums of cash to parents with worse financial situations and who do not own their home (p. S218). Parents give greater financial assistance to their less well off children and children give more to less well off parents. These results hold for both incidence of transfer and amounts (p. S184). Does not state preferred specification? Does not focus findings on fixed effect results.

Scanlon & Page-Adams (2000). NSAF. US Households with children. GLM and hierarchical regressions. Child's extracurricular activities & behavioral problems. Homeownership. Homeownership is positively associated with decreased behavioral problems and increased participation in extracurricular activities among adolescents, however the relationships are not found, or the associations are less strong for black, Hispanic, and poor children (p. 11).

Williams (2003) PSID. 2,936 children age 0-12 who live with their biological or adopted mothers. OLS regression, ANOVA, logistic and ordered logit regressions in hierarchical blocks. Cognitive development/academic achievement, physical health, and socio-emotional behavior/schooling variables. Income, wealth, homeownership. Household income and wealth are positively associated with child cognitive development, health, and behavioral development. The manner in which these economic status variables associated with child development outcomes differs substantially by race (p. 88-89). Household income and wealth influence child outcomes in cognitive development, health, and behavioral development/schooling, but the manner in which wealth influences outcomes differs by race. Assets are mediated by intermediate variables regarding neighborhood characteristics and parental quality. Benefits of assets extend to both low and high-income households (pp. 88-89).

Zhan & Sherraden (2003) NSFH. 591 children 12-18 years old in female-headed households (p. 195). OLS regression. Mother's academic expectations of child, child's academic performance, and whether the child graduates from high school. Mother's savings, homeownership. Homeownership is positively and statistically significantly associated with mother's expectations and child's academic performance, but not statistically significantly associated with child's high school graduation. Savings are positively and statistically associated with child's high school graduation and mother's expectations, but not statistically significantly associated with child's academic performance (p. 199, 200, 201). Single mother's assets are positively associated with mother's expectations and with child's educational achievement. Mother's expectations are positively associated with child's outcome and partially mediates the relationship between assets and the child's outcome. The relationship between income and child's educational achievement disappears when assets are included in the regression, suggesting that models including income but not assets are underspecified (p. 204, 205).

Health and Psychological Well-Being

Bynner & Despotidou (2001) National Child Development Study (U.K.). 12,000 British individuals born in 1988. OLS regression. Health at age 33: malaise, general health, smoking. Savings and investments at age 23. Asset holding is associated with positive health outcomes (p. 3). Health appeared to be strongly asset dependent (p. 16). Endogeneity.

Moore et al. (2001) IDA survey data. 318 current and former IDA participants. Descriptive. Psychological effects as measured by thoughts, feelings, and plans for the future (p. 19, 20). IDA Program Participation, which includes saving accounts, economic education, expectations for saving behavior, incentives to save, staff and peer support, etc (p. 46) Current participants agreed or strongly agreed that they felt more confident about their futures (93 percent), more economically secure (84 percent), and more in control of their lives (85 percent) because they had IDAs (p. 46). Effects on planning were somewhat less common with about 60 percent of respondents saying they were more likely to make educational and retirement plans (p. 47). Cross-sectional survey data.

Yadama & Sherraden (1996) PSID. 2,871 individuals who were heads of household in both 1968 and 1972 (p. 5). Simultaneous estimation of a path model of directly observed variables (p. 5). (1) Attitudes and behaviors as measured by efficacy, prudence, horizons, connectedness, and effort in 1972. (2) Assets as measured by house value and savings in 1972. (1) Assets as measured by housevalue, savings, and income in 1968. (2) Attitudes and behavior as measured by prudence, efficacy, horizons, connectedness, and effort in 1968. (1) Five of the 10 coefficients relating the two asset variables and five attitude and behavior variables were significant, all in the expected direction. Savings had positive effects on prudence, efficacy, horizons, and connectedness; house value had a positive effect on horizons (pp. 9-10). (2) Four of the 10 coefficients relating attitude and behavior to assets were significant: prudence had a positive effect on house value but not savings, efficacy had positive effect on savings, connectedness had a positive effect on savings, and effort had a positive effect on house value (p. 10). Results support the proposition that assets have a positive effect on expectations and confidence about the future; influence people to make specific plans with regard to work and family; induce more prudent and protective personal behaviors, and lead to more social connectedness with relatives, neighbors, and organizations (p. 3). Limited timeframe. Simultaneous estimation of a path model of directly observed variables (p. 5). It is not clear how the model is identified.

Rossi & Weber (1996) GSS, NSFH, & ANES. 1,500 (GSS), 13,000 (NSFH), 13,000 (ANES) US Households. Descriptive OLS and logit regressions. Personal well-being: 11 measures of health and psychological well-being (p. 14). Homeownership. Homeowners regard themselves as having a greater sense of well-being than renters, but only marginally so. The issue of causality prohibits any claim that ownership leads to a greater sense of well-being (p. 15). Claims for some social and individual benefits from homeownership are only weakly supported. Strong differences between owners and renters exist only in demographic characteristics (p. 1). Endogeneity.

Adverse Consequences

Dietz and Haurin (2003) Literature review. US households. Literature review. The impact of homeownership on private and social micro-level behavior. Homeownership. Homeowners are less mobile than renters due to higher transaction costs and perhaps because of greater ties to their neighborhood and community. This lack of mobility may negatively impact the ability of households to respond to changes in the local labor market, but this result is not firmly established (p. 439). There is substantial evidence that homeownership has important effects on some household behaviors and outcomes. However, much of the past 30-year's literature on consequences of homeowning is deficient from a theoretical or econometric perspective (abstract).

Goetzmann & Spiegel (2002) Office of Housing Enterprise Oversight housing price indices. Sales of homes in the United States from 1980-2000. Descriptive analysis comparing returns on financial investments. Returns on home sales & financial investments. Returns on housing, returns on financial investments such as S&P 500, Treasury bills, inflation, T-bills, etc. (1) Housing had a 4.2 percent nominal annual rate of return (appreciation) over the past 21 years; the consumer price index was 3.7 percent for the same time period (p. 259). It is dangerous for homeowners to devote too much of their wealth to an asset that has low historical return and a serious risk of loss over multiple-year horizons. If the government actively encourages homeownership, it has the responsibility to inform potential homeowners of the risks (p. 272). Doesn't consider implicit rent.

McCarthy, Van Zandt & Rohe (2001) Literature review. US households and housing markets. Literature review. Economic benefits and costs of homeownership on an individual and societal level. Homeownership. (1) High homeownership rates cause an inflexibility in large portions of the population and capital stock, leading to efficiency losses in the economy. Behavioral studies on tenure choice are much more readily available than research on the benefits and costs of homeownership. Although most households, including lower- and moderate-income households, benefit through home purchase, there has been little research that carefully measures and compares the relative benefits of ownership across socioeconomic groups (p. 44). Promoting homeownership as an across-the-board policy may not be the best route. Future research should focus on ways to preserve homeownership, develop a better understanding of the housing market, and examine how homeownership affects other economic variables.

Moore et al. (2001) IDA survey data. 318 current and former IDA participants. Descriptive. Economic effects as measured by (1) decreasing consumption (p. 21). IDA Program Participation, which includes saving accounts, economic education, expectations for saving behavior, incentives to save, staff and peer support, etc (p. 46) (1) Thirty percent of participants said they had less money for leisure, eight percent said they had to give up food or other necessities (p. 20). (2) nine percent said that they felt more stressful about the future (p. 19). (3) 35 percent said they were less likely to save outside their IDAs (p. 47). Cross-sectional survey data with self-reported effects.

Reid (2004) PSID and 55 low-income homeowners in Seattle, Washington. Low-income homeowners. Logit regressions (discrete time hazard rate) and qualitative interviews. Exit from home ownership to renting. Race & income. (1) In the first three years of home ownership, low-income homeowners are particularly at risk of moving from homeownership to renting (p. 20). (2) For low-income homeowers, the financial returns to homeownership even after 10+ years were lower than the return on treasury bills (p. 29). (1) Homeowners run a significant risk of returning to renting in the first years of homeownerhsip. This may be because they lack the savings to cope with crises such as unemployment or health problems (p. 32). (2) Low-income homeowners do not see the same levels of house price appreciation as do higher income homeowners (p. 28)

Race, Income Group, and Location

Baker (2005) Potential low-income homeowners. Calculates the costs of renting vs. homeownership in 3 hypothetical housing market situations. Total costs incurred from homeownership or renting. Homeownership, relative to renting may not be as desirable for low-income families for three reasons. First, low-income families do not have tax liabilities that are high enough to benefit from mortgage interest deductions. Second, the median period of homeownership is only four years meaning that transaction costs are higher for low-income families. Third, it is likely that in areas that have experienced a significant run up in housing prices, in real terms, homebuyers will sell their houses for less than they paid (p. 4). In a scenario where home prices rise in step with inflation, a typical homeowner loses (relative to renting) an amount equal to approximately 25 percent of their total rent over 4 years (p. 2).

Baker & Baribeau (2003) Potential low-income homeowners. Calculates the potential losses possible when purchasing a home during a real estate bubble. Total costs incurred from homeownership or renting. While homeownership might be desirable in normal times, encouraging moderate income families to purchase homes during a housing bubble may not be optimal. The current situation implies that the price of many homes low-income families would seek to buy are likely to fall in value when the real estate bubble collapses. This would lead to enormous losses for low-income families (p. 2).

Case and Marynchenko (2002) OFHEO and Case-Shiller sales indexes; American Housing Survey (AHS) home value data. Housing markets by region, specifically in Chicago, Boston and Los Angeles; households in these regions. Descriptive (though briefly mentions "exploratory regressions" (p. 255)). Housing price appreciation. Homeownership. For low-income households in Chicago and Boston, homeownership has been a good investment for asset accumulation since 1987 and the early 1980s, respectively. This can also be said for Chicago, Boston, and Los Angeles in 1995. There have been, however, significant periods of decline in Boston and Los Angeles that have led to losses and periods of negative equity for low-income households (p. 248, 252, 255). (1) Whether homeownership is a worthwhile investment depends on the time of purchase, conditions in the regional economy, and local supply and demand. (2) Since home purchase is particularly and especially leveraged among low-income households, these households face substantial shifts in equity accumulation when housing prices change. (3) Appreciation is an important component in the overall return to housing, but from the data, no general conclusion can be reached on whether homeownership is a good or bad strategy for asset accumulation (p. 255).

Dietz and Haurin (2003) Literature review. US households. Literature review. The impact of homeownership on private and social micro-level behavior. Homeownership. (1) Econometric problems in the existing research include omitted variable bias and unobserved factors that influence the tenure decision and subsequent behavior. Most pre-1990 studies are econometrically unreliable. (2) Existing papers report mixed results regarding homeownership benefits and geographical location, race, and the value of housing. (3) Homeowners are less mobile than renters, which may negatively impact their labor market responses. (4) Homeownership has positive effects on child outcomes, though literature on household fertility and divorce is sparse, and literature on health is econometrically lacking. There are many gaps in the existing literature about the consequences of homeownership. These gaps include a lack of analysis on the impact of structure type; the impact of homeownership in developing countries; the effect of homeownership by race, ethnicity and income; an examination of how tax law changes affect behavior; and other demographic, social and political consequences, particularly including negative consequences. Also, older results must be reevaluated using updated techniques and data (p. 439-440).

Duda & Belsky (2001) Raw data used to construct the Case, Schiller, and Weiss repeat sales indices. Low-income and other homeowners in 4 MSAs. Comparisons of home purchase and resale values. Net profit on resold homes. Type of home. Superior performance of resellers can be attributed to both the timing of purchase and sale of a home. Low-cost homes were generally purchased during downturns and sold during upturns (23). Homeownership was relatively less risky for those purchasing low-cost rather than more expensive homes (purchased after 1982 and resold by 1999). Superior performance of low-cost home investments can be attributed to both the timing of purchase and sale of the home. In a single home purchase/home sale round a not insignificant number of households sell their homes for a loss, implying that it is important to help owners weather downturns, etc (p. 23). Only looks at homes both purchased after 1982 and resold by 1999.

McCarthy, Van Zandt & Rohe (2001) Literature review. US households and housing markets. Literature review. Economic benefits and costs of homeownership on an individual and societal level. Homeownership. (1) Lower-income and minority families hold larger portions of their wealth in housing, and so are more susceptible to extreme price movements, and also face higher costs (p. 43). Behavioral studies on tenure choice are much more readily available than research on the benefits and costs of homeownership. Although most households, including lower- and moderate-income households, benefit through home purchase, there has been little research that carefully measures and compares the relative benefits of ownership across socioeconomic groups (p. 44). Promoting homeownership as an across-the-board policy may not be the best route. Future research should focus on ways to preserve homeownership, develop a better understanding of the housing market, and examine how homeownership affects other economic variables.

Nichols (2005) American Housing Survey (AHS). 9,711 U.S. households with fixed rate mortgages who have not refinanced or moved (from AHS 1985-2002). (1) Calculate compound annual rate of return that incorporates appreciation rate, implicit rent, mortgage contract, and mortgage income tax deduction (p. 23); (2) Logistic regressions of the probability of negative total dollar return to homeownership (p. 31); (3) OLS regressions of measures of housing return (conditional on positive return) and the appreciation rate (p. 31). Total returns on housing. Homeownership. (1) The mortgage interest tax deduction increases the returns to homeownership for higher-income borrowers. (2) Single-family homes provide higher rates of return. (3) Households that have low income, low education and are black have a higher probability of negative return and lower rates of return and appreciation. (4) The mortgage contract augments the negative effect of low income, low education, minority status and property type on returns to housing.

Reid (2004) PSID and 55 low-income homeowners in Seattle, Washington. Low-income renters and homeowners. Logit regressions, comparisons of means before and after home purchase, and qualitative interviews. Changes in neighborhood characteristics, changes in home value, & exit from home ownership to renting. Homeownership. The benefits of homeownership are not distributed evenly across class or race (p. 33). (1) Homeownership brings no significant improvements in neighborhood characteristics for low-income white households, marginal improvements for middle- and high-income white and minority households, and the greatest improvements for low-income minority households (p. 18). (2) The financial returns to homeownership are extremely small for low-income minorities and whites, and middle-income minorities. In contrast, the returns are higher for middle- and high-income whites (p. 29). (3) The risk of homeowners returning to the rental market is extremely high in the first 3 years for low-income respondents, high for middle-income respondents, and stable for high-income respondents (p. 20).

Scanlon & Page-Adams (2000). NSAF. US Households with children. GLM and hierarchical regressions. Child's extracurricular activities & behavioral problems. Homeownership. Homeownership is positively associated with decreased behavioral problems and increased participation in extracurricular activities among adolescents, however the relationships are not found, or the associations are less strong for black, Hispanic, and poor children (p. 11).

Williams (2003) PSID. 2,936 children age 0-12 who live with their biological or adopted mothers. OLS regression, ANOVA, logistic and ordered logit regressions in hierarchical blocks. Cognitive development/academic achievement, physical health, and socio-emotional behavior/schooling variables. Income, wealth, homeownership. Household income and wealth's association with child cognitive development, health, and behavioral development differs substantially by race (p. 88-89). Household income and wealth influence child outcomes in cognitive development, health, and behavioral development/schooling, but the manner in which wealth influences outcomes differs by race. Assets are mediated by intermediate variables regarding neighborhood characteristics and parental quality. Benefits of assets extend to both low and high-income households (pp. 88-89).

Appendix Table.
Empirical Studies of Effects of Asset Holding
Author	Data Source	Sample/Study Population	Method	Outcomes Analyzed	Key Asset Explanatory Variables	Findings	Author’s Principal Conclusions	Key Weaknesses²
Economic Well-Being
Bynner & Despotidou (2001)	National Child Development Study (U.K.).	12,000 British individuals born in 1988.	OLS regression	Labor market experience at age 23-33: years in full-time employment, years unemployed, self-employed, .…	Savings and investments at age 23.	Asset holding is associated with positive employment outcomes (p. 3).	Labor market experience appears to be strongly affected by assets. The coefficient on asset measures is especially strong when the outcome is unemployment (p. 16).	Endogeneity.
Carasso and McKernan (2005)	Survey of Consumer Finances	US households.	Literature review.	Assets, liabilities, and networth	Homeownership.	In 2001, homeowners had total asset holdings 16 times greater than renters ($210,400 vs $12,700); debt 12 times greater than renters (69,400 vs 6000) but were 4 times less likely to be delinquent on debt, and net worth 36 times greater than renters (171,700 vs 4,800).	Homeownership status made a larger difference in net worth than income, education, single-family status, and minority status.
Cho (1999)	NLSY.	363 women experiencing marital disruption between 1985 and 1989.	Instrumental variables regression where assets are instrumented with their own lags. Seemingly unrelated regression.	Log total family income, log per capita family income, log earned income, log welfare income, & log private transfers.	Log financial assets, log home equity, log value of business or real property.	Financial assets are positively associated with the economic well-being of women one year after marital disruption. Home equity and value of business/real property are not statistically, significantly related.	Financial assets (as measured in the post-marital disruption year) have a positive effect on income/earnings and reduce welfare receipt. Asset accumulation could be a positive strategy for women and children at risk of marital disruption (abstract).	Instrument for assets with poor instruments? Include other endogenous variables in model (human capital, social support). Could these bias instrumented variable? It is strange that they use per capita income and total family income as dependent variables. Presumably in many cases the husband was a contributor to these variables before divorce so it isn't surprising that they fall.
Goetzmann & Spiegel (2002)	Office of Housing Enterprise Oversight housing price indices.	Sales of homes in the United States from 1980-2000.	Descriptive analysis comparing returns on financial investments.	Returns on home sales & financial investments.	Returns on housing, returns on financial investments such as S&P 500, Treasury bills, inflation, T-bills, etc.	(1) Housing had a 4.2 percent nominal annual rate of return (appreciation) over the past 21 years; the consumer price index was 3.7 percent for the same time period (p. 259).	It is dangerous for homeowners to devote too much of their wealth to an asset that has low historical return and a serious risk of loss over multiple-year horizons. If the government actively encourages homeownership, it has the responsibility to inform potential homeowners of the risks (p. 272).	Doesn't consider implicit rent.
Jencks & Mayer (1989)	Chicago survey.	1,617 non-hispanic blacks and whites pooled from surveys in 1983 and 1985.	Regressions (likely OLS).	Number of material hardships: couldn't afford food, no insurance, unmet medical care, housing problems, & cut-off utilities.	Home ownership and access to credit as measured by ability to borrow $500 if needed.	(1) Homeowners reported fewer hardships than tenants with the same income and needs. Homeownership may be a proxy for unmeasured economic resources and unmeasured forms of household efficiency (p. 108). (2) All else equal, a family's ability to borrow $500 in the event of an emergency did as much to reduce hardship as tripling family income (p. 109).	Poverty statistics do not provide reliable information about material hardship. Family income is not the only critical determinant of material well-being--it explains only 14 percent of the variance in material hardship. Broader measures of economic resources explain only a little more. Better measures of both savings and credit would improve our ability to predict hardship, but probably not enough to alter basic findings (p. 111).
McCarthy, Van Zandt & Rohe (2001)	Literature review.	US households and housing markets.	Literature review.	Economic benefits and costs of homeownership on an individual and societal level.	Homeownership.	(1) House prices appreciated an annual average of 3.5% (city-based) and 4% (suburban) from 1987-1997, however this rate is very uneven across areas and over time (p. 23). (2) Homeowners hold a larger-than-optimal portfolio share in housing, exposing owners to higher overall portfolio risk (p. 43).	Behavioral studies on tenure choice are much more readily available than research on the benefits and costs of homeownership. Although most households, including lower- and moderate-income households, benefit through home purchase, there has been little research that carefully measures and compares the relative benefits of ownership across socioeconomic groups (p. 44). Promoting homeownership as an across-the-board policy may not be the best route. Future research should focus on ways to preserve homeownership, develop a better understanding of the housing market, and examine how homeownership affects other economic variables.
Mills et al. (2006)	Random assignment IDA survey data.	840 randomly assigned IDA program participants (treatment) and non-participants (control) individuals.	Experimental program design with OLS and probit regressions.	Homeownership and equity, non-retirement financial assets, business ownership and equity, enrolling in courses, computer purchase, retirement savings, home improvement, net worth.	Eligibility to participate in the IDA program	Eligibility for the IDA program: (1) Raised homeownership rates by almost 10 percentage points over 4 years for black renters, but reduced financial assets and business ownership. (2) Had no effects on homeownership for white renters, but their business equity rose. (3) Overall had no statistically significant effect on net worth, among other findings.	The IDA program: (1) Raised homeownership rates by almost 10 percentage points over 4 years for black renters, but reduced financial assets and business ownership. (2) Had no effects on homeownership for white renters, but business equity rose. (3) Home owners use the IDA in different ways than renters (abstract).
Moore et al. (2001)	IDA survey data.	318 current and former IDA participants.	Descriptive.	Economic effects as measured by (1) decreasing consumption, (2) purchasing a home or starting/expanding a business, and (3) increasing employment (p. 21).	IDA Program Participation, which includes saving accounts, economic education, expectations for saving behavior, incentives to save, staff and peer support, etc (p. 46)		(1) Thirty percent of participants said they had less money for leisure, eight percent said they had to give up food or other necessities (p. 20), 35 percent said they were less likely to save outside their IDAs. (2) Almost 75 percent said they were more likely to purchase or renovate a home and fifty-seven percent said they were more likely to start or expand a business. (3) Forty-one percent said they were more likely to increase work hours and 61 percent said they were more likely to increase their income in other ways (p. 47).	Cross-sectional survey data with self-reported effects.
Nichols (2005)	American Housing Survey (AHS).	9,711 U.S. households with fixed rate mortgages who have not refinanced or moved (from AHS 1985-2002).	(1) Calculate compound annual rate of return that incorporates appreciation rate, implicit rent, mortgage contract, and mortgage income tax deduction (p. 23); (2) Logistic regressions of the probability of negative total dollar return to homeownership (p. 31); (3) OLS regressions of measures of housing return (conditional on positive return) and the appreciation rate (p. 31).	Total returns on housing.	Homeownership.	(1) The annualized average return to homeownership is 1.9% based on appreciation alone, 9.8% based on appreciation and implicit rent; 8.5% based on appreciation, implicit rent, and mortgage contract, and 8.6% based on appreciation, implicit rent, mortgage contract, and mortgage income tax deduction (table 5.1, p. 23). (2) Single-family homes provide higher rates of return.
Raphael & Rice (2002)	SIPP.	16-65 year old non-disabled individuals living in households with zero to three cars.	OLS and Instrumental variables regressions (2SLS) where car ownership is instrumented by state insurance rates and gas taxes. OLS and 2SLS estimates are similar for employment and hours worked outcomes, but differ for the wage outcome.	Employment, hours worked, log wages.	Car ownership.	Car ownership increases employment and hours worked, but not necessarily wages (p. 124).	Large positive differences in employment rates and hours worked between those who do and do not own cars in both OLS and instrumented regressions. Instrumenting eliminates the positive impact car ownership has on wages in OLS regression (abstract).
Reid (2004)	PSID and 55 low-income homeowners in Seattle, Washington.	Low-income renters and homeowners.	Descriptive comparisons of means before and after home purchase.	Changes in home value and exit from homeownership to renting.	Homeownership.	For low-income homeowners, the financial returns to homeownership even after 10+ years averaged only 3 percent per year. For middle- and high-income whites, the returns averaged 5 percent per year (p. 29).	Low-income homeowners do not see the same levels of house price appreciation as do higher income homeowners (p. 28). For low-income homeowners, the financial returns to homeownership even after 10+ years were lower than the return on Treasury bills.
Rossi & Weber (1996)	GSS, NSFH, & ANES.	1,500 (GSS), 13,000 (NSFH), 13,000 (ANES) US Households.	Descriptive OLS and logit regressions.	8 measures of household finances, including assets and liabilities.	Homeownership.	(1) Total household income of homeowners is more than $17,000 greater than renters (p. 12). (2) Homeowners have about $6,000 more in savings and about $5,000 more in mutual funds than renters (p. 12). (3) Homeowners are more likely to have credit card debt, installment debt, and personal bank loans. Homeowners are less likely to have education loans and to have bills more than 90 days overdue (p. 13).	Claims for some social and individual benefits from homeownership are only weakly supported. Strong differences between owners and renters exist only in demographic characteristics (p. 1).
Sullivan (2005).	PSID & SIPP.	9,400 (SIPP) and 15,666 (PSID) Working-age household heads with steady employment.	OLS and Instrumental variables regressions where earnings are instrumented with unemployment spells (p. 6). Assets are not instrumented for but are measured at an initial period (p. 7).	Changes in unsecured debt & changes in food and housing consumption.	Unemployment induced earnings losses. Low-asset holdings as measured by non-positive financial assets, non-positive total gross assets, and total gross assets to earnings ratio of less than 0.12 (p. 7).		Households with assets use unsecured debt to borrow when faced with temporary shortfalls in earnings. Households with low-assets do not borrow--likely because they do not have sufficient access to unsecured credit--and their consumption falls (abstract).
Social Well-Being
DiPasquale & Glaeser (1999)	General Social Survey & German Socio-Economic Survey.	1,500 U.S. households and 13,000 German households	OLS and instrumental variables regressions where homeownership is instrumented with the average homeownership rate by income quartile, race, and state (p. 356). Individual-level fixed effects.	Social capital and good citizenship as measured by: number of nonprofessional organizations affiliated with, political knowledge, voting, participation in community planning, gardening, gun ownership, and church attendance.	Homeownership.		Both in the U.S. and in Germany, homeownership is strongly correlated with variables that attempt to measure good citizenship and social capital. A large portion of the effect of homeownership comes from lower mobility rates for homeowners. Standard economic incentives (both the effects of ownership and tenure) influence investment in social capital (p. 383).
Haurin, Dietz & Weinberg (2003)	Literature review.	US households.	Literature review.	The impact of neighborhood homeownership rates on residents.	Homeownership.		Theoretical discussions of neighborhood homeownership effects supports the hypothesis that homeownership rates can affect behavior. However, relatively few studies have rigorously empirically tested the theoretical hypothesis put forth. Attention should be given to determining causal and noncausal correlated effects and determining whether feedback effects are present. Further examination of neighborhood effects has relevance for social welfare policies.
Moore et al. (2001)	IDA survey data.	318 current and former IDA participants.	Descriptive.	Perceived social and civic effects as measured by (1) more problems and more good relationships with family, (2) more problems with neighbors and more involvement in neighborhood, and (3) more respected in community (p. 21).	IDA Program Participation, which includes saving accounts, economic education, expectations for saving behavior, incentives to save, staff and peer support, etc (p. 46)		(1) About half of current participants said they were more likely to have good relationships with family members, (2) about one-third said they were more likely to be involved in their neighborhoods, and (3) about one-third said they were more likely to be respected in their communities (p. 47).	Cross-sectional survey data with self-reported effects.
Reid (2004)	PSID and 55 low-income homeowners in Seattle, Washington.	Low-income renters and homeowners.	Comparisons of means before and after home purchase.	Changes in neighborhood characteristics	Homeownership.		Neighborhood benefits for shifting from renting to owning were minimal for low-income whites, but substantial for low-income minorities (p. 18).
Rossi & Weber (1996)	GSS, NSFH, & ANES.	1,500 (GSS), 13,000 (NSFH), 13,000 (ANES) US Households.	Descriptive OLS and logit regressions.	(1) 22 measures of sociability (p. 16). (2) 64 measures of marriage and family issues (p.19).	Homeownership.	(1) Homeowners are not consistently more likely to be members of social networks (p. 17). (2) Homeowners-renter differences on marriage and family behavior and value issues are not very great (p. 20).	Claims for some social and individual benefits from homeownership are only weakly supported. Strong differences between owners and renters exist only in demographic characteristics (p. 1).
Civic Engagement
Bynner & Despotidou (2001)	National Child Development Study (U.K.).	12,000 British individuals born in 1988.	OLS regression.	Citizenship and values: voting, political interest, political cynicism, work ethic.	Savings and investments at age 23.	Although voting showed no association with assets, political interest is positively associated with asset holding (p. 16).	Although voting showed no relation to assets, political interest is likely to be greater with asset holding (p. 16).	Endogeneity.
DiPasquale & Glaeser (1999)	General Social Survey & German Socio-Economic Survey.	1,500 U.S. households and 13,000 German households	OLS and instrumental variables regressions where homeownership is instrumented with the average homeownership rate by income quartile, race, and state (p. 356). Individual-level fixed effects.	Social capital and good citizenship as measured by: number of nonprofessional organizations affiliated with, political knowledge, voting, participation in community planning, gardening, gun ownership, and church attendance.	Homeownership		Both in the U.S. and in Germany, homeownership is strongly correlated with variables that attempt to measure good citizenship and social capital. A large portion of the effect of homeownership comes from lower mobility rates for homeowners. Standard economic incentives (both the effects of ownership and tenure) influence investment in social capital (p. 383).
Rossi & Weber (1996)	GSS, NSFH, & ANES.	1,500 (GSS), 13,000 (NSFH), 13,000 (ANES) US Households.	Descriptive OLS and logit regressions.	27 measures of political behavior (p. 24).	Homeownership.	(1) Homeowners and renters are not very different in general political interest (p. 23). (2) Homeowners are almost consistently more engaged in local politics than renters but the activism does not extend to all area (p. 23).	Claims for some social and individual benefits from homeownership are only weakly supported. Strong differences between owners and renters exist only in demographic characteristics (p. 1).	Endogeneity.
Child Well-Being
Aaronson (1999)	PSID.	5,143 children reaching the age of 17 between 1975 and 1993.	Probit regressions and instrumental variables regressions where homeownership is instrumented with the average homeownership rate by income quartile, race, and state (p. 366).	High school graduate by age 19.	Homeownership.		Some of the homeownership effect found by Green & White (1997) is driven by family characteristics such as residential stability that are correlated with homeownership. Nevertheless, in the sense that homeownership increases residential stability, it appears to be correlated with education attainment (p. 356).	Endogeneity of residential mobility and imperfect controls for the endogeneity of homeownership.
Bynner & Despotidou (2001)	National Child Development Study (U.K.).	12,000 British individuals born in 1988.	OLS regression	Parenting at age 37: number of books child has, frequency of reading to child, frequency child reads.	Savings and investments at age 23.	Asset holding is not associated with parenting outcomes as measured by child's reading.	Parenting outcomes did not appear to be improved by the presence of assets (p. 16).	Endogeneity?
Dietz and Haurin (2003)	Literature review.	US households.	Literature review.	The impact of homeownership on private and social micro-level behavior.	Homeownership.	The literature contains solid studies that report positive direct and indirect effects of homeownership on child outcomes (p. 439).	There is substantial evidence that homeownership has important effects on some household behaviors and outcomes. However, much of the past 30-year's literature on consequences of homeowning is deficient from a theoretical or econometric perspective (abstract).
Green & White (1997)	PSID, 1980 PUMS, & High School and Beyond.	17 or 18 year old youths.	Probit models and bivariate probit models that instrument for the endogeneity of homeownership with the relative cost of owning versus renting (p. 455). The correlation coefficients from the bivariate probit estimates suggest that homeownership is not an endogenous variable for the two outcomes considered (p. 456).	Attending school at age 18 & having a child by age 18.	Homeownership, housing tenure.	Homeownership decreases the likelihood that children of homeowners have children and increases the likelihood that children of homeowners are in school at age 18. Findings are particularly important for low-income households (p. 441).	Children of homeowners are less likely to have children and more likely to be in school at age 18. The dollar benefit per low-income household of parents being homeowners is at least $31,000 (p. 441).
Haurin, Parcel, & Haurin (2001)	NLSY79 and NLSY Children.	Mothers age 14-22 in 1979.	Random effects regressions with an unknown instrument to control for the endogeneity of homeownership (p.3).	(1) Child's home environment as measured by emotional and cognitive/physical scores. (2) A measure of child's behavior problems. (3) Child cognition as measured by mathematics and reading scores (p. 10).	Homeownership.	Homeownership is positively and statistically, significantly associated with an improved home environment, but only marginally statistically, significantly associated or not statistically, significantly associated with child cognition and child behavior (p. 13-14).	Children of homeowners have better home environments, higher cognitive test scores, and fewer behavior problems than do children of renters. Owning a home leads to a 13 to 23 percent higher quality home environment, all else equal (abstract). Homeownership increases cognitive math test scores 9 percent and cognitive reading scores 7 percent relative to renters. Homeownership reduces the index of child behavior problems by about 3% (pp. 13-14).	Not clear that controls for endogeneity. First stage regressions not shown and not clear what instrument is.
Henretta (1984)	PSID.	Individuals in the 1980 PSID who are heads of household and were sons or daughters in previous waves.	Logit and OLS regressions.	Child's homeownership and child's home value.	Parent's income, homeownership, and home value for the last year the child was in the parent's home.	(1) Parent's homeownership is positively and statistically, significantly associated with child's homeownership, for both whites and blacks (p. 134). (2) Parent's home value is positively and statistically associated with child's home value for whites, but not blacks (p. 135). (3) Parental gifts have no statistically significant association with child's equity or mortgage for whites, but do have a positive and statistically significant association for blacks (p. 137).	Tentatively interprets the findings to suggest that the effect of parental home value reflects a socialization process in which the child's living situation affects the child's expectations concerning the proper or appropriate standard of living (p. 138). For blacks however, parental gifts (and not just child's expectations) appear to play a bigger role in determining a child's homeownership and value choices (p. 138).	Endogeneity of parental income, homeownership, and home equity.
Kane (1994)	CPS.	Average sample of 500 dependent black and 3,250 white 18-19 year olds per year (p. 883).	Full information maximum likelihood probit models to control for the endogeneity of parent's education (p. 887).	Child's high school graduation and college entry.	Parent's homeownership.	Parent's homeownership is positively and significantly associated with child's high school graduation and college enrollment for blacks and whites (p. 888, 894, 896, 909).	From 1973 to 1988, college costs drove enrollment rates down while increases in parental education levels of blacks exerted upward pressure on increased educational attainment by blacks (p. 878).	Does not control for the endogeneity of family resources (income, homeownership, employment).
McGarry & Schoeni (1995)	HRS.	HRS respondents and their children who are not living at home.	Logit, OLS, and fixed effect regressions.	Whether parents transfer cash to their children and the amount. Whether children transfer cash and time to their parents and the amount of each.	Child's total income and homeownership status, parent's financial situation and homeownership status.	(1) Wealthy parents transfer larger sums of cash to their children, but the finding is not statistically significant in the fixed effects model (p. S206, S209). (2) Parents transfer larger sums of cash to lower income children (p. S208). (3) Wealthy children transfer larger sums of cash to their parents, but the finding is not statistically significant in the fixed effects model (p. S219). (4) Children transfer larger sums of cash to parents with worse financial situations and who do not own their home (p. S218).	Parents give greater financial assistance to their less well off children and children give more to less well off parents. These results hold for both incidence of transfer and amounts (p. S184).	Does not state preferred specification? Does not focus findings on fixed effect results.
Scanlon & Page-Adams (2000).	NSAF.	US Households with children.	GLM and hierarchical regressions.	Child's extracurricular activities & behavioral problems.	Homeownership.		Homeownership is positively associated with decreased behavioral problems and increased participation in extracurricular activities among adolescents, however the relationships are not found, or the associations are less strong for black, Hispanic, and poor children (p. 11).
Williams (2003)	PSID.	2,936 children age 0-12 who live with their biological or adopted mothers.	OLS regression, ANOVA, logistic and ordered logit regressions in hierarchical blocks.	Cognitive development/academic achievement, physical health, and socio-emotional behavior/schooling variables.	Income, wealth, homeownership.	Household income and wealth are positively associated with child cognitive development, health, and behavioral development. The manner in which these economic status variables associated with child development outcomes differs substantially by race (p. 88-89).	Household income and wealth influence child outcomes in cognitive development, health, and behavioral development/schooling, but the manner in which wealth influences outcomes differs by race. Assets are mediated by intermediate variables regarding neighborhood characteristics and parental quality. Benefits of assets extend to both low and high-income households (pp. 88-89).
Zhan & Sherraden (2003)	NSFH.	591 children 12-18 years old in female-headed households (p. 195).	OLS regression.	Mother's academic expectations of child, child's academic performance, and whether the child graduates from high school.	Mother's savings, homeownership.	Homeownership is positively and statistically significantly associated with mother's expectations and child's academic performance, but not statistically significantly associated with child's high school graduation. Savings are positively and statistically associated with child's high school graduation and mother's expectations, but not statistically significantly associated with child's academic performance (p. 199, 200, 201).	Single mother's assets are positively associated with mother's expectations and with child's educational achievement. Mother's expectations are positively associated with child's outcome and partially mediates the relationship between assets and the child's outcome. The relationship between income and child's educational achievement disappears when assets are included in the regression, suggesting that models including income but not assets are underspecified (p. 204, 205).
Health and Psychological Well-Being
Bynner & Despotidou (2001)	National Child Development Study (U.K.).	12,000 British individuals born in 1988.	OLS regression.	Health at age 33: malaise, general health, smoking.	Savings and investments at age 23.	Asset holding is associated with positive health outcomes (p. 3).	Health appeared to be strongly asset dependent (p. 16).	Endogeneity.
Moore et al. (2001)	IDA survey data.	318 current and former IDA participants.	Descriptive.	Psychological effects as measured by thoughts, feelings, and plans for the future (p. 19, 20).	IDA Program Participation, which includes saving accounts, economic education, expectations for saving behavior, incentives to save, staff and peer support, etc (p. 46)		Current participants agreed or strongly agreed that they felt more confident about their futures (93 percent), more economically secure (84 percent), and more in control of their lives (85 percent) because they had IDAs (p. 46). Effects on planning were somewhat less common with about 60 percent of respondents saying they were more likely to make educational and retirement plans (p. 47).	Cross-sectional survey data.
Yadama & Sherraden (1996)	PSID.	2,871 individuals who were heads of household in both 1968 and 1972 (p. 5).	Simultaneous estimation of a path model of directly observed variables (p. 5).	(1) Attitudes and behaviors as measured by efficacy, prudence, horizons, connectedness, and effort in 1972. (2) Assets as measured by house value and savings in 1972.	(1) Assets as measured by housevalue, savings, and income in 1968. (2) Attitudes and behavior as measured by prudence, efficacy, horizons, connectedness, and effort in 1968.	(1) Five of the 10 coefficients relating the two asset variables and five attitude and behavior variables were significant, all in the expected direction. Savings had positive effects on prudence, efficacy, horizons, and connectedness; house value had a positive effect on horizons (pp. 9-10). (2) Four of the 10 coefficients relating attitude and behavior to assets were significant: prudence had a positive effect on house value but not savings, efficacy had positive effect on savings, connectedness had a positive effect on savings, and effort had a positive effect on house value (p. 10).	Results support the proposition that assets have a positive effect on expectations and confidence about the future; influence people to make specific plans with regard to work and family; induce more prudent and protective personal behaviors, and lead to more social connectedness with relatives, neighbors, and organizations (p. 3).	Limited timeframe. Simultaneous estimation of a path model of directly observed variables (p. 5). It is not clear how the model is identified.
Rossi & Weber (1996)	GSS, NSFH, & ANES.	1,500 (GSS), 13,000 (NSFH), 13,000 (ANES) US Households.	Descriptive OLS and logit regressions.	Personal well-being: 11 measures of health and psychological well-being (p. 14).	Homeownership.	Homeowners regard themselves as having a greater sense of well-being than renters, but only marginally so. The issue of causality prohibits any claim that ownership leads to a greater sense of well-being (p. 15).	Claims for some social and individual benefits from homeownership are only weakly supported. Strong differences between owners and renters exist only in demographic characteristics (p. 1).	Endogeneity.
Adverse Consequences
Dietz and Haurin (2003)	Literature review.	US households.	Literature review.	The impact of homeownership on private and social micro-level behavior.	Homeownership.	Homeowners are less mobile than renters due to higher transaction costs and perhaps because of greater ties to their neighborhood and community. This lack of mobility may negatively impact the ability of households to respond to changes in the local labor market, but this result is not firmly established (p. 439).	There is substantial evidence that homeownership has important effects on some household behaviors and outcomes. However, much of the past 30-year's literature on consequences of homeowning is deficient from a theoretical or econometric perspective (abstract).
Goetzmann & Spiegel (2002)	Office of Housing Enterprise Oversight housing price indices.	Sales of homes in the United States from 1980-2000.	Descriptive analysis comparing returns on financial investments.	Returns on home sales & financial investments.	Returns on housing, returns on financial investments such as S&P 500, Treasury bills, inflation, T-bills, etc.	(1) Housing had a 4.2 percent nominal annual rate of return (appreciation) over the past 21 years; the consumer price index was 3.7 percent for the same time period (p. 259).	It is dangerous for homeowners to devote too much of their wealth to an asset that has low historical return and a serious risk of loss over multiple-year horizons. If the government actively encourages homeownership, it has the responsibility to inform potential homeowners of the risks (p. 272).	Doesn't consider implicit rent.
McCarthy, Van Zandt & Rohe (2001)	Literature review.	US households and housing markets.	Literature review.	Economic benefits and costs of homeownership on an individual and societal level.	Homeownership.	(1) High homeownership rates cause an inflexibility in large portions of the population and capital stock, leading to efficiency losses in the economy.	Behavioral studies on tenure choice are much more readily available than research on the benefits and costs of homeownership. Although most households, including lower- and moderate-income households, benefit through home purchase, there has been little research that carefully measures and compares the relative benefits of ownership across socioeconomic groups (p. 44). Promoting homeownership as an across-the-board policy may not be the best route. Future research should focus on ways to preserve homeownership, develop a better understanding of the housing market, and examine how homeownership affects other economic variables.
Moore et al. (2001)	IDA survey data.	318 current and former IDA participants.	Descriptive.	Economic effects as measured by (1) decreasing consumption (p. 21).	IDA Program Participation, which includes saving accounts, economic education, expectations for saving behavior, incentives to save, staff and peer support, etc (p. 46)		(1) Thirty percent of participants said they had less money for leisure, eight percent said they had to give up food or other necessities (p. 20). (2) nine percent said that they felt more stressful about the future (p. 19). (3) 35 percent said they were less likely to save outside their IDAs (p. 47).	Cross-sectional survey data with self-reported effects.
Reid (2004)	PSID and 55 low-income homeowners in Seattle, Washington.	Low-income homeowners.	Logit regressions (discrete time hazard rate) and qualitative interviews.	Exit from home ownership to renting.	Race & income.	(1) In the first three years of home ownership, low-income homeowners are particularly at risk of moving from homeownership to renting (p. 20). (2) For low-income homeowers, the financial returns to homeownership even after 10+ years were lower than the return on treasury bills (p. 29).	(1) Homeowners run a significant risk of returning to renting in the first years of homeownerhsip. This may be because they lack the savings to cope with crises such as unemployment or health problems (p. 32). (2) Low-income homeowners do not see the same levels of house price appreciation as do higher income homeowners (p. 28)
Race, Income Group, and Location
Baker (2005)		Potential low-income homeowners.	Calculates the costs of renting vs. homeownership in 3 hypothetical housing market situations.	Total costs incurred from homeownership or renting.			Homeownership, relative to renting may not be as desirable for low-income families for three reasons. First, low-income families do not have tax liabilities that are high enough to benefit from mortgage interest deductions. Second, the median period of homeownership is only four years meaning that transaction costs are higher for low-income families. Third, it is likely that in areas that have experienced a significant run up in housing prices, in real terms, homebuyers will sell their houses for less than they paid (p. 4). In a scenario where home prices rise in step with inflation, a typical homeowner loses (relative to renting) an amount equal to approximately 25 percent of their total rent over 4 years (p. 2).
Baker & Baribeau (2003)		Potential low-income homeowners.	Calculates the potential losses possible when purchasing a home during a real estate bubble.	Total costs incurred from homeownership or renting.			While homeownership might be desirable in normal times, encouraging moderate income families to purchase homes during a housing bubble may not be optimal. The current situation implies that the price of many homes low-income families would seek to buy are likely to fall in value when the real estate bubble collapses. This would lead to enormous losses for low-income families (p. 2).
Case and Marynchenko (2002)	OFHEO and Case-Shiller sales indexes; American Housing Survey (AHS) home value data.	Housing markets by region, specifically in Chicago, Boston and Los Angeles; households in these regions.	Descriptive (though briefly mentions "exploratory regressions" (p. 255)).	Housing price appreciation.	Homeownership.	For low-income households in Chicago and Boston, homeownership has been a good investment for asset accumulation since 1987 and the early 1980s, respectively. This can also be said for Chicago, Boston, and Los Angeles in 1995. There have been, however, significant periods of decline in Boston and Los Angeles that have led to losses and periods of negative equity for low-income households (p. 248, 252, 255).	(1) Whether homeownership is a worthwhile investment depends on the time of purchase, conditions in the regional economy, and local supply and demand. (2) Since home purchase is particularly and especially leveraged among low-income households, these households face substantial shifts in equity accumulation when housing prices change. (3) Appreciation is an important component in the overall return to housing, but from the data, no general conclusion can be reached on whether homeownership is a good or bad strategy for asset accumulation (p. 255).
Dietz and Haurin (2003)	Literature review.	US households.	Literature review.	The impact of homeownership on private and social micro-level behavior.	Homeownership.	(1) Econometric problems in the existing research include omitted variable bias and unobserved factors that influence the tenure decision and subsequent behavior. Most pre-1990 studies are econometrically unreliable. (2) Existing papers report mixed results regarding homeownership benefits and geographical location, race, and the value of housing. (3) Homeowners are less mobile than renters, which may negatively impact their labor market responses. (4) Homeownership has positive effects on child outcomes, though literature on household fertility and divorce is sparse, and literature on health is econometrically lacking.	There are many gaps in the existing literature about the consequences of homeownership. These gaps include a lack of analysis on the impact of structure type; the impact of homeownership in developing countries; the effect of homeownership by race, ethnicity and income; an examination of how tax law changes affect behavior; and other demographic, social and political consequences, particularly including negative consequences. Also, older results must be reevaluated using updated techniques and data (p. 439-440).
Duda & Belsky (2001)	Raw data used to construct the Case, Schiller, and Weiss repeat sales indices.	Low-income and other homeowners in 4 MSAs.	Comparisons of home purchase and resale values.	Net profit on resold homes.	Type of home.	Superior performance of resellers can be attributed to both the timing of purchase and sale of a home. Low-cost homes were generally purchased during downturns and sold during upturns (23).	Homeownership was relatively less risky for those purchasing low-cost rather than more expensive homes (purchased after 1982 and resold by 1999). Superior performance of low-cost home investments can be attributed to both the timing of purchase and sale of the home. In a single home purchase/home sale round a not insignificant number of households sell their homes for a loss, implying that it is important to help owners weather downturns, etc (p. 23).	Only looks at homes both purchased after 1982 and resold by 1999.
McCarthy, Van Zandt & Rohe (2001)	Literature review.	US households and housing markets.	Literature review.	Economic benefits and costs of homeownership on an individual and societal level.	Homeownership.	(1) Lower-income and minority families hold larger portions of their wealth in housing, and so are more susceptible to extreme price movements, and also face higher costs (p. 43).	Behavioral studies on tenure choice are much more readily available than research on the benefits and costs of homeownership. Although most households, including lower- and moderate-income households, benefit through home purchase, there has been little research that carefully measures and compares the relative benefits of ownership across socioeconomic groups (p. 44). Promoting homeownership as an across-the-board policy may not be the best route. Future research should focus on ways to preserve homeownership, develop a better understanding of the housing market, and examine how homeownership affects other economic variables.
Nichols (2005)	American Housing Survey (AHS).	9,711 U.S. households with fixed rate mortgages who have not refinanced or moved (from AHS 1985-2002).	(1) Calculate compound annual rate of return that incorporates appreciation rate, implicit rent, mortgage contract, and mortgage income tax deduction (p. 23); (2) Logistic regressions of the probability of negative total dollar return to homeownership (p. 31); (3) OLS regressions of measures of housing return (conditional on positive return) and the appreciation rate (p. 31).	Total returns on housing.	Homeownership.	(1) The mortgage interest tax deduction increases the returns to homeownership for higher-income borrowers. (2) Single-family homes provide higher rates of return. (3) Households that have low income, low education and are black have a higher probability of negative return and lower rates of return and appreciation. (4) The mortgage contract augments the negative effect of low income, low education, minority status and property type on returns to housing.
Reid (2004)	PSID and 55 low-income homeowners in Seattle, Washington.	Low-income renters and homeowners.	Logit regressions, comparisons of means before and after home purchase, and qualitative interviews.	Changes in neighborhood characteristics, changes in home value, & exit from home ownership to renting.	Homeownership.		The benefits of homeownership are not distributed evenly across class or race (p. 33). (1) Homeownership brings no significant improvements in neighborhood characteristics for low-income white households, marginal improvements for middle- and high-income white and minority households, and the greatest improvements for low-income minority households (p. 18). (2) The financial returns to homeownership are extremely small for low-income minorities and whites, and middle-income minorities. In contrast, the returns are higher for middle- and high-income whites (p. 29). (3) The risk of homeowners returning to the rental market is extremely high in the first 3 years for low-income respondents, high for middle-income respondents, and stable for high-income respondents (p. 20).
Scanlon & Page-Adams (2000).	NSAF.	US Households with children.	GLM and hierarchical regressions.	Child's extracurricular activities & behavioral problems.	Homeownership.		Homeownership is positively associated with decreased behavioral problems and increased participation in extracurricular activities among adolescents, however the relationships are not found, or the associations are less strong for black, Hispanic, and poor children (p. 11).
Williams (2003)	PSID.	2,936 children age 0-12 who live with their biological or adopted mothers.	OLS regression, ANOVA, logistic and ordered logit regressions in hierarchical blocks.	Cognitive development/academic achievement, physical health, and socio-emotional behavior/schooling variables.	Income, wealth, homeownership.	Household income and wealth's association with child cognitive development, health, and behavioral development differs substantially by race (p. 88-89).	Household income and wealth influence child outcomes in cognitive development, health, and behavioral development/schooling, but the manner in which wealth influences outcomes differs by race. Assets are mediated by intermediate variables regarding neighborhood characteristics and parental quality. Benefits of assets extend to both low and high-income households (pp. 88-89).

Where to?

Top of Page
Main Page of Report | Contents of Report

Home Pages:
Human Services Policy (HSP)
Assistant Secretary for Planning and Evaluation (ASPE)
U.S. Department of Health and Human Services (HHS)

Last updated: 03/13/2009

Effects of Holding Assets on Social and Economic Outcomes of Families

Appendix: Empirical Studies of Effects of Asset Holding

Where to?

Appendix:
Empirical Studies of Effects of Asset Holding