2. Consumer Borrowing Behavior of African-American Homeowners in U.S
2.9 Hypotheses Testing
2.9.2 Hypotheses Testing Using SCF Data
In this section,first the study uses the SCF data to explain the importance of itemization tendency in explaining racially disparate borrowing behavior, which is hypothesis (d) in section 2.6.210. Then, using
some of the credit profile measures generated from SCF survey questions, study again attempts to explain to what extent difference in borrowing response arises from the differential credit profile of the respondents as suggested by hypothesis (c) in section 2.6.2.
In the SCF data, the study finds that there is a differential borrowing response across races in terms of carrying HELOC, in terms of refinancing, in terms of refinancing to lower interest rate and in terms of carrying credit card. There does not seem to be any difference across races in terms of refinancing to cash out equity.
To explain each of the borrowing behavior, the study first inserts the black dummy by itself to em- phasize the size of the difference, and then it inserts the black dummy along with an elaborate set of demographic and behavioral controls to explain the black dummy. Last, the study inserts the variable(s) relevant to the hypothesis in addition to the standard controls to evaluate their contribution in explaining 1 0Itemization may be endogenous, affected by refinancing and collaterlaized borrowing behavior. It is however likely that
itemization is primarily determined by exogenous factors like. deductions for real estate taxes and state and local income taxes and would only marginally be affected by choices about refinancing or getting a second mortgage or heloc.
the difference. The elaborate set of variables used as standard control from the SCF data set is year since house purchase, age of household head, dummy for male household head, dummy for household head being married, dummy for household head being divorced, grades completed by household head, controls for behavioral characteristics like self confessed propensity to shop and risk taking ability, dummy for being financially constrained, household earned income and dummies for observation being from survey years 2001 and 2004.
In table O.1, study attempts to explain the significant black dummy in the HELOC equation. The standard controls considerably explain the back dummy however the dummy remains negative significant at one percent level. Additionally, controlling for the itemization behavior of the respondent does not contribute much in explaining the black dummy either. However the regression suggests that those who itemize are more likely to carry HELOC. In table O.2, the study attempts to explain the racial difference in refinancing behavior. With the introduction of the standard controls, the significance of the black dummy drops from one percent to five percent level. After introducing the itemization dummy on the top of standard controls, the black dummy ceases to be significant. The dummy for itemization is highly positively significant suggesting that homeowners who itemize are very likely to refinance. Table O.3 takes a closer look at the tendency to refinance to lower mortgage interest rate. The black dummy continues to be highly negative significant in the regression even after controlling for itemization tendency along with elaborate controls. The high positive significance of itemization suggests that those who itemize are more likely to refinance to lower mortgage interest rate. Similarly in table O.4, the dummy for itemization along with standard controls contributes little in explaining the black dummy. The itemization dummy is however again positive significant in this regression.
The three credit profile dummy indicators used to control for credit characteristics of the respondent were created from SCF survey questions. First a dummy indicating that of all the various loan or mortgage payments made during last year by the respondent, payments on any of the loans sometimes made later or missed. Second dummy indicates whether the respondent ever declared bankruptcy in his/ her credit history. Third dummy indicates whether the respondent never had a checking account. In table P.1, in an attempt to explain the significance of black dummy in HELOC equation, study found that though controlling for credit profile in addition to standard controls lower the significance of black dummy from one percent to the five percent level, it does not completely explain the black dummy. The negative significance of the dummy indicating ever bankrupt suggests that respondents with a history of bankruptcy are less likely to carry HELOC. In table P.2, credit profile measures along with the standard controls do completely explain the black dummy in the refinance equation. The strong negative significance of never checking dummy suggests that the respondents with no checking account are less likely to refinance.
Surprisingly, the ever bankrupt dummy indicator turned out to be positive significant in this regression. In table P.3 in an attempt to explain the refinancing decision to lower interest rate, the black dummy continues to be negative significant at one percent level even after controlling for credit profile measures along with standard controls. As seen previously in table M.2, the dummy for never checking is negative significant where as that for ever bankrupt is positive significant. Last, in attempt to explain the negative significance of black dummy in the credit card borrowing equation in table P.4, we see that though all the three individual credit profile measures are highly negatively significant, they contribute little to explain the significance of the black dummy in that regression. Even after taking the elaborate standard controls and controlling for the credit profile measures, the black dummy continues to be negative significant at one percent level, the magnitude of the coefficient is however somewhat smaller with the controls in place.
2.10 Conclusion
This chapter has pointed out to the disparate consumer borrowing behavior that exists among African- American homeowners which is very different from their White counterpart. The analytical framework in chapter one suggested thatfinancially constrained homeowners aiming to smooth consumption should demonstrate a strong positive consumption response to housing capital gains, and collateralized borrowing should be the ideal borrowing mechanism. The probit and ivprobit regressions in this chapter using three recent survey years of SCF found no significant tendency to carry collateralized borrowing instruments and to lower the use of non-collateralized borrowing instruments among black homeowners experiencing house price appreciation. In terms of debt balance, the theoretical intuition was that there should be some tendency to consolidate non-collateralized debt in other collateralized debt instruments following increases in home equity. For Blacks we did not see any consistent attempt to lower credit card debt or increase HELOC borrowing in response to increased house prices although there is some tendency to increase cash out refinancing following house price appreciation. The findings essentially suggested that black homeowners are not responding to their increased home equity through increased consumption or borrowing and not consolidating their high interest unsecured debt through low interest debt collateralizing home equity.
The FMF data from a 2005 survey with its rich source of information on householdfinancial knowledge, tenure status, socio-demographic and economics characteristics is used in this chapter to take a deeper look into the causes of this racially disparate borrowing behavior. The descriptive analyses using this data suggested that lack of response among black homeowners is not stemming from their belief that they have not experienced any housing windfall as the blacks have reported more house price appreciation on an average. The descriptive results also suggested in terms offinancial sophistication, black homeowners are
not any less savvy than their white counterparts. The factors which stood out to be as big characteristic differences across races is their differential credit profile and their differential exposure to discrimination in the real estate market. All indicators of credit history suggested that black homeowners have a worse credit profile than their white counterparts. There is also some evidence of a possibility/ perception of existence of market inefficiencies in the form of racial prejudice. An interesting aspect of the FMF data is however that unlike SCF it did notfind any significant lack of refinancing tendency among black homeowners.
The study next tested the five hypotheses generated from the SCF findings using the SCF and FMF data. Thefirst hypothesis was that blacks experiencing high house price appreciation behave differently from the rest in their community. The test using FMF data showed that blacks experiencing high ap- preciation are unresponsive towards carrying second mortgage/ HELOC as well as credit card where as the rest in the community are less likely to carry either of the borrowing instruments. Therefore, though there is some difference from the rest in their community but blacks experiencing high appreciation are not more likely to borrow as their white counterparts. Second, the study tested using FMF data whether differences in financial sophistication can explain the borrowing difference. The test suggested that fi- nancial sophistication difference do not seem to explain the racial difference. Next, using the FMF data differential credit profile was tested as a possible culprit driving the difference. Again, credit history do not seem to explain the entire racial gap in borrowing tendency though there is some evidence that poor credit profile of black homeowners preventing them from carrying a non-collateralized borrowing instrument like a credit card. The FMF data was subsequently used for testing the discrimination story. Having an actual discrimination experience or perception of discrimination in the real estate market did explain the gap in second mortgage/ HELOC borrowing. However, it did not explain the gap in credit card borrowing. Therefore, the factor which stood out to be the most important in explaining the gap using the FMF data was discrimination exposure or perception followed by the differential credit profile. The study subsequently tested using SCF whether differential tendency of itemizing taxes across races can explain the differential borrowing behavior. The test results suggest that though difference in itemization tendency does contribute significantly in explaining the gap but does not explain the entire racial gap. The study subsequently used the credit profile measures available in SCF to test the effectiveness of credit history in addressing the racial differences in borrowing behavior of homeowners. The SCF results sug- gest that though disparate credit profile across races is responsible for the disparate borrowing behavior of homeowners but on its own can not completely explain the difference.
To conclude this chapter makes an essential contribution in highlighting an important behavioral difference in terms of borrowing across non-Hispanic White and African-American homeowners. The
chapter also points out at the possible causes leading to this behavioral difference. Differential credit profile across the homeowners in the two races, exposure to discrimination or perception of existence of discrimination in the real estate market and differential tax itemization tendency across the homeowners in the two races turned out to be the important factors driving the disparate borrowing behavior. The study however suggests that no single factor can entirely explain the borrowing gap across the races, it is likely the result of an interplay of multiple factors.
CHAPTER 3
DISSERTATION SUMMARY
The dissertation contributed toward enhancing our understanding of the consumer borrowing behavior of U.S. homeowners who experienced house price shocks with a special emphasis on the differences in borrowing behavior across non-Hispanic whites and African-Americans. The dissertation is split in two main essays. The first essay developed an analytical model to generate a theoretical intuition as to how homeowners experiencing housing capital gains are expected to respond in terms of consumption or borrowing. The hypotheses generated from this model are then subsequently tested on a sample of white homeowners using three recent survey years of Survey of Consumer Finances. In the second essay, the study using the three recent survey years of SCF data and a recent survey data of Fannie Mae Foundation highlighted the racially different borrowing behavior of African-American homeowners. The second essay subsequently goes on to develop and test the hypotheses on the possible factors driving these racially disparate behaviors using the two data sets. The essay identifies a few factors which are possibly contributing to the two races to react differently to their house price shocks.
The essential contribution of chapter one is in studying the response of white homeowners who experi- enced a significant change in their level of home equity holding in the recent past. The theoretical model in this chapter aimed at analyzing the circumstances under which increased equity in homes results in increased consumption by homeowners, concluded that liquidity constrained households and households that smooth consumption over time are the likely candidates to react strongly to any housing wealth wind- fall. In addition, collateralized borrowing should be the instrument of choice for cashing out equity from homes and spending on consumer goods. The empirical analysis highlights the fact that there has been a very high tendency to cash-out refinance among white homeowners experiencing gains in home equity. There is also evidence of an increase in the balance in HELOC and second mortgage debt, though the magnitude of these increases is quite moderate. The balance in non-collateralized debt instruments like credit card and education loan did not go down as was predicted by the theoretical framework. Therefore, there is not much of an evidence of any attempt on the part of the white homeowners to consolidate in terms of secured borrowing instruments. There has been some drop in auto loan balance though, but that is quite insignificant. Therefore an interesting finding is that though white homeowners seemed to have
adjusted their consumption level upwards in response to increased home equity, but they have not made any significant attempt to bring down their level of high interest unsecured debt. This behavior suggests an overall increase in the leverage of a substantial fraction of American households who witnessed a rise in their house prices.
In chapter two of the dissertation, the focus is on the behavior of African-American homeowners. Given that on average African-Americans are more likely to have a smaller asset base and more volatile income than the non-Hispanic whites, the theoretical model suggests that Blacks should react strongly to any house price shock, as they are more likely to befinancially constrained. This chapter has pointed out to the disparate consumer borrowing behavior that exists among African-American homeowners which is very different from their white counterpart. The probit and ivprobit regressions in this chapter using three recent survey years of SCF found no significant tendency to carry collateralized borrowing instruments and to lower the use of non-collateralized borrowing instruments among black homeowners experiencing house price appreciation. In terms of debt balance, the theoretical intuition was that there should be some tendency to consolidate non-collateralized debt in other collateralized debt instruments following increases in home equity. For Blacks we did not see any consistent attempt to lower credit card debt or increase HELOC borrowing in response to increased house prices although there is some tendency to increase cash out refinancing following house price appreciation. The chapter suggests that financially constrained Blacks refinance to only cash out equity where as constrained whites achieve the dual objec- tives of lowering interest rate and cashing out equity from refinancing. Thefindings essentially suggested that black homeowners are not responding to their increased home equity through increased consump- tion or borrowing and are not consolidating their high interest unsecured debt through low interest debt collateralizing home equity. This chapter also used data from a 2005 survey sponsored by Fannie Mae Foundation of 501 homeowners with rich source of information on householdfinancial knowledge, tenure status, socio-demographic and economics characteristics divided equally across African-Americans and non-Hispanic whites to take a deeper look into the causes of this racially disparate borrowing behavior. In this chapter the new data has been utilized to explore the differences in household economic charac- teristics, financial sophistication, perceptions about occurrences and sources of discrimination, and also the proportion offinancially constrained households across the two races. The descriptive analyses using this data suggested that lack of response among black homeowners is not stemming from their belief that they have not experienced any housing windfall as the blacks have reported more house price appreciation on an average. The descriptive results also suggested in terms of financial sophistication, black home- owners are not any less savvy than their white counterparts. The factors which stood out to be as big characteristic differences across races is their differential credit profile and their differential exposure to
discrimination in the real estate market. All indicators of credit history suggested that black homeowners have a worse credit profile than their white counterparts. There is also some evidence of a possibility/ perception of existence of market inefficiencies in the form of racial prejudice. The study next tested alternative hypotheses on the factors generating disparate borrowing behavior across non-Hispanic White homeowners and their African-American counterparts using the SCF and FMF data. Differential credit profile across the homeowners in the two races, exposure to discrimination or perception of existence of discrimination in the real estate market and differential tax itemization tendency across the homeowners in the two races turned out to be the important factors driving the disparate borrowing behavior. The study however suggests that no single factor can entirely explain the borrowing gap across races, it is likely the result of an interplay of multiple factors.
The results from this analysis will be of importance to policy makers aiming to make homeownership lucrative among minorities. The access to increased equity associated with housing capital gains should be a major incentive to become a homeowner. The current study suggests that there are reasons to believe that this particular incentive may not be present among African-Americans. There is then scope for policies to help overcome minority homeowners, obstacles of the nature of poor credit profile and racial prejudice to reap the benefits of the increased equity in their homes. The findings in this study should also help financial institutions marketing various types of collateralized borrowing instruments such as cash out refinancing, home equity loans, and lines of credit to better target their products to a sizable section of the market, namely African-American homeowners. The current study suggests that