# Influence of Household Attitudes on the Joint Mobility-Homeownership Decision

^{[ to cite ]:}

Steven D. Silver (1986) ,"Influence of Household Attitudes on the Joint Mobility-Homeownership Decision", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 263-267.

^{[ direct url ]:}

http://acrwebsite.org/volumes/6500/volumes/v13/NA-13

The relationship of household attitudes to the major consumption-saving decision of homeownership is addressed. Findings of logit models of ownership are reviewed and the joint dependency between ownership and expected mobility is.noted.

This joint dependency is then modeled in a conditional logit specification with attitude factors as explanatory variables. Results of the model estimation support hypotheses on direct and indirect influences of attitudes on ownership. The potential policy relevance of the findings is discussed.

INTRODUCTION

The relevance of attitudes to major household consumption decisions has been discussed in studies that now span several decades (e.g., Katona. 1951; Pickering 1981). There is, however, limited empirical support for relationships hypothesized in these studies. The present study will consider the relationship between attitudes and the consumption-savings decision of homeownership. This decision is generally discussed as choice of ownership or contract rent forms of housing tenure and is among the most important household decisions in the consumption of housing services and the management of household financial assets.

The mainstay of existing studies estimate single equation logit or probit models of homeownership. Results of these studies have indicated income, the relative price of ownership and rental fores of tenure, family size, and age and sex of the household to be significant explanatory variables (e.g., Kain and Quigley 1975; Li 1977; Struyk 1978). The influence of explanatory variables other than income and price is generally interpreted in terns of institutional effects such as discrimination in housing and financial markets and taste differences. Tastes are, however, never explicit variables in these specifications.

This study instead proposes that the choice of ownership or rental tenure is influenced by socially learned normative orientations of the household in which attitudes are important elements. We consider the relationship between homeownership and household attitudes in a framework of household production theory (e.g., Becker 1976) where households do not consume goods and services they acquire in market transactions but enter them into the production of higher order social commodities such as "health status" or "prestige."

Following several generations of social research, we argue that the utility households derive from social commodities depends on its values and attitudes. Given this dependence of utility on attitudes, the derived demand for goods and services will also include attitudes as relevant explanatory variables. In the case of homeownership, we may model demand as discrete choice between ownership and rental tenure, and specify a "random utility" model of choice (e.g., McFadden 1974) which under certain independence and distributional assumptions is estimable in simple logit form.

Recent studies have taken up critical interdependencies between choice of housing tenure and other household decisions such as ability (Boehm 1981; Krumm 1984) and the quantity of housing services consumed (Gillingham and Hagemann 1983). Within the framework of the present study, we believe that modeling the joint dependency between tenure choice and expected mobility can contribute to the explanation of effects of attitudes on choice.

The joint dependency between homeownership and mobility arises because search and transactions costs of changing residences are greater for owners than for renters. The cost of mobility will therefore depend on the households type of tenure. Similarly, the cost of a tenure type to the household will depend on its ability.

We have previously hypothesized direct effects of attitudes on choice of housing tenure. Relationships between attitudes, career aspirations and occupational mobility rates (e.g., Sewell et al, 1973) also suggest direct effects of attitudes on expected mobility. If attitudes affect expected mobility, the joint mobility-ownership dependency leads us to hypothesize that attitudes also have indirect effects on tenure through mobility.

If in modeling, the joint dependency between ownership and expected ability, we specify their reciprocal influence in an interaction term then the indirect effect of attitudes on ownership should be evidenced in this term. From previous discussion, we expect the interaction between ownership and expected mobility to be negative (i.e., expected ability should decrease the probability of ownership and vice versa). If attitudes increase expected ability as hypothesized, we would expect the inclusion of these explanatory variable to significantly increase the hypothesized negative interaction between the probability of ownership and expected ability.

In the next section, we will test this hypothesis in the estimation of a conditional logit specification (e.g., McFadden 1974; Nerlove and Press 1973) of the joint dependency between choice of tenure and expected ability. Boehm (1981) develops a similar specification and we take up more differences in testing the significance of the interaction tern.

Conditional Logit Model of Tenure Choice and Expected Mobility. Re begin by specifying both choice of tenure and expected ability as binary endogenous variables, and further assume "random utility" models of both tenure choice and expected mobility are applicable.

We can then write the following simultaneous equations for the joint dependency between tenure choice and expected mobility.

EQUATION (1.1), (1.2), (1.3) and (1.4)

From equations (1.1) to (1.4), we can derive the following structural equations for tenure choice and mobility.

Since the parameter a_{i} as defined above, equals the log odds ratio between tenure choice and expected ability, we use it as a measure of the strength of the interaction between tenure choice and expected. [From (1.1) to (1-4) EQUATION Thus, a_{i} can be interpreted as the logarithm of the ratio of the odds of a household owning given it expects to move to the odds of a household owning given that it does not expect to move.]

From equation (2.1) we observe that if a=0, Pr(T_{i}=1|EM_{i}=1) is less than Pr(T_{i}=1|EM_{i}=0). This is consistent with previous discussion which maintained that expected ability should decrease the probability of ownership. Thus, we predict the sign of the estimated value of a_{i} to be negative.

We derive estimates of the logit coefficients B_{1}, B_{2} and B_{3} from maximum likelihood estimation of equations (2.1) and (2.2).

PROCEDURES

The Data Sets The observations used in the estimation of logit model of tenure choice were drawn from the Panel Study for Income Dynamics (PSID) for the year 1968 to 1972 (cf., Survey Research Center 1972). Since the PSID was initially based on a conceptual model relating attitudes to a household's income relative to its needs, a range of attitude and behavior items were included with income and expenditure data on participant households.

Measures of Attitudes in the PSID. To specify attitude measures in the PSID, eight items were selected from the original survey on the basis of their judged relatedness to social class and the household's utility for social commodities. [While these items include judgments of one's own past behavior as well as attitudes, we identify the items and derived factors as attitude measures for consistency with the Survey Research Center's grouping and simplicity in exposition.]

A principal components analysis of the attitude ideas yielded a three-factor solution accounting for 54 percent of the total item variance. The first attitude factor, goal setting (GS), includes educational items of expectations for children, preferred task difficulty level, and the frequency of implementing and completing plans. The second factor, future orientation (FO), is based on frequency of thinking about the future, planning ahead, and preference for saving rather than immediate consumption items. The task persistence factor (TP) includes general satisfaction/dissatisfaction and frequency of implementing and completing Plans.

The dimensions of these attitude factors have been related to social class membership by a number of researchers (e.g., Blau and Duncan 1967; Cofer and Appley 1964; Keller and Zavalloni 1964). The relationship between these attitude factors and social class was directly investigated by regressing an assigned household Socio-Economic Index (SEI: Duncan 1961) scores [A procedure implemented by Featherman, Sobel and Dickens (1975) was used to approximate SEI scores on occupational prestige from standard Bureau of Labor Statistics codes of occupation, industry of employment, and the sex, race and age of the household head.] on the attitude factors. These factors were found to account for 14 percent of the variation in SEI scores across household.

Measures of Income, Price and Sociodemographic Variables. The PSID reports annual household income over the years 1967-1971 and identifies household residence by county. This allows well-specified income and price variables to be constructed for each household.

We define 1968 permanent income as a four-year unweighted average of measured income before taxes in constant 1968 dollars plus the imputed rental value of o owner-occupied housing for the years-1967 to 1970.

The relative price of own and rent tenure was defined as where P_{o} and P_{r} are BLS indices of the costs of owning and renting a home, respectively, for a family of four at a moderate living standard in the household is residence county (cf., U.S. Bureau of Labor Statistics 1972), s = the percentage of each dollar spent on housing services which is deductible from taxable income during the year and t = the marginal tax rate of the household in the year (c.f., Rosen 1979).

We also include the age, education (ed), race, and veteran status (vet) of the household head, family size, and categorical measures-of whether young children are present (ch<18), and wealth (non-wage income: NWY (1) and (2)) in the joint tenure-mobility equations.

Finally, a set of variables that are hypothesized to influence expected mobility are included. These are expected change in family size (ECFS (1) and (2)), expected change in job (ECJ), and whether or not the household is on speaking teras with half of the neighbors or has relatives within walking distance (NT). The specification of the joint tenure-mobility equations to be estimated is

P(T,EM) = f(Yp, Po/r, ed, age, GS, FO, TP, famsiz, ch<18, NWY(1), NWY(2), NT, ECJ, ECFS(1), ECFS(2), Vet, Caucas, married)

RESULTS AND DISCUSSION

Logit coefficients and asymptotic "t" statistics obtained from the estimation of equations (2.1) to (2.2) are presented in Table 1. [Since some variables were included in the JTE because of their singular influence on either tenure choice or expected mobility, not all variables in Table 1 have predicted signs for both tenure choice and expected ability.] Our discussion of the effects of independent variables in the joint tenure equations will concentrate on the B_{2} and B_{3} columns of the logit coefficients and the difference B_{1}-B_{2}-B_{3}.

From equations (2.1) and (2.2) of the previous section, we observe that B_{2} and B_{3} represent the direct effects of independent variables on Pr(T_{i} = 1|EM_{i}), tenure choice given expected mobility, and Pr(EM_{i} = 1|T_{i}), expected mobility given tenure choice, respectively. Correspondingly, B_{1}-B_{2}-B_{3} represents the contribution of the independent variables to the interaction of tenure choice and expected mobility.

Considering the coefficients of the conditional tenure choice function (B_{2}) first, we observe that income, price, the presence of young children, non-wage income, attitude factors of goal setting and future orientation, and the age, race and marital status of the household heat are significant. Results for the sociodemographic variables are generally consistent with previous studies (see, for example, Li 1978) and for reasons of brevity are not discussed here.

Although signs of all three attitude factors are positive in the conditional tenure choice equation, only coefficients of the goal setting and future orientation factors are significant. Constituent items of both these attitude factors have clear links to social class and relevant value orientations.

The conditional tenure choice equation was also estimated with coefficients of the attitude factors set to zero. Comparisons of the log likelihood values for this estimation and the fully parametized tenure choice equation in Table 1 indicate that attitude factors significantly (p<.01) increase model fit to the data.

Turning nest to the conditional expected mobility function, pr(EM_{i}=1|T_{i}), coefficients for income, education, age, neighborhood ties and attitude factors of goal setting and task persistence are significant in Table 1.

Among attitude factors, goal setting and task persistence have different influences on expected mobility. Goal setting has an expected positive effect on expected mobility which may result from its hypothesized influence on occupational aspirations. Task persistence evidences a negative effect on expected mobility. The high frequencies of planning ahead and completing plans that define this factor may result in less frequent but more carefully planned adjustments to consumption through moving.

Finally, in considering the interaction between tenure choice and mobility, we observe that permanent income, non-wage income over $500, and the goal setting attitude factor have significant influences on the magnitude of this interaction (B_{1}-B_{2}-B_{3} in Table 1). The significant contribution of goal setting to the interaction of tenure choice and expected mobility is consistent with study hypotheses. In addition to the direct effects of the goal setting factor on tenure choice, this attitude factor apparently has an indirect effect on choice through its effect on expected mobility.

The coefficient of the interaction parameter, a, evaluated at population means of sociodemographic characteristics (X), has the predicted negative sign in P(T_{i} = 1|EM_{i}) and is significantly different from zero (a = -2.2, p < .05). This was tested by a t statistic that differs from the one Boehm (1981) has used in a similar model specification. [We suggest the denominator of Boehm's test statistic must be multiplied by (1/N)^{1/2} to be defined as a "t". However, Boehm's (1981) definition of t_{a} also assumes the B coefficients are known with certainty and the elements of X are random. This definition does not take the variability of the estimated Bs into account. We consider a more appropriate assumption for the definition of t to be that the elements of X (sociodemographic characteristics of households) are fixed and the B coefficients are random, and therefore define the "t" ar above.] We define the test statistic as

The goal of this research was to demonstrate the relationship between household attitudes and the major consumption-saving decision of homeownership. Estimation of a conditional logit model in which tenure choice and expected mobility were endogenous variables indicated positive effects of goal setting and future orientation attitude factors on the probability of homeownership. Indirect effects of goal setting on homeownership through expected mobility were also obtained. These results support hypotheses on the interrelatedness of attitudes, expectations and consumption.

While earlier work on attitudes and consumption provides a foundation for the present inquiry, we believe that we now benefit from more definitive product or service class specific modeling of complex demand effects of attitudes.

Finally, we briefly note potential policy applications of this research. Since 1968, a major end of federal housing policy has been to increase homeownership among disadvantaged households as a means of increasing their housing quality and improving neighborhood quality in urban areas. These programs have focused on affecting the relative price of owning and renting largely through direct grants and subsidies. While income and price are major influences on the probability of ownership in our results, ownership is also found to be significantly affected by attitudes and expectations. More definitive investigation of the influence of these latter variables may yield policy methods to augment or guide constrained grant programs in increasing homeownership. For example, it may be that when households are grouped by attitude factors, income grants are efficient policy in some, while subsidies to the stability of neighborhoods which decrease expected mobility are efficient in others. These policy issues would be best considered in data with measurement instruments specifically designed for the investigation.

REFERENCES

Becker, Gary S. (1976), Economic Approach to Human Behavior. Chicago: University of Chicago Press.

Blau, Peter and Otis D. Duncan (1967), The American Occupational Structure. Nev York: Wiley.

Boehm, Thomas P. (1981), "Tenure Choice and Expected Mobility: A Synthesis," Journal of Urban Economics, 10, 375-389.

Cofer, Charles and M. a. Appley (1964), Motivation: Theory and Research (second edition). Nev York: 1964.

Featherman, David, Michael Sobel and David Dickens (1975), -A Manual for Coding Occupations and Industries into Detailed 1970 Categories and a Listing of 1970-based Duncan Socioeconomic and NORC Prestige Scores," Working Paper No. 75-1, Center for Demography and Ecology, University of Wisconsin.

Gillingham, Robert and Robert Hagemann (1983), "Cross-Sectional Estimation of a Simultaneous Model of Tenure Choice and Rousing Services Demand," Journal of Urban Economics, 14, 16-39.

Kain, John F. and John M. Quigley (1972), "Housing Market Discrimination, Some Ownership, and Savings Behavior," American Economic Review, 62, 263-277.

Katona, George (1951), Psychological Analysis of Economic Behavior. Nev York: McGraw-Hill.

Keller, Suzanne and Marisa Zavalloni (1964), "Ambition and Social Class: A Respectification, Social Forces, 43. 176-188.

Krumm, Ronald J. (1984), "Housing Tenure Choice and Migration," Journal of Urban Economics, 16, 259-271.

Li, Mingche (1977), "A Logit Model of Homeownership," Econometrica, 45, 1081-1097.

McFadden, Daniel (1974), "Conditional Logit Analysis of Qualitative Choice Behavior," in Frontiers in Econometrics, ed. by P. Zarembka. Nev York.

Nerlove, M. and S. J. Press (1973), Univariate and Multivariate Log-Linear and Logistic Models," Rand Corporation Report R-1306, Santa Monica, CA: Rand Corporation.

Pickering, J.F. (1981), -A Behavioral Model of the Demand for Consumer Durable-," Journal of Economic Psychology, 1, 59-77.

Rosen, R. S. (1979), "Housing Decisions and the U.S. Income Tax: An Econometric Analysis," Journal of Public Economics, 11, 1-23.

Sewell, W. R., A. O. Haller and G. W. Ohlendorf (1973), "The Educational and Early Occupational Attainment Process: Replication and Revision," American Sociological Review, 35, 1014-1027.

Survey Research Center, 1968-1972 (1972), Five Thousand American Families Patterns of Economic Progress, Vols. 1 and 2. Ann Arbor, MI: Institute for Social Research.

U.S. Department of Labor, Bureau of Labor Statistics (1972), "Three Budgets for an Urban Family of Four Persons, 1969-70,- supplement to Bulletin 1570-5, Washington, D.C.

----------------------------------------

Tweet
window.twttr = (function (d, s, id) { var js, fjs = d.getElementsByTagName(s)[0], t = window.twttr || {}; if (d.getElementById(id)) return; js = d.createElement(s); js.id = id; js.src = "https://platform.twitter.com/widgets.js"; fjs.parentNode.insertBefore(js, fjs); t._e = []; t.ready = function (f) { t._e.push(f); }; return t; } (document, "script", "twitter-wjs"));