American Economic Review: Papers & Proceedings 2011, 101:3, 366370

http://www.aeaweb.org/articles.php?doi=10.1257/aer.101.3.366

The New Deal, Race, and Home Ownership

in the 1920s and 1930s
By Trevor M. Kollmann and Price V. Fishback*
The gap between white and black homeownership rates has remained at roughly 26 percent
for the past century (William Collins and Robert
Margo 2011). Federal housing policies long
have promoted homeownership, but minorities
are said to have benefitted little from these policies. Several policies began in the New Deal and
often included nondiscrimination statements.
Yet, state and local officials often had latitude
to indulge their own discriminatory attitudes in
administering the programs. For example, the
Home Owners Loan Corporation (HOLC) was
said to have discriminated because the agency
typically gave neighborhoods with black,
Hispanic, or Asian infiltration the lowest ratings when evaluating loan quality. The Federal
Housing Administration (FHA) is said to have
relied on these ratings in choosing to insure only
a small share of loans in low-rated neighborhoods (Charles Abrams 1955; Kenneth Jackson
1985). On the other hand, Amy Hilliers (2003)
study of Philadelphia shows that the HOLC
refinanced a number of loans for blacks and a
reasonable number of loans in the low-rated districts. We examine the extent to which blacks
and whites benefitted from 1930s housing policies based on a new pseudo panel that matches
households from the censuses of 1920, 1930,
and 1940 with policy measures in 460 state economic areas (SEAs).

indirectly influenced housing. To keep people

in their homes, the HOLC purchased over
1 million troubled nonfarm mortgages from
lenders and then refinanced the mortgages on
advantageous terms between 1933 and 1936.
The program simultaneously propped up home
ownership and replaced the toxic loans on lenders books. Both Charles Courtemanche and
Kenneth Snowden (forthcoming) and Fishback
et al. (forthcoming) find that HOLC loans
increased the number of home owners in counties with fewer than 50,000 people but did not
find an effect in larger areas.
Race-specific information on the distribution
of HOLC funds (and other New Deal funds) is
not available, so the estimated effect for blacks
is a reduced-form function of three relationships,
Receipt, Impact Given Receipt, and Spillover.
Receipt represents the extent to which blacks
within the SEA received the HOLC funds. Impact
Given Receipt is the effect on home ownership
for those blacks who received refinancing, which
was likely positive. Spillover is the spillover
effect on black home ownership when whites
received HOLC loans. Even if all of the HOLC
money went to whites, the replacement of toxic
assets held by private lenders freed up loan funds.
If a share of freed funds was subsequently lent
to blacks for mortgages, black home ownership
would have risen. The spillover effect might have
been negative if lending to white borrowers raised
the demand for white homes in ways that raised
the relative return of lending to whites.
Twenty-seven state governments in 1933 and
1934 sought to keep people in their homes with
mortgage moratoria that prevented lenders from
foreclosing on delinquent home loans where borrowers had some chance of repaying later (Robert
Skilton 1943). The laws had the immediate effect
of allowing homeowners to stay in their homes.
When the moratoria ended, however, interest
rates rose and loan terms tightened more than in
states without moratoria (Randal Rucker and Lee
Alston 1987). The net long-term effect on home
ownership rates depended on the strength of the

I.Programs and Their Anticipated Effects

The programs of the 1920s and 1930s

included programs designed to keep people in
their houses, policies promoting new opportunities for housing, and New Deal programs that
*Kollmann: Department of Economics, University of
Arizona, Tucson, AZ 85721 (e-mail: kollmann@email.arizona.edu). Fishback: Department of Economics, University
of Arizona, Tucson, AZ 85721 (e-mail: pfishback@eller.
arizona.edu). The authors thank William Collins, Tomas
Cvrcek, Robert Margo, Jonathan Rose, and Kenneth
Snowden for comments.
366

VOL. 101 NO. 3

The New Deal, Race, and Home Ownership in the 1920s and 1930s

contrasting effects and whether the new restrictions hit blacks or whites harder.
Among the programs providing new housing
opportunities, the FHA began providing repayment guarantees to lenders for nonfarm home
repair loans in 1934 and for nonfarm home
mortgages in 1935. As with the HOLC, the
expectation is that the impact given receipt was
to raise home ownership rates for blacks, and the
spillover effect could be positive or negative.
The Public Works Administrations (PWA)
public housing grants cleared slums and built new
and higher quality rental public housing projects
for the poor. To the extent that poor homeowners
were attracted to the new apartments, the homeownership rate would have fallen.
In the farm sector the Federal Land Banks
(FLB) began providing low interest amortized mortgages to farm owner/operators in
1917 and Congress provided a series of crop
and feed loans with ad hoc appropriations in
response to crises. By 1929 the FLBs were
involved in about 19 percent of farm mortgages. The New Deal reorganized the farm
mortgage program and regularized production
and emergency crop and field loans under the
Farm Credit Administration (FCA) and provided aid to poor farmers through the Farm
Safety Administration (FSA). All programs
were said to be targeted at farmers struggling to
obtain private loans. To the extent that low risk
farm borrowers obtained access to the loans,
the programs might have had a perverse effect.
If low risk borrowers crowded high risk borrowers out of federal programs, private lenders
faced a pool of high risk borrowers that would
have led them to constrain credit.
The payments that the Agricultural Adjustment
Administration (AAA) made to farm owners to
take land out of production potentially had mixed
effects on homeownership among the black farm
population. Narrative evidence suggests that farm
owners obtained a disproportionate share of the
payments at the expense of their tenants. When
farm owners took land out of production, their
demand for labor likely fell, leading to fewer
opportunities for share croppers, farm workers,
and lower income tenants. Given that whites in
agriculture were more likely to start the decade
of the 1930s as farm/homeowners while blacks
dominated the ranks of the farm workers, croppers, and tenants, the AAA likely raised white
and reduced black homeownership rates.

367

II. Data and Estimation

We develop a pseudo panel with repeated

cross sections of households obtained from 1
percent samples of the 1920, 1930, and 1940
census provided by the Integrated Public Use
Microdata Series (Steven Ruggles et al. 2008).
The households are matched with aggregate
measures of the New Deal provided by the
Office of Government Reports (1939), Farm
Credit Administration (various years), Federal
Farm Loan Board (various years), census data
on demographics (Michael Haines 2004), and
government estimates of economic activity
in 460 SEAs (US Department of Commerce
Bureau of Economic Analysis 2010; William
Martin 1939). We estimate separate probit equations of the form below for nonfarm and farm
households on samples that pool both black and
white households:
Ownijt =f(NDjt, Xjt, Yijt, SEA, YEAR,
Bi NDjt, Bi Xjt,
Bi Yijt, Bi SEA, Bi,
Bi YEAR, ijt).
Ownijt is equal to one if household i in SEA j
in year t owns a home, and zero otherwise. The
vector NDjt is composed of measures of various
government programs in SEA j for the period
since the last census year. By including a vector
of SEA fixed effects, the coefficient of the per
capita spending on a New Deal program captures the relationship between changes between
the 1920s and the 1930s in the New Deal per
capita measure and changes in average home
ownership within the SEA between 1920, 1930,
and 1940. A vector of characteristics for household i in SEA j in year t (Yijt) controls for the
household heads age, gender, marital status,
literacy, labor force status, and occupation (12
categories), as well as the households family
size, number of children, number of employees,
spouses labor force participation, and location
in a metropolitan area. A vector of YEAR fixed
effects controls for nationwide shocks in 1930
and 1940.
The Roosevelt administration distributed
more funds to areas facing greater economic

368

MAY 2011

AEA PAPERS AND PROCEEDINGS

Table 1Marginal Effects and Percentage Change in Probability of Homeownership Associated with the Change
in the Mean Policy Variable between the 1920s and 1930s
Change in probability
Nonfarm households
Farm households
Policy variable
FHA loans insured per 1930
nonfarm homeowner
HOLC Loans 193336 per 1930
nonfarm homeowner
PWA public housing grants
193339 per 1930
nonfarm population
Public works grants 193339
per person in 1930
Relief grants 193339 per
person in 1930
AAA grants 193337 per
farmer in 1930
Federal farm mortgage loans
per farm population in prior
census
Federal farm nonmortgage
loans per farm person in 1930
Federal nonfarm, nonhousing
loans, 193239 per person
in 1930
State mortgage moratorium,
193334
Average state income since
previous census per person
in prior census
Average state crop income
Since last census per farm
person in prior census
Actual change in
homeownership rate

Marginal effects
Nonfarm households
Farm households

White

Black

White

Black

White

1.82*

1.16

0.75

1.08

0.32

1.10

0.07

2.80*

0.37*

0.25

0.05

0.18

0.00

0.27

0.35

0.12

0.24

0.94

1.07

0.00

0.89

0.71*

0.47

0.95*

1.00

0.33

1.10

0.11

0.08

0.11

1.00

0.007

0.005

0.20

0.50

2.10*

1.30

0.002

0.005

1.03*

0.58

0.12*

0.07

0.100*

0.054*

0.099* 0.063a

0.93*

1.56*

0.06

0.04

0.144*

0.237*

0.029

5.0

3.2

1.6

0.7

0.023*
0.004

Black
0.015*

White

Black

0.018*

0.026

0.013

0.002

0.062*a

0.650* 0.443

0.321

1.150*a

0.000

0.027

0.032

0.13

0.010

0.040

0.068*

0.008

0.62*

0.81

0.000

0.088*a

0.039*

0.050

0.86*

1.44

0.375*

0.246

0.546*

0.888

0.473* 0.499a

0.010

0.191

0.641

0.010

0.092

0.021*

0.013

0.018

Notes: Full probit results are reported in the online data appendix (http://www.aeaweb.org/articles.php?doi=10.1257/
aer.101.3.366). Marginal effects are reported in thousands of dollars and are adjusted for inflation using 1967 as the base year.
* Significant at the 10 percent level.
a

Interaction effect is statistically different from zero at the 10 percent level.

distress. This practice might lead to a negative

omitted variable or endogeneity bias in the New
Deal coefficients. To reduce the negative omitted
variable bias, we include the vector Xjt composed of the average per capita state income, the
average state crop income from 18 crops in the
period between census years (US Department
of Agriculture National Agricultural Statistics
Service 2010), and the percentages of blacks,
foreign born, and illiterates in the area in year
t. All dollar values were deflated by the CPI
(1967=100) (US Department of Commerce
Bureau of the Census 1975, series E-135).

The coefficients on interactions between the

various vector components and a black dummy
Bi measure the black-white difference in relationships, and ijt is a stochastic error term.
III. Results from Probit Analysis

Table 1 reports the marginal effects for policy

variables derived from the probit estimates with
standard errors clustered at the SEA level and
the change in the probability of home ownership
associated with the change in the policy mean
between the 1920s and the 1930s. The last row

VOL. 101 NO. 3

The New Deal, Race, and Home Ownership in the 1920s and 1930s

provides context for the changes in probability by showing the actual change in ownership
rates between 1930 and 1940 in the sample. In
most cases statistical tests cannot reject the null
hypothesis of no difference between the black
and white coefficients.1 Typically, the marginal
effects for blacks (based on the combination of
the coefficients for whites and the black interaction term) were less precisely estimated than for
whites.
FHA insurance and PWA public housing provided new housing opportunities. The
spread of FHA mortgage insurance contributed
to increases in nonfarm home ownership that
more than offset the declines contributed by
introductions of public housing. An increase in
FHA insurance of loans from $0 in the 1920s
to roughly $800 (1967$) in loans insured per
nonfarm home owner in 1930 was associated
with a 1.83 percent rise in the probability of
white home ownership and a 1.17 percent
increase in the probability of black home ownership. On the other hand, PWA public housing
and slum clearance appears to have funneled
some home owners into renting in the projects. The rise in PWA public housing spending per nonfarm person from $0 in the 1920s
to $5.60 (1967$) in the 1930s was associated
with a probability of home ownership that was
0.36 percent lower for whites and 0.25 percent
lower for blacks.
State mortgage foreclosure moratoria had
the most success at maintaining homeownership among farm households with little impact
on nonfarm households. In moratoria states,
the probability of white farm households owning their home was 1.36 percent higher and the
probability for black farm households was 0.863
percent higher. The early 1930s reductions in
foreclosures in moratoria states apparently outweighed the impact of tightened credit later.
In contrast, the AAA program was associated with lower white farm home ownership
and higher black farm home ownership. The
rise in AAA spending per capita from $0 in the
1920s to $160 (1967$) per member of the farm
population in the 1930s lowered the probability
of home ownership for white farm h ouseholds
1

The farm interaction coefficient was statistically significant for nonfarm HOLC and public housing programs, and
the nonfarm interaction was statistically significant for the
AAA and nonmortgage farm loans.

369

by 0.6 percentage points and raised it by 0.80

percentage points for blacks. The white result
seems consistent with a consolidation of farms
into larger units associated with the AAA
that contributed to the outmigration found by
Fishback, William C. Horrace, and Shawn
Kantor (2006). The positive result for blacks is
consistent with a scenario where black workers,
tenants, and croppers were harmed by a decline
in labor demand when land was taken out of production. Those without land were more likely
to move, so that the blacks remaining on farms
were more likely to be owners.
Proponents of the farm credit programs for
farm mortgages would have been unpleasantly
surprised by findings of a negative relationship
with white farm ownership. The increase in
average farm mortgage loans per farm population from $9 in the 1920s to $25 (1967$) in the
1930s was associated with a 0.86 percent reduction in the probability of white farmers owning
a home. In contrast, the change was associated
with 1.4 percent higher black farm homeownership. The reduction in white farm homeownership is consistent with a setting where low risk
borrowers crowded out high risk borrowers in
the federal program, but more work needs to be
done to document whether that is the reason for
the negative effect.
The HOLC purchase and refinancing of toxic
nonfarm mortgages was negatively but statistically insignificantly related to nonfarm ownership for both blacks and whites. The results differ
from the positive impact on home ownership
found for smaller counties by Courtemanche
and Snowden (forthcoming) and Fishback et al.
(forthcoming). One reason for the difference
may be that this study includes large urban areas,
whereas the prior studies found weak effects in
larger counties.
In summary, the tests for different relationships between housing policies and black and
white homeownership in most cases show no
statistically significant differences. In most cases
the relationships for blacks are smaller than for
whites and the black effects are imprecisely estimated. Nonfarm white home ownership rates
were higher in areas with more FHA mortgage
insurance and fewer grants for public housing
and slum clearance. White farm homeownership
rates were higher where states passed mortgage
moratoria, the AAA spent less, and there were
fewer federal farm loans.

370

AEA PAPERS AND PROCEEDINGS

REFERENCES
Abrams, Charles. 1955. Forbidden Neighbors: A

Study of Prejudice in Housing. Port Washington, NY: Kennikat Press.

Collins, William, and Robert Margo. 2011. Race
and Home Ownership from the End of the
Civil War to the Present. American Economic
Review, 101(3): 35559.
Courtemanche, Charles J., and Kenneth A.
Snowden.  Forthcoming. Repairing a Mort-

gage Crisis: HOLC Lending and Its Impact on

Local Housing Markets. Journal of Economic
History.

Farm Credit Administration. Various years from

1933 to 1939. FCA Annual Report. Washington, DC: United States Government Printing
Office.
Federal Farm Loan Board. Various years from

1917 to 1932. Annual Report. Washington, DC: United States Government Printing
Office.
Fishback, Price V., William C. Horrace, and
Shawn Kantor. 2006. The Impact of New

Deal Expenditures on Mobility during the

Great Depression. Explorations in Economic
History, 43(2): 179222.

Fishback, Price V., Alfonso Flores-Lagunes, William C. Horrace, Shawn Kantor, and Jaret
Treber.  Forthcoming. The Influence of the

Home Owners Loan Corporation on Housing

Markets during the 1930s. Review of Financial Studies.

Haines, Michael R., and the Inter-University Consortium for Political and Social Research. 2004.

Historical Demographic, Economic, and

Social Data: The United States, 1792000.
ICPSR (2896).
Hillier, Amy E. 2003. Who Received Loans?
Home Owners Loan Corporation Lending
and Discrimination in Philadelphia in the
1930s. Journal of Planning History, 2(1):
324.

MAY 2011

Jackson, Kenneth T. 1985. Crabgrass Frontier:

The Suburbanization of the United States. New

York: Oxford University Press.
Martin, William. 1939. National Wealth and Its
Geographic Distribution, 19191938. In Conference Board Studies in Enterprise and Social
Progress, 11422. New York: National Industrial Conference Board.
National Archives Depository II, College Park,
MD, Office of Government Reports, RG 44.

Report No. 10-1939.

Rucker, Randal R., and Lee J. Alston. 1987.

Farm Failures and Government Intervention:

A Case Study of the 1930s. American Economic Review, 77(4): 72430.

Ruggles, Steven, Matthew Sobek, Trent Alexander, Catherine Fitch, Ronald Goeken, Patricia
Kelly Hall, Miriam King, and Chad Ronnander.

2008. Integrated Public Use Microdata

Series: Version 4.0 (Machine-readable database). Minneapolis, MN: Minnesota Population Center.
Skilton, Robert H. 1943. Mortgage Moratoria
since 1933. University of Pennsylvania Law
Review and American Law Register, 92(1):
5390.
United States Department of Agriculture National
Agricultural Statistics Service. 2010. Field

Crop Data. Data and Statistics. http://www.

nass.usda.gov/Statistics_by_Subject/index.
php?sector=CROPS. (accessed December 10,
2010)

United States Department of Commerce Bureau

of Economic Analysis. 2010. Personal Income

Dataset, 1929Present. Regional Economic

Accounts
(SA1-3).
http://www.bea.gov/
regional/spi/default.cfm?selTable=summary
(accessed December 10, 2010).

United States Department of Commerce Bureau

of the Census. 1975. Historical Census of the

United States: Colonial Times to 1970. Washington, DC: United States Government Printing Office.