OCCUPANCY DISCOUNTS IN THE U.S. RENTAL HOUSING MARKET*
| Date | 01 November 1983 |
| Published date | 01 November 1983 |
| DOI | http://doi.org/10.1111/j.1468-0084.1983.mp45004003.x |
| Author | Robert C. Marshall,J. Luis Guasch |
OCCUPANCY DISCOUNTS IN THE U.S. RENTAL
HOUSING MARKET*
Robert C. Marshall and J. Luis Guasch
Recently in the urban literature it has been conjectured that landlords
in the US rental housing market offer rent discounts to tenants who
remain in their units for more than one contract period.1 Specifically,
landlords ask rents of these tenants lower than those they would have
asked if the units were vacant. Furthermore, the discount is claimed to
increase with each additional contract period (typically one year) that
the tenant remains in the unit, hence the terminology 'length of resi-
dency' discount. Typically two explanations are offered for the exist-
ence of discounts. First, since vacancies are costly in both foregone
income and marketing costs (i.e. showing the unit to searchers, etc.)
landlords will try to induce current tenants to remain in their units by
asking discounted rents. Second, landlords use their rent setting power
to screen tenants by asking discounted (inflated) rents of tenants who
reveal themselves to be desirable (undesirable). There may also be
upward pressures on reset contract rents. Guasch and Marshall (1983,
subsequently referred to as GM) have theoretically modelled landlord
price setting behaviour accounting for both the downward and upward
pressures on reset rents.2 GM discuss conditions for the existence of
discounts, and, in addition, the nature of discounts. In particular they
show that not only may length of residency discounts but also 'sit' dis-
counts may be offered by landlords. The sit discount is offered upon
the first contract renegotiation and, unlike the length of. residency
discount, does not increase in future contract renegotiations. In prin-
ciple, either, neither or both of these discounts may be offered by
landlords. The objective of this paper is empirically to analyse residency
discounts and provide unbiased estimates of these discounts using panel
data from the Annual Housing Survey (AHS, discussed in Section II).
* We are extremely grateful to Robert F. Engle for numerous useful suggestions during the
course of this research. Also, David F. Hendry, John Knight, Mark Machina, and Hal White
provided helpful comments on an earlier version of this paper. Of course, all errors are the
authors' responsibility. Robert C. Marshall received support from the Shell Oil Company
Foundation for part of the time during which this research was done.
'See Barnett (1979), Follain and Malpezzi (1980), Goodman and Kawai (1982), and
Noland (1980).
2GM pose their models within a search theoretic framework. Their results indicate that
landlords may charge premiums to undesirable sitting tenants. However, in a world where all
tenants are homogenous in characteristics relevant to the landlord they find that premiums
will generally not be charged.
357
358 BULLETIN
Not only are the magnitudes of possible discounts of interest but so
also are their statistical significance. The hypotheses to be tested are as
follows:
H0: The sit discount equals zero
H1: The sit discount does not equal zero
H0: The length of residency discount equals zero
H1: The length of residency discount does not equal zero.
Note that for both (1) and (2) two-tailed tests will be employed. In
other words, if significant rent premiums are discovered the null will be
rejected.
Several empirical studies claim to have found evidence of significant
length of residency discounts. These studies rely on cross section data
where models like equation (1) are estimated by ordinary least squares
(OLS).
r=Xß1+'yTn1+u1 (1)
where j denotes the ith unit, R. = rent of the ith unit, r = logR1,
X. = vector of unit j specific characteristics, Tn = length of time in
years that the current tenant has occupied the ith unit (e.g. zero if less
than one year, one if more than one year but less than two, etc.), and
it is usually assumed that u1 '-N(0, u I).
The coefficient y of Tn1 is claimed to measure the length of residency
discount which is offered by landlords. Table 1 records some previous
OLS estimates of 'y. We have also estimated with OLS a similar speci-
fication using panel data from the AHS and have obtained a coefficient
on tenant tenure (in years) of comparable magnitude and significance
to that of previous studies:
rj = X - 0.010 Tn, + ÊJt (lA)
(3.26)
TABLE 1
Estimates of the Length of Residency Discount3
* Significant at the 5 per cent level.
Follain and Malpezzi (1980) and Goodman and Kawai (1982) estimated the length of
residency discount for several cities. The estimates presented here are averages of those esti-
mates over non-rent-controlled cities. However, for nearly every city in both studies the
estimated discount was found to be negative and significant at the 5 per cent level.
y
Follain and Malpezzi (1980) _0.010*
Goodman and Kawai (1982) _0.011*
Barnett (1979), Lowery (1982), Noland (1980) _0.038*
To continue reading
Request your trial