Most previous research on household financial obligations has focused on debt burdens, without considering all financial obligations, such as rent payments and vehicle leases. My research updated two studies of a broad measure of financial obligations burdens (Hanna, Yuh, & Chatterjee, 2012a; 2012b) by adding alimony and child support payments to the measure of financial obligations, and by using additional explanatory variables, including expectations variables, and a measure of financial literacy. The purpose of this research was to investigate the effects of household characteristics on the ratio of a broad measure of financial obligations payments to income.
I found that homeowner households and renter households had very different income distributions and distributions of the financial obligations ratio, so in all of my analyses, I conducted separate analyses of homeowner households and renter households.
One limitation of the previous studies analyzing household financial obligations is that they considered only an arbitrary dummy variable for whether households had a high financial obligations ratio. My research used alternate methods, in order to go beyond the previous research. I used three different statistical models in my analyses.
In Study 1, I used logistic regressions to estimate the effects of household characteristics on whether households had a financial obligation over income ratio over 40%, also referred to as having a heavy burden. I used the 2016 Survey of Consumer Finances (SCF) with logistic regressions, separately for homeowners and for renters. . In 2016, about 23% of all households had a heavy financial obligations burden. About 13% of homeowner households and 40% of renter households had a heavy financial burden. My results for the logistic regressions controlling for income were generally similar to the results reported by Hanna et al. (2012b). In addition, some of the explanatory variables I added that were not used in the previous research had significant results. For the logistic regression for homeowners, there were positive effects for having an education loan, being credit constrained, and having alimony or child support payments. The positive effect of being credit constrained was contrary to expectations of some economists, and probably reflects the fact that credit constrained households were constrained because of having a heavy debt burden. There were similar effects for the logistic regression for renters, but of those additional variables, only the dummy variable for having alimony or child support payments was statistically significant.
In order to obtain more insights than my Study 1 investigation of having a ratio more than an arbitrary threshold (40%), in Study 2, I used Ordinary Least Squares (OLS) regressions on the natural log of the ratio. For the OLS regression on the log of the ratio, for homeowners, there were positive effects for having an education loan, being credit constrained, and having alimony or child support payments, just as in the comparable logistic regression in Study 1. For the OLS regression on the log of the ratio, for renters, there were positive effects for having an education loan, for being credit constrained, and having alimony or child support payments.
OLS regressions provide analyses centered around the mean of the dependent variable, so to obtain better multivariate analyses of my skewed dependent variable, in Study 3 I conducted quantile regressions on the financial obligations over income ratio. In each quantile regression, I examined results at the 10th, 25th, 50th, 75th, and 90th percentiles for renter households and for homeowner households. I focused my discussions on results for the 75th percentile quantile regressions, because that level represented a danger point for renters and approaching a danger point for homeowners. The results for homeowners, were generally similar to the results for comparable analyses in Study 1 and Study 2.
The empirical results for homeowners across all three studies were that having an education loan, being credit constrained, having alimony or child support payments, and having child under 18 were positively associated with financial obligations. Additionally, compared to having current income the same as normal income, homeowners with current income higher than normal tended to have a lower financial obligations ratio, and homeowners with current income lower than normal income tended to have a higher ratio. For renter households, having alimony or child support payments was positively associated with the financial obligations ratio.
For both homeowners and renters, for all three studies, the lack of an effect of financial literacy on having the financial obligations burden suggests that having a higher financial obligation over income ratio was not necessarily a mistake. For homeowners, it is possible that lenders control the amount of much of the debt that can be taken on, so that financial literacy might not be a factor in the outcomes. For renters, even though credit had become less restricted over the decades (Lyons, 2003), after the Great Recession, lenders might also have controlled the amount of debt that could be taken on, making the decision-making by households less crucial.
Because income is in the denominator of the financial obligations ratio, it seemed possible that regressions controlling for income might have most of the explanatory power in the income variable, without many significant effects for other household characteristics. Therefore, regression models are presented without controlling for income, as a comparison. For many of the analyses have approximately the same or even more variables with significant effects with income controlled. In regressions for renters, while there are fewer significant effects when income (included as the log of income) is controlled, there are still many significant effects. Furthermore, more insights can be obtained from the regressions with income controlled, for instance, for the effects of education and of racial/ethnic status.
It is obvious that more research is needed to fully explore household decisions in taking on financial obligations. The effects of some household characteristics, for instance, racial/ethnic differences, should be explored with choice models, for the choices between homeownership and renting. There is a strong relationship between racial/ethnic group and tenure choice (Kusisto, 2019; Schlesinger, 2019; Kusisto & Eisen, 2019). Kusito (2019) reported that the homeownership rate for Black households had reached its lowest point in modern history. Kusisto and Eisen (2019) reported although the Hispanic homeownership rate had reached a 50 year low in 2015, since then Hispanic homeownership has grown faster than White homeownership.
My findings for racial/ethnic differences in the financial obligations burden have mixed effects for homeowners and renters. For homeowners, my logistic regression controlling for income shows that Hispanic and Asian/others are slightly more likely to have a heavy burden than Whites, but the difference between Blacks and Whites is not significant. My OLS regression on the log of the financial obligations ratio, controlling for income has no significant effects for the racial/ethnic variables. My quantile regression for homeowners, controlling for income (Table 7.4) also has no significant effects for any of the quantiles for the racial/ethnic variables. However, the results for renter households are more consistent. My logistic regression controlling for income has significant positive effects for each of the racial/ethnic variables, and I found similar results for my OLS regression on the log of the ratio. The estimated effects in the OLS regression are substantial, with Black households having a financial obligations ratio over 10 percentage points higher than otherwise similar White households, Hispanic households having a ratio almost 26 percentage points higher than White households, and Asian/other households having a ratio over 16 percentage points higher than White households. The quantile regression for renters, controlling for income, does not have significant effects for the Black or the Asian/other variables at any quantile, but does have significant positive effects for the Hispanic variable for the 10th, 25th, and 50th percentiles. The effect at the 50th percentile of being Hispanic is almost a 6 percentage point higher financial obligations ratio. Without geographic information on the households, we cannot know from SCF datasets whether some of the differences among renters are due to geographic location, as discussed by Hanna et al. (2012a). Discrimination by landlords and lenders could contribute to some of the effects, along with limited information about real estate markets and lenders by immigrants, which would especially affects Hispanic and Asian households. However, another issue that could be explored is the tenure status, as renters have much higher financial obligations ratios than homeowners, so the tenure choice is important in ascertaining the effects of racial/ethnic status on the financial obligations burden.
Therefore, future studies on the choice models of household decisions are needed. Use of panel datasets would also help provide more insights, in order to ascertain what levels of a financial obligations ratio would be more likely to lead to negative consequences requires analyses of panel datasets.
Future research should use other research approaches such as experimental or mixed methods. Various research approaches may provide different insights in household decision-making process when taking financial obligations. Further research may help us better understand consumer behaviors in financial obligations, and how decision-making process in household finance could help households improve their financial well-beings like prepare for income uncertainties.
Financial educators and personal finance textbooks have debt-income guidelines that suggest that households have debt payments no more than 40% of income, or similar guidelines (e.g., Greninger et al., 1996). While financial practitioners and financial educators generally agree on these guidelines, they should be reconsidered in terms of the idea of the financial obligations burden rather than the narrower debt payment burden. Logically, any payment that has involves some type of obligation should be considered along with debt payments.
It seems obvious that having a high financial obligations payment to income ratio poses risk for households, but given that I did not find any evidence that, controlling for income, education and financial literacy were negatively related to the financial obligations ratio, so I did not find evidence that bounded rationality contributes to having a high financial obligations ratio.