Abstract
We show that financial constraints lead to spatial misallocation and contribute to racial disparities in housing and wealth accumulation. Using bunching and difference-in-differences designs, we document that down payment constraints disproportionately limit the ability of Black households to access housing in high-opportunity areas. We build a dynamic life-cycle model to examine the long-term wealth effects of these leverage distortions on group differences in wealth accumulation. Black households are more affected by financial and spatial frictions, limiting wealth building opportunities. Improving mortgage access and housing supply in high-opportunity areas helps reduce racial wealth disparities, emphasizing the need for access to geographic opportunities rather than homeownership alone.
1. Introduction
Standard models of spatial equilibrium (e.g., Rosen, 1979, Roback, 1982) assume that any durable advantages to living in particular regions should be arbitraged away through moving. However, a large literature points to persistent differences in access to opportunity across areas in the form of labor market prospects and human capital accumulation (e.g., Chetty and Hendren, 2018, Bilal and Rossi-Hansberg, 2021, Boustan, 2016). In this paper, we argue that down payment constraints act as a key friction generating spatial misallocation, rationing households with limited initial resources out of more expensive housing markets. Because high-cost areas typically offer better jobs, schools, and intergenerational prospects, geographic sorting leads to persistent differences in access to wealth building opportunities.
We combine quasi-experimental evidence with a new spatial life-cycle model to show how financial constraints can result in persistent group differences in wealth and access to geographical opportunities. We do so by building and calibrating a rich dynamic model which generates realistic choices for migration, home ownership, and mortgage leverage decisions. In the model, households with low starting wealth and worse initial conditions remain persistently disadvantaged because down payments requirements create frictions to accessing high-opportunity areas. Leverage constraints therefore generate a spatial poverty trap that sustains historically determined differences in outcomes between groups.
We apply this framework to a particularly salient and well-measured example of persistent inequality: racial differences in wealth and homeownership. We begin by presenting empirical evidence that down payment constraints differentially bind for Black households, distorting borrowing, home purchase, and location choices. Black households tend to start life with less wealth and are more likely to grow up in under-resourced neighborhoods, making it difficult to come up with the down payment necessary to buy homes in high-opportunity areas (see, e.g., Bhutta et al., 2020, Chetty and Hendren, 2018). We document a striking stylized fact in the form of a racial leverage gap, with Black borrowers taking on substantially more leverage when purchasing homes. Black households are more likely to reach the maximum allowable leverage limit, which is suggestive of a tighter overall borrowing constraint.
We then implement two reduced form empirical strategies to show that leverage constraints are more likely to bind for Black households, distorting their location and housing choices away from neighborhoods with better income prospects. Both approaches exploit regulatory limits on loans insured by the Federal Housing Administration (FHA). FHA loans come with less stringent down payment requirements — 3.5% instead of 20% for conventional mortgages — but are subject to maximum loan caps so can only be used for relatively inexpensive homes. These caps are set yearly at the county level, generating variation in the size of the down payment requirement across the housing stock. Our first strategy is a bunching estimator showing that Black borrowers disproportionately cluster precisely at the FHA loan cap, indicating a greater distortion in borrowing relative to a frictionless benchmark.
Our second strategy focuses on a natural experiment created by a major reduction in FHA caps, which occurred when temporary measures put in place during the global financial crisis were rolled back in 2014. This unforeseen policy reversal caused down payment requirements to increase sharply in many high-cost areas, while access to leverage was effectively unchanged in low-cost areas. Difference-in-differences estimates indicate sizable impacts on home buying and location choices. After losing access to high leverage mortgages, the share of new mortgage originations to Black borrowers in affected areas dropped by roughly 8 percent. These prospective borrowers did not switch to the rental market, leading to a decline in the overall Black population. We show that high-cost areas provide better income prospects and test scores, highlighting the disproportionate impact of down-payment requirements on access to opportunity for Black households.
Motivated by this evidence, we build a dynamic model to evaluate and quantify the role of financial constraints in perpetuating racial group differences in wealth and housing. The economy consists of high- and low-opportunity areas, which are populated by overlapping generations of heterogeneous risk-averse households that are divided into Black and white demographic groups. Throughout their life-cycles, households choose to either purchase housing or rent in one of the two types of areas. Households have an intergenerational wealth accumulation motive with voluntary bequests, which are redistributed to the next cohort within the same racial group and create an incentive for the current cohort to move to opportunity. Purchases are financed with long-term defaultable mortgages that are subject to the same down payment constraint across areas.
The down payment requirement is lower (3.5%) for relatively small mortgages below a fixed geography-specific loan cap to match the structure of FHA mortgages. It is higher for mortgages above the cap (20%). Households also face idiosyncratic moving and homeownership shocks, which capture residual exogenous motives for relocating and owning (including moving frictions, discriminatory barriers, and location-specific housing quality). The two areas differ in their loan caps, their levels and price elasticities of housing supply, and in their income processes which endogenously depend on the local composition of households through the presence of high-productivity workers via an agglomeration externality. The two groups differ in their initial wealth, income processes, and the probabilities of being born in each area. In equilibrium, differences in house prices and rents arise endogenously across areas as a result of local housing supply and demand.
The central friction we analyze comes from down payment requirements. Low-wealth agents, many of whom are current and future Black borrowers, cannot access homeownership because of high prices in high-opportunity areas. As a result, Black households are caught in a spatial version of a poverty trap: they cannot afford down payments to own housing in high-opportunity areas, and hence are limited in their ability to accumulate wealth and afford down payments to begin with.
This new framework accounts for spatial and racial heterogeneity in the data from which life-cycle models typically abstract (see, e.g., Gomes (2020) for a survey). The model generates 2 × 2 cross-sectional distributions over individual state variables for the two area types and demographic groups, which are key for evaluating the effects of spatial misallocation on wealth accumulation across groups.
We calibrate the model using indirect inference to match our quasi-experimental estimate of the elasticity of Black borrowing to the level of the down payment constraint, which is obtained from our difference-in-differences approach. To do so, we replicate the same experiment in the model that we examine in our reduced form analysis, a change in the loan cap in high-opportunity areas, as part of our calibration. This is a numerically challenging step which significantly improves the realism of the model.1 The model matches targeted differences in income, homeownership, and moving rates across groups and areas. In our calibration, income differences arise due to both the endogenous spatial income shifter and endogenous skill sorting across areas. One component of the difference in income across areas is due to the causal effect of place while the other is due to sorting of higher-productivity workers to higher-income areas (see, e.g., Bilal and Rossi-Hansberg, 2021, Card et al., 2025). Overall, the model is able to closely match racial differences in leverage and more than 75% of the racial gap in wealth, despite not targeting these moments. We then use the model as a laboratory and run several counterfactual experiments to quantify the role of financial constraints as a driver of racial disparities in U.S. data.
The first counterfactual experiment demonstrates the importance of leverage constraints by relaxing the down payment requirement. Specifically, we compare our baseline model with an economy in which the loan cap is raised in the high-opportunity area, which allows borrowers in the high-opportunity area to purchase more expensive homes with as little as 3.5% down. Relaxing the constraint has positive effects for Black households across financial and real measures, reducing Black–white gaps in wealth, income, homeownership, leverage, and consumption. On average, Black household wealth is higher by 9.6%. To help contextualize the effect of financial constraints in terms of spatial mobility, we show that a 15% reduction in the costs to moving to high-opportunity areas is necessary to generate a comparable increase in Black wealth.
The key mechanism is a flow of Black households to high- opportunity areas. This result underscores the main insight of our paper: the presence of leverage requirements adversely impacts Black borrowers and leads to spatial misallocation, which in turn persistently impairs income prospects and wealth building. Importantly, our estimates account for equilibrium price adjustments, and we find that house prices grow much more than rental prices in high-opportunity areas. Reductions in the wealth gap are in part driven by an influx of Black homeowners. Due to a complementarity between the individual and location-specific components of the income process, high productivity Black households particularly benefit in this counterfactual.
High home prices in high-opportunity areas are at the core of the spatial distortion created by down payment requirements. Our second set of counterfactual experiments examines the role of spatial constraints, in the form of housing supply restrictions, in exacerbating this distortion (see also Hsieh and Moretti, 2019). We consider an economy where the level of housing supply is increased by 10% in high-opportunity areas, relative to the baseline model. This modification corresponds, for example, to less stringent regulatory requirements on zoning. Our contribution is to show that the impact of changes in housing supply is strongly heterogeneous across demographic groups. The expansion — and corresponding decline in home prices — results in 1.7% higher average wealth for Black households, more of whom are able to overcome the down payment requirement and purchase homes (or rent more cheaply) in high-opportunity areas. The consequences are different for white households because they are more likely to own homes in the baseline. The reduction in home prices actually increases their average wealth by less, further reducing the racial gap. Furthermore, we show that increasing only rental housing supply in high-opportunity areas can also address the spatial mismatch and increase Black income, but has a less pronounced impact on Black wealth and actually reduces Black homeownership.
Finally, our third set of counterfactual experiments combines the first two modifications to consider the interaction of financial and housing supply constraints. We simultaneously relax the loan cap and increase housing supply by 10% in high-opportunity areas. A higher level of housing supply alleviates one of the main drawbacks of relaxing leverage constraints: an increase in prices due to higher housing demand. As a result of the complementarity between the two modifications, the increase in the average wealth of Black households (12.4%) is larger than the sum of the changes that occur in each experiment individually (11.3%), largely owing to their much higher presence in high-opportunity areas in the combined experiment.
Our results are robust to various alternative specifications of the baseline model. First, the effects of relaxing the FHA loan cap are nearly identical when introducing discrimination in mortgage rates. The spatial misallocation due to leverage constraints generates persistent wealth gaps even absent explicit racial discrimination in the financial system. Our results remain comparable when idiosyncratic moving and homeownership shocks are the same across groups, which shows that preference differences are not the main driver of racial disparities in the model. We also that show that our findings are stronger when households have access to a higher rate of return on financial assets as a complementary way to accumulate wealth. Higher returns make it easier to build the down payment necessary to access housing, pointing to important complementarities between housing and financial assets. Finally, extending the model to allow for Payment-to-Income (PTI) limits in addition to LTV limits leads to very similar conclusions.
There are two important caveats to our analysis. First, our conclusions should not be construed as advocating for the unrestricted expansion of access to leverage. The results highlight important tradeoffs between down payment requirements and considerations of equity across groups. However, analyzing the implications for the optimal design of mortgage policy would require taking into account a range of factors that go beyond the scope of our model, particularly the consequences for financial stability.2 Nevertheless, the model does account for the effects of financial constraints on house prices and default risk, and we find that the effects on credit risk vary substantially across areas. Default rates increase when the FHA down-payment requirement is lowered (from 3.5% to 1%), which disproportionately impacts low-opportunity areas. Alternatively, relaxing the FHA cap by $75,000 in the high-opportunity area actually improves spatial allocation and incomes. This, in turn, helps borrowers absorb shocks and lowers average default rates. These findings suggest that while increasing leverage may add to household risk, all else equal, it is also critical where borrowers locate.
Second, the reduced form analysis exploits variation in FHA limits and down payment constraints in the model that replicate the FHA system. While these choices are useful for identification, they do not imply that the distortions we examine are only a consequence of the availability of FHA lending (or lack thereof). Given the distribution of wealth for Black households, even a 3.5% down payment requirement puts a large fraction of the housing stock out-of-reach (see Appendix Figure A.I). As such, the spatial distortion we highlight is first and foremost a consequence of down payment requirements and relevant even within areas that are entirely eligible for the FHA. A related concern is that, in principle, the FHA system relaxes credit score requirements alongside leverage constraints. However, average credit scores for FHA borrowers have consistently exceeded 660 since the financial crisis, suggesting that a large fraction of FHA borrowers have the option to access mortgage lending through conventional channels, and that leverage is the key driver of demand for FHA loans.3
Related literature
Our paper contributes directly to several broad literatures. The first is a resurgence of work studying the Black–white wealth gap and the role of housing. While there has long been both empirical and theoretical work considering disparities in housing wealth (see, e.g. Gyourko et al., 1999, Charles and Hurst, 2002, Collins and Margo, 2011, Garriga et al., 2017, Stein and Yannelis, 2020), including older work examining FHA borrowing by race (e.g. Canner et al., 1991), a new wave of studies using rich historical microdata has brought new insights into both the historical persistence of the racial wealth gap overall (Derenoncourt et al., 2022, Boerma and Karabarbounis, 2023, Bartscher et al., 2022) and the nature of housing gaps faced by Black borrowers (Bayer et al., 2021, Bayer et al., 2014, Eldemire et al., 2022). This literature has emphasized specific barriers to the accumulation of housing wealth for Black households based on differences in house price appreciation (Kermani and Wong, 2021, Kahn, 2021, Wolff, 2022), property tax assessments (Avenancio-Leon and Howard, 2022), refinancing propensities (Gerardi et al., 2023, Gerardi et al., 2021a), and credit supply (Fuster et al., 2022). Recent studies have also explored the role of racial disparities in mortgage access, with mixed results—Ghent et al. (2014) and Giacoletti et al. (2022) show evidence of discrimination in pricing and approvals and Bartlett et al. (2022) finds evidence of disparities in interest rates, while Bhutta and Hizmo (2021) argues rate differences can be accounted for by racial differences in the take-up of mortgage points.
We add to this literature by highlighting the racial leverage gap, and analyzing its consequences for wealth accumulation using a dynamic model that accounts for home price responses and endogenous moving decisions. Combined with our reduced-form evidence, the model allows us to quantify a new channel that perpetuates wealth differences: the spatial misallocation generated by leverage constraints. By analyzing the role of leverage, our paper also relates to recent work that has emphasized the ambiguous effects of financial variables on wealth inequality, focusing particularly on interest rates (e.g. Gomez and Gouin-Bonenfant, 2020, Greenwald et al., 2021).
Second, we add to the macro-finance literature that analyzes the impacts of financial constraints in life-cycle models with heterogeneous households and incomplete markets. This includes Cocco (2005), Ortalo-Magne and Rady (2006), Corbae and Quintin (2015), Greenwald (2018), Gete and Zecchetto (2018), Chen et al. (2019), and Greenwald et al. (2020). We depart from existing models by introducing a new type of 2 × 2 heterogeneity across geographic areas and demographic groups, which accounts for spatial and racial differences in the data that these models typically abstract from. Endogenizing prices and location decisions in this context is a challenging exercise, which we tackle using methods from the dynamic demand literature. The resulting richness is key for evaluating the real effects of financial and spatial constraints for long-run outcomes, which would be difficult to measure and identify in the data. Another contribution of our work is to significantly improve the quantification of this class of models by calibrating the model to match an empirically identified elasticity, which is endogenous in our setting. This approach can help improve the realism of recent spatial macro-finance models with heterogeneous agents (e.g., Favilukis and Van Nieuwerburgh, 2021, Favilukis et al., 2023, Mabille, 2023) and with identification in macro-finance more broadly (e.g. Nakamura and Steinsson, 2018).
Many of these papers explicitly focus on collateral constraints and inequality. On the housing side, Favilukis et al. (2017), Kaplan et al. (2020) emphasize the role of down payments constraints in limiting housing access for poor households, thereby contributing to inequality. Kiyotaki et al., 2024, Kiyotaki et al., 2011 highlight the role of down payment constraints on wealth and housing consumption. We complement these papers by focusing on the spatial consequences of down payment constraints, and the resulting misallocation due to lost income generation and wealth building prospects. On the firm side, Midrigan and Xu (2014) analyze how collateral constraints limit firm entry decisions and drive misallocation. Chaney et al. (2012) have estimated this channel empirically, focusing on firms’ real estate collateral. Constrained entrepreneurship decisions have also been studied for smaller firms, as in Schmalz et al. (2017). While this literature has demonstrated the relevance of collateral constraints for inequality in the distribution of firms on the corporate finance side, we focus on the role of down payment requirements for homeownership and location choice.
Finally, the persistence of a racial wealth gap in the data is at odds with the predictions of workhorse frameworks such as infinite-horizon models in which initial conditions dissipate in steady state. Theoretical and empirical work has emphasized the role of self-saving to overcome financial constraints (e.g., Moll, 2014, Blattman et al., 2020), suggesting the possibility of long-run convergence for agents who begin with low initial wealth. Our findings suggest a possible resolution of this tension by highlighting the role of leverage constraints, which can generate persistent wealth differences through spatial misallocation.
The paper proceeds as follows. In Section 2, we present stylized facts on the Black–white leverage gap. In Section 3, we present quasi-experimental evidence on the contribution of down payment requirements to the spatial allocation of Black households. Section 4 describes our dynamic model of housing choice and Section 5 discusses the calibration. Section 6 reports the results and Section 7 provides robustness around these estimates. We conclude in Section 8.
2. Data and stylized facts: The Black–white leverage gap
We begin by documenting the main stylized fact that motivates our analysis: Black borrowers have substantially higher leverage than white borrowers at the time of mortgage origination. We exploit recent changes in Home Mortgage Disclosure Act (HMDA) data reporting to accurately and comprehensively measure this racial leverage gap. We show that higher leverage comes because Black households make smaller down payments in dollar terms, and that it is facilitated by mortgages originated through the Federal Housing Administration (FHA), which are disproportionately used by Black borrowers. The differential use of high leverage mortgages and the FHA suggests that leverage constraints — the maximum size of home a buyer can purchase with a given down payment — bind more tightly for Black households.
2.1. Data
We combine several sources of micro-data. Our primary source is loan-level HMDA data. HMDA captures close to the full universe of mortgage originations and contains comprehensive information on race and ethnicity. Crucially for our analysis, HMDA began to include home prices and loan-to-value ratios in 2018, allowing a direct window into leverage differences by race in recent years. Our benchmark sample focuses on owner-occupied, first-lien, new origination mortgages. We supplement this with American Community Survey (ACS) 5-year estimates at the census tract and county level. We use a series of additional datasets for the calibration of our model. To connect information on borrowers over time and measure moving rates, we use Infutor data (as discussed in Diamond et al., 2019). We also use the Current Population Survey (CPS), the Survey of Consumer Finances (SCF), and the Survey of Consumer Finances Plus (SCF+) as described in Kuhn et al. (2020).
2.2. The racial leverage gap
Panel A of Fig. 1 presents the racial leverage gap: Black borrowers have strikingly higher leverage ratios at mortgage origination. This plot shows the distribution of combined loan-to-value ratios at origination for Black and white borrowers from HMDA in 2018. A substantial fraction of Black borrowers — roughly 60% — have initial combined loan-to-value-ratios (CLTV) above 95 (implying a down payment of less than 5%). This stands in contrast to less than 30% of white borrowers. Indeed, the median CLTV for Black borrowers is 96.5 (vs. 90 for white borrowers). These differences persist and even grow beyond origination. For example, the median LTV for Black borrowers with mortgage debt in the SCF+ in 2016 is roughly 66, compared to 52 for white borrowers.4
The presence of large Black–white differences in leverage shows that racial housing gaps go beyond well-studied differences in homeownership. A disproportionate share of black borrowers take effectively the maximum leverage available in the U.S. mortgage system (an initial CLTV of 96.5). This suggests that Black households are more likely to be close to their leverage limits.5
Fig. 1. The Black–white leverage gap.
Notes: Panel A plots the distribution of leverage at origination for Black and white borrowers. Panel B plots the share of borrowers by race and ethnicity across the leverage distribution. Data includes all owner occupied, first lien, new purchase mortgages, excluding VA, FSA, and RHS loans in the 2018 HMDA data with combined loan to value ratios from 20–100. In Panel A Black and white categories are inclusive of Hispanic households, while in Panel B these categories refer to non-Hispanic households.
Appendix Table A.I shows that leverage differences are robust to controlling for geography, income, or other borrower characteristics (although wealth is not observable in our data). This is not to suggest that the leverage gap represents a causal effect of race. Differences in leverage likely reflect pre-existing and historically determined disparities in wealth and access to capital that go beyond current income.6 Racial disparities also persist when analyzing down payments in dollar terms—Black borrowers typically purchase homes with much smaller down payments, and are much more likely to post less than $10,000 when purchasing a home. This confirms that the leverage gap is not a consequence of Black households choosing more expensive homes.
In the presence of a down payment requirement, available wealth determines the set of possible housing and location choices for prospective homeowners. As a result, the very presence of a leverage gap suggests that down payment requirements have differential spatial consequences for Black households. There are two potential concerns with this interpretation. First, higher leverage by Black borrowers could potentially reflect higher preferences for debt or other demand side factors. Second, it could reflect supply-side factors, like the availability of FHA loans in Black neighborhoods. An examination of the wealth distribution in the SCF data helps to mitigate these concerns. Panel A of Appendix Figure A.I shows the fraction of households with enough liquid wealth to post the required down payment at various points in the national wealth distribution. A large fraction (nearly 70%) of Black individuals appear constrained in their ability to purchase a house in the 25th percentile of the national distribution, and less than 10% have the wealth to meet the down payment requirement for the median home. Panel B of this figure indicates that constraints also bind within MSAs.
2.3. The FHA provides the dominant channel for high leverage loans
The Federal Housing Administration (FHA) is the largest source of high leverage loans for all borrowers, including Black households. Panels A and B of Appendix Figure A.III show that the majority of very high leverage loans are originated through the FHA (and that nearly all FHA loans are high leverage). In our 2018 sample, FHA loans represent under 2 percent of mortgages with initial CLTV below 80 but nearly 70 percent of those with initial CLTV over 95.
The FHA system was created in the wake of the Great Depression, when private lenders typically required much higher down payments for private mortgages. In its current form, the FHA provides approved lenders with 100% guarantees against default for qualifying loans. In exchange for an upfront fee and recurring insurance payment, borrowers with credit scores above 580 may make down payments as low as 3.5% (an initial LTV of 96.5).7 While it is possible to get a high leverage loan through a conventional channel (including conforming loans sold to Fannie Mae or Freddie Mac) doing so requires costly private mortgage insurance that varies substantially with borrower risk. There is a significant clustering precisely at the limit of 96.5 for FHA loans, while the modal conventional loan has an initial CLTV of 80.
Given the relatively high leverage taken by Black borrowers, the FHA is the key origination channel. As panel C of Appendix Figure A.III shows, more that 50% of loans to Black households in our 2018 sample were through the FHA, compared to roughly 20 percent of loans to white households. While the FHA allows borrowers a relatively low-cost way of accessing high leverage loans, only certain loans qualify. Perhaps the most important constraint is that the FHA imposes county-specific loan caps that limit the amount a household is able to borrow. As it currently stands, these caps are set at 115 percent of last year’s median home price for the local area subject to a nationwide floor ($356,362 for the year 2021) and a nationwide ceiling ($822,375 in 2021).8 As a consequence, the relaxed down-payment requirement enabled by the FHA is only relevant for a portion of the housing stock.
3. Reduced form evidence: Leverage constraints bind more for black households
We next show direct evidence that leverage limits differentially distort the borrowing, purchase, and location choices of Black borrowers, with real consequences for access to opportunity. The presence of a leverage constraint forces borrowers to make large down payments to access homeownership. The upfront burden tends to be largest in geographic areas with strong labor markets, good schools, and high intergenerational mobility. Leverage constraints may therefore generate spatial rationing on the basis of current wealth, rather than productivity or permanent income. We exploit variation in the down payment requirement generated by FHA loan caps using bunching and difference-in-difference approaches. Ultimately, we also produce moments from this estimation that help calibrate our model.
3.1. Down payment requirements distort loan sizes for black borrowers
We begin by showing that Black households are more likely to choose a loan precisely at the FHA cap, generating excess bunching for Black versus white borrowers. Below the cap, most borrowers qualify to put as little as 3.5 percent down, but lenders typically require larger down payments — often 20 percent — for loans above the cap. This generates a kink in the down payment requirement at the county-specific loan cap. The concentration of borrowers at the threshold indicates that the leverage constraint disproportionately binds for Black households, and that loan sizes are differentially restricted relative to a world with no leverage limit.
We present graphical evidence of this excess bunching in Fig. 2. The solid lines and dots present the distribution of originated loans for Black and white borrowers in $10,000 intervals, relative to the county-specific FHA cap (which is normalized to 0). The dashed lines represent estimates of the counterfactual distribution for each group in the absence of the cap, calculated following Chetty et al. (2011) and explained in more detail below.9
Fig. 2. Differential bunching at county FHA limits for black households.
Notes: Solid lines and dots show the fraction of Black and white households in with mortgages in each $10,000 interval surrounding the county specific FHA limit. Dashed lines denote counterfactual distributions constructed following Chetty et al. (2011), with the excluded bunching region defined as the $10,000 above or below the limit itself. Data includes all owner occupied, first lien, new purchase mortgages, excluding VA, FSA, and RHS loans in the 2010–2020 HMDA data.
A first observation is that there is more mass in the left portion of the distribution for Black borrowers. These households tend to choose smaller loans (relative to the FHA cap) but the proportion of white borrowers begins to exceed that of Black borrowers for loans roughly $50,000 below the limit. Following this trend, the counterfactual distributions indicate that a substantially greater share of white borrowers would choose loans in the vicinity of the FHA cap in the absence of a limit.
Click to read more