The Mission of the Manhattan Institute is
to develop and disseminate new ideas that
foster greater economic choice and
individual responsibility.

Civic Report
No. 36 May 2003


Rent Control and Housing Investment: Evidence from Deregulation in Cambridge, Massachusetts

Endnotes

  1. According to initial findings from the 2002 New York City Housing and Vacancy Survey, there are 3,209,000 housing units in New York City, of which 2,085,000 are rented. 1,065,000 are under rent stabilization, and 60,000 are rent controlled. 686,000 private units are unregulated, and the remaining 274,000 rental units include Public Housing, Mitchell-Lama, In Rem, HUD-regulated, Article 4, Loft Board units. Source: NYC Department of Housing Preservation and Development.
  2. Rent control began in Cambridge in 1971, shortly after the 1970 Rent Control Enabling Act. The initial acts controlled the rents of most units built prior to January 1, 1969, the major exceptions being cooperatives and owner-occupied two or three-family homes. In addition to limiting rent increases, the law also limited the circumstances under which a landlord could remove a tenant from a unit and required a certificate of eviction in such a case. In 1981, citing the removal of 10 percent of controlled rental units in the city between 1970 and 1980 and a vacancy rate below 1 percent, the city passed additional regulation limiting the removal of units from the market.
  3. Boston, Lynn, Somerville, and Brookline also adopted rent control. In 1976, the planned expiration date, the state legislature allowed some jurisdictions to extend rent control under a home-rule petition. Nonetheless, Lynn deregulated in 1974, as did Somerville in 1979. Boston adopted vacancy decontrol in 1974, and Brookline decontrolled many of its units by 1991. Cambridge alone kept the strictest form of rent control.
  4. Maximum rents were set, in general, at 1967 levels. Future adjustments to the maximum rent level were allowed in order to provide owners a “fair net operation income.” Such rent changes could be positive or negative. Increases were allowed for capital improvements to the units (upgrades, as distinguished from standard maintenance) and changes in operating expenses, including taxes. If landlords failed to perform ordinary maintenance and repairs, or if the units became deteriorated, maximum allowable rents could be adjusted downward. Provisions were also made for general adjustments to rent levels for any particular class of rental units. These rules provided the city with substantial power to limit rents, resulting in 1994 rents that were substantially below market.
  5. Only 7 percent of rent control households applied and qualified for these transitional extensions. This small response reflected in large part the increased occupational status and incomes of residents benefiting from rent control in Cambridge (Pollakowski, 1997).
  6. The correlation coefficient calculated is between growth in investment (1995-98 investment divided by 1993-94 investment) and percent of rental housing previously under rent control. For the 10 primary rental neighborhood used for most of the calculations, the correlation coefficient is .15. For all 13 neighborhoods, it is .30. The remainder of this study focuses on 10 of 13 neighborhoods, those which had a significant amount of available rental housing.
  7. When more than one permit was issued for a specific address in a given year, this activity is represented as one “permit” in the data set used for this research. These multiple permits often occur because costs are initially underestimated. In these cases, the costs for the individual permits are summed to obtain the total cost for each building in each year. Since building permit cost is at the core of this work, it is important to create a cost variable that represented total cost per building per year.
  8. Permit data used covers all buildings having four or more units and at least one rental unit. Excluded are public housing and tax-exempt properties, such as dormitories.
  9. These building-level results exploit our data set to the fullest since they are based on address-matching of the assessor’s file, the rent control file, and the building permit file, along with neighborhood-level data. This analysis covers most renovation and repair investment requiring a building permit in buildings with four or more units. A small number of exceedingly large jobs were not included in this analysis. These jobs were largely reconstruction projects that dwarfed typical renovations in terms of cost by several orders of magnitude. The selection of these projects for separate consideration is described in Appendix B.
  10. The sample size is not exactly six times the number of buildings because of a small number of missing observations and because a small number of very large projects were treated separately (Appendix B).
  11. The city Assesor’s Department defines residential properties within the following main categories: single-family, two-family, three-family, four- to eight-unit apartment building, nine-unit or larger apartment building, and condominium. The regression variables divide the largest buildings into two further categories: 9- to 16-unit apartment building and 17-unit apartment building.
  12. The estimated Model 2 is statistically significant (F-test) at the .01 level. The adjusted R2 of .002 looks low, but the nature of what is being explained must be remembered. For a majority of building/year observations, cost per unit is zero. There are also many observations that are low but varying. Finally, having a substantially higher explanatory power would entail being able to explain, for example, why an investment took place in 1998 instead of 1997. Some, but not all, of these issues could be partially addressed with more building-specific data. The estimated standard errors in parentheses in Table A-3 are somewhat higher than we would like. It is for this reason that four different models are estimated and compared.
  13. Ideally, out-of-sample prediction would be used. That is, half of the sample would be used to estimate the model, and the other half used for simulation purposes. Issues of sample size and precision made this option impractical.
  14. 10 of the 13 neighborhoods had a significant amount of available rental housing, and this study focuses on them. Table A-1 provides summary data for all 13 neighborhoods.

 


Center for Civic Innovation.

EMAIL THIS | PRINTER FRIENDLY

CR 36 PDF (195 kb)

WHAT THE PRESS SAID:

Rent Control's Costs
by Henry Olsen, New York Post, 5-21-03
New York's Self-Destruction: By vetoing rent control, Pataki can help save the city.
OpinionJournal, 5-19-03

SUMMARY:
This study, authored by Massachusetts Institute of Technology (MIT) economist Dr. Henry O. Pollakowski, tracks the effects of rent deregulation on construction and repair-related housing investment in Cambridge, MA since rent control ended there in 1994. Dr. Pollakowski finds that rent deregulation in Cambridge led to a 20% increase in construction and repair related investment in formerly rent-controlled buildings. This study suggests that similar deregulation in the New York City housing market should lead to significant new investment in both affluent and modest income neighborhoods, thereby increasing housing quality.

TABLE OF CONTENTS:

Executive Summary

About the Author

Acknowledgements

Introduction and Overview

The City of Cambridge and Rent Control

The Data

Figure 1: Total Investment in Existing Cambridge Rental Buildings with 4 or More Units

The Models

Table 1: Percentage of Post-Deregulation (1995-1998) Investment in Formerly Rent Controlled Buildings Attributed to Deregulation

The Benchmark Model

Simulation of Renovation Investment Due to Rent Decontrol

Model Variation 1

Model Variations 2 and 3

Conclusion

Appendix A: Cambridge Neighborhoods and Housing Stock

Figure A-1: Map of Cambridge Neighborhoods

Table A-1: Distribution of Rental Buildings by Structure Type

Figure A-2: Median Household Income by Neighborhood (1990)

Figure A-3: Number of Controlled Apartment Buildings by Neighborhood

Table A-2: Description of Variables Used in Regressions

Table A-3: Four Regression Models

Appendix B: Very Large Renovation/Reconstruction Projects

Figure B-1: Total Costs by Regulation Status, Very Large Observations

Endnotes

 


Home | About MI | Scholars | Publications | Books | Links | Contact MI
City Journal | CAU | CCI | CEPE | CLP | CMP | CRD | ECNY
Thank you for visiting us.
To receive a General Information Packet, please email support@manhattan-institute.org
and include your name and address in your e-mail message.
Copyright © 2009 Manhattan Institute for Policy Research, Inc. All rights reserved.
52 Vanderbilt Avenue, New York, N.Y. 10017
phone (212) 599-7000 / fax (212) 599-3494