IMPACT OF VACANCY DECONTROL IN NEW YORK
CITY: THE FIRST ESTIMATES FROM THE 1996 HOUSING AND VACANCY SURVEY
author wishes to acknowledge the financial support of the New York City Rent
Guidelines Board without associating this organization with any of the conclusions
reached and to thank Doug Hillstrom, Michael Schill, and Ben Scafidi for helpful
comments at the formative stages of the study and Greg Loya and Neil Seftor
for excellent research assistance.
this paper was not completed until after the decision concerning the immediate
future of rent regulation was made, preliminary results of the analysis were
disseminated a month before the decision at a conference at New York University's
School of Law sponsored by its Center for Real Estate and Urban Policy and the
New York City Rent Guidelines Board, and they were reported in an article in
the New York Times on May14, 1997.
some units drop out of the housing stock and hence out of the Survey between
survey years and others were added to account for newly built units, the overwhelming
majority of units surveyed in a year were in the preceding Housing and Vacancy
Survey. The preliminary version of the 1996 Survey available for this analysis
did not contain the indentification numbers that would make it possible to match
units in the1996 Survey with units from the 1993 Survey. So it was impossible
to usedata from the two most recent surveys to predict how many households of
each type would move between 1996 and1998.
of the information needed to calculate gross rent is collected or imputed for
each apartment. However, if the calculated gross rent exceeds $2500 per month,
the exact amount is not reported in the publicly available data.
rent discount is the excess of the market rent of the unit over its actual rent.
To estimate the rent discount in 1991, data from the 1991 Housing and Vacancy
Survey were used to estimate the market rent of each rent regulated unit in
the same way as data from the 1996 Survey were used to estimate these magnitudes
forthat year. So the details of the estimation of market rents for 1991 will
not be provided.
determine the appropriate values of the dummy variables, the 1996 rent discount
and gross rent were deflated to account for the change in the Consumer Price
Index for the New York metropolitan area between 1991 and 1996.
otherwise noted, all means and medians in this analysis are weighted. Essentially,
the sample was obtained by dividing the population of housing units into groups
and then selecting different fractions of the units in each group at random.
(This is a simplified description of the procedure for reasons that are not
important to understand.) The weight for an observation is the inverse of its
probability of selection. These weights must be used to obtain unbiased estimators
of population means and medians. For example, if the sample consisted of observations
on two households Bonnie and Clyde from different groups and there are 90 households
in Bonnie's group and 10 in Clyde's group and Bonnie's income is $40,000 and
Clyde's is $20,000, an unbiased estimate of the population mean is $38,000 [=(90
x $40000 + 10 x $20000)/100]. The unweighted mean $30,000 is a biased estimate.
results of this method were reported at a conference on the New York City's
housing market and rent regulations at New York University's School of Law in
retrospect, it would have been better to limit the analysis to rent stabilized
results reported at the conference were medians.
When the estimated market rent of an apartment is below actual rent, market
rent is clearly underestimated.
This is the difference between the estimated coefficients of Moved Into Unit
1996 and Moved Into Unit 1986-1990 from Table 1.
The most refined method incorporates all of the aforementioned adjustments.
Unfortunately, the only systematic evidence on these two matters refers to a
very different rent regulation and housing market. In a paper entitled "Welfare
Analysis of Rent Control with Side Payments: A Natural Experiment in Cairo,
Egypt" that will appear soon in Regional Science and Urban Economics, Stephen
Malpezzi found that the bribes used to obtain possession of controlled apartments
and the extra maintenance undertaken by their occupants eliminated all of the
benefit of rent regulation to tenants as a group, though tenants who moved less
frequently gained and others lost.
Several studies of rent control in New York City in the 1960s found that the
benefit of rent control to occupants of controlled units was only 50 to 70 percent
of the excess of market over controlled rent on this account alone. See Edgar
O. Olsen, "An Econometric Analysis of Rent Control," Journal of Political
Economy 80 (November/December 1972): 1081-1100 and Richard Ault and Richard
Saba, "The Economic Effects of Long-Term Rent Control: The Case of New
York City," Journal of Real Estate Finance and Economics 3 (March 1990):
25-41. Unfortunately, there are no estimates that compare the tenant benefit
with the rent discount for rent stabilized apartments.
It would have been possible to do an analysis of vacancy decontrol implemented
in the spring of 1994 rather than 1996 without making this assumption. The 1996
Survey tells us which households moved into their rent regulated units within
the preceding two years. The excess of the market over the actual rents of the
apartments occupied by these households is their maximum loss from vacancy decontrol
implemented in the spring of 1994. In retrospect, this probably would have been
the better approach.
Owners of rent regulated apartments prefer tenants with reasonably short stays
because they are allowed greater rent increases when vacated units are reoccupied.
We deleted from the sample households who said that they had negative or zero
income because their reported incomes give a distorted view of their standard
of living. This eliminated three percent of the sample.
With additional time, it would have been possible to check some of these conjectures
using the data from the Survey.
The market rent of an apartment in a given location is a reasonable measure
of its overall desirability. The mean of the estimated market rents of regulated
apartments is substantially greater than the mean rent of unregulated units
in each borough except Staten Island.
This section deals with changes in the market rent of apartments with specified
characteristics rather than changes in the average market rent due to changes
in the characteristics of the housing stock. To say that rent levels in the
unregulated sector will ultimately fall as result of vacancy decontrol means
that the rents of units with each combination of characteristics will fall.
See Dirk W. Early and Edgar O. Olsen, "Rent Control and Homelessness,"
Thomas Jefferson Center for Political Economy, Discussion Paper 298, September
1997, Table 3.
The values of all variables for each city are contained in Early and Olsen's
Table A.1, thereby making it possible for the reader to replicate the regression
results and the estimate of the long-run effect of vacancy decontrol on the
level of market rents reported here.