Arnott 1987
Arnott 1987
Arnott 1987
RICHARD ARNOTT
Queen's University
Introduction
This survey will review the current state of the art in neoclassical microeconomic
modelling of the housing sector in developed, mixed economies' as a basis for
both describing its operation and identifying the appropriate role of government.
Housing has a set of intrinsic characteristics which make it significantly
different from any other good. As a result, the operation of the housing market is
significantly different from that of any other market. These points are elaborated
in Section 1. Models of the housing market can conveniently be grouped into two
classes. Those that treat the housing market as being competitive are surveyed in
Sections 2 and 3. The rest attempt to capture some of the imperfections in the
housing market, and are discussed in Section 4.
This survey has several related themes. First, the competitive theory of housing
markets is reasonably sophisticated and well-developed. Second, there is no well-
worked-out, imperfectly competitive or non-competitive theory of the housing
market, only partial models. Third, it is hard to ascertain the adequacy of the
competitive theory in explaining a particular housing market phenomenon or the
effects of a particular housing policy because of the absence of well-articulated
alternative models. Fourth, since the appropriate role of government in the
housing sector depends crucially on the competitiveness of the market, which is
hard to judge, existing theory provides little guidance concerning desirable
government intervention. And fifth, though further development of the competi-
tive theory of housing markets is needed, the highest priority item on the re-
search agenda should be the development of imperfectly competitive and non-
competitive theories of the housing market.
*The author would like to thank Steffen Ziss and Salman Wakil for assistance with the preparation
of the survey. Very helpful comments on a previous draft from the following are gratefully
acknowledged: Ralph Braid, Jan Brueckner, Masahisa Fujita, Lawrence Jones, Yoshitsugu
Kanemoto, Edwin Mills, Edgar Olsen, David Pines, Lawrence Smith, Konrad Stahl, and Mahlon
Straszheim.
'Thus, no attention is given to models of the housing sector for less developed or Eastern Bloc
countries. nor to non-neoclassical models of the housing market.
Handbook of Regional and Urban Economics, Volume II, Edited by E.S. Mills
((1987, Elsevier Science Publishers B.V
960 R. Arnott
2This section and the next draw heavily on Rothenberg (1978) and Stahl (1985) which provide
excellent discussions of the peculiar characteristics of housing and the housing market.
3
A frequently-cited statistic is that the number of housing units constructed annually typically
constitutes only about 2% of the housing stock.
4
Recall that in the Arrow-Debreu model of perfect competition, there are profit-maximizing, price-
taking firms with independent, convex production sets, utility-maximizing, price-taking households
with independent, convex, consumption sets, no transactions costs, symmetric information, and a
complete set of state-contingent markets.
5In a perfectly competitive economy, a typical household would choose to consume housing at a
variety of locations. This, however, is inconsistent with observation. Thus, competitive models of
spatial housing markets adapt the Arrow-Debreu assumptions by requiring that each household
consume housing at only one location.
Ch. 24: Economic Theory- and Housing 961
etc. - are present. 6 Finally, a non-competitiue economy is one in which there are
some agents who are price setters.
Some economists refer to the set of institutions relating to the provision of
housing services as the "housing sector" and separate the sector into the housing
markets and non-market institutions. In this survey, the term "housing market"
will be used in its popular sense as synonymous with "housing sector".
The most natural way to describe a housing unit is to exhaustively list its
characteristics. Modelling the housing market, however, requires a relatively
simple description of housing technology and demand. As a result, housing
economists typically describe a housing unit in terms of one or several indexes of
characteristics. At the extreme of aggregation, a housing unit can be described in
terms of the number of units of housing service it provides as measured by, say,
the rent it commands at a point in time. With an intermediate degree of
aggregation, a housing unit may be described in terms of its location, floor area,
and quality, and the structural density of the building in which it is located.
Viewing housing in this way, one may distinguish four housing production
processes. The first is construction whereby land and capital are combined to
produce the housing unit. The second is maintenance which entails the gradual
application of capital to an existing unit with fixed floor area, location, and
structural density, to slow the rate of quality deterioration. An improvement in
quality is often called upgrading, and a deterioration, downgrading. The third,
rehabilitation, is similar to maintenance except that it involves the sudden
application of capital to produce a discontinuous increase in quality. The fourth
is conversion which involves changing the size of units in a building with location
and structural density fixed. With downward conversion, unit size is reduced,
while with upward conversion, it is enlarged. Conversion almost always occurs
discontinuously.The fact that changes in structural density, unit size, and quality
may occur discontinuously indicates that there are nonconvexities in the
technology of housing production.
Because of the thinness of the housing market, a household is unlikely to find
any housing unit close to ideal and therefore has a strong incentive to modify the
unit to suit its tastes. Relatedly, housing services are produced by combining the
housing unit with personal furniture and accessories. Also, because of transaction
6
The market is imperfect in the sense that it does not conform with the assumptions of the Arrow-
Debreu model of perfect competition. The presence of market imperfections may, but do not
necessarily, imply market inefficiency.
962 R. Arnot
costs, routine maintenance is usually done more cheaply by the household itself.
For all these reasons, household time and money are typically important inputs
into the production of housing services.
Due to lags in construction and to the relatively small effect of annual
construction on the total stock of housing, housing supply responds only
partially to cyclical movements in demand. Rent therefore tends to move pro-
cyclically. The amortized costs of construction, meanwhile, bear little systematic
relation to the business cycle. Thus, the rate of housing construction is sensitive
to the state of the macroeconomy and is in fact one of the most volatile
components of the business cycle.
Because of the length of approval and construction lags, as well as the
durability of housing, builder expectations play a critical role in the timing of
construction. If these expectations are wide of the mark, the resultant excess
demand or supply in the market for housing can cause local housing booms and
busts. 7
When a household rents or purchases a housing unit, it obtains not only the
physical unit but also because of spatial fixity, a neighborhood, a set of public
services, and tax obligations. It also acquires a set of legal rights and obligations,
of which those related to security of tenure are particularly important.
The type of housing a household acquires is sensitive to its income and
demographic characteristics, 8 and depends as well on relative prices, of course.
The location of the unit is strongly influenced by household members' job
locations. Furthermore, housing is such an important item of consumption that
there is likely significant simultaneity between the household's three major sets of
decisions - household composition (formation, dissolution, number of children,
etc.), job choice, and housing consumption.
The flow of services from a good is normally obtained through its purchase.
For the housing market, however, renting is typically as common as owning, and
in many countries co-operative ownership is common. A household must
therefore make a tenure choice decision.9 Since almost all housing market
policies are targetted either on renters or on a particular class of owners, it is
7
Such booms and busts are often ascribed to speculative activity. Since speculation is so poorly
understood, however, it is probably best to look first to market fundamentals for an explanation of a
boom or a bust.
8In housing market theory, the number of households is typically treated as exogenous. Yet the
number and type of households (through children leaving home, divorce and separation, choice in
family size, elderly parents living apart from their children, etc.) is sensitive to economic conditions
generally and is also influenced by housing policy. On this subject, see Smith et al. (1985).
9
The mode of tenure is the form of contract under which a household occupies a housing unit.
Ch. 24: Economic Theory and Housing 963
Many imperfections are evident in the way housing units are exchanged. First,
due to the thinness of both households and housing units in characteristics space,
as well as mobility costs on both sides of the market, both the suppliers and
demanders of housing have some market power.
Second, while in a competitive market prices adjust instantaneously to clear
markets, in the housing market there is a negative relationship between the level
of demand and the vacancy rate. This implies that prices do not adjust
instantaneously and that some other adjustment mechanism must be at work.
Some of the explanations of housing market adjustment which treat vacancy
rates are discussed in Section 4. All are based on imperfections in the housing
market.
Third, the regulations affecting housing unit production, exchange and
consumption (e.g. building codes, zoning, noise by-laws), as well as the variety
and complexity of housing contracts, suggest that both asymmetries in
information and externalities are significant.
Fourth, a peculiar set of capital market institutions have developed in each
country vis-a-vis home purchase. In the Anglo-Saxon world, the principal
borrowing instrument is the mortgage. Several characteristics of the mortgage
instrument - the requirement that the house be used as collateral, and credit
rationing" - provide strong evidence that housing capital markets are strongly
2
Because the lender is unable to identify the riskiness of the borrower, adverse selection problems
[see Diamond and Rothschild (1978)] arise. Since the lender is unable to properly monitor the actions
of the borrower that affect the probability of default, there are also problems of moral hazard [again
see readings in Diamond and Rothschild (1978)].
' 3The term "filtering" is used to refer either to the phenomenon whereby a residential building is
occupied by households of increasingly low relative income as it ages, or to the change in quality of a
residential building as it ages. According to the second definition, downward filtering is equivalent to
downgrading and upward filtering to upgrading. Typically, but not always, most housing units
constructed are of high to medium quality and of large to medium size, and are modified as they age
for use by lower-income households via downward conversion and filtering.
' 4 For reasons of space, there is no discussion of housing policy per se.
15
For a survey of welfare economics up to 1970, see Arrow and Scitovsky (1969). This survey
covers the basic welfare theorems and classical imperfections such as technological externalities. There
is no good survey of welfare economics with non-classical imperfections - asymmetric information,
transactions costs, incomplete markets, etc. However, Arnott and Stiglitz (1986a, b), which treat
welfare economics when moral hazard is present, provide a flavor of how non-classical imperfections
affect the welfare properties economies. The general result is that, in the presence of non-classical
imperfections, competitive equilibrium (if it exists) is essentially never constrained (i.e. subject to the
information technology) efficient, and the potential scope for welfare-improving government in-
tervention is very large. Other pertinent articles are contained in Diamond and Rotschild (1978).
Ch. 24: Economic Theory and Housing 965
Up to the early 1960s, the dominant form of housing market analysis was
Marshallian. The housing market was viewed as the market for housing services,
with equilibrium being determined by a Walrasian interaction of demand and
supply. The analysis was diagrammatic and discursive rather than algebraic. It is
easy now to dismiss this work as primitive. Yet to view the housing market as a
competitive market for housing services was an impressive feat of simplification.
Furthermore, this mode of analysis is still predominant in policy discussion, and,
in the hands of a skilled practitioner with a good understanding of the operation
of the market, can be quite effective and illuminating.
As an example, consider the application of this mode of analysis to the study of
the effects of a rent subsidy of s. Suppose that prior to the rent subsidy, the
housing market is in long-run equilibrium with the price of housing services equal
to the long-run supply price, (minimum long-run average cost). Equilibrium
quantity is qb, and the pre-subsidy demand curve Db. Now the (unanticipated)
subsidy is introduced. The immediate effect is to shift up the demand curve from
Db to D. Because of construction lags and the durability of structures, the
instantaneous supply curve is vertical. In response to the increased demand, there
is an immediate price rise from to p,. Suppliers gradually respond (the supply
curve rotating around (qb, p) from So to S to S) to the price rise, causing the
supply curve to become increasingly elastic. Eventually, the price falls back to
and the quantity of housing services provided by the market increases to q. The
analysis could be made more sophisticated: If moving costs were treated, housing
consumers would respond sluggishly to the price rise and as time proceeds
Ch. 24: Economic Theory and Housing 967
PRODUCER
PRICE
housing demand would become more elastic. Also, the housing market could be
separated into submarkets and the interaction between the submarkets treated.
Furthermore, the demand and supply elasticities over time could be estimated,
and the time path of the equilibrium price and quantity of housing services
forecast.
In Cities and Housing, a modern classic, Muth formalized and extended the
Marshallian analysis. He analyzed only the stationary state. The housing
producer at a particular location chooses the capital-land ratio, k, to maximize
profits. The quantity of housing services provided when k units of capital are
applied to a unit area of land is Q(k), where Q'>O and Q"<O (primes denote
derivatives). Muth terms Q the housing production function. Where the rents on
land at this location, capital and housing services are R, Pk, and p respectively,
the housing producer's problem, in the absence of government, is to maximize
profits per unit area of land, i.e.
d(pkk/R) dR
-=(u-l) Pkk >-->0 as a<l, and, (6c)
dh OKh dp (6d)
df. pdf df
At less accessible locations: (a) the capital-land ratio is lower; (b) land rent is
reduced; (c) the land share in housing is larger/smaller according to whether the
elasticity of substitution between land and capital in housing production is
greater/less than one; and (d) housing consumption is higher/lower according to
whether the compensated elasticity of demand for housing is greater/less than
one.' 7
"TAnother noteworthy result is that, under reasonable conditions, the population density gradient
(population vs accessibility) can be shown to be bowed-in towards the origin. This has provided the
theoretical basis for the estimation of urban density gradients.
Ch. 24: Economic TheorY and Housing 969
A few comments are in order. First, as mentioned previously, the Muth model
described a long-run stationary state. It has never been successfully modified to
treat adjustment dynamics. Relatedly, it does not provide an explicit treatment of
housing durability. Second, much of the empirical work on housing markets has
gone into estimating the parameters of the Muth model. Since however, actual
housing markets are unlikely to be close to a long-run stationary state, the
parameters estimated are likely to be biased. Third, the Muthian production
function is really a composite function, capturing not only the relationship
between construction costs and structural density (floor-area ratio), but also
households' tastes vis-a-vis structural density [see Grieson (1974)].
One can criticize the Muthian model for its excessive simplicity, but it was the
first formal, general equilibrium model of the housing market, and almost all the
subsequent mainstream housing market theory has evolved from it.
During the seventies, there was rapid, sustained, and systematic progress in
modelling competitive housing markets. There were essentially three separate
lines of development that are now being integrated. First, residential location
theory or the "new urban economics" provided a formal and workable general
equilibrium model with differentiated space. Second, non-stationary, spatial,
durable housing market models were developed. And third, Sweeney (1974a,
b) constructed an aspatial, stationary-state, general equilibrium model of
durable housing with endogenous maintenance and quality. The full and
successful integration of these models would solve for the non-stationary,
dynamic general equilibrium of an economy in which there is differentiated space
and durable housing with endogenous quality and maintenance. Serious
empirical estimation and policy implementation of these models is only
beginning.
In the late sixties and early seventies, the new urban economics was founded.
Important early contributions were Mills (1967, 1972), Beckmann (1969), Solow
and Vickrey (1971), Solow (1972), Mirrlees (1972), and Wheaton (1974). All
drew on the seminal work of Alonso (1964) and Muth (1969) which
incorporates material from earlier papers, e.g. Muth (1961).
Consider the following very simple model. There is a fixed number of identical
households, N, each of which derives utility, u, from lot size, T, and a numeraire
good, C. Different households live on annular lots at different distances from the
970 R. Arnott
(8a) states that development should take place when the benefits from postponing
development (the interest saved on construction costs) equal the costs (the rent
foregone). (8b) indicates that the increase in the discounted revenue from adding
a unit of capital should just cover the cost. Land value is bid up to the point
where the present value of profit is zero when the profit-maximizing development
decisions are made.
Combine (8a) and (8b) to give
Q'(k(T))k(T) r-j(T) (8c)
Q(k(T)) r
where g(T) is the average future growth rate of rents at time T and is implicitly
defined by the equation
21
The second-order conditions reduce to g > r(sk)(l -Sk) where g(p/aT)/p and Sk is capital's
share in housing production.
Ch. 24: Economic Theory and Housing 973
sgn
(dx
( t7ap a ap ax
=sgnQ _ __ (I Ob)
adfO"xa
ax aT '
where
p(Tx)
(T X (t,x) e -- T)dt = _-(T, x)' (10c)
is the discounted rent from a unit of housing at location x at time T, and g(T, x) is
defined by (9) with the dependence on x explicit.
To illustrate these results, consider a few scenarios. First, suppose that trans-
port costs and utility change in such a way that rents grow exponentially at a
constant rate everywhere in the city throughout time. It follows from (8c) that
structural density will be the same at all locations. From (10a), meanwhile,
sgn(dT/dx) =sgn(1 -ask). In accordance with the results of almost all empirical
studies [reviewed in McDonald (1981)] it is assumed that a<l, in which case
dT/dx >O. Thus, the city expands outwards at a constant structural density.
Second, suppose that rent grows at the same exponential rate at all locations at
each point in time, but that the rate may vary over time. From (10b),
sgn(dk/dx)=sgn(0/0T), while from (10a), sgn(dT/dx)=sgn(l-ask). The city
expands outwards with structural density depending on the average future
growth rate of rents. Thus, one explanation for the stylized fact that housing
density tends to fall away from the center is that the growth rate of housing rents
tends to fall as a city matures.
Finally, suppose that rents grow at a constant exponential rate over time at
each location in a city, but at different rates at different locations. Then, from
(10b),
sgn(d ) =sgn(0),
it may become profitable to leave the land there vacant for a period and
construct later at a higher density. This possibility provides one explanation for
leap-frog development [Ohls and Pines (1975)].
In the above discussion only the supply side of the market was treated. On the
demand side of all the models in this branch of the literature, households
maximize utility subject to a contemporaneous budget constraint. Putting to-
gether the demand and supply sides in an economy with commuting costs to a
single center generates a non-stationary, monocentric city with durable housing
in which the spatial structure of the city depends in a complex way on consumer
tastes, housing technology, and the time paths of income, the price of capital,
population, the interest rate, and transport costs as a function of distance.
Non-stationary durable housing models have shown that the durability of
housing can significantly affect the spatial structure of a city. Not only do these
models permit a richer set of spatial structures than does the Muth model, but
also the economic determinants of spatial structures are quite different. The set of
papers by Fujita (1976a,b, 1982, 1985) on this class of models merit special
mention for their high quality and systematic development.
The next stage of development was the treatment of demolition and recon-
struction [Brueckner (1980, 1981)]. At any site, there may be a sequence of
buildings. Index the buildings on the site by i, with i=1 denoting the first
building, etc. If D(k) is the cost of demolition per unit area of land at time t for a
building with capital-land ratio k, then the modified landlord-builder's problem is
max
{Ti, kit
E { j,
df
p(t)Q(k)e-rdt Pkkje ri D(ki)e-i +
1}.
i
The first-order conditions are similar to (8a) and (8b). But in deciding on T for
i1l, the builder will take into account the costs of demolishing the current
building on the site and the rent foregone from it. In deciding on ki, he will take
into account the lifetime of the building and future demolition costs. When this
treatment of the builder's problem is embedded in a general equilibrium spatial
model, one obtains a succession of waves of development. Since the waves can
move inwards or be discontinuous because of leap-frogging, however, the spatial
structure that evolves can be very complex.
The life cycle of a house and the phenomenon of filtering (see fn. 12) had been
discussed previously in the literature [e.g., Grigsby (1963)] but Sweeney (1974a,
b) was the first to model the process. He viewed the housing stock as a com-
Ch. 24: Economic Theory and Housing 975
V=max (O,
max(j'' (p -mi)e-rdt+Ve-rxm))). (13)
From (13), the profit-maximizing maintenance level, provided it is positive, is
given by
-er) =0.(14)
22
Sweeneyalso treated the case where C =k,E, k a constant.
23
Since Sweeney's model is non-spatial, housing units are abandoned rather than demolished.
976 R. ArnoHt
The pace of research on competitive models of the housing market in the eighties
Ch. 24: Economic Theory and Housing 977
has been slack. But some progress has been made in integrating the various lines
of development of the seventies.
Arnott, Davidson, and Pines (1983, 1986) (ADP) provided a description of the
supply side of the housing market in stationary state that is similar to Sweeney's,
but they incorporate space, treat quality as continuous, and provide a more
general specification of the maintenance technology. On his site, the landlord-
builder may construct either a single, infinite-lived building, or a sequence of
buildings described by a periodic cycle of construction, downgrading, and
demolition. More specifically, he chooses construction quality (q0), lifespan (z-
possibly infinite), terminal quality (q,), structural density (), and maintenance
over the life of the building (m(t) where t is building age) to maximize the present
value of profits. Where p(q) is housing rent per unit of floor space of quality q, L
is land value, and K(q0 , ) are construction costs, discounted profits are
When the building is infinite-lived, = oo and (1 -e-rt)- =1. And when there is
an infinite sequence of finite-lived buildings, (-e-t) -
1 gives the ratio of
discounted net revenue from the sequence of buildings to those from the first
building." The maintenance technology is characterized by the function
=g(q, m). (16)
This specification of the technology permits upgrading or downgrading, but is
restrictive in assuming that a building's rate of deterioration does not depend on
its age per se. 2 s
Subject to reasonable restrictions on the construction and maintenance techno-
logies, ADP provide a complete characterization of the solution and a rich set of
comparative static results. It is shown that a single, infinite-lived building
provides greater profits than a sequence of buildings when construction costs are
high relative to maintenance, and that the sequence of buildings is more
profitable when construction costs are relatively low. Structural density typically
increases as housing rents and land values increase, but this need not always
occur.
Arnott, Braid, Davidson, and Pines (1986) (ABDP) embedded the above
supply model in a stationary general equilibrium, monocentric city model. On the
demand side, they improved on Sweeney in permitting a household to choose
both the quality and floor area of the housing unit, but treated only a single
-
241 +e-r+e-'+e-3' . =(1
-- e) '
25
A rehabilitation technology may also be introduced, R(q, q; p) which describes the cost per unit
area of floor space in a building of structural density p of rehabilitation from quality q to quality q.
978 R. Arnott
26In all the competitive housing models without externalities constructed to date, competitive
equilibrium exists and is efficient. The natural conjecture is that the same will be true of more complex
and realistic models. Uniqueness of equilibrium has not been addressed.
Practically no work has been done in investigating the existence and properties of competitive
equilibrium in an economy with housing in which there are externalities. [Stahl (1980), however, has
investigated the landlord-builder's problem in a Sweeney-type model in which there are externalities.]
The general literature on externalities prompts the conjecture that externalities may cause either non-
existence or a multiplicity of competitive equilibria.
27Arnott (1985) is in the process of constructing such a simulation model.
28
That is, if one had only aggregate information on the housing stock, one could probably
rationalize it on the assumption of a convex housing production technology.
29
1t is interesting to note that continuous upwards adjustment of structural density is common in
cities in the Middle East.
Ch. 24: Economic Theory and Housing 979
over quality and floor area. This aggregation was chosen in the light of policy
concerns and household tastes in developed Western countries. In other si-
tuations, however, different aggregations might be preferable. For example, in a
less developed country, the availability of basic amenities - heat, running water,
sanitary facilities - might be of more concern than quality. In any event,
construction of housing models which describe housing in terms of a different set
of aggregate characteristics may prove insightful.
Whatever set of aggregate characteristics housing is described in terms of,
empirical implementation will require the construction of aggregate characteristic
indexes from the set of basic housing characteristics. Since aggregation inevitably
entails aggregation loss, the question naturally arises as to whether it might not
be useful to model housing in terms of its basic characteristics. This is the
approach taken in the hedonic price literature. The basic theory was developed
by Rosen (1974), and it has been extensively applied in empirical work on
housing [see Follain and Jimenez (1985) for a review of the literature]. Let
z = (z i, ... z) be a vector of housing characteristics and p(z) be the hedonic price
function - the market price for a given bundle of basic characteristics. On the
demand side, the household is viewed as choosing a bundle of basic characteris-
tics, i.e. maxx z u(x,z; c) subject to y=x+p(z), where x is a composite of
nonhousing goods, y is household income, and is a taste parameter. On the
supply side, firms produce bundles of basic characteristics to maximize profits, i.e.
max, =p(z) -c(z; fB) where c(z; /f) is the cost of producing z given factor prices fl.
The interaction of supply and demand generates the market-clearing hedonic
price function.
Since any practical housing market theory must be expressed in terms of basic
housing characteristics or aggregate housing characteristics, which are indexes
formed from basic housing characteristics, reliable estimation of hedonic price
functions and of the underlying supply and demand parameters is essential for
any sound, quantitative analysis of the housing market. Hedonic theories of the
housing market, which describe housing in terms of only its basic characteristics,
are therefore valuable in providing a sound theoretical basis for empirical
estimation. They are not, however, particularly useful for conceptual analysis.
First, the detail required in the description of housing tends to obscure basic
insights. For example, while in some contexts, it might be of interest to know how
a particular housing policy will affect the equilibrium number of two-car garages
by households with particular financial and demographic characteristics, such
information would usually be distracting. Second, description of the technology
for the production of housing characteristics which treats durability and "joint
clayness" (that certain characteristics, once bundled, can only be unbundled at
very high cost) is very difficult. For this reason, hedonic price theories of housing
are most useful in providing a conceptual basis for the determination of a
temporary equilibrium in which supply is fixed.
980 R. Arnott
To sum up: In recent years, considerable progress has been made in modelling
competitive housing markets. A central issue has been the level of aggregation at
which to model the housing market. Early models examined the market for
housing services, possibly with submarkets differentiated by location, and hous-
ing and tenure types. These models were criticized as being too crude to address
many housing policy concerns. One reaction was to disaggregate as much as
possible and to treat the housing market as a set of markets for housing
characteristics. Doing so provides a useful theoretical basis for empirical work,
but a distractingly detailed description of the market. One of the main trends of
theoretical housing research over the past fifteen years has been to develop
models in between these two extremes of aggregation.
Another major issue has been the treatment of housing durability. Earlier
models ignored durability or treated it crudely, distinguishing between the
market for housing stock and the market for housing services, a flow. Current
models, meanwhile, are explicitly dynamic and treat construction, demolition,
rehabilitation, and maintenance. Relatedly, earlier models, by neglecting dy-
namics and durability, failed to satisfactorily treat the dynamic spatial evolution
of the housing stock. In recent models, not only does history matter, but also the
current spatial form of the housing market depends on expectations concerning
the future. Finally, with the elaboration of the new urban economics, models of
the housing market can be incorporated into general equilibrium spatial models
of the city, thereby permitting formal treatment of the interlinkages between
housing, transportation, pollution, firm location, etc.
We are now close to having an analytical model of competitive housing
markets that is dynamic and non-stationary, in which space is differentiated,
households choose unit size and quality, and landlord-builders, through con-
struction, demolition, rehabilitation, and maintenance, choose the structural
density and quality of their housing over time. There are still several problems to
be solved 30 and further refinements to be made. 3t Nevertheless, it is fair to say
30
1n the corresponding stationary-state model, existence of equilibrium where there is a continuum
of differentiated consumers and a continuum of differentiated locations has yet to be proved. Since,
however, standard proofs can be used to demonstrate existence with an arbitrarily large but finite
number of differentiated consumers and locations, this problem appears to be mathematical rather
than economic. Relatedly, efficiency of competitive equilibrium has yet to be proved. Since, however,
there are no readily identifiable externalities, the conjecture is that competitive equilibrium is efficient.
Uniqueness has not yet been investigated, but in view of the nonconvexity of the production
technology, the possibility of multiple equilibria appears likely.
Extension to a non-stationary environment will prove mathematically difficult, but is unlikely to be
conceptually troublesome.
3'Further refinements to be made include: (i) further disaggregation in the description of the
commodity housing; for example, various theorists have suggested that it may be useful to treat
durability as another aggregate housing characteristic; (ii)allowing the maintenance technology to be
dependent on building age; and (iii) providing households with lifetime rather than single-period
budget constraints.
Ch. 24: Economic Theory and Housing 981
Over the past fifteen years, general theorists have devoted a lot of attention to the
behavior of imperfectly competitive and non-competitive economies (in which the
assumptions of the Arrow-Debroeu model are relaxed). There have been three
principal overlapping lines of development - asymmetric information, search
theory, and strategic behavior. This work is slowly being incorporated into
models of housing markets, but much remains to be done.
The discussion that follows is selective.
Similarities between the vacancy rate in housing markets and the unemployment
rate in labor markets have frequently been noted, but only recently have theories
of unemployment been adapted to treat housing vacancies.
The analogy between the housing vacancy rate and the unemployment rate
cannot be perfect since in the housing market, vacancies are a manifestation of
the excess supply of a commodity, while unemployment entails the excess supply
of an input. Furthermore, in the labor market, job vacancies and unemployment
32
Computational costs impose a practical constraint on the feasible complexity.
982 R. Arnott
occur together, while.in the housing market, there are vacancies but no one
without housing. Yet much of the work that has been done on the microfoun-
dations of unemployment can be modified fairly straightforwardly to treat
vacancy rates in the housing market.
Search theory [see Mortensen (1984) and McCall (1982) for an overview of
the literature] provides the basis for one important strand of unemployment
theory [see Diamond and Yellin (1985) for a recent example]. Just as workers
search for a higher-paying job or one that better matches their tastes and
abilities, so households search for more suitable housing at a lower price. Because
search is costly, landlords face downward-sloping demand curves. In con-
sequence, they will set price, p, above marginal cost, c. Freedom of entry and exit,
meanwhile, will drive profits to zero. Thus, p(l-v)=c, where v is the vacancy
rate. 33
The stylized facts concerning vacancy rates are not well-documented [but
see Rosen and Smith (1983)], yet it appears to be the case that vacancy rates are
lower the larger the housing market and are counter-cyclical. The former stylized
fact is consistent with search-based theories - the smaller the housing market, the
thinner is it likely to be, and consequently the less elastic the demand facing each
landlord. The latter stylized fact is harder to explain. In good times, households
have the disposable income to upgrade their housing and to pay moving costs,
while in bad times, because of moving costs, households are apt to stay put rather
than move to cheaper housing. Thus, there is likely to be considerably more
turnover in the rental housing market in periods of prosperity. For a given
vacancy rate, therefore, there will be more households searching over the same
number of vacant units in good times than in bad, an effect which by itself results
in the landlord facing less elastic demand by searchers in good times and
generates pro-cyclicity of the vacancy rate. There are other effects, which pre-
sumably more than offset this one. Most important is probably the demand
elasticity of sitting tenants. In bad times, sitting tenants, because they can ill
afford to move, are almost captive, while in good times they are mobile. Thus, the
overall demand (sitting plus prospective tenants) facing a landlord may well be
less elastic in troughs than in peaks. It has also been suggested that landlord and
tenant enter into a long-term implicit contract, whereby the less risk-averse
landlord provides his tenants with insurance against cyclical fluctuations in
demand by setting rents below their instantaneous profit-maximizing levels in
upturns and above in downturns3 4 [see Azariadis and Stiglitz (1983) for a review
of the implicit contracts literature which analyzes the analogous phenomenon in
33
Guasch and Marshall (1985) provide a search-based model of vacancy rates.
34
The quantitative significance of this effect in North America is questionable because mobility
rates are so high (the average rental household moves on average every two or three years) but it may
be important in Europe where mobility rates tend to be significantly lower.
Ch. 24: Economic Theory and Housing 983
the context of labor markets]. Finally, it should be noted that rent controls, of
one form or another, are a significant determinant of vacancy rates in a large
proportion of Western housing markets.
35
1tis assumed that the landlord can monitor a tenant's level of care only imperfectly.
3
6Eldor, Pines, and Schwartz (1985) point out that, depending on the covariance of the return from
housing with other assets and labor income, the purchase of housing may reduce portfolio risk.
984 R. Arnott
rate of return given the average level of care of renters [Henderson and
loannides (1983)].
5. Owning offers greater security of tenure.
6. Government policy may encourage homeownership. The deductibility of
mortgage interest and property tax payments, and the non-taxation of
imputed rent and capital gains on housing, in computing income tax
payable, encourages homeownership by the rich, since with progressivity the
value of the deductions rises with income.
Several papers have examined mobility costs in the housing market [Weinberg,
Friedman, and Mayo (1981), Hanushek and Quigley (1979), and Venti and Wise
(1984)]. All focused on the effect that, because of mobility costs, households do
not instantaneously adjust their housing unit choice. This has significant impli-
cations for the process of adjustment of the housing market to exogenous and
policy changes. Mobility costs also influence the search technology, 37 and, as
noted earlier, affect the nature of vacancy rate adjustment.
It has already been noted that imperfections in capital markets affect the
household's tenure choice and housing consumption decisions. They also in-
fluence builder's supply decisions. General theorists have only recently started
analyzing the form of capital markets [Gale and Hellwig (1984)]. As a result,
capital market imperfections have typically been treated in an ad hoc fashion. In
the context of housing, the capital market imperfections facing renters have been
captured by assuming that they are subject to a period-by-period, rather than an
intertemporal, budget constraint. Homeowners or prospective homeowners,
meanwhile, are assumed to maximize utility, subject to being able to borrow only
via a mortgage contract, the form of which is taken as exogenous. 3 s
37
Because of mobility costs, a household does not search continuously through time. nstead, it
makes a decision to initiate search when it is sufficiently dissatisfied with its current housing that it
anticipates the gains from search to exceed mobility costs (or, more precisely, the optimal time to
initiate search occurs when the expected marginal benefit of postponing initiating search (which
includes the expected benefit of postponing moving costs) equals the expected marginal costs].
38
1t has previously been noted that this procedure was followed in Artle and Varaiya (1978) and
Brueckner (1986). There is also a body of literature on the effects of the mortgage tilt (that mortgage
payments are constant in nominal rather than real terms, or more generally are imperfectly adjusted
to changes in the inflation rate) - see Arvan and Brueckner (1985).
Ch. 24: Economic Theory and Housing 985
While there has been little explicit theoretical modelling of imperfectly com-
petitive housing markets, a number of empirically-based housing simulation
models have been developed which incorporate imperfect features of the market
[Anas (1982, 1983) Anas et al. (1984), DeLeeuw and Struyk (1977), Engle et al.
(1972). Forrester (1969), Ingram et al. (1971), Kain et al. (1976), Lowry (1964),
Kain and Apgar (1985), Behrig and Goldrain (1985), and Wegener (1985)]. The
strength of such models is in positive analysis - their ability to forecast. Their
weaknesses stem from the ad hoc nature of many of the assumptions concerning
individual behavior and market adjustment, and their complexity. Because of
their complexity and their lack of strong theoretical foundations, it is often
difficult to understand the simulation results. Also, due to their lack of a common
theoretical base, it is hard to compare the models. Finally, since there are features
of the models which are not solidly-based on individual maximizing behavior,
they are poor tools for welfare analysis. Such models will become more useful as
their theoretical bases are strengthened.
As was argued earlier, the highest priority item on the research agenda in the
economic thery of housing markets is the development of imperfectly competitive
and non-competitive models of the housing market. This is essential for the
analysis of appropriate government intervention in the housing market. It is also
necessary in order to reverse the unfortunate divergence between theoretical and
empirical work that has occurred during the last decade.
References
Alonso, W. (1964) Location and land use. Cambridge, Mass.: Harvard University Press.
Anas, A. (1978) 'Dynamics of urban residential growth', Journal of Urban Economics, 5:66-87.
Anas, A. (1982) Residential location markets and urban transportation: economic theory, econometrics,
and policy analysis with discrete choice models. New York: Academic Press.
Anas, A. (1983) 'The Chicago area transportation land use analysis system', mimeo.
Anas, A., et al., (1984) 'The Swedish housing market: structure, policy issues, and modelling, The
Council on Urban and Regional Planning', The Technological Institute, Northwestern University,
working paper.
Arnott, R. (1980) 'A simple urban growth model with durable housing', Regional Science and Urban
Economics, 10:53-76.
Arnott, R. (1985) 'HOPSIM: a housing policy simulation model', mimeo.
Arnott, R., R. Braid, R. Davidson and D. Pines (1986) 'A general equilibrium spatial model of housing
quality and quantity', mimeo.
Arnott, R., R. Davidson and D. Pines (1983) 'Housing quality, maintenance and rehabilitation',
Review of Economic Studies, 50:467-494.
Arnott, R., R. Davidson and D. Pines (1986) 'The spatial aspects of housing quality, quantity, and
maintenance', Journal of Urban Economics, 19:190-217.
Arnott, R. and F. Lewis (1979) 'The transition of land to urban use', Journal of Political Economy,
87:161-169.
Arnott, R. and J.E. Stiglitz (1986b) 'The welfare economics of moral hazard', Institute for Economic
Research, Queen's University, Discussion Paper 635.
Arnott, R. and J.E. Stiglitz (1986a) 'Moral hazard and optimal commodity taxation', Journal of Public
Economics, 29:1-24.
986 R. Arnott
Arrow, K. and T. Scitovsky (1969) Readings in welfare economics. Homewood, Ill.: Richard D. Irwin.
Arrow, K. and F. Hahn (1971) General competitive analysis. Amsterdam: North-Holland.
Artle, R. and P. Varaiya (1978) 'Life cycle consumption and ownership', Journal of Economic Theory,
18:35-58.
Arvan, L. and J. Brueckner (1985) 'Risk sharing in the adjustable-rate loan market: are existing
contracts efficient?', Office of Real Estate Research, no. 22. Champaign-Urbana: University of
Illinois.
Atkinson, A. and J. Stiglitz (1980) Lectures on public economics. New York: McGraw-Hill.
Azariadis, C. and J. Stiglitz (1983) 'Implicit contracts and fixed price equilibria', Quarterly Journal of
Economics, 98:1-22.
Beckmann, M. (1969) 'On the distribution of urban rent and residential density', Journal of Economic
Theory, 1:60-67.
Behrig, K. and G. Goldrian (1985) 'The Ifo housing market model', in: K. Stahl, ed.. Microeconomic
models of housing markets. Berlin: Springer-Verlag.
Braid, R. (1981) 'The short-run comparative statics of a rental housing market', Journal of Urban
Economics, 10:280-310.
Braid, R. (1984) 'The effects of government housing policies in a vintage filtering model', Journal of
Urban Economics, 16:272-296.
Brueckner, J. (1980) 'A vintage model of urban growth', Journal of Urban Economics, 8:389-402.
Brueckner, J. (1981)'A dynamic model of housing production', Journalof Urban Economics, 10:1-14.
Brueckner, J. (1986) 'The down payment constraint and housing tenure choice: a simplified expo-
sition', Regional Science and Urban Economics, forthcoming.
Debreu, G. (1959) Theory of value: an axiomatic analysis of economic equilibrium. New Haven: Yale
University Press.
DeLeeuw, F. and R. Struyk (1977) 'Analyzing housing policies with the Urban Institute model', in:
G.K. Ingram, ed., Residential location and urban housing markets. Cambridge, Mass.: Ballinger.
Diamond, P. and M. Rothschild (1978) Uncertainty in economics. New York: Academic Press.
Diamond, P. and J. Yellin (1985) 'Pricing and distribution of money holdings in a search economy',
Department of Economics, M.I.T., working paper 370.
Eldor, R., D. Pines and A. Schwartz (1985) 'The determinants of the household demand for hedging
instruments', The Foerder Institute for Economic Research, Tel-Aviv University, working paper no.
1-85.
Engle, R., F. Fisher, J. Harris and J. Rothenberg (1972) 'An econometric simulation of intra-
metropolitan housing location: housing, business, transportation, and local government', American
Economic Review, 62:87-97.
Follain, J. and E. Jimenez (1985) 'Estimating the demand for housing characteristics', Regional Science
and Urban Economics, 15:77-109.
Forrester, J.W. (1969) Urban dynamics. Cambridge, Mass.: M.I.T. Press.
Fujita, M. (1976a) 'Toward a dynamic theory of land use', Papers of the Regional Science Association,
37:133-165.
Fujita, M. (1976b) 'Spatial patterns of urban growth: optimum and market'. Journal of Urban
Economics, 3:209-224.
Fujita, M. (1982) 'Spatial patterns of residential development', Journal of Urban Economics, 12:22-52.
Fujita, M. (1983) 'Urban spatial dynamics: a review', Sistemi Urbani, 3:411-475.
Fujita, M. (1985) 'Existence and uniqueness of equilibrium and optimal land use: boundary rent curve
approach', Regional Science and Urban Economics, 15:295-324.
Fujita, M. and H. Ogawa (1982) 'Multiple equilibria and structural transition of non-monocentric
urban configurations', Regional Science and Urban Economics, 12:161-196.
Gale, D. and M. Hellwig (1984) 'Incentive-compatible debt contracts: the one-period problem',
London School of Economics discussion paper no. 84-97.
Grieson, R.E. (1974) 'The economics of property taxes and land values: the elasticity of supply of
structures', Journal of Urban Economics, 1:367 381.
Grigsby, W. (1963) Housing markets and public policy. Philadelphia, Pa.: University of Pennsylvania
Press.
Guasch, J. and R. Marshall (1985) 'Age of rental units and vacancy characteristics', Regional Science
and Urban Economics, 15:403-419.
Ch. 24: Economic Theory and Housing 987
Hanushek, E. and J. Quigley (1979) 'The dynamics of the housing market: a stock adjustment model
of housing consumption', Journal of Urban Economics, 6:90-111.
Harrison, D. and J.F. Kain (1974) 'Cumulative urban growth and urban density functions', Journal of
Urban Economics, 1:68-98.
Hartwick, J., U. Schweizer and P. Varaiya (1976) 'Comparative statics of a residential economy with
several classes', Journal of Economic Theory, 13:396-413.
Henderson, J.V. (1977) 'Externalities in a spatial context', Journal of Public Economics, 7:89-110.
Henderson, J.V. and Y. Ioannides (1983) 'A model of housing tenure choice', American Economic
Review, 73:98-113.
Hochman, 0. and D. Pines (1982) 'Costs of adjustment and the spatial pattern of a growing, open
city', Econometrica, 50:1371-1392.
Ingram, G., J. Kain and T. Ginn (1971) The Detroit prototype of the NBER simulation model. New
York: National Bureau of Economic Research.
Jones, L. (1985) 'Wealth and housing', Faculty of Commerce, University of British Columbia, mimeo.
Kain, J., W. Apgar and T. Ginn (1976)'Simulation of the market effects of housing allowances', vol. 1:
description of the NBER urban simulation model, mimeo.
Kain, J. and W. Apgar (1985) 'The Harvard urban development simulation model', in: K. Stahl, ed.,
Microeconomic models of housing markets. (Berlin: Springer-Verlag).
Kanemoto, Y. (1980) Theories of urban externalities. Amsterdam: North-Holland.
Lowry, I.S. (1964) 'A model of metropolis', RM-4035-RC, Rand Corporation.
McCall, J.J. (1982) The economics of information and uncertainty. Chicago: The University of Chicago
Press.
McDonald, J. (1981) 'Capital-land distribution in urban housing: a survey of empirical estimates',
Journal of Urban Economics, 9:190-211.
Mills, E.S. (1967) 'An aggregative model of resource allocation in a metropolitan area', American
Economic Review, 57:197-211.
Mills, E.S. (1972) Studies in the structure of the urban economy. Baltimore: Johns Hopkins University
Press.
Mirrlees, J.A. (1972) 'The optimum town', Swedish Journal of Economics, 74:114-135.
Mortensen, D. (1984) 'Job search and labor market analysis', Northwestern University, Center for
Mathematical Studies in Economics and Management Science, mimeo.
Muth, R. (1961) 'Economic change and rural-urban land conversion', Econometrica, 29:1-23.
Muth, R.F. (1969) Cities and housing. Chicago: The University of Chicago Press.
Ohls, J. and D. Pines (1975) 'Discontinuous urban development and economic efficiency', Land
Economics, 51:224-234.
J. Quigley (1979) 'What have we learned about urban housing markets', in: P. Mieszkowski and M.
Straszheim, eds., Current issues in urban economics. Baltimore: Johns Hopkins.
Rosen, K. and L. Smith (1983) 'The price adjustment process for rental housing and the natural
vacancy rate', American Economic Review, 73:779-786.
Rosen, S. (1974) 'Hedonic prices and implicit markets', Journal of Political Economy, 82:35-55.
Rothenberg, J. (1978) 'Urban housing markets and housing policy', in: J. Bernstein and W.G. Mellon,
eds., Selected readings in quantitative urban analysis. Oxford: Pergamon Press.
Smith, L., et al., (1985) 'The demand for housing, household headship rates, and household
formation', Urban Studies, 21:407-414.
Solow, R. (1972) 'Congestion, density and the use of land in transportation', Swedish Journal of
Economics, 74:161-173.
Solow, R. and W. Vickrey (1971) 'Land use in a long, narrow city', Journal of Economic Theory,
3:430-447.
Stahl, K. (1980) 'Externalities and housing unit maintenance', University of Dortmund Working Papers
in Economic Theory and Urban Economics, 800.
Stahl, K. (1985) 'Microeconomic analysis of housing markets', in: K. Stahl, ed., Microeconomics models
of housing market. Berlin: Springer-Verlag.
Sweeney, J. L. (1974a) 'Quality, commodity hierarchies and housing markets', Econometrica, 42:147-
167.
Sweeney, J.L. (1974b) 'A commodity hierarchy model of the rental housing market', Journal of Urban
Economics, 1:288-323.
988 R. Arnott
Sweeney, J.L. (1974c) 'Modelling housing markets: commodity hierarchies in a spatial setting',
Stanford Journal of International Studies, 9:167-197.
Venti, S. and D. Wise (1984) 'Moving and housing expenditure: transportation costs and disequilib-
rium', Journal of Public Economics, 23: 207-243.
Wegener, M. (1985) 'The Dortmund housing market model: a Monte Carlo simulation of a regional
housing market', in: K. Stahl, ed., Microeconomic models of housing markets. Berlin: Springer-Verlag.
Weinberg, D., J. Friedman and S. Mayo (1981) 'Intraurban residential mobility: the role of
transactions costs, market imperfections, and household disequilibrium', Journal of Urban
Economics, 9:332-348.
Weiss, Y. (1978) 'Capital gains, discriminatory taxes, and the choice between renting and owning a
house', Journal of Public Economics, 10:45-55.
Wheaton, W. (1974) 'A comparative static analysis of urban spatial structure', Journal of Economic
Theory, 9:223-237.
Wheaton, W., (1979) 'Monocentric models of urban land use, contributions and criticisms', in: P.
Mieszkowski and M. Straszheim, eds., Current Issues in Urban Economics. Baltimore: Johns
Hopkins.