Housing Affordability and Spatial Price Variations in the United States

Author(s): Frederick P. Stutz and Arthur E. Kartman

Source: Economic Geography , Jul., 1982, Vol. 58, No. 3 (Jul., 1982), pp. 221-235
Published by: Taylor & Francis, Ltd.

Stable URL: https://www.jstor.org/stable/143511

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Taylor & Francis, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to
Economic Geography

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
 E C ON OMICa GEOGR6APHY
VOL. 58 JULY, 1982 No. 3

HOUSING AFFORDABILITY AND SPATIAL PRICE VARIATIONS

IN THE UNITED STATES

FREDERICK P. STUTZ

ARTHUR E. KARTMAN

San Diego State University

The nation is in the midst of a severe housing affordability crisis wlich has meant
spiraling housing prices, extremely low vacancy rates, and a cutback in construc-
tion. Following a theoretical formulation and regression analysis, we conclude that
loan rates and construction costs are not principal determinants of national house
price fluctuation. A moderate climate and per capita income, surrogates for shift-
ing demand, accounted for the largest share of data variability while environmen-
tal restrictions, though not a formal part of the analysis, appeared important. Local
level policy tools to ease the affordability problem include land use regulation,
building code and permit regulations, and the use of tax exempt status of interest on
municipal bonds.

According to the U.S. Bureau of the all center around structural characteris-
Census, the median price of a new, single tics of the house and its improvements,
family housing unit in the United States the external neighborhood characteris-
rose from $23,000 in 1970 to $64,600 in tics, the location, and some form of insti-
1980, an average annual compounded tutional behavior [5; 16].
growth rate of 10.7 percent [24]. Much of There generally have been problems
this growth occurred in the second half of when comparing variables accounting
the decade, but by 1980 a sharp decline in for different levels of prices using the
price appreciation appeared (figures in- cross section technique versus those taken
dicate that from 1979 to 1980 the rise was through time [2]. Studies of housing price
only 2.7 percent [24]). In spite of this levels are also complicated by the fact
recent decline in appreciation, the expe- that housing is not a homogeneous good
rience of the 1970s has created concern [14]. Instead, housing consists of a bundle
about the affordability of housing. of attributes, the average price of which
Since price is the most reliable index of we observe in the aggregate [5].
housing market performance despite its Any attempt to address the issue of
many measurement difficulties, it merits affordability must begin with a recogni-
special attention [18; 29]. All studies to tion of the multi-dimensional nature of
date which have focused on spatial varia- the topic. Changes in national median
tions in housing prices and affordability home prices over time cover only one
have dealt with housing submarkets at aspect. Another is the matter of regional
only the local or regional levels [3; 8; 21]. variation in the level of housing prices.
The results vary with the area studied, but Such variations, documented in Table 1

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
222 ECONOMIc GEOGRAPHY

with Federal Home Loan Bank Board existing) in various Standard Metropoli-
data on the average transaction price of tan Statistical Areas (SMSAs) for the first
owner-occupied housing (both new and two months of 1981, may be caused by
long-term structural differences between
TABLE I
local housing markets and/or by differen-
AVERAGE PRICE OF HOUSING IN UNIrED STATrES cGrIES' tial rates of change in price in the recent
Average l'rice past.' Finally, one must carefully distin-
Jan.-Feb. 1981 guish between the prices of owner-oc-
Atlanta $ 94,000 cupied units and rents charged by land-
Baltimore 67,000 lords. Higher prices in the former case
may not necessarily represent an afforda-
Boston-Lawx-rence-Lowell 70,500
bility problem for the purchaser, provided
Chicago-Gary 68,300
these prices are reflected in the ultimate
Cleveland-Akron-Lorain 67,250 resale value. Rather, the problem is one of
Columbus 87,000 inadequate cash flow in the short run.
I)allas-Fort Worth 111,550 Higher rents, however, may represent an
Denver-Botilder 80,300
affordability issue for tenants.
Tlhe purpose of this paper is to present
I)etroit-Ann Arbor 78,900
a general discussion of affordability and
Honolulu 109,800
some of the factors which have contrib-
I Iotliston-Galveston 83,850 uted to differential levels in the prices of
Indianapolis 71,450 owner-occupied housing. Differential
Kansas City 64,300
levels of housing prices leading to re-
gional variations pinpoint areas of varia-
Los Angeles-Long Beach-Anaheim 118,450
tion in housing affordability (Figure 1).
Louisville 56,750
The identification of such areas are of
Miami-Fort Lauderdale 81,900 strong geographic interest. But more impor-
N ilxautkee-Racine 74,300

Minneapolis-St. Pautl 83,700

I The emphasis in this paper is on present levels of
New York-Newark-Jersey City 86,800
housing prices in the United States and not rates of
P'hiladell)hiai-WN'ilmington-'T'renton 63,900 change over time per se. Rates of change is another
dimension to the affordability question; however,
Phoenix 103,250
due to the Federal HIone Loan Bank Board's policy
Pittsburgh 47,500 change in 1979, which lumped new and existing
home price information together with homes whose
Portland, OR 65,200
purchases were financed by conventional mort-
Rochester, NY 55,650 gages, it was impossible to show compatible rates of
St. Louis 69,600 change prior to and after 1979. The data from 1979
to 1981 present too small a growth period to be of
Salt Lake City-Ogden 73,600 much value. However, because of month-to-nmonth
San Diego 114,200 variability of average prices, the data psed are two-
month averages. To determine if 1981 price levels
San Francisco-Oakland-San Jose 134,450
were representative of a longer period of time, a
Seattle-Tacoma 74,100 rank correlation of housing prices city-by-city for
1979 and 1981 was performed. It yielded R=0.71,
lamp)a-St. Petersburg 73,650
suggesting moderate stability in home prices over
Washington, D).C. 112,500 the short run. An R of 0.77 for 1979 compared to
1982 data further suggest that tendency. While the
U.S. 75,200
1979 data represent a time slice, they are representa-
T'he data were taken
tive of a longer period of time, and any explanatory f
Federal Home Loan Bank Board. They are limited to
variables identified are probably relatively stable
homes whose p)lrchases were financed by a conventional
mortgage (i.e., not FIIA insured nor VA guaranteed and for other years. From this preliminary check it
include loans made by "all major lenders"). Because of the appears that there is a strong temporal autocorrela-
month-to-month volatility of average prices, the data are
tion among housing prices from year to year, and
simple averages of two monthly averages. No attempt
was made to adjust the data for regional differences in that during this time period, prices have not been
housing quality nor for differential rates of quality change.highly cyclical spatially.

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
 HOUSING AFFORDABILITY 223

EE1~

z In*4

~~~~zAKJ~~~~~~~~.JLA0
Q~~~~~~~~

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
224 ECONoMic GEOGRAPHY

tantly, housing affordability affects busi- rural elderly. Today, however, the prob-
ness and household migrations in these lem affects everyone without permanent
areas, as well as rental vacancy levels, residence. Families or individuals with an
municipal tax revenues, and building indus- annual income of less than $35,000 usually
try activity. The difficult question of the cannot qualify to purchase a home in
spatial affordability problem is, then, many states. Those who do not own a
which areas and what groups are suffer- home presently are being priced out of
ing the most. In the final analysis, some the market, as an average income will not
people are paying more than they should qualify an individual or family to pur-
for housing and more than they can chase an average priced home.
afford. A number of political, social, and eco-
A discussion of the problem of afforda- nomic events occurred over a short
bility and of the factors which contrib- period of time creating the present situa-
uted to the rapid escalation in housing tion. On the supply side, during the
prices in the 1970s is presented first. Then period from 1970 to 1972, housing de-
a general framework, based on economic mand in the country began to exceed
theory and designed to offer an explana- housing production dramatically. The
tion for geographic variations in the mood for construction was never better.
prices of owner-occupied housing, is de- But housing starts and housing produc-
veloped and discussed. Next, a prelimi- tion fell sharply between 1973-1977, as a
nary regression model is constructed and result of an increased difficulty in getting
tested. Some implications for public pol- credit, associated with the nationwide
icy to ease the affordability problem are recession during that period.
presented. Meanwhile, on the demand side, house-
hold formations started to escalate, re-
gional variation in demand became evi-
THE PRESENT HOUSING CRISIS- dent, and an increased number of people
began to migrate to the South and West
SUPPLY AND DEMAND
from the Midwest, primarily as a result of
The nation is in the midst of a severe environmental and employment factors.
housing crisis which has meant spiraling Also, sociological and demographic con-
housing prices and extremely low va- ditions meant that more people were now
cancy rates. Further, a cutback in con- looking for homes. The post-World War
struction is contributing to the downturn II baby boom began to mature, from the
in the national and state economies. For college boom years of the 1960s to the
example, in the nation's most populous apartment boom of 1967-72. Finally, it
state, California, it is estimated that the reached into the single-family housing
building deficit has reached 300,000 market in the late 1970s. At the same time,
units [1]. The result of this decreased there was a decrease in household size,
supply has led to a steep escalation in partly created by an increase in both
housing costs from an average of $30,400 young and elderly singles living alone and
for a single-family detached house in by an increased divorce rate as a result of
southern California in 1970 to $124,100 in increased time pressures associated with
1980, a three-hundred percent increase two-worker households-a situation
[1]. Indeed, housing prices have in- created in part by the high cost of housing
creased in most western countries in real itself. This decrease in household size also
terms (discounting for inflation and in- increased somewhat the number of homes
come growth [6]). needed. There was also a tendency for
Up to the mid-1970s, the problem of children to "leave the nest" sooner, and
finding and affording a house was con- enter college or the occupational world.
fined chiefly to the lowest income groups, This meant an even greater demand for
minority and inner city groups, and the housing.

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
 HOUSING AFFORDABILITY 225

Coupled with these demographic fac- waste water treatment facilities, and
tors was the expectation, starting again in street lights). This threw an increased
1975 and continuing to the first quarter of burden on the building industry to pay
1980, that housing was a good investment for these improvements and to include
and the best hedge against inflation. This the cost of them in the price of homes.
increased the demand, as many small In the late 1970s, when the recession
investors entered the market [7]. They was over and credit was again available
expected to capitalize on the tremendous to home builders and buyers, the builders
appreciation that was occurring in hous- began to feel the impact of the environ-
ing values. This expectation of price in- mental laws that had multiplied, but were
crease helped to send prices upwards. essentially unproven. Time for process-
Boom conditions similar to those ini- ing permits and plans was greatly length-
tially observed in California appeared in ened, during which time inflation drove
an increasing number of other single- the price of homes and contracts up-
family housing markets. In the spring of wards.
1978, when sales activity and price in- As home building lagged, demand in-
creases in some areas had moderated, creased through the real growth in wages
buyers were reportedly camping out wait- and salaries and through the increase of
ing for the opening of desirable new women entering the labor force. This
tracts in the Miami-Fort Lauderdale area. resulted in a large number of two-income
Some builders in Phoenix were selling out households. Due to women's rights move-
homes before they put in streets and ments, women got equal credit standing
curbs. Phoenix, Denver, Minneapolis-St. and this further increased the demand.
Paul, and Seattle, which all set records for Property taxes during this period of
single-family permits in 1977, reached rapidly increasing home values likewise
new peaks in early 1978 [1]. Increases in increased several hundred percent to
house prices were still accelerating in keep pace with the wild escalation in
many areas. Prices were rising at annual housing prices. This resulted in property
rates of 20-25 percent in Denver and owner rebellions across the nation. The
Phoenix and 15-20 percent in Chicago, result was property tax limiting measures
Dallas, and Houston [I]. which drastically reduced the property
During the same period, the late 1970s, tax. Proposition 13 in California in 1978
there was another set of circumstances rolled back assessments to 1975 levels and
reducing further the supply of housing. limited tax assessments to a 2 percent
New environmental concerns, beginning increment per year until the home was
with the National Environmental Policy resold. The movement to cut property
Act (NEPA) of 1969, required environ- taxes soon spread across the country.
mental impact reports for federal proj- Shortly after California's vote, an Asso-
ects. Later, in 1971, the California Envir- ciated Press-NBC poll revealed 75 per-
onmental Quality Act (CEQA) and sim- cent of the American public felt property
ilar acts in Massachusetts and Hawaii taxes were too high and should be low-
required environmental impact assess-ered, even at the expense of government
services. Within six weeks, 30 states had
ments for every major project, including
housing. Coastal zone permits were also similar measures in the form of legislative
needed, adding yet another bureaucratic bills. This wave of property tax cutting
hurdle to the housing process [7]. spread across the country stimulating
Along with the environmental move- new demand for housing [7].
ment came a decreasing interest and abil- This further complicated the housing
ity on the part of local government to picture because in those states where tax
finance public and community services rollbacks occurred, municipalities
and maintain them (e.g., parks and further determined that single family d
schools, flood control channels, roads, opments no longer paid their own way

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
226 ECONOMIC GEOGRAPHY

for services that they required. ANALYTICAL FRAMEWORK:

City govern-
ments no longer had sufficient revenues to HOUSING PRICE DIFFERENTIALS
provide needed services. As a result, cit-
ies further curbed their interest in approv- A useful method of explaining the basis
al of new housing developments. of housing price differentials is the supply-
An important development has oc- demand framework. This approach is
curred as a result of skyrocketing home based on the idea that the actual price at
prices coupled with a rollback in prop- which any item is sold is the outcome of a
erty taxes. There has been a downturn in complex set of factors which determine
intra-urban residential mobility among the behavior of buyers and sellers. De-
homeowners in those states with tax roll- mand factors are those which determine
backs. By moving laterally into compara- the number of people who want to buy,
ble housing, one must buy and sell at the the amount each wants to buy, and the
going market prices. This increases as- maximum price each is willing to pay.
sessed valuations, and thus taxes, two- to Supply factors determine the number of
four-fold. Coupled with this immobility sellers, how much each will offer for sale,
has come a reduction in migration due to and the minimum price necessary to
extremely high costs of housing. Local induce each seller to offer one item for
firms are focusing their recruitment strate- sale. Except in very special cases, neither
gies almost entirely on the local labor set of factors is the sole determinant of the
pool [12]. price at which an item is sold. For exam-
The apartment market in some locales ple, even if there is only one seller, a
has also been affected by property tax monopolist, the ability of the monopolist
rollbacks. With apparent savings accru- to set the price is limited by how much
ing to owners of single family as well as money or income buyers have ("what the
multiple family dwellings came rent con- market will bear').
trol initiatives from tenants. These initia- Key demand factors in the housing
tives lowered rents to 1975 levels and set market are population (both the number
strict limits on rent increase. These new of peop "le and the household size), infla-
limitations reduced the incentive on the tion adjusted or "real" income per person
part of builders and investors to construct or per household, and, in the case of
and purchase new units [15]. Expecta- owner-occupied housing, the expectation
tions were that low rents would cause a home buyers have of the future change in
major negative cash flow problem (mort- home prices. Important supply factors
gages and repairs exceeding rents) and are the availability of items needed in
make the return on the investment unprofit-building new housing units (land, labor,
able. This reduced the supply of apart- and construction equipment and mate-
ments, further reducing vacancy rates. rials) and construction standards or build-
Many cities have vacancy rates below 5 ing codes. Availability of financing may
percent, which limits mobility and filter- be viewed either as a demand factor
ing up or down the price/size apartment (long-term mortgages) or as a supply fac-
ladder. tor (construction loans).
At the same time, inflation cut into As an example of the operation of a
people's real purchasing power, but de- demand factor, consider an increase in
mand continued to outrun supply. As a the number of households in an area due
result homes doubled and tripled in value to in-migration. If vacant housing of the
in a six-year period in Atlanta, Honolulu, desired kind is limited, the price of exist-
Dallas, Phoenix, San Francisco, San ing housing will be bid up. This will
Diego, Los Angeles, and Washington, induce new construction and create new
D.C. [1]. Presently, price increases have demand for land, labor, and raw mate-
sagged due to high interest rates, but the rials. If these items are in abundant sup-
demand is keeping prices high. ply, their prices should remain constant.

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
 HOUSING AFFORDABILITY 227

If, however, supplies are limited, their Qd2 = al + a2P2 + a3Y2 +

prices will be bid up. This increase in a4HH2 + avr (1)
builder cost will lead to even further
increases in housing prices with the a- where,
mount depending on how willing home
buyers are to pay the higher prices. In- Qd2 =quantity of housing de-
creases in income and heightened expec- manded in some time period
tation of future increases in home prices which is arbitrarily called "2";
will have similar effects.
On the supply side, consider a case of P2 = relative price of housing in
vacant land where the zoning is suddenly time 2;
changed from residential to nonresiden-
tial by the local authorities. This reduc- Y2 = household income in time pe-
tion in available supply will tend to raise riod 2;
land prices and, hence, builder costs. The
ability of builders to pass on the higher HH2 = number of households in time
costs depends on the willingness of buy- period 2; and
ers to pay them. The more willing the
buyers are to pay, the higher the demand r mortgage credit cost (interest
pressures remain on land, labor, etc.; and rate)
the more limited these items are relative For the purpose of analysis, we shall
to the demand, the greater the upward assume that price has a negative effect on
pressure on home prices. Similar re- Qd2 (a2 < 0), income has a positive effect
sponses occur for reduced vacancies, (a3 > 0), each additional household de-
higher labor, equipment and materials mands one additional housing unit, other
prices, and tighter building codes. things equal ( a4 = 1), and higher interest
Figure 2 contains a flow chart designed rates have a negative effect (a5 < 0). The
to show the operation of these effects. constant term in equation (1), a,, repre-
Differential rates of change in home pricessents the effects of all other demand fac-
between cities are likely the effect of tors. If a given housing market has a bet-
differential rates of change in the demand ter environment or more job opportu-
and supply factors. As a general rule, we nities, the value of a, would be higher
hypothesize that increases in home prices than in a city with poorer perceived
will be more rapid in areas with: (1) high characteristics.
rates of in-migration; (2) high rates of Equation 2 is a hypothetical market
household income growth; (3) increased supply equation for one housing market.
expectations of rapid future growth in
prices; (4) declining vacancy levels; (5)
increases in restrictions placed upon sup- Q s2 = A + /2 P2 (2)
plies of land and other inputs; (6) increas-
ingly tighter building codes. Lower rates where Qs2 equals market quantity sup-
of price increase should appear with the plied in period 2 and P2 is the same as in
opposite conditions. All these rates of the demand equation. We shall assume /2
change or changes in rates of change may > 0, the higher the price, the more hous-
be easily converted so that one may con- ing supplied. Higher prices mean greater
sider causes of price level differentials. profits which tend to increase activity by
This method is presented below in a for- builders and, over the long run, attract
mal supply-demand framework. more people to the building of new
Equation (1) represents a hypothetical homes. In addition, additions and altera-
housing demand equation for one market. tions to the existing housing stock (such as
extra rooms) becomes more profitable.
The constant term, ,81, reflects the influ-

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
228 ECONOMIC GEOGRAPHY

' .-0 .-e

J. -JO

D CW > m a; XcoiL j

,;:: m O< E. -

0) 0)O .

~~ OLL ~ M

0--E

co

0 .

L1 w~o-
? O -OD ?C

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
 HOUSING AFFORDABILITY 229

ence of all other supply factors. For Hypothesis 1 stated that home prices
example, an area with higher permit fees, should be higher in areas of high in-
longer permit processing times, and migration. SMSA suburban population
higher land costs will have less housing change, new building starts and permits
supplied at any price. These higher costs (average), and employment growth were
mean lower builder profits at a given used as variables to test hypothesis 1. Of
price and, thus, builders would be less the three variables, only suburban popu-
likely to build. lation change had a significant simple
The actual price of housing in any correlation with house price (r=0.33,
market is determined by the interaction n=30). It did not enter the stepwise
of demand and supply factors. To deter- regression due to its collinearity with
mine price using equations (1) and (2), we solar radiation (Table 2). Apparently in-
define equilibrium to be where Qd2 = Qs2,migration is one of a number of factors
and solve for P2. This leads to contributing to increasing home prices,
but it can be linked to the third factor of
environmental amenities.

P2 = _ ) (Y2 +
taken from Table 1. Housing is obviously a heter-
ogeneous commodity with each housing unit
(;)H ) HH2 + (a ) r (3) consisting of a bundle of attributes. However, no
adjustment was made for the fact that individual
cities may have somewhat varying degrees of

If 32> 0 and a2 < 0, then (/32-a2) > 0. average bundles of attributes. While the hedonic
technique is a widely used device to accomplish
Thus, a larger number of households
such a task at the intraurban level, the lack of a
would be associated with higher prices. sample of attributes which would match the
Also, if a3 > 0 and a5 < 0, then [a3/(132- price data of Table 1 made it impossible to apply
ac2) ] > 0 and [a5/(/32-a2) ] < 0. Higher this technique. The data at the national level
income and lower interest rates would be simply do not exist.
b. Mean temperature difference: National Ocean-
associated with higher prices. Any factor
ographic and Atmospheric Administration
not explicitly considered in equation (3), (NOAA) measured difference between summer
such as climate or building code require- and winter daytime highs. "Map of Solar Energy
ments, would lead to higher prices if it in the U.S. and Southern Canada," EN Map
Corporation, Boulder, Colorado, 1980.
increased a, and/or decreased 81.
c. Solar radiation: N.O.A.A. measured mean daily
amount of solar radiation in BTUs/ft.2/da (see
PRELIMINARY STATISTICAL ANALYSIS reference above).
d. Vacancy: U.S. and regional vacancy estimates
Data were gathered for supply and
are based on U.S. Census Bureau monthly sur-
demand factors to test the hypotheses veys of 52,000 housing units. Total vacancies in
concerning the statistical variation in SMSAs are estimated, using idle electric meter
house price levels of the 30 largest cities in data compared to available vacancy data from
the 1960 and 1970 Census of Housing and the
the United States. The analysis is prelimi-
HUD-Census Bureau Annual Housing Surveys.
nary because data were not available for The estimates represent a weighted average of
each factor for each city, nor were they the ratio and the numerical difference between
available for identical time periods. The idle meter counts and the census data. Vacancy
dependent variable, single-family house data of individual SMSAs may be revised from
time to time as new data becomes available.
price, was averaged for the first two
e. Total population: U.S. Bureau of the Census,
months of 1981 for each of the cities. Current Population Reports, Series P-25, No.
Other time periods, or the growth of 810, "Estimates of the Population of Metropoli-
prices over time, may yield different tan Areas," U.S. G.P.O., Washington, D.C., 1979.
f. Median family income: total money income for
results.2
families-1978; U.S. Bureau of the Census,
Current Population Reports, "Money Income of
2 The variables that were used in the analysis are: Families and Persons in the U.S." U.S. G.P.O.,
a. Price of housing: this is the transaction price 1950.

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
230 ECONOMIC GEOGRAPHY

TABLE 2

EXPLAINING HOUSING PRICES WITH MULTIPLE REGRESSION

Stepwise Regression Standard Multiple Regression

Standardized
Added Regression Beta Coeff. Standard
Step Variable Entered Multiple R R2 R2 Coeff. (Beta) Error Sig.

Step I V3 Solar Radiation 0.61 0.37 0.37 0.97 0.61 0.29 .01
Step 2 V6 Per Capita Income 0.73 0.53 0.16 0.09 0.42 0.04 .01

Step 3 V2 Mean Temperature

Diff. 0.77 0.61 .08 -0.70 -0.30 0.25 .05
Step 4 V8 Loan Rates 0.79 0.64 .03 -1.63 -0.17 0.88 .10

Step 5 V5 Total Popolation 0.81 0.66 .02 0.02 0.12 0.01 .20
Constant = 228.47

Hypothesis 2 stated that home prices with house prices, but due to its collinear-
would be higher in areas of high income ity with mean temperature and solar
growth. Per capita income was used to radiation it does not enter the model.
measure this hypothesis. Not surpris- Hypothesis 3 stated that increased expec-
ingly, per capita income showed a signifi- tations of rapid future growth in prices
cant relationship to house price and ac- create higher home prices. Measuring
counted for 16 percent of the total varia- expectations of rapid increases in future
tion in the data (Table 2). As expected, house price is difficult, if at all possible,
the average income in a region affects the and was not attempted directly. Areas
price of homes. The income level of a high in amenities, as measured inversely
region is actually a surrogate for the by mean temperature difference between
number of other cost of living variables, summer and winter and as measured by
including not only housing but also food, solar radiation, were used for two rea-
clothing, transportation, and utilities.3 Vari- sons. First it allowed the opportunity to
able 10, utilities, also correlates highly include purely geographic factors. Sec-

employed persons, domestic servants, unpaid

g. Suburban growth: U.S. Bureau of the Census,
family workers, and members of the armed
Current Population Reports, Series P-25, No.
forces are excluded. These account for an addi-
810, "Estimates of the Population of Metropoli-
ditional 20 percent of the labor force. Employ-
tan Areas," U.S. G.P.O., Washington, D.C., 1979.
ment estimates are based on establishment sur-
h. Loan rate: rates are for fixed-rate 30-year mort-
veys for middle of reporting month. Unemploy-
gages. Dates for renegotiable-rate mortgages
ment estimates are based on household surveys,
(RRM) were 1/2 to 1 percent lower, Advance
hence affect SMSA of residence rather than
Mortgage Corporation, U.S. Housing Markets,
SMSA where last employed; data is from U.S.
March 1981.
Department of Labor, establishment data and
i. Building starts and permits: data for U.S., Re-
state and local employment services.
gions, and all SMSAs from U.S. Department
of Commerce, Census Bureau, Construction
Statistics Division. Data for Los Angeles-Orange 3We, of course, are not concluding that sunshine
County from Real Estate Research Council of and temperature explain house price variation.
Southern California. These two environmental variables were used as
j. Industrial utility rates: utility rates for selected easy-to-measure surrogates of variations in regional
industries, 1980. San Diego Gas and Electric, demand for housing related to perceptions of loca-
Annual Report, 1981. tional amenities. Sunshine and temperature were
k. Union wage rates: union hourly wage rates and also statistically related to building starts and sub-
employer insurance, pension, vacation, and urban growth rates, the latter two of which are also
other fund payments for selected building trades related to housing demand. This "demand-push/
for January 1981, U.S. Department of Labor. price rise" explanation is perhaps the most widely
1. Employment growth: employment data covers accepted perspective, at least among academic
civilian non-farm wage and salary workers. Self- researchers [23].

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
 HOUSING AFFORDABILITY 231

ond, these variables were included be- ing codes and environmental restrictions
cause of the continual media build-up of create increases in home prices. Data for
migration and expected housing price all cities on such codes and restrictions (to
surges in high amenity areas such as the be added as a dummy) variable could not
so-called "sunbelt," especially in those be obtained, and thus the hypothesis
areas of the sunbelt on or near the sea- could not be tested. However, judging
coast. Prices have surged in these areas. from recent studies, the effect is expected
While distance that the sunbelt city lies to be significant.4
from the seacoast is another measure of Reviewing the test results in relation to
this high amenity, the variable mean the model, we have:
temperature variation, to a large extent, Null hypotheses: coefficients all equal
accounts for it. to zero.
The results for hypothesis 3 show that Alternate hypotheses:
solar radiation (a surrogate for regional
demand) did in fact explain 37 percent of hypothesis 1: coefficient of V5 > 0,
the variation in home prices and it en-
tered on step 1 of the regression (Table 2).
I > O0
,82 - a2
Mean temperature difference, another surro-
gate for demand, was also significant and hypothesis 2: coefficient of V6 > 0,
yielded the second highest simple corre-
lation with house price. 3 > 0
Hypothesis 4 stated that low vacancy fl2 - a2
rates would contribute to high home
prices. While the simple correlation was hypothesis 3: coefficient of V3 > 0;
weak but significant (r=0.30, n=30), the V2 > 0, al > 0;
vacancy factor was less important than
expected, perhaps due to its collinearity hypothesis 4: coefficient of V4 < 0,
with solar radiation.
Hypothesis 5 stated that house price is /31 < 0
pushed upward by restricted supplies of
land, capital, and other inputs. To test this
hypothesis 5: coefficient of V8 < 0,
a5
hypothesis, data were gathered on con-
ventional loan rates, industrial utility bill ,82 - a2 < 0.
change, and union wage rates in the build-
ing trades. Data on the availability of All coefficient estimates are significant
developable land was not obtainable, and statistically except for that of coefficient
this part of the hyposthesis could not be V5, total population.
tested. However, four of the five most The agreement between both sides of
expensive cities in the nation-San Fran- Table 2 is striking. Stepwise regression is
cisco, Los Angeles, San Diego, and Hono- useful to show the order of entry of varia-
lulu all have land shortage constraints bles in the equation and adds one new
which probably contribute to their high variable to the problem at each step.
cost of houses.
While loan rates and industrial utility 4While Hloustoun [9] and Miller [13] discuss the
effect of environmental restrictions on home prices,
bills displayed significant simple correla-
Black [4] of the research department at the Urban
tions of r = 0.49 and r = 0.41, respectively, Land Institute in Washington, D.C. has done an
only loan rates was able to explain two empirical study to determine how important they
percent of the total variation in the data. are in the cost of new homes. A group of 30 cities
were each ranked using a Delphi technique. Cities
Even this variable did not have much
ranked high in environmental restrictions were also
explanatory power. Collinearity with high in home prices. There is growing evidence now
other variables was again the problem. to suggest that this relationship is widespread [20;
Hypothesis 6 stated that tighter build- 22; 27].

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
232 ECONOMIC GEOGRAPHY

However, standard multiple regression ous governmental action [5]. During the
shows the interrelationship of independ- Carter administration, a Task Force On
ent variables with the dependent varia- Housing Costs was created to investigate
bles, with each independent variable en- the matter and to make specific policy
tered holding other variables constant. recommendations [26]. In localities
However, in both regressions, the varia- where the price rise was larger than aver-
bles entered in the same order. This age, there were and still are additional
agreement points to the rigid underlying calls for strong state and local action.
structure of the variables in this instance. A major federal policy proposal, which
In sum, geographic factors of amount is now becoming a reality, is for a series of
of sunshine and temperature modera- new mortgage instruments which
tion/proximity to the ocean, surrogates would supplement, if not replace, the
for regional demand shifts, were the most traditional fixed term, fixed interest rate
important variables in this preliminary mortgage. Such instruments include one
analysis for explaining the level of hous- or more of the following: variable interest
ing prices. Solar radiation and mean rate, variable term, graduated payment
temperature difference together ac- schedule, and shared appreciation. They
counted for the largest proportion of the have the dual goals of lowering the initial
total variation in the data. While there is a monthly payment burden of home-
collinearity factor between the two of r - buyers and insuring a continuing supply
0.46, it is less than expected. Per capita of mortgage funds for housing.
income was also important in explaining An entire group of possible federal pol-
housing price spatial variation, and this icies to ease the affordability problem
suggests that prices reflect "what the would provide subsidies and additional
market will bear." Not surprisingly, per tax benefits to the housing market. In-
capita income was significantly related to cluded are tax exemptions for savings
the total population of the SMSA. If a accounts earmarked for down payments
larger sample was used which included on houses and an exemption status for
more medium and small cities, the total pension funds made available for re-
population of cities would likely have had duced rate mortgages [17].
a more important effect on house price. A final federal policy would be an
It appears that loan rates and labor con- effective anti-inflation program. To the
struction costs are not principle determi- extent home demand represents a desire
nants of house price fluctuation due to for a hedge against inflation, lower infla-
their relative spatial uniformity. How- tion would reduce this demand and re-
ever, availability of land and building duce the upward pressure on home
codes and environmental restrictions, prices.
although not measured, are thought to be Policy tools at the local level include
important. These latter factors no doubt land use regulation, building code and
favored the sunbelt, and their presumed permit regulations, and the use of tax
collinearity with solar radiation and mean exempt status of interest on municipal
temperature difference probably helped bonds. Since the cost of land constitutes
the latter factors to account for the largest 20 to 40 percent of the cost of a new
share of data variability. home, reductions in land costs can have a
significant impact on prices [13]. To this
AFFORDABILITY AND IMPLICATIONS FOR end, many local governments have been
PUBLIC POLICY urged to allow increased densities, up-
zoning of certain areas, use of zero-lot
The rapid rise in the purchase price of lines, and reduced set-back requirements
owner-occupied housing in the 1970s was, and street widths. Some cities have been
and still is, perceived by many to be a selling or leasing city-owned land at
serious national problem requiring vigor- below market rates with the proviso that

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
 HOUSING AFFORDABILITY 233

it be used for new housing. Under pres- (renters who want to buy but are and will
sure from the housing industry, the effec- be unable to do so). The issue, however,
tiveness of environmental regulation and is more complex and requires an under-
the permit process is being examined. standing of the causes of the price rise.
Cities have been using their ability to bor- On the one hand, one may argue that
row at below-market interest rates and the price rise is the result of unrestricted
then channeling the money and market forces (increased numbers of
interest rate savings through financial households and higher incomes) and thus
intermediaries to buyers and builders of has some built-in self-correcting tenden-
new homes. This use of mortgage rev- cies. Higher prices will induce more
enue bonds, however, is under attack and homes to be built and more of the existing
it is possible that Congress may remove stock of rentals to be converted into
the income tax exemption on the interest. owner-occupied units. Such induced in-
A proper assessment of these and other creases in supply likely will not cause
policies must begin with the question of prices to return to old levels, but the long-
whether high and increasing housing term rise in price is really the result of
prices are a problem. In the case of limitations of resources for building hous-
owner-occupied housing, higher prices ing. To those who counter that renters
are actually a benefit to the 65 percent of will be hurt as fewer rental units are
U.S. households already living in their available, one may counter that higher
own homes [25]. Higher prices represent rents will induce more construction of
a gain in wealth which can be realized rental units. Further, the real problem is
upon resale or refinancing. For those who not the higher rents or purchase prices
are young and/or of limited incomes, ris- but rather the low level of income which
ing prices represent both a hope and a makes rents and prices unaffordable. In
curse. The hope is that if such individuals such as case, the appropriate public pol-
can buy a home they will be able to get on icy is to formulate an effective income
the bandwagon. The curse is that as supplement program to make housing
prices rise faster then income or the abil- more affordable for those deemed to be
ity to save for a down payment, the likeli- needy.
hood of being able to buy diminishes. On the other hand, one may argue that
One problem with this analysis is that at least part of the rise in housing prices
the housing price spiral observed in the has been the result of public policy at
1970s probably cannot be sustained in- various levels. Income tax advantages for
definitely. After all, if today's $80,000 owner-occupied housing and failure to
home increased in value at an annual curb inflation have served to make owner-
compounded rate of 20 percent, it would occupied housing relatively more attrac-
be worth nearly $500,000 in 10 years. tive for tenants than rental housing, thus
While the economy has shown remark- fueling demand. At the local level, land
able and unforeseen ability to adjust to use regulations, building codes, and en-
higher prices in the 1970s, figures of such vironmental restrictions have raised the
magnitude seem improbable. If the price cost of building and occupying housing
spiral does end, there may be drastic con- and have also pushed up prices. To the
sequences for those who borrowed the extent that costs of these policies, in terms
maximum possible in expectation of of housing prices, are not commensurate
future price rises. with the benefits received, these policies
Another problem with the above view should be considered for elimination.
is that it suggests that the housing price It is a fact of life that neither view offers
problem is basically an income class issue a cost-free solution. Expanded income
with the haves (current homeowners and supplement programs at the federal level
those who will buy in the near future) would put added strain on the federal
benefiting at the expense of the have-nots budget and thus contribute to inflation.

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
234 ECONoMic GEOGRAPHY

Further inflation may fuel a further rise in Regulatory Conditions," Urban Land Institute
housing prices. Reduced income tax ad- unpublished report, 1982.
vantages will create capital losses for a 5. Bourne, Larry S. The Geography of Housing.
majority of the population as they find New York: Halsted Press, 1981.
the resale value of their homes slipping. 6. Bourne, L. S. and R. J. Hitchcock (eds.). Urban
To the extent local land use regulations, Housing Markets: Recent Directions in Re-
building codes, and environmental re- search and Policy. Toronto: University of
strictions confer real benefits on society, Toronto Press, 1978.

these benefits will be lost. Finally, an 7. English, John W. and Gray E. Cardiff. The
effective anti-inflation policy which re- Coming Real Estate Crash. New York: Warner
sult in lower home prices may lead to Books, 1979.
higher unemployment and would cer- 8. Goodman, J. L., Jr. "Hedonic Prices, Price In-
tainly hurt those who bought homes ex- dices and Housing Markets," Journal of Urban
pecting continued inflation. Economics, 5 (1978), pp. 471-84.

Appropriate public policy is that for- 9. Houstoun, Lawrence O., Jr. "Market Trends
mulated after careful analysis of the Reveal Housing Choices for the 1980s," Journal
causes of higher housing prices and deter- of Housing, 21 (1981), pp. 73-9.

minations that the benefits of the new 10. Kaplan, Marshall A. "Housing and Economic
policy (policies) outweighed the costs, Developments-A Quarterly Review," Federal
and that the policy change was seen as Home Loan Bank Board Journal, 19 (1981), pp.
30-3.
equitable by most of the population.
Throughout the process, the possibility 11. Kartman, Arthur E. "Demand for Housing: the
Neglected Supply Side," University of Washing-
of no change in existing policy should
ton Business Review, 30 (1971), pp. 16-25.
always remain a viable alternative. Unfor-
tunately, the affordability of housing, 12. Los Angeles Times, Wed., May 20, 1981, p. C9.

regardless of how multifaceted and diffi- 13. Miller, Jay, "Assessing Residential Land Price
cult to measure, is even more a problem Inflation," Urban Land, 23 (1981), pp. 16-20.
for specific minorities. Especially hard
14. Mckay, D. Housing and Race in Industrial
hit are racial minorities, the elderly on Society. London: Crown Helm, 1977.
fixed incomes, the poor, and residents in
15. Navarro, Peter. "Rent Control and the Land-
the older, inner cities. Some of these peo-
lord's Loophole: Condo Conversions in Cam-
ple are present home owners faced with bridge, Mass," Urban Land, 22 (1980), pp. 6-9.
increased costs of property taxes and
16. Palm, Risa. "Homeownership Cost Trends,"
maintenance. Few of the policies dis-
Environment and Planning, 4 (1977), pp.
cussed above are designed to ease the 795-804.
affordability problem for these groups.
17. Passou, Shirley S. "Public Pension Funds Create
a Secondary Market for Housing Finance,"
Journal of Housing, 21 (1981), pp. 88-92.

LITERATURE CITED 18. Quigley, J. M. "Housing Markets and Housing

Demand," Urban Housing Markets. Edited by
1. Advance Mortgage Corporation. U.S. Housing L. S. Bourne and J. R. Hitchcock. Toronto:
Markets, March, 1981. University of Toronto Press.

2. Berry, Brian J. L. "Ghetto Expansion and Single- 19. Real Estate Research Council of Southern Cali-
Family Housing Prices, Chicago 1968-1972," fornia. Quarterly Report, 10 (1981).
Journal of Urban Economics, 3 (1976), pp.
397-423.
20. Rice Center for Community Design and Re-
search. The Cost of Delay Due to Government
3. Berry, B. J. L. and R. Bednary. "A Hedonic Regulation in the Houston Housing Market.
Model of Prices and Assessments for Single Urban Land Institute Research Report No. 28,
Family Homes: Does the Assessor Follow the Washington, D.C., 1979.
Market or the Market Follow the Assessor?"
21. Rosen, H. S. "Estimating Inner-City Differences
Land Economics, 51 (1975), pp. 21-40.
in the Price of Housing Services," Urban Stud-
4. Black, J. Thomas. "Residential Land Prices and ies, 14 (1978), pp. 351-55.

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms
 HOUSING AFFORDABILITY 235

22. Seidel, S. R. Housing Costs and Government 26. U.S. Department of Housing and Urban Devel-
Regulations. New Brunswick, N.J.: Rutgers opment. Final Report of the Task Force on
University, 1978. Housing Costs. Washington, D.C.: U.S. Govern-
ment Printing Office, 1978.
23. Scheffman, D. "Some Evidence on the Recent
Boom in Land and Housing Prices," Urban 27. Urban Land Institute and Gruen Gruen Asso-
Housing Markets. Edited by L. S. Bourne and J. ciates. Effects of Regulation on Housing Costs:
R. Hitchcock. Toronto: University of Toronto Two Case Studies. U.L.I. Research Report, No.
Press, 1978. 27, Washington, D.C., 1977.

24. U.S. Bureau of the Census. Construction Re- 28. Wall Street Journal, June 2, 1977.
ports, Series C25, "New One-Family Houses
Sold and For Sale," various issues. 29. Wilkinson, R. K. and C. A. Archer. "Measuring
the I)eterminants of Relative House Prices,"
25. U.S. Department of Commerce, and U.S. Depart- Environment and Planning, 5 (1973), pp.
ment of Housing and Urban Development. 357-63.
Current Housing Reports, Series H-150-78, Part
A, "General Housing Characteristics for the
United States and Regions: 1978." Washington,
D.C.: U.S. Government Printing Office, 1980.

This content downloaded from

196.191.53.170 on Fri, 24 Jun 2022 11:27:22 UTC
All use subject to https://about.jstor.org/terms