• Keine Ergebnisse gefunden

African-American Mayors, Home Ownership and Mortgage Lending

5.1 Introduction

Chapter 5

African-American Mayors, Home

applications even after controlling for credit score proxies and demograph-ics. Second, black renters exhibit a 20 percentage points lower likelihood to initiate a mortgage application in the first place than white renters.1 I con-tribute to the literature by addressing these two frictions in the US mortgage market and explore whether local political leadership is able to generate a favorable environment for financial commitment, mortgage access and home ownership.

This paper analyzes the short and long run effects of local black political leadership on both mortgage access and home ownership transition of black households between 1990 and 2016. To address endogeneity of political lead-ership, I employ a static and dynamic regression discontinuity (RD) design to analyze interracial elections in US cities. This strategy compares housing market outcomes in US cities where a black candidate barely won a mayoral election with housing market outcomes in cities where a black candidate barely lost.

The RD design takes advantage of three main datasets. First, I comple-ment existing records on mayoral elections with information on the name, party affiliation, vote return and the race for each of the top two mayoral candidates. This results in a total dataset consisting of 1,083 mayoral elec-tions between 1990-2016 in 905 US cities. Second, loan-level application data from the Home Mortgage Disclosure Act (HMDA) contain rich bor-rower information on the applicant’s income, sex, loan amount, location of the house, whether the loan volume was accepted or denied and most im-portantly the race of the applicant. Additionally, the HMDA data structure allows me to disentangle loan supply from loan demand by using multiple bank-city lending relationships to exploit a within city lending comparison that absorbs city-specific credit demand changes. Third, the Panel Study of Income Dynamics (PSID) is a Longitudinal survey of US families and tracks

1Black households might be discouraged to apply due to systematic racial differences in e.g. down payment constraints, uncertainty about income streams or demographic status and supply-side borrowing constraints. See Charles and Hurst (2002) for a well structured description on why home ownership constraints might differ by race.

home ownership transitions of households, housing wealth and mortgage ap-plications over time.

Before turning to causal effects, I first show the existence of an electoral mortgage cycle for US cities that elected an African-American mayor for the very first time. Raw bank-level correlations demonstrate not only that the number of accepted mortgage applications from black applicants increase between 10 to 19% in the post-election period but also that the black-white acceptance differential increases by 3% one year after the election. Static RD estimates confirm this tentative evidence with a positive treatment effect of 11 percentage points increase of black mortgage acceptance rates in the year after black mayors took office. The dynamic RD design indicates long run effects on black mortgage acceptance rates that are most pronounced around four to six years after a black mayor gets elected. Interestingly, these long-run effects are significant around the transition between mayoral term periods and might indicate re-election effects. I find that banks accept more mortgage applications from black borrowers in the upper part of the income distribution while no significant treatment effects can be found in the lower part of the income distribution. Also, black debt-to-income ratios increase for the higher income groups. Evidence on local political leadership affecting home ownership transitions is still to be done as soon as the PSID data access is available.

Establishing a channel to rationalize these findings is challenging. Given that US cities have always been confronted with racial discrimination in housing markets (Appel and Nickerson, 2016), a black mayor should be more concerned about housing conditions for the black population than a white mayor. As a result one might expect newly elected black mayors to prioritize the elimination of such frictions in direct and indirect ways. One possibility is that black incumbency leads to a perception change since it provides con-crete information that disproves the fears and expectations of many white residents and also loan officers. Because it is very hard to provide empirical evidence for this explanation, I concentrate on channels where data availabil-ity is given. The first African-American mayor of Atlanta, Maynard Jackson,

pressured white-run banks to appoint black individuals as executives and used deposits of city money to exert pressure on these financial institutions (Bayor, 2001). In a first step, I examine this narrative evidence by collecting bank-level data on city deposits to investigate such apolitical-pressure chan-nel. Second, I exploit the Census Building Permits Survey data at the city level between 1988-2010 to analyze whether the results might be driven by housing supply expansion. Third, FDIC Data on minority bank ownership and bank types will give insight into theproximity channel: a mayor’s soft power and leverage on banks is higher if a depository institution is owned by a peer or if it is a community bank. Finally, data on (CRA) bank ex-aminations will enable me to investigate the reputation channel: assuming that banks are concerned about their reputation, I hypothesize that discrim-inating banks would act against their prior as soon as a black politician got elected.

The first strand of literature on economic effects of local political leader-ship has concentrated exclusively on “aggregate” city policy outcomes such as public spending, employment, education or crime rates (Ferreira and Gy-ourko, 2009; Ferreira and GyGy-ourko, 2014; Hopkins and McCabe, 2012; Mey-ersson, 2014). The second strand of literature focuses on the impact of hard political power, such as US federal laws or regulations, on mortgage lend-ing outcomes. Despite numerous efforts of legislative acts2 to expand credit access and reduce discrimination in the mortgage market, evidence on the success of these government actions is mixed (Agarwal et al., 2014a; Agar-wal et al., 2016b; AgarAgar-wal et al., 2017; Bayer et al., 2017; Munnell et al., 1996). This paper is mostly related to the third strand of literature on the nexus between soft political power and the mortgage market. Akey et al.

(2017) show that ascension to the chairmanship of US Senate committee is associated with a large reduction in the availability of consumer credit in the ascending Senator’s state. Antoniades and Calomiris (2016) exploit the US presidential election in 2008 to show that voters punish Presidential

2See for example the Fair Housing Act of 1968, the Equal Credit Opportunity Act of 1974, the Community Reinvestment Act of 1977 or the Home Mortgage Disclosure Act of 1975.

candidates for local mortgage supply contractions but do not reward them for local mortgage supply expansions. Chavaz and Rose (2016) demonstrate that receivers of the 2008 liquidity assistance program TARP increased bank lending by 23% to 60% more in areas located inside their home representa-tive’s district than elsewhere. I contribute to this third strand of literature in two distinct ways. First, no attention has been paid to political influence on bank lending at the very local level: city mayorships. This is an important angle since political power might be most effective in municipal environments where spatial proximity between politicians and banks is closest. Second, no understanding has been established on whether and how politicians have an impact on mortgage access and home ownership transitions of minority groups. Since historically disadvantaged groups face higher uncertainty, es-pecially in the context of long-lasting and large financial commitments, it is of relevance if political leadership can create a comfort zone for their voters.

This paper reveals important implications. First, political participation matters. Since hard political power is only partially effective in reducing mortgage market frictions (e.g. Agarwal et al., 2014a; Agarwal et al., 2016b;

Agarwal et al., 2017; Bayer et al., 2017; Munnell et al., 1996), I show that soft political power can be a complementary tool for alleviating these market imperfections. From a policy perspective, this means that the mortgage market can be a useful wealth accumulation tool given certain constraints for politicians to redistribute wealth and income at the local level. Second, the evidence for political influence on easy credit might assign politicians a role in the housing boom-bust cycle.

The paper is structured as follows. Section 5.2 describes the three main datasets, explains the RD design and tests for the validity of the research design. Section 5.3 presents the results for the electoral mortgage cycle, the short run static RD effects and the dynamic RD effects in the long run.

Section 5.4 concludes.