Abstract
The Great Depression led to dramatic changes in financial regulation and home financing. Policymakers opted for insurance programs—deposit guaranty and mutual mortgage insurance—as safe and stable ways to manage the critical financial sector. An analysis of data collected in support of the Residential Security Maps of the Home Owners’ Loan Corporation shows that the age, upkeep, and price of housing along with mortgage availability are robust predictors of mortgage risk in these maps while the presence of non-white and foreign-born residents represent more modest estimators of mortgage risk. The system of mortgage insurance established in this period sowed the seeds of center city economic deterioration.
The reputation of the Residential Security Maps, assembled by the Home Owners’ Loan Corporation (HOLC) from 1935 to 1940, is formidable and to many compelling. Kenneth Jackson discovered these remarkable maps and the extensive documentation that accompanied them some thirty years ago and provided the first succinct discussion of their importance in both suburban development and central city decline that proceeded apace in the postwar era in America. Colored vividly, these maps provide striking assessments of the risk lenders engaged in mortgage lending are exposed to across different neighborhoods of cities. Accompanying these maps, and indeed providing empirical justifications for the risk classification of neighborhood housing markets, are Area Description Sheets that are frequently stunning, sometimes crude, discussions of these individual communities, their housing stock, and most pointedly their residents. The narratives included on these supporting documents are all too frequently racist, anti-Semitic, and rife with Nativist sentiments.
It should hardly be surprising that these maps captured the attention and imagination of urban scholars. Kenneth Jackson argued that with these maps the HOLC “initiated the practice of ‘redlining.’” The HOLC created a uniform system to assess the risk—for lenders—of local neighborhood housing markets. He emphasized that the agency’s risk assessment rating “undervalued neighborhoods that were dense, mixed, or aging.” 1 Guy Stuart adds that HOLC created a neighborhood rating system that accurately reflected and indeed “institutionalized the racial and class attitudes of the real estate industry.” 2 Kevin Fox Gotham notes that HOLC initiated a national and consistent system of real estate appraisal that both assessed the creditworthiness of an area’s housing stock and played an important role (along with the actions especially of the Federal Housing Administration [FHA]) in what he aptly terms the “racialization” of the real estate market. 3 Indeed, numerous analysts have highlighted the fact that the HOLC, along with other federal agencies, was “an agent of segregation,” a concrete example of the racial bias of federal housing programs, a reality that would become increasingly strident and important in the late 1940s and subsequent decades. 4 All in all, these maps and the extensive research and data collection that supported the assignment of mortgage risk grades, as Colin Gordon summarizes, “embody all the salient neighborhood information required for the intelligent operation of mortgage lending activities.” 5
Ironically, HOLC’s city survey program that created this extensive collection of Residential Security Maps only took place after the agency had completed the explicit task that Congress assigned to the Corporation in the legislation that was signed by the president on June 13, 1933: to refinance homes in foreclosure. Nor is this the only paradox. The HOLC did not use its elaborate system of maps that assigned risk by the locations of properties in its own financing activities. Many, perhaps most, properties that the HOLC successfully refinanced were in communities the Corporation later graded as “definitely declining” or “hazardous.” Additionally, the Corporation refinanced homes without overt regard to the ethnicity or race of applicants and certainly made loans to both foreign-born and African American applicants. 6
By conducting the city survey program, a program that engaged thousands of HOLC employees and local consultants 7 over five years, the HOLC went far beyond its primary charge of home financing. 8 Especially important was the Federal Home Loan Bank (FHLB), whose members were almost exclusively Savings and Loan Associations (S&Ls) that worked directly with and formulated mortgage policy recommendations for all federally regulated financial institutions—banks and S&Ls alike. HOLC agents were, a historian of Brooklyn notes, throughout the late 1930s “surveying and mapping urban communities across the nation and working with banks and building investors to enact mortgage guidelines that guaranteed profit.” Chartered by Congress to refinance and thus save the nation’s home mortgage market, the HOLC conducted the city survey program across the country for different purposes: to minimize the risk to all home financing lenders (banks and S&Ls alike), assure the profits of these institutions, and “positioned the federal government as the primary agent of segregation.” 9
The HOLC was a mundane agency that performed remarkable tasks. It certainly achieved, far beyond expectations, its core objective of refinancing hundreds of thousands of mortgages that were in foreclosure and in doing so probably saved the American real estate and home financing market when it was at its nadir. It performed this task dramatically and efficiently: the Corporation expanded from only a handful to nearly twenty thousand employees in the course of a few months. It accepted and processed over a million refinancing applications within its first two years. Perhaps most surprising (at least to its initial detractors), it did not lose money but actually turned a small surplus that it eventually returned to the U.S. Treasury. 10 Equally remarkable is HOLC’s development of the Residential Security maps.
There are two issues that have not been adequately addressed in the literature on historical redlining. First, why did the HOLC create the City Survey Program and devote the substantial resources necessary to conduct this five-year program? The development of these remarkable maps that, ironically, were not widely disseminated, as is commonly assumed, 11 was neither explicitly included in the enabling legislation nor in any clear manner central to HOLC’s mission or its legislative mandate. At the direction of the FHLB, the HOLC, as Douglas Rae observed, moved far beyond its legislative mandate by initiating the sweeping City Survey Program that created these stunning maps. Yet, no one has ventured an answer why the FHLB directed the HOLC to do so.
Second, much is assumed about the Residential Security Maps. Most scholars have appropriately noted the important role that race played in the underwriting protocols of the Federal Housing Administration and the appraisal practices developed by the HOLC. These practices have been persuasively traced not only to the prevailing ideology of the American real estate industry, but directly to two highly influential real estate authors of the era—Fredrick Babcock and Homer Hoyt, who held high-level positions, respectively, in the HOLC and the FHA. 12 Finally, those scholars who have read the Area Description Sheets accompanying HOLC’s map have rightly noted that they are rife with racist remarks and assertions that presume that property values are doomed to dire deflation with the entry of non-white residents into a housing market. From this, it is frequently assumed that race is the rudimentary component of these maps.
These two issues form the subject of this article. I will speak to this first question by surveying the extant literature on the activities of the HOLC but also by consulting the HOLC and FHLB Board meeting minutes and resolutions during the period when the Corporation ceased accepting applications for refinancing. The explanation, I argue, comes from the fact that the HOLC was restricted to refinancing homes that were in foreclosure. HOLC was not permitted to refinance homes whose mortgages were in distress—a much larger number of properties. As the Corporation successfully completed its legislative mandate, it was all too obvious to the FHLB Board that the dire problems of the American housing financing market persisted and represented an ominous and continuing threat to the entire housing sector. HOLC’s extensive mapping program provided guidance to all newly federally regulated financial institutions on how they should proceed in a new, revolutionized home financing market that was structured around long-term, low interest rate, amortized mortgages—the sole type of mortgage that would be eligible for insurance programs that were promulgated by the Federal Deposit Insurance Corporation (FDIC).
The second issue will be addressed by an analysis, presented here for the first time, of data collected from the Area Description Sheets of these security maps of numerous metropolitan areas from all regions of the country, including the largest ten cities at that time. These HOLC data sheets provide a rich source of data, both quantitative and qualitative, that can be used to empirically analyze the numerous factors that were used by the HOLC to assign mortgage risk. Data from over 3,600 data sheets have been collected. Analysis of these data shows that many factors along with race were determinants of HOLC’s mortgage risk grade. My conclusion is surprising: while race certainly informed the development of these maps, the presence of non-whites and foreign-born ethnics are rather modest (albeit clearly negative) correlates of the final HOLC mortgage risk grade. Other features of neighborhood housing markets, especially the age, quality, upkeep, repair, and price of housing, in addition to the willingness of private sources to make mortgage investments into a neighborhood’s housing market, are even more robust predictors of HOLC’s grade of mortgage risk.
To conduct this analysis, first a survey of New Deal interventions into both banking regulation as well as policies that dramatically overhauled home mortgage financing will be summarized. Following, I will discuss the activities of the HOLC (one of the many policy interventions and the creation of potent, permanent agencies created to implement these policies) in first refinancing tens of thousands of mortgages that were in foreclosure and then, expanding far beyond its initial legislative mandate, in developing, through detailed analysis, its Residential Security Maps. These maps, I argue, represent a set of programmatic guidelines for the home mortgage financing industry to follow resulting from these revolutionary changes in both financial institution regulation and the creation of a new, rigid, and insured system of mortgage financing. Following this, I will discuss, in some detail, the process the HOLC followed in developing these remarkable geographic documents. The instructions the HOLC provided to its analysts as well as the data collected to support the mortgage risk grades assigned in these maps will be described. The data from the Area Description Sheets that accompanied each HOLC map will then be analyzed to show the numerous empirical factors that shaped the assignment of the mortgage risk grades that form the central core of HOLC’s maps.
The New Deal’s Revolution of the American Home Mortgage Market
The creation of deposit guaranty was one of the centerpieces of New Deal intervention into the activities of financial institutions in the country. This was, as well, the basis for the new and extensive regulation of these institutions. This new regime of financial regulation in general and deposit insurance in particular mandated that all financial institutions invest in ways that were sanctioned by agencies of the national state—initially the FDIC and then the FSLIC—as sound, safe, and prudent. With the constant and routine examination of the balance sheets and portfolios of all financial institutions that participated in deposit insurance, the collective risk of all loans and investments was scrutinized to assure that they were not in any significant degree so careless so as to threaten the ongoing viability of the financial institution. This guaranteed, in turn, that the insurance on deposits in these thrift institutions was actuarially sound. 13
With deposit insurance, “moral hazard” became a responsibility that no longer was entrusted to private investors or depositors in American financial institutions, but to agencies of the federal government. To fulfill these responsibilities, federal agencies imposed sweeping regulation on those financial institutions that were covered under these agencies. 14 The scope of the enforcement powers of federal banking agencies was, as one economist characterized them, “remarkably extensive.” Sanctions were potentially draconian, including the seizure of an institution, placing it in receivership, revocation of charter, the expulsion from the Federal Reserve (or later the FHLB), and termination of deposit insurance. 15 At the time, deposit guaranty was considered the most controversial element of the bill. Indeed, the most strident critics of deposit insurance argued that such measures did not assure “guaranty” but constituted the “spoilation or redistribution of wealth.” 16 Deposit guaranty was a centerpiece of New Deal financial regulation. It was premised on the assurance that all of the nation’s financial institutions were engaged in cautious investment activities. In housing financing, home mortgages that were covered under mortgage insurance (or mortgages explicitly designed to meet all the criteria of insured mortgages) were defined as safe and sound investments that were approved by regulators. All other possible mortgages that did not meet these criteria would meet with the disapproval of regulators. Regulated financial institutions, in other words, were given little choice but to finance homes using mortgages that were insured.
The second element in the New Deal’s revolution in home financing—the creation and elements of a mortgage program—is equally important. Mortgage insurance, and the Federal Housing Administration to administer this program, was created under Title II of the National Housing Act (1934). Insurance would only be available for mortgages that conformed to the new criteria of such loans included in this Act and sanctioned by the FHA—measures that covered the overall financial structure of the mortgage as well as exacting construction and community standards. Specifically, mortgages were require to be long term (twenty years in the original legislation), low interest rate, high loan to value ratio, and fully amortizing where low monthly payments would pay off the principal and interest of the loan simultaneously over the entire life of the loan. 17
In the legislation, and then extensively (and in exceedingly detailed form) in the FHA regulations that were promulgated in their Underwriting Manuals in 1935-1938, the federal government set the explicit conditions for mortgages to be covered under this insurance plan—conditions that could not be violated. In standardizing the mortgage instrument, state agencies, consciously or not, took discretion out of the hands of loan committees of S&Ls and all financial institutions. By taking away these elements, insured mortgages (and those that were modeled on this new standard) removed those rudiments of lending that allowed lending officers to deal with perceived increased risk—such as increasing the down payment or requiring collateral, increasing the interest rates, or adding fees.
With these factors removed, lending became akin to a zero-sum game. Much of the housing built before the advent of mortgage insurance in 1934 was rendered difficult if not impossible to finance using mortgages that could meet the rigid and inflexible provisions of the Act and the requirements of FHA appraisal standards. Moreover, nearly all housing developed after the imposition of this program, housing built to the exacting standards of the FHA, could by definition be financed using this new standard of home financing.
This new mortgage standard worked against the financing of such a large portion of the standing housing stock for at least three reasons.
First, this new standard for mortgages extended the time horizon of home lending considerably. The calculus of repayment over a relatively brief period (as compressed as three to five years) typical of home lending throughout the 1920s was replaced by the necessity of creating a repayment schedule that would extend for twenty years or more. Homes financed with such long-term loans would, in other words, have to maintain or increase their value over the entire (long) life of the mortgage. Second, under the initial underwriting guidelines provided in such great detail in the FHA Underwriting Manuals, 18 a large proportion of the already standing housing stock would not meet the basic requirements for mortgage insurance for one or more reasons. The 1940 Census of Housing and Population showed that much of the American housing stock was of poor quality and a notable proportion was dilapidated and unfit for habitation. 19 Finally, many homes, perhaps most, were in neighborhoods that real estate professionals at the time believed would be subject to changes—both economic and social—that would undermine their long-term value. 20
HOLC Activities
The HOLC was created at the conclusion of the first 100 days of legislation of the New Deal to refinance home mortgage loans that were in foreclosure. Much of HOLC’s activities have now been repeatedly documented: the agency was empowered to exchange government bonds for delinquent mortgages and indeed by all measures was exceedingly important to arresting the freefall of the American housing market during the early years of the New Deal and saving the homes of hundreds of thousands of families that almost assuredly would have lost their homes. 21 Not only did the HOLC perform this crucial emergency role of refinancing (finally totaling about one in five urban mortgages throughout the country), its enabling legislation, drafted by the General Council of the FHLB and a former chief executive of an Atlanta S&L, mandated the use of a dramatically different structure of mortgage. HOLC refinancing homes used long-term (initially fifteen years, but later extended to another possible ten years under the Mead-Berry Act), fully amortized mortgages where the principal and interest were paid simultaneously. This both lowered the monthly cost of a home mortgage and rendered it stable for the long life of the loan. 22
Frequently overlooked, however, were the limits placed on HOLC’s authority in its enabling legislation. HOLC could only refinance properties that were one- to four-family homes that were used solely for residential purposes. Loans could not exceed $14,000 nor surpass 80 percent of the appraised value of the property, and no funds could be expended on taxes, maintenance, or incidental expenses. Importantly, the properties had to be in foreclosure. 23 This final point became increasingly important to the Corporation as it quickly attracted applications and began to engage in its core role of refinancing foreclosed properties. HOLC was only permitted (as per legislation) to accept applications for two years after the president signed the bill into law—until June 1935. The Corporation quickly discovered that the demand for their refinancing activities was extraordinary: it received over 1.8 million applications seeking total refinancing for $6.2 billion. C. Lowell Harriss, a Columbia University economist who wrote an NBER monograph on the HOLC, estimated that nearly 40 percent of all qualifying properties in the country (by size, price, and characteristics dictated in the legislation) applied for HOLC relief and that the Corporation ended up refinancing some 20 percent of the nation’s non-farm-owner–occupied properties. 24
After the Corporation ceased accepting applications, its role changed. In part, this was because the HOLC was now the largest holder of mortgages in the country. The HOLC increasingly became involved in the mundane activities of managing such a large portfolio under conditions of continuing economic distress. While the HOLC was clearly a bulwark against the collapse of the nation’s housing market (and not insignificantly buttressing the financing of innumerable thrifts and S&Ls), the economic situation and very high levels of unemployment continued. Many of the same factors, particularly job loss, that had created the financial and the foreclosure crisis remained, and the Corporation now found itself dealing with defaults on the loans it had made. Not only did a significant number of homes come into the possession of the Corporation through foreclosure, HOLC and FHLB officials were increasingly aware of the serious financial problems that persisted throughout much of the nation’s housing market. 25
HOLC diversified administratively to grapple with these problems and expanded its research and policy purview. When the Corporation stopped accepting applications, Harriss reports that the HOLC took two immediate steps. First, it wrote both pending applicants and their lending institutions asking them to work out refinancing arrangements that the HOLC could no longer process. Second, HOLC and other agencies within the FHLB set up informal committees in all areas of the country in an effort to bring applicants and lending institutions together. 26 For its own borrowers, the Corporation created offices to handle the mechanics of loan payments, delinquencies, deficiencies, foreclosures, and the management and resale of repossessed properties. 27
At the same time, the FHLB Board also created what it titled the Mortgagee Rehabilitation Unit. This bureau housed within the HOLC came about to deal with two problems, both outside the responsibilities assigned to it by Congress. This unit was charged with dealing with the wider issues of the U.S. real estate and home finance market, a set of concerns that were outside the legislative purview of either the FHLB or the HOLC. 28 First, the crisis in America’s housing market extended far beyond the thousands of homes that were in actual foreclosure. One contemporary estimated that nearly one-half of the urban home mortgages were in some degree of default. 29 Second, institutions that were not members of the FHLB—primarily life insurance companies and mutual savings banks—held most of these delinquent mortgages. As the nation’s largest holder of home mortgages, the FHLB Board sought to offer assistance, on the basis of HOLC’s refinancing experience, to other institutions that were heavily invested in housing finance. 30
During the course of 1935, the FHLB Board redefined the role and functions of this unit. One of the first moves by the Board (in March 1935) was to approve the hiring of six additional field agents, all assigned to the Mortgagee Rehabilitation Unit and given responsibility over a region of the country. These agents were provided the right to act on behalf of the HOLC and were charged with working directly with mortgage lenders that had large numbers of mortgages that were in default and identify ways to refinance these loans, presumably using appraisal methods and financing structures the HOLC had found successful. 31
The Board finally revised the roles and functions of HOLC’s Mortgagee Rehabilitation Unit in late summer, 1935. The Unit was to collect data and information to be shared with all agencies within the FHLB, as well as the S&Ls that were members of the FHLB, and provide the “greatest possible assistance . . . in restoring normal service to the home finance and in the relieve of distress.” To do this, it was assigned four tasks. First, the Mortgagee Rehabilitation Unit was to conduct nationwide surveys of the nation’s real estate situation and mortgage problems. Second, it was to assess the extent to which private lenders were able to meet their local mortgage lending “responsibilities.” Third, based on their surveys, the Unit was to prepare reports for the Board and agencies within the FHLB on “real estate conditions” in all cities across the country along with recommendations on the best means of improving local policies and practices of public and private lending as well as “bringing about a more general uniformity and higher standard of service at lower costs to home owners.” Finally, this bureau of the HOLC was to “make careful and detailed studies of communities or areas wherein there has been a general breakdown of the functioning of mortgage lending institutions . . . and work with local lending institutions to relieve the situation.” These studies and the recommendations that ensued formed the basis of HOLC’s City Survey Program, the extensive program that in turn created the numerous Residential Security Maps. 32
HOLC’s Development of the Residential Security Maps
The instructions for data collection were included on the reverse side of each of these data collection forms. These directions and the information collected on these forms formed the basis of the final risk assessment code of HOLC’s maps. The data culled by HOLC appraisers on the numerous neighborhood housing markets in the cities analyzed in HOLC’s city survey program reflected a focus on housing quality and upkeep, the amenities of each neighborhood’s housing stock, the advantages as well as those factors that would adversely affect the long-term value of homes, the demographics of the population, the occupations and income of residents, and each area’s ability to attract and maintain mortgage investment flows. 33 Notably, the race and ethnicity (i.e., the “foreign born” population) and, importantly, if zoning provisions and/or racially restrictive covenants to “protect” the character of the community were in place were also to be documented. Clearly those factors, which affected the long-term health and viability of each housing submarket, were of the utmost concern to HOLC administrators and were conveyed to the appraisers in each city (and each area) as critical in making their assessments. 34
The information on these Area Description Sheets was both qualitative and quantitative but uniformly focused on those factors that both determined a neighborhood’s present and long-term attractiveness “from a residential standpoint.” Of utmost concern were measures of the character, upkeep, and present value of housing in the community. HOLC appraisers were instructed to provide the predominant characteristics of the community’s housing stock, the type of housing (single family, duplexes, and apartment buildings), the average age of housing, the repair of the stock (excellent, good, fair, or poor), the percentage of homes occupied by owners, and recent average prices for homes sold (during 1935-1937). Reflecting the new and restrictive dictates of the FHA’s criteria of housing that could be insured under its mortgage insurance program, single-family homes that had been recently constructed, those homes that were well maintained, that had all or most of the most recent amenities (large numbers of bedrooms and one or more fully equipped bathrooms, up-to-date plumbing and heating systems), and maintained relatively high price levels were highly regarded and received the highest grades. Understandably, with housing construction of these newest, most amenity rich, and high value homes in scarce supply since the peak of construction in 1926, few housing submarkets in the HOLC survey received high ratings. In contrast, a great deal of America’s housing stock, much older and lacking many of these basic amenities, and indeed, overcrowded and not infrequently formerly single-family housing that had been subdivided, and was of much lower sales value, received the lower assessment in HOLC’s grades. Numerous data were collected on these sheets that address directly housing quality and characteristics: the prevailing type of housing in the community (single family, duplex, etc.), the average sales price of homes in the area, the age of the housing stock, the general state of upkeep and repair (excellent, good, fair, or poor), and finally, the percentage of homes occupied by owners. And, importantly, for home sales that had occurred (in a clearly down market in 1935-1937), HOLC appraisers were asked to assess the ability of each housing market to attract home mortgage financing from conventional sources of investment (banks, S&Ls, and other lenders). Such assessments were categorically ordered as good (or excellent), ample, restricted, or completely lacking. Predictably, these evaluations were closely correlated with the final mortgage risk assessment grade: those communities receiving the highest HOLC grades (first or second grade—“best” or “very good”) were those individual housing markets that were successful in attracting sufficient levels of financing from conventional lending sources; those that were rated as restricted, highly restricted, or unable to access these sources of financing for home sales were rated much lower (third and fourth grade areas, “declining” or “hazardous”).
HOLC appraisers were required as well to provide a brief narrative of the advantages and liabilities of these neighborhoods. These would frequently place each neighborhood in the topography of the metropolitan area and note specific attractive features (e.g., lake or river views or higher elevations) or features that were quite unfavorable (e.g., proximity to stockyards or factories and frequently low-lying, flood prone areas). Equally if not more important was the neighborhood’s array of community amenities such as parks, schools, churches, and shopping facilities and an area’s basic transportation and utilities (water, electricity, and telephone networks). 35 Even at this late date, many areas of American cities had imperfect (or unsafe) water supplies and did not have paved roads or access to sewer lines. 36 As local real estate and investment specialists, HOLC evaluators were aware that neighborhoods were not only collections of housing that varied widely in terms of quality, type, and upkeep, they were locations that bundled access to the variety of public and private goods across the city. 37 Those neighborhoods that had both high-quality indigenous facilities and the very best access to the network of the city’s basic amenities (and importantly, lacking propinquity to the numerous spatially located disadvantages throughout the metropolitan area) received the highest risk grades. Communities suffering from a lack of or poor access to the city’s public and private goods and in close proximity to noxious facilities were rated by the HOLC as declining or hazardous housing markets.
Equally important to HOLC appraisers were the presence or lack thereof of zoning and protective covenants. The instructions for these Area Description Sheets required HOLC researchers to note the “absence of zoning or restrictions for [the] protection of the neighborhood.” Such deficiencies were considered a distinct negative feature for a housing submarket and almost uniformly translated into a downgrading of the overall risk assessment of a neighborhood. Mirroring the strident approval of zoning and racially restrictive covenants included in the now public FHA Underwriting Manuals, neighborhoods that were zoned solely for residential use as well as “protected” by covenants on individual properties that barred conveyance for occupancy by non-whites were prerequisites for high HOLC risk grades (“best” or “good”). Communities either not covered by zoning restrictions, or having zoning designations that permitted mixed residential and commercial (or industrial) use, as well as those without covenants regulating the use of residential properties were apt to receive the low HOLC risk assessment grades.
Additionally, HOLC instructions for the risk grades assigned on these maps concerned the characteristics of the population, their occupations, and the estimated income of residents. Prominently, HOLC appraisers were to “estimate percentage of Negro families” [italics in original] and note “any threat of infiltration of foreign born, Negro, or lower grade population.” Such concerns were in concert with the prevailing wisdom of prominent real estate analysts, notably Frederick Babcock and Homer Hoyt, the expressed positions of real estate interest groups (especially the important National Association of Real Estate Boards), and individuals throughout the real estate industry, many of whom held positions in the HOLC, the FHLB, and the FHA, as well as those retained by the HOLC that prepared the residential security maps, on the relationship between the racial composition of a community and real estate values. 38 Formulated in the previous twenty years, since African Americans entered Northern cities with the first Great Migration, the American real estate industry firmly held the simple, crude, and straightforward belief that the entry of a single non-white into a neighborhood’s housing market inevitably translated into the decline of that housing market in terms of price, upkeep, and quality.
Finally, appraisers were provided the opportunity to include a narrative of their descriptions and characteristics of the community. It is in these sections that local HOLC evaluators provided much of the vivid and frequently distasteful images of communities that many analysts have highlighted in their analysis of HOLC’s treatment especially of those communities, which were both rated as the very worst risks and those that were home to non-white residents. The overt racism of HOLC’s agents is frequently all too obvious in these brief final justifications of the poor grade assigned to many center city communities. 39
Empirical Determinants of HOLC’s Residential Security Maps
Many factors are related to the risk assessment grade HOLC appraisers assigned to individual neighborhoods. The basic zero-order correlations between all the variables collected and analyzed here from the Area Description sheets are provided in Table 1. We can reasonably make a few tentative conclusions from these basic results. Notable is the fact that there are two prominent factors (presence of either African Americans or foreign-born ethnics) that are negatively related to the HOLC grade. Other factors, however, are also quite strongly related to the mortgage risk grade. HOLC appraisers were advised to consult the results of the numerous Real Property Inventories that had been conducted in major cities across the country. 40 The results were eye opening. Over 2.6 million housing units were surveyed in the initial set of Inventories (for sixty-four cities) and over 15 percent lacked basic indoor plumbing, overcrowding was widespread, and about 15 percent of owner-occupied housing was in need of major repair while over 24 percent of renter-occupied homes were in a similar state of disrepair. 41 The spectrum of housing quality and upkeep is evident in the results. Housing quality is the most robust positive correlate to HOLC’s grade: newer, well-built, and maintained homes are concentrated in those communities receiving the best grades while older, frequently subdivided, and poorly maintained properties were largely located in areas with poor risk grades. Following from this, housing price is similarly associated with the assessment of HOLC analysts. More valuable homes are found in areas with the best grades while least expensive housing is found primarily in submarkets with poor grades.
Correlation Coefficients (Pearson’s r), All Variables Included in Analysis. a
Data were coded from the Area Description Sheets for: Atlanta, Baltimore, Boston, Canton (Ohio), Chicago, Cleveland, Denver, Detroit, Hartford, Kansas City, Los Angeles, Milwaukee, New York, Philadelphia, Pittsburgh, Portland (Oregon), Sacramento, San Francisco, San Jose, St. Joseph (Missouri), St. Louis, Stockton (California), Syracuse, Toledo, Troy (New York), and Youngstown.
Mortgage availability: “Ample” = 4, “Ample FHA” = 3, “Limited” and “Selective” = 2, and “None”= 1.
Housing quality: “Excellent” or “Very Good” = 4, “Good” = 3, “Fair” = 2, and “Poor” = 1.
Occupancy by “Negroes” is treated as dummy variables: Any Negroes = 1, none = 0.
Occupancy by “Foreign born” groups is treated as dummy variables (any foreign born = 1, none = 0).
Source: National Archives and Records Administration II: RG 195.
Age of housing was clearly an important factor. 42 Along with African Americans and foreign-born ethnics, the average age of housing is another, and stronger, negative correlate of HOLC risk assessment grade. New housing is generally of higher quality, has all of the basic amenities (heating, lower per person density, indoor plumbing, and better upkeep), and makes up a very large proportion of the housing stock of those neighborhoods with high grades by HOLC while communities with the lowest grades are made up largely by the oldest housing stock.
The strongest positive correlate of the risk grade, not surprisingly, is the analysts’ qualitative assessment of a neighborhood housing market’s ability to attract high and sufficient levels of mortgage lending from private, conventional lending institutions. Measured against the demand for home lending based on sales during a period of modest economic improvement (1935-1937), HOLC appraisers judged whether these sources were adequate and from private, conventional lenders or simply unavailable. Those individual housing markets that sustained high levels of private home loans were also those neighborhoods judged to be first grade risks, while the many neighborhoods where home lending was nonexistent were graded as poor risks. Overall, the correlations among these variables are relatively modest: the factors that the HOLC used to determine mortgage risk grades are not strongly intercorrelated.
These same variables are shown descriptively as they are grouped by HOLC’s risk assessment grades in Table 2. This highlights the differences HOLC’s appraisers found as they graded the neighborhood and housing markets over the course of the City Survey Program. Most striking, perhaps, is the high percentage of neighborhoods graded as “hazardous” and “definitely declining,” which total nearly two-thirds (62.3 percent) of the areas graded, while only about one-third of the housing markets in this sample are considered “good” or “best” (36.7 percent). Age of housing is directly related to the mortgage risk grade. Not surprising, homes in the most distressed areas were far older than housing in the better housing markets. Nationally, homes in those communities deemed hazardous for mortgage investment were three times older (31.6 years) than those in the best graded housing markets (only 10.1 years). A much higher proportion of the housing stock in the worst areas was renter occupied (nearly half) in contrast to very high owner occupancy in the very best HOLC graded communities (over 90 percent). Housing price and residents’ annual income rise predictably across the hierarchy of HOLC grades. The average housing value in the red, or hazardous, areas was less than one-fourth the value of homes in the very best, or green, areas (about $3,100 in the hazardous areas compared to over $16,000 in the green or best graded housing markets). Occasionally, the differences are even more stark—in some smaller Midwest cities where housing values are listed on the HOLC summary documents as low as $300 to $550 in red, hazardous neighborhoods while the average housing value in the best, or green, neighborhoods is more than $21,250. Housing prices show a clear stepwise progression from the worst to best HOLC graded areas: those communities judged as “hazardous” by the HOLC appraisers are populated by the poorest residents in the city and occupy the least valued housing stock. As the HOLC grade improves, so too do the income levels and value of housing. The better and very best graded communities have noticeably higher income residents and elevated home values.
Summary of Information Collected on Home Owners’ Loan Corporation (HOLC) Area Description Sheets, Selected Cities. a
See notes to Table 1.
Source: National Archives and Records Administration, II, RG 195, Entry 39.
The aggregation of African American and foreign-born residents across communities with different risk grades is especially conspicuous. Both are clearly concentrated in the hazardous and definitely declining neighborhoods. The foreign-born population represents more than 15 percent of communities the HOLC deemed declining or hazardous, the yellow and red neighborhoods, yet barely 3 percent of the better risk neighborhoods. For blacks, the results are even more jarring: while nearly 15 percent of the residents in the worst communities are black, barely 0.1 percent are residents in the very best housing markets, and these are likely domestics in upper income households.
Finally, the assessment of the quality of homes closely tracks understandably to the overall HOLC grade for these community-level housing markets. Communities graded as fourth grade real estate score only 1.6 (or somewhere between poor and fair upkeep of the housing stock) while neighborhoods with the better HOLC risk assessments score 3.0 (good) or better. Mortgage availability also varies in a predictable fashion across the HOLC neighborhood types. In the poorest quality, high-risk neighborhoods, HOLC researchers found that mortgage resources were scarce and many times simply unavailable on any terms. Perhaps surprisingly, even during the depths of the Depression, HOLC appraisers report that there were sufficient mortgage monies in communities with first grade real estate. Indeed, this is clearly tautological: HOLC mortgage risk assessment is closely related to the ability of specific local housing markets to attract (or not) sufficient mortgage resources to adequately facilitate the real estate market of the mid-1930s.
The summary presented in Table 2 suggests that there are several factors that contributed to the assignment of mortgage risk by community. Particularly striking is the fact that while race is certainly an important component of the risk assigned to specific areas, other factors appear to be even more influential. To assess this proposition, the Residential Security Maps for Chicago, St. Louis, and Cleveland—all large and important industrial cities at this time—were selected for more detailed examination. The objective here is simply to visually examine, in these three cities, the relationship between race, mortgage availability, residential upkeep, age of housing, and the final HOLC risk assessment.
The distribution of poor HOLC mortgage grades (third and fourth quality real estate markets) and similarly the presence of non-whites in Chicago are shown in Figure 1.
Chicago Residential Security Map: Poor Home Owners’ Loan Corporation (HOLC) grade neighborhoods and the location of black residents.
There is a perfect overlap geographically of the presence of even a single non-white resident in a Chicago neighborhood and a red or hazardous HOLC grade. Especially striking is the contrast between the very expansive areas that were graded poorly by the HOLC compared to the much smaller geography of areas in which there were any black residents. Indeed, the map in the right panel reflects the largest possible geography of Chicago’s black community at this time. If an area has even a single African American residing in the neighborhood it is shaded on the map in the right panel. Clearly, as in all American cities at this time, the African American community was confined to a very small area in Chicago and all of these communities have adverse HOLC mortgage risk grades—indeed nearly all communities with even a single African American family is graded as “hazardous” by HOLC evaluators.
Simply put, redlining surely affected all black areas, but its effects were much more widespread than the neighborhoods populated by any non-whites. Those communities that represent the highest risk to mortgage lenders (third or fourth grade real estate) comprise an area far larger than the expanse of the then extant ghetto, or even of communities where blacks represent a small proportion of the population. Many all white areas are also graded as high mortgage risks by HOLC appraisers. Indeed, in Chicago there are over four hundred all white areas that are graded as third or fourth grade real estate markets, or many times the number of similarly graded neighborhoods that have even a single black resident. This is replicated in each of these three Midwest cities. Together in Chicago, Cleveland, and St. Louis, all neighborhoods with even a single black family are in communities that were graded “hazardous” or “definitely declining,” and most (nearly 85 percent) were in those communities determined to be red or “hazardous” mortgage risks. At the same time, however, the far majority (68 percent) of the neighborhoods graded red or yellow—“hazardous” or “definitely declining”—had no black residents at all.
African Americans, even after the first Great Migration, only represented a modest minority of the population of Northern cities at this time. Indeed, in Chicago, as of the 1940 census, the total non-white population of the city totaled just 8.3 percent of the city’s total population—in Cleveland and St. Louis the comparable figures were only 7.2 percent and 11.2 percent, respectively. As in all Northern cities, Chicago’s black community lived in but a few neighborhoods where they lived at very high population densities. The communities that the HOLC identified as housing markets that represented a poor risk for long-term mortgages (the third and fourth grade areas) represented a very wide swath across the metropolitan Chicago community. It was far larger than the extant black ghetto at this time. The racial composition of these Chicago housing markets does not provide an adequate reason for most neighborhoods that were assessed as poor mortgage risks. Other factors are clearly involved.
These can be tentatively analyzed by examining maps of mortgage availability (in St. Louis), housing repair and upkeep (in Cleveland), and age of housing (again using St. Louis) and comparable maps of the HOLC grade in the Residential Security Maps. In each case, it is clear that there is an apparently close relationship between these factors and the risk grade assigned by HOLC. This will be confirmed in the results of the regression analysis that will follow the presentation of these maps.
Maps of both mortgage availability and the risk assessment grade for St. Louis are shown in Figure 2. Close examination of both maps shows that mortgage availability (between 1935 and 1937) closely approximates the final risk grade assigned by HOLC. Especially striking is the fact that within the central city of St. Louis mortgage resources were not generally available while the opposite is the case for the suburban areas. Equally the case is the fact that the majority of neighborhoods within the city are graded as a moderate or high risk (third or fourth grade) for mortgage investments. Where the HOLC investigators discovered that mortgage lenders were not willing to invest in the mid-1930s were also neighborhoods graded as fourth rate (hazardous) real estate markets. In contrast, in those areas evaluators found that building and loan associations, some banks, insurance companies, and even individuals were making mortgage loans in the depressed real estate market of the 1930s were graded as the very best mortgage risk for lending institutions.
St. Louis Residential Security Map: Mortgage availability and Home Owners’ Loan Corporation (HOLC) risk assessment.
These basic relationships—the dearth of mortgage availability in central city neighborhoods and its relative abundance in suburban sites and the redundant relationship between mortgage lending in the depressed housing markets of the 1930s and HOLC’s final risk grade—is duplicated in city after city. In the case here, similar comparisons between the HOLC’s assessment of mortgage availability and the final risk grade would show very similar results in Cleveland or Chicago. The small, compact character of St. Louis urbanization displays this relationship quite vividly.
The strength of the relationship between mortgage financing availability during this period and the mortgage risk grade is demonstrated in the cross-tabulation between these two factors. There is a close relationship between these two variables, especially in lower risk grades. It is especially striking to note the near perfect relationship between those St. Louis housing markets where there was mortgage availability assessed as “none” and the risk grade of “hazardous” (ten of eleven of these cases). These relationships are somewhat weaker for communities with limited, somewhat limited, and ample levels of mortgage financing with their respective HOLC grades, but overall the relationship is both significant and rather robust (r = .668, p < .001).
The comparison of the two Cleveland maps in Figure 3 shows a close relationship between the grade assigned by HOLC appraisers and their assessment of housing quality and upkeep for these communities. This is visually conspicuous: the map of the HOLC neighborhood grade from first grade to fourth grade real estate in the left panel is closely approximated by the map showing the housing quality. In general, in neighborhoods where the appraisers deem housing quality to be “dilapidated” or “poor,” the neighborhood grade for mortgage risk is “hazardous”/fourth rate real estate; for communities where housing quality is judged “excellent,” the mortgage risk grade is quite high. The two maps, in other words, largely mimic each other. The basic zero-order relationship is strong and positive (r = .744, p < .001). The very close association between the upkeep and quality of the housing stock and the final risk grade is striking. Over 75 percent of communities with very worst quality housing stock were rated as well as the worst risk for home mortgages. At the other end of this scale nearly 70 percent of those markets where the housing stock was judged as “good” or “excellent” by HOLC analysts were also rated as minimal mortgage risks (second and first grade real estate).
Cleveland Residential Security Map: Home Owners’ Loan Corporation (HOLC) risk assessment and quality of housing.
One final figure, again using St. Louis, illustrates one final point that is vividly demonstrated in these maps is the relationship between the average age of housing and HOLC’s assessment of mortgage risk. This is shown in Figure 4. A facsimile of the St. Louis Residential Security Map is shown in the left panel while the distribution of these same areas by age of homes in provided on the right. Again, it is clear that average housing age closely approximates the mortgage risk grade. The oldest housing, much of it dating to the late nineteenth century, is concentrated within the city of St. Louis and these areas are largely classified as “hazardous” or “definitely declining” housing markets. Most of the then existing suburban areas were composed of homes that had been much more recently built and these communities were largely determined by the HOLC to be advantageous places in which mortgages could be made.
Average age of housing and Home Owners’ Loan Corporation (HOLC) grade in St. Louis neighborhoods.
The results of a regression analysis are provided in Table 3. Separate equations are shown—one for all cases in the sample and individual results separately for Chicago, Cleveland, and St. Louis. Together, the data recorded on the Area Description Sheets are robust predictors of HOLC’s mortgage risk grade. In each equation the adjusted R2 exceeds .70 and is significant in all equations.
Ordinary Least Squares (OLS) Estimates (Standardized Beta Coefficients) of Home Owners’ Loan Corporation (HOLC) Mortgage Risk by Neighborhood, National Sample, and Three Metropolitan Areas, Dependent Variable HOLC Grade a (t Statistic in Parentheses).
Note: *, **, and *** indicate significance at the 90 percent, 95 percent, and 99 percent levels, respectively.
HOLC grade is coded as follows: “Hazardous”/Red = 1; “Definitely Declining”/Yellow = 2; “Still Desirable”/Blue = 3; and “Best”/Green = 4.
For the entire sample, the appraisers’ judgment of the overall quality and repair of the housing stock of a neighborhood is a strong and consistent predictor of the final HOLC grade. While this was ostensibly a subjective judgment by the investigators of the basic, overall state of a neighborhood’s housing stock, it closely approximates mortgage risk. Not surprisingly, the far majority of neighborhoods judged to be at highest risk had the poorest quality housing while markets representing the lowest mortgage risk were primarily composed of homes in excellent or good repair. In all likelihood more than the physical condition or appearance of the housing stock was being judged. HOLC consultants and agents involved in this program were largely composed of professionals from the local real estate industry and were, as developers, salespersons, and financiers of real estate, knowledgeable of more than the basic structures of the housing stock of a community. They were also cognizant of the amenities, types of construction, the planning (or lack thereof) of neighborhoods, as well as public schools, parks, and other facilities. Their assessment was more comprehensive than a simple grading of the housing stock. As such, it is hardly unexpected that the estimation housing quality and repair closely matches the overall HOLC grade.
Along with housing quality, availability of mortgages (during the 1935-1937 period) is a strong predictor of the HOLC grade. HOLC appraisers (many times clearly consulting local bankers and officials of building and loan associations) determined whether or not and to what degree mortgage resources were available for home purchase. In neighborhoods ultimately graded as red, “hazardous” mortgage risk areas, HOLC officials most frequently determined that no bank or other financial institutions in the local area was willing to lend in that specific neighborhood. As the HOLC grade of the housing market improved, the general availability of home mortgage resources increases. Given that the expressed intention of these maps is to determine the mortgage risk to the lender, this measure is tautological—if a neighborhood cannot attract mortgage investments then too in that neighborhood buyers will be unable to secure mortgage financing. Risk, in other words, is defined by the past performance of financial institutions, and furthermore their lending behavior during a period of notable distress in the economy. This qualitative assessment by HOLC appraisers is, for the entire sample, a strong predictor of the mortgage risk grade (b* = .271, p < .001).
Two additional measures are moderate and positive predictors of the HOLC grade—average housing price and the proportion of owner-occupied housing. Areas with diminished property values are also the most distressed communities and are treated as riskiest for mortgages. The more desirable and higher priced properties are located in neighborhoods with little long-term risk to a mortgage investment. The relationship between home price and HOLC’s grade is modest but positive (b* = .157, p < .001). Finally, the percentage of owner-occupied homes is a more modest yet positive and significant correlate of the HOLC mortgage risk grade. To HOLC, ownership was in itself an important positive contributor to the ongoing viability of a community. The more distressed residential markets have, in their view, a higher proportion of renters, a sign of the weakness of the market, and stronger real estate markets have a high proportion of owner-occupied housing. As a predictor of the HOLC grade, ownership is a modest but significant correlate of the mortgage risk assessment (b* = .083, p < .001).
Other factors, also documented on the Area Description Sheets, are negative estimators of the risk grade assigned by HOLC. Three are age of housing and the presence of both non-white and foreign-born residents in the neighborhood (both of which are entered as dummy variables: presence of black or foreign-born residents = 1, none = 0). From the growing literature on redlining and the activities of the FHA and HOLC we anticipate that race will be a particularly important predictor, and a negative one, of the mortgage risk of communities. In fact, of the three variables that are negative correlates of HOLC grades, race is the most modest, yet statistically significant, predictor of HOLC grade. Among real estate professionals in this period, there was a strongly held belief in the “neighborhood life cycle.” Housing, built initially for the highest income groups, is passed down the socioeconomic ladder as the city expands geographically and the upper and middle classes exit from the most central (and oldest) sites to outlying areas. With increasingly diminished upkeep and, in many cases, the subdivision and more intensive use, the age of housing translates into poorer quality and much less valuable real estate. From the data collected in the HOLC data sheets, age is indeed negatively and moderately related to the mortgage risk of a neighborhood. In fact, age is comparably robust to average housing price (b* = –.125, p < .001). Finally, the presence of black and foreign-born residents are both modest and negative estimators of the HOLC mortgage security grade.
Conclusion
The HOLC was one of many newly formed, potent, and innovative agencies involved in housing and finance during the first years of the New Deal. The HOLC and the FHA dealt directly with housing issues. The HOLC intervened to refinance a very large proportion of the American housing stock and did so in only a few years while the FHA was responsible for implementing a mortgage insurance program and in so doing creating a new standard mortgage instrument. Both agencies were involved as well in the mapping of mortgage risk across America’s cities. HOLC’s efforts are well (although probably not fully) documented in records that are available at the National Archives. The maps and other documents of a similar project at the FHA assigning a numeric mortgage risk grade by block is not available, although at least one summary map of this project in Chicago has been uncovered. 43
With overwhelming Democratic majorities in both houses of Congress coupled with relatively weak legislative oversight over the increasingly numerous and large national agencies that formed the heart of the first New Deal, both the HOLC and the FHA enjoyed initiative and a high degree of autonomy. Indeed, the FHA, like the FDIC and the FSLIC, was not dependent upon Congressional appropriations, further enhancing their autonomy. This independence allowed these agencies to develop the implementation of policies that had sweeping consequences. For the HOLC particularly, creating the Residential Security Maps had two purposes: to direct the underwriting criteria of the FHLB and to provide to all other newly regulated financial institutions that were engaged in home mortgage lending a detailed guide for their contemporary and future mortgage loan investment decisions.
HOLC’s City Survey Program was conducted by a staff (and consultants) that largely came from the real estate and finance industry. They were aware that a surprising portion of the housing stock of America’s cities was inadequate, lacked amenities, and suffered much from years of deferred maintenance. They were also cognizant that mortgage insurance had created a new standard mortgage financial instrument, one that dramatically extended the normal length of a mortgage to a minimum of twenty years. This new standard mortgage (by definition a “safe and sound” investment in the eyes of vigilant and energetic regulators) was essentially the only kind of home purchasing loan that a regulated financial institution could offer. Much of the housing stock already built by the onset of the Depression could not meet the minimal standards of mortgage insurance requirements. Bluntly, the lending behavior of regulated financial institutions would have to change. The Residential Security Maps provided a set of spatially specific directions on how investment decisions ought to be made by these newly regulated financial institutions in the new world of home financing.
The Residential Security Maps recognized, in other words, a very different mortgage market, a market that would direct lending toward new construction and only with great hesitation make similar loans on much of the housing that was already standing. S&Ls and all regulated institutions in general were going to have to abandon their old investment and lending habits. In particular, because of the many requirements of mortgage insurance, many neighborhoods would have to be abandoned. Older and obsolete housing (and neighborhoods) in less desirable areas could not be refinanced.
The HOLC Residential Security Maps were constructed from an empirical enterprise. Data from HOLC’s Area Description Sheets that were an integral part of the Residential Security maps are analyzed here for the first time. Based upon these data culled from this source, it is clear that the mortgage risk assessment grade was based not only (or even most importantly) on the race and ethnicity of the residents of an individual community but on a bundle of factors. Consistently and emphatically, the presence of non-whites in a community is a negative factor in determining the risk grade.
However, the analysis presented here shows that a set of factors, one of which is race, were together robust predictors of the risk assigned by HOLC analysts. The appraisers’ judgment of the quality and upkeep of homes is consistently in each of the equations the strongest correlate with the final HOLC risk grade. Somewhat less robust is the assessment of the availability of mortgages primarily from conventional financing sources in the mid-1930s. HOLC records show, for instance, that in a select set of cities where the agency conducted a “re-survey” in the final two years of the program that bankers were shown drafts of the maps and queried on where their institution (and others to their knowledge) were making home mortgage loans—and on what terms. Clearly, this assessment was also a strong predictor frequently comparable to housing quality in the equations. Finally, the average age of homes in a community, one final indicator of obsolescence, is only a modest correlate with the risk grade determined by HOLC analysts. It was, however, consistently a stronger estimator of grade than race or the presence of foreign-born residents.
New Deal agencies were responsible in restructuring home finance. This revolution in home financing mediated by mortgage insurance created a new way and new channels in which mortgage capital would flow across metropolitan space. HOLC’s Residential Security Maps are both empirical and normative. Based upon the structure of neighborhood housing development that had been built and expanded with the explosive population and economic growth of America’s industrial metropolitan centers, HOLC analysts applied the new rules of mortgage insurance and home financing that were the centerpiece of New Deal housing policy over the already present housing development and neighborhood structure of American cities. With this, they provided a set of home mortgage flow predictions. They saw the nation’s cities as a set of highly differentiated and primarily obsolete housing markets and provided the nation’s financial institutions a set of normative statements (a range of mortgage risk grades) that should (and would) guide their investment decisions in the newly created system of home mortgage financing. Empirically rooted and analytically astute, HOLC’s Residential Security Maps not so much initiated the practice of home mortgage financial redlining as accurately predict, across the built environments of American metropolitan complexes, where mortgage capital would and would not flow. HOLC’s Residential Security Maps represent a narrative on the new landscape of investment decisions—investments to be made and those to be avoided.
Footnotes
Acknowledgements
The author thanks Colin Gordon for graciously sharing the ArcView polygons of the St. Louis Residential Security Map. Several individuals read and provided valuable comments on the article, including Marty Abravenal, Ira Katznelson, Keith Reeves, Carol Nackenoff, Bob Ibanez, and two anonymous reviewers. The views and opinions expressed in this article are solely those of the author and do not reflect those of the Department of the Treasury or the U.S. government.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.