Rent Regulations` Pricing Effect in the
Transcription
Rent Regulations` Pricing Effect in the
Journal of Housing Research • Volume 10, Issue 2 Mae Foundation 1999. All Rights Reserved. 267 q Fannie Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation Dirk W. Early and Jon T. Phelps* Abstract Rent controls, designed to lower the cost of housing for renters, may have the perverse effect of increasing rents for tenants in the unregulated sector. Although new construction is exempt from current rent control laws, a reduction in supply of rental housing will occur if investors are wary of future controls affecting their units. Also, if controls reduce landlord maintenance, total housing supply in a market will fall. Using 1984 to 1996 data from the American Housing Survey, this study examines the role of rent controls in determining the variations in prices of uncontrolled rental housing across metropolitan areas. The results suggest a positive and statistically significant relationship between the introduction of rent control and price in the uncontrolled sector. However, the link between controls and prices declines through time and may completely disappear after 20 to 30 years with no new construction subject to controls. Keywords: Rent control; Housing price indices; Housing markets Introduction Rent controls are ordinances that regulate the rate at which nominal rents for housing units are allowed to rise. New York, the city with the longest continuous history of rent controls in the United States, enacted controls in November 1943 on authority of the U.S. Emergency Price Act of 1942 to protect against excess rent increases brought on by World War II (Gyourko and Linneman 1989). Several other major cities also imposed controls at that time. These first-generation rent controls were very restrictive and typically froze nominal rents. All jurisdictions eliminated controls during the 1950s and 1960s except New York City, which retained first-generation style rent controls on units built before 1947 and occupied since 1971 (New York State Division of Housing and Community Renewal 1997). During the inflationary 1970s, less restrictive second-generation rent controls were enacted across several urban areas, including New York City, to protect tenants from rapid rent increases. Instead of freezing nominal rents, these controls allow for annual increases in rents, exempt new construction from regulation, and permit rent increases in response to improvements by *Dirk W. Early is an Assistant Professor of Economics at Southwestern University, Georgetown, TX. Jon T. Phelps is Controller at CyBerCorp.com, Austin, TX. The authors are grateful for the comments and suggestions from Edgar Olsen and two anonymous referees. Sara Barnes offered incredible research assistance during the preliminary drafts of this study. This research was initially funded through the Mundy Faculty Fellowship Fund at Southwestern University. The generous support of the Brown Foundation made the final verson of this article possible. Any errors are the sole responsibility of the authors. 268 Dirk W. Early and Jon T. Phelps landlords (Arnott 1995). Though rent controls exist in some 200 communities in six states, major metropolitan areas, such as Boston,1 Los Angeles, San Francisco, New York City, Newark, and San Jose, hold the vast majority of rent controlled housing units (U.S. Department of Housing and Urban Development 1991). Rent controls are controversial because of the possible adverse effects they can have on the housing market. Lower maintenance by owners, longer tenancy rates, mismatching of tenant needs with unit characteristics, and higher prices in the uncontrolled sector are some distortions that rent controls purportedly cause. The last effect is the focus of this study. Examining the role rent control plays in determining the price of housing in the uncontrolled sector is crucially important. Studies of the benefits and costs of rent control to tenants and owners of controlled units should take into account any distortions that rent controls cause in the uncontrolled market. However, all empirical studies of the benefits of rent control fail to control for possible changes in the price of rental housing service in the uncontrolled sector (Ault and Saba 1990; Gyourko and Linneman 1989; Linneman 1987; Olsen 1972). If rent controls lead to an increase in rents in the uncontrolled sector, past measures of the benefits are exaggerated and ignore the cost to tenants in uncontrolled housing. This study examines the effects of rent controls on rental housing prices in the uncontrolled sector. Previous research examining the link between rent control and the price of rental housing is described in the next section. This is followed by a discussion of the hedonic model used to construct a price index for uncontrolled rental housing. The results section describes the estimation of a reduced-form equation describing the price of uncontrolled rental housing as a function of—among other things—the existence of rent controls in the urban area. The article concludes with a brief summary of the findings and suggestions for further research. Previous Studies Rent control can influence the price of housing in the uncontrolled sector by altering the supply of and demand for rental housing service in a metropolitan area. Hubert (1993) examines the role of rationing in the controlled sector and its effect on prices in the free sector. Tenants unable to obtain controlled housing must find housing in the free sector. If the free sector has higher than average demands for housing, prices in that sector will rise. Hubert (1993) concludes that prices in the free sector vary according to the method used to ration units of controlled housing. Landlords may also reduce the maintenance of their controlled units if unable to offset fully their costs by raising rents or if the administrative costs of approving rent increases are prohibitively expensive. Kutty (1996) develops a dynamic infinite horizon model to derive the optimal landlord maintenance path under various rent control ordinances. She shows that under most regulations landlord maintenance will fall, although building codes, tenant maintenance, and certain second-generation features can mitigate some of the deleterious impacts of rent control on housing quality. If a reduction in maintenance occurs, the quality 1 Massachusetts eliminated rent controls in 1994. Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation 269 of controlled units falls, and tenants with the strongest tastes for housing may move to the free sector, pushing up rents in the uncontrolled sector. However, as Olsen (1988) points out, tenant maintenance may offset some reduction in landlord maintenance, lessening this effect. Using panel data from New York City from 1978 to 1987, Moon and Stotsky (1993) estimate the conditional probability that a sound unit becomes substandard. They find that the probability of a unit deteriorating in quality increases with the size of the implied subsidy from rent control. In a similar study, Gyourko and Linneman (1990) examine the probability that a unit of housing in New York City is substandard as a function of its rent control status. Their results suggest that, holding all other factors constant, controlling the rent of a unit increases the probability that the unit will become dilapidated. Controls also can influence the supply of housing in the free sector. Although new construction is typically exempt from current rent control regulations, potential investors in rental housing may be wary of future controls affecting their units. This concern may be greater in metropolitan areas with a history of controlling rents. If this is true, investors in rental housing in areas with rent control ordinances will require a risk premium to compensate for this uncertainty, pushing up rents in the uncontrolled sector (Olsen 1972). Furthermore, owners of rental housing may convert their properties to alternative uses if controls are established or if they fear future controls. Hohm (1983) surveyed landlords in San Diego County when a ballot measure was being debated that would have controlled rents in the area. He found that nearly 70 percent of landlords were thinking of lowering their investment in rental housing because of the threat of implementing rent controls. Although higher rent in the uncontrolled sector is a commonly mentioned cost of rent regulations, few empirical studies have examined the role of rent control in determining rent in the unregulated sector. The first empirical estimates of the effect of controls on prices in the uncontrolled sector were from Fallis and Smith (1984), who examined the price of rental housing in the controlled and uncontrolled sectors using a rent-forecasting model. Analyzing data from 1969 to 1978, they find that rent controls in Los Angeles effectively lowered the rate at which rents rose in the controlled sector and increased the rate at which rents rose in the uncontrolled sector. Caudill (1993) examined the role of controls on New York City housing prices in 1968. His estimates suggest that if controls were removed, rents in the uncontrolled sector of New York City would fall by 22 to 25 percent (Caudill 1993). More recently, Early and Olsen (1998) estimate a simultaneous-equations model explaining housing consumption, vacancy rates, and price per unit of housing service, using data across 44 urban areas. They find that controls increase the price of rental housing service in the uncontrolled sector. Malpezzi (1996), using data across 54 large metropolitan areas, finds a positive and statistically significant relationship between the existence of rent control and median contract rent. However, he does not restrict his examination to rents in the uncontrolled sector. A casual examination of the data used in this study shows that the price of uncontrolled rental housing is higher in metropolitan areas with rent controls. In 1996 dollars, the mean rent of a standard unit of housing in metropolitan areas without rent regulations is roughly $450. In areas with rent controls the mean rent of a similar unit in the uncontrolled sector is more than $700. The goal of this article is to determine whether rent controls are responsible for these higher prices. In the next section, we describe the model used to estimate the causes of variations in prices of uncontrolled rental housing. 270 Dirk W. Early and Jon T. Phelps Method of Estimation A Simple Model of the Urban Housing Market The model developed here draws from the methodology presented by Malpezzi, Chun, and Green (1998); Malpezzi (1996); and Ozanne and Thibodeau (1983). Consider an urban housing market made up of six subsectors: the demand for rental housing service in each of the uncontrolled and controlled sectors, the supply for each of those sectors, the demand for owner-occupied housing, and the supply of owner-occupied housing. These subsectors are linked through the rationing of controlled units and the decision of each household either to rent or purchase housing. Because prices are not free to adjust in the controlled sector, no market clearing equilibrium for controlled housing will occur. Instead, the features of that sector are determined by which units are controlled and the mechanism used to allocate the short supply of controlled housing. Of interest is whether rent controls affect the price of housing in the uncontrolled sector. Therefore, a simple model of the market for unregulated housing is developed that, in its reduced form, will allow for an estimate of the role played by rent controls in determining prices in the free sector. Let the following equations represent the demand for uncontrolled rental housing and owner-occupied housing, respectively: QRD 4 D R (PHR ,P X ,Y R ,D R ,N R ,T,eRD) (1) O ,P X ,Y O ,D O ,N O ,T,eOD) QOD 4 D O (PH (2) R where QD R represents the quantity of rental housing service demanded, and PH its price, in the uncontrolled sector; PX, the price of nonhousing goods; YR, the mean income, and DR, the demographic characteristics of renters in that sector; NR, the number of renters, and T, the property tax rate. The equation explaining the demand for owner-occupied housing is similarly represented. Demographic characteristics are included to control for differences in housing tastes across groups. Both demand equations also include an assumed normally distributed mean zero error term, eRD, and eOD, capturing unobserved characteristics and the heterogeneity of preferences across urban areas. Supply of uncontrolled rental and owner-occupied housing depends on the corresponding price of housing (PHR or PHO), a measure of construction costs (C), whether the metropolitan area is adjacent to a large body of water (G) that is an impediment to growth and an urban amenity, interest rates (i), whether rent controls have been established in the metropolitan area (RC), and the property tax rate. The equations describing the supply of rental and owner-occupied housing are: QRS 4 S R (PHR ,C,G,i,RC,T,eRS) (3) and S O 4 S O (PH ,C,G,i,T,eOS). QO (4) Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation 271 As with the demand equations, error terms are included to capture unobserved characteristics and the heterogeneity across metropolitan areas. The uncontrolled rental and owner-occupied markets are linked by the choice between owning and renting, conditional on not entering the controlled sector. The fraction of households in the metropolitan area that rent uncontrolled housing would depend on the income in the metropolitan area (Y), metropolitan area demographics (D), the price of rental and owneroccupied housing, the interest rate, the property tax rate, and the existence of rent control. Since the burden of property taxes passed on to the consumer may be different for renters and owners, the property tax rate is included. The tenure choice relationship is given by: O ,i,T,RC,e N ), N R 3 N 4 N(Y,D,PHR ,PH (5) where NR is the number of uncontrolled renter households and N is the total number of households in the metropolitan area. The Reduced-Form Equation Because the above equations are simultaneously determined, several variables are endogenous. Most obvious are the prices, quantities, and fraction renting. The tenure choice decision means that the sector-specific demographic variables are also endogenous. Deriving the reduced-form solution for the price of rental housing in the uncontrolled sector—the focus of this article—yields: PHR 4 fn(Y,D,P X ,N,i,T,C,RC,e P ). (6) It is expected that larger incomes and higher prices of other goods will lead to increased prices of rental housing in the uncontrolled sector. Cities with larger populations are expected to have higher prices. The interest rate, tax rate, and construction cost index are expected to be directly related to the price of uncontrolled rental housing, since they measure the cost of producing housing. As discussed earlier, the existence of rent control is expected to increase the price of housing in the uncontrolled sector. The metropolitan area demographic variables measure the fraction of households headed by blacks and Hispanics, the fraction between ages 25 and 34, and average household size. Demographic characteristics are included to capture variation in tastes across household types. These characteristics also might alter prices other than through their influence on the demand for housing. The inclusion of race and ethnic composition may pick up the effects of discrimination in housing. Rent increases in unregulated markets are typically less for sitting tenants than when a unit becomes vacant. These tenure discounts will be less prevalent for the most mobile groups in a housing market. Since young people move more frequently, they face higher rents. Hence, a larger fraction of the population between the ages of 25 and 34 is expected to put upward pressure on prices in the uncontrolled sector. Larger households may have to pay a higher price per unit of housing service to compensate landlords for higher rates of depreciation. Furthermore, large households may have stronger tastes for housing. Therefore, a larger average household size is expected to lead to higher prices of rental housing. The dependent variable used in reduced-form equation 6 is the price of uncontrolled rental housing. Existing measures of the price of rental housing across urban areas are not re- 272 Dirk W. Early and Jon T. Phelps stricted to uncontrolled housing and, therefore, could not be used in this study. The next subsection describes the data and the hedonic model we use to estimate a price index for uncontrolled rental housing. The Data and Hedonic Estimation The primary data sources are the metropolitan area files of the American Housing Surveys (AHS) for the years 1984 to 1996 (U.S. Department of Commerce 1985–97). The AHS collected data across 49 areas. Generally, each area is surveyed every four years, with 7 to 13 areas surveyed in any one year. Nearly all of the metropolitan statistical areas (MSAs) were surveyed more than once over the 13 years covered by this study. Starting in 1993, the AHS reduced the number of areas surveyed in their metropolitan area files and offset this by including large samples for a few major MSAs in their national sample. In 1995, the survey information for five MSAs was taken from the national file, including two areas with rent control, New York and Los Angeles (U.S. Department of Commerce 1996). Combining the areas from the MSA and national AHS files yields a total of 141 observations across 49 MSAs. Table 1 lists the MSAs used in this study, the years in which they were surveyed, and their rent control status. All published rental housing price indices use information on units in the controlled and uncontrolled sectors and, therefore, are unsuitable for this study. Estimates of the price of rental housing service in the uncontrolled sector are estimated using data from the AHS. The procedure follows closely the hedonic specification developed by Thibodeau (1992, 1995) in his calculation of price indices—except that the sample used here is restricted to uncontrolled rental housing. The natural log of the rent of a unit is assumed to be a function of the following characteristics: number of bathrooms, bedrooms, and other rooms; number of units in the building; dwelling age; type of heating equipment; the existence of air conditioning; unit and neighborhood quality; race of the household head; persons per room; length of tenure; utilities included in the rent; and month of the interview. Race of the head of the household is included as a proxy for neighborhood characteristics, such as segregation by race, that are not captured by the measures of neighborhood quality included above. Because higher density can lead to higher rates of unit depreciation, it seems reasonable that landlords would charge higher rents to larger households. The number of persons per room is included to capture this effect. Table A.1 in the appendix describes the explanatory variables used in the hedonic regressions. Hedonic regressions, using unsubsidized, unregulated units, are estimated to predict the rent of a unit as a function of these characteristics. Price indices are found by estimating the mean rent of a constant quality unit across metropolitan areas. The estimated rent of a unit of housing with the average characteristics of rental housing units in 1995 is used as the price per unit of uncontrolled rental housing service in this study. These prices are given in the last column of table A.2.2 2 Although not used in this study, consistent with Thibodeau (1995), the price of rental housing service is also found for three distinct qualities of housing: substandard units, standard housing (housing at least three years old and not substandard), and new construction (housing less than three years old and not substandard). These price indices are reported in table A.2 because they may be of interest to others studying market-determined prices of rental housing. Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation 273 Table 1. Metropolitan Areas, Survey Years, and Rent Control Status Metropolitan Statistical Area Anaheim–Santa Ana, CA Atlanta, GA Baltimore, MD Birmingham, AL Boston, MA Buffalo, NY Charlotte–Gastonia–Rock Hill, NC–SC Chicago, IL Cincinnati, OH–KY–IN Cleveland, OH Columbus, OH Dallas, TX Denver, CO Detroit, MI Fort Worth–Arlington, TX Hartford, CT Houston, TX Indianapolis, IN Kansas City, MO–KS Los Angeles–Long Beach, CA Memphis, TN Miami–Fort Lauderdale, FL Milwaukee, WI Minneapolis–St. Paul, MN–WI New Orleans, LA New York–Nassau–Suffolk, NY Newark–Northeastern New Jersey, NJ Newport News–Hampton, VA Norfolk–Virginia Beach–Newport News, VA Oklahoma City, OK Paterson–Clifton–Passaic, NJ Philadelphia, PA Phoenix, AZ Pittsburgh, PA Portland, OR–WA Providence, RI Rochester, NY Sacramento, CA Salt Lake City, UT San Antonio, TX San Bernardino–Riverside–Ontario, CA San Diego, CA San Francisco, CA San Jose, CA Seattle–Everett, WA St. Louis, MO–IL Tacoma, WA Tampa–St. Petersburg, FL Washington, DC–MD–VA Survey Years Rent Control 1986, 1990, 1994 1987, 1991, 1996 1987, 1991 1984, 1988, 1992 1985, 1989, 1993 1984, 1988, 1994 1995 1987, 1991, 1995b 1986, 1990 1984, 1988, 1992, 1996 1987, 1991, 1995 1985, 1989, 1994 1986, 1990, 1995 1985, 1989, 1993, 1995b 1985, 1989, 1994 1987, 1991, 1996 1987, 1991 1984, 1988, 1992, 1996 1986, 1990, 1995 1985, 1989, 1995b 1984, 1988, 1992, 1996 1986, 1990, 1995 1984, 1988, 1994 1985, 1989, 1993 1986, 1990, 1995 1987, 1991, 1995b 1987, 1991 1984, 1988, 1992 1984, 1988, 1992 1984, 1988, 1992, 1996 1987, 1991 1985, 1989, 1995b 1985, 1989, 1994 1986, 1990, 1995 1986, 1990, 1995 1984, 1988, 1992 1986, 1990 1996 1984, 1988, 1992 1986, 1990, 1995 1986, 1990, 1994 1987, 1991, 1994 1985, 1989, 1993 1984, 1988, 1993 1987, 1991, 1996 1987, 1991, 1996 1987, 1991 1985, 1989, 1993 1985, 1989, 1993 no no no no yes yesa no no no no no no no no no no no no no yes no no no no no yes yes no no no yes no no no no no no no no no no no yes yes no no no no yes Source: U.S. Department of Housing and Urban Development, 1991. a Because rent controls in Buffalo only apply to the 1.4 percent of units built before 1947, estimates were also produced considering Buffalo free of controls. The difference between those estimates and the estimates presented in this study were trivial. b From the 1995 national American Housing Survey. 274 Dirk W. Early and Jon T. Phelps Next, we turn to the estimation of the reduced-form regressions to test whether the existence of controls affects the price in the uncontrolled sector. The dependent variable, determined by the estimation of the hedonic model described above, is the price of uncontrolled rental housing. Nearly all of the independent variables were constructed using data from the AHS. Ideally permanent income would be used in a study of the housing decision of a household. Unfortunately, only measures of current income are available. To control for downturns in the economy, which may prevent current income from being a good proxy for permanent income, a dummy variable indicating the data were collected during a recession year, 1990 or 1991, was included in the regression using information across all 13 years (National Bureau of Economic Research 1999). Three additional variables were derived using information from sources other than the AHS: the price of nonhousing goods and services, the construction cost index, and whether the metropolitan area borders a large body of water. The price of other goods is constructed using the American Chamber of Commerce Research Association’s (ACCRA) Cost of Living Index across urban areas (ACCRA 1984 to 1996) and the Bureau of Labor Statistics’ (BLS) Cost of Living Index across time (BLS 1998). Each year ACCRA produces a price index that measures the cost of a bundle of goods in each metropolitan area relative to the average cost of that bundle across all metropolitan areas in that year. A price index for all goods other than housing was derived using the price index for all goods, the weight-assigned housing expenditure in its calculation, and the price index for housing. Because the average price index constructed by ACCRA is 100 every year, they are not comparable across time. Therefore, the consumer price index from the BLS is used to control for changes in average prices across time. The cost of providing housing is the Boeckh building cost index (E. H. Boeckh 1984 to 1996). A map was used to determine whether a large body of water bordered the metropolitan area. Several variables were transformed to their natural logarithms, namely the uncontrolled rental housing price index, the price of nonhousing goods, the number of households, average household size, and the construction cost index. The aggregate measure of income is the mean of the natural logarithm of income. Table 2 reports the means and standard deviations of all variables used in the estimation procedure. Results The reduced-form equation for the price of rental housing in the uncontrolled sector is estimated using data from the AHS across 13 years, 1984 to 1996. Using the entire sample, a random-effects regression on the panel of data was estimated. These results are presented in the second column of table 3. To test whether the effects of rent control have changed through time, regressions were run separately for three periods: 1984 to 1987, 1988 to 1992, and 1993 to 1996. Because nearly all of the MSAs were sampled only once within each period, ordinary least squares methodology was employed instead of relying on panel data estimates. The results of these three regressions are also reported in table 3. The results suggest that the introduction of rent controls increased the price of uncontrolled rental housing. The panel estimates imply that, holding all other factors constant, the existence of rent control increases rents in the uncontrolled sector by more than 13 percent. It appears, however, that this effect has diminished through time. The coefficient on the Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation 275 Table 2. Means, Standard Deviations, and Descriptions of the Variables Variable Rental housing price index Log of rental housing price indexa Mean income Mean of the log of incomeb Mean (Standard Deviation) $402.49 (137.27) 5.94 (0.33) $38,153.14 (7,493.07) Description Price index for rental housing in the uncontrolled sector Natural logarithm of the price of rental housing service in the uncontrolled sector Average annual household income 10.21 (0.21) Mean of the natural logarithm of household income Price of nonhousing goods 85.59 (15.32) Constructed from the ACCRA Cost of Living Index and the BLS Consumer Price Index for All Urban Consumers Log of the price of nonhousing goodsb 4.43 (0.17) Natural logarithm of the price of nonhousing goods Number of households Log of the number of householdsb Black or Hispanicb 911,540.6 (723,598.7) 13.50 (0.65) 19.43 (10.69) Number of households Natural logarithm of the number of households Fraction of households headed by Blacks or Hispanics Youngb 22.65 (3.07) Fraction of households headed by persons ages 25 to 34 Household size 2.65 (0.12) Average number of persons in a household Log of household sizeb 0.97 (0.05) Natural logarithm of the average number of persons in a household Property tax rateb 1.37 (0.67) Average of property tax paid divided by property value Mortgage interestb 9.11 (0.65) Average interest rate on 30-year, fixed-rate mortgages issued in the last two years Rent controlb 0.18 (0.38) Existence of rent control in the metropolitan area Construction cost index 1,500.49 (291.63) Boeckh building cost index for apartments, hotels, and office buildings Log of the construction cost indexb 7.30 (0.19) Natural logarithm of the Boeckh building cost index for apartments, hotels, and office buildings Geographic restrictionb 0.55 (0.50) Dummy variable for the MSA bordering a large lake, bay, or ocean (1 4 yes, 0 4 otherwise) Economic downturnb 0.17 (0.38) Dummy variable for surveyed in recession years, 1991 or 1992 (1 4 yes, 0 4 otherwise) Note: Unless otherwise mentioned, variables were found using American Housing Survey data. a Dependent variable. b Included explanatory variables. 276 Dirk W. Early and Jon T. Phelps Table 3. Reduced-Form Regression Results Coefficient (Standard Error) Variable Intercept Mean of the log of income Log of the price of nonhousing goods Log of the number of households Black or Hispanic Young Log of household size Tax rate Interest rate Geographic boundary/ urban amenity Log of construction cost index Economic downturn Existence of rent controls N adj-R2 Panel Estimation 1984–96 16.397*** (1.126) 0.578*** (0.087) 0.172 (0.131) 0.015 (0.032) 0.003 (0.002) 0.010* (0.005) 0.157 (0.340) 10.013 (0.023) 0.018 (0.024) 0.190*** (0.040) 0.654*** (0.124) 10.022 (0.027) 0.131** (0.053) 141 0.859 X2 4 713.15 OLS Estimation 1984–87 17.177*** (2.327) 0.579** (0.165) 0.372 (0.348) 10.007 (0.047) 10.001 (0.003) 0.006 (0.009) 0.356 (0.556) 10.007 (0.039) 0.064 (0.062) 0.147*** (0.050) 0.618** (0.257) — — 0.148** (0.063) 47 0.758 F 4 14.09 OLS Estimation 1988–92 18.666*** (1.495) 0.797*** (0.185) 0.202 (0.220) 0.010 (0.039) 0.002 (0.002) 10.002 (0.009) 0.274 (0.453) 10.036 (0.029) 0.040 (0.108) 0.194*** (0.041) 0.648*** (0.167) — — 0.109* (0.064) 56 0.854 F 4 30.24 OLS Estimation 1993–96 17.137** (3.471) 0.675*** (0.218) 0.310 (0.457) 0.008 (0.061) 0.006 (0.003) 0.025 (0.015) 10.816 (0.729) 10.001 (0.037) 10.050 (0.121) 0.202*** (0.061) 0.691** (0.251) — — 0.084 (0.087) 38 0.779 F 4 12.83 Note: Dependent variable 4 log of the price of rental housing in the uncontrolled sector. *p < 0.10. **p < 0.05. ***p < 0.01. existence of rent control using data from 1984 to 1987 is statistically significant and suggests that controls increase the price in the uncontrolled rental market by nearly 15 percent. Results using data from 1988 to 1992 and from 1993 to 1996 suggest a positive relationship between rent control and prices in the uncontrolled sector; however, the effect is smaller. Furthermore, the hypothesis that rent controls did not alter the price of rental housing in the 1993 to 1996 sample cannot be rejected. The reduction in the importance of rent control over time fits well with the notion that the supply of uncontrolled housing falls when investors are concerned that future ordinances will control the rents of new construction. It seems reasonable that the probability of new controls being implemented would decrease as the number of years since implementation increases. If investors become less wary of future controls, they will be more willing to supply housing. This may explain the lack of a statistically significant relationship between controls Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation 277 and prices in the free sector in the late 1990s, nearly 30 years since rents on existing units were controlled in most areas. One concern in the interpretation of these results is the relationship between the price of rental housing and a city’s decision to implement rent controls. If areas with high growth rates of housing costs are more likely to introduce controls, the existence of rent control variables is correlated with the error terms in the above equations. The possible endogeneity of rent control casts some doubt on the reliability of the above results that suggest rent controls increase the price in the uncontrolled sector. However, if the high general inflation of the 1970s—experienced by all the urban areas examined in this study—combined with a stronger propensity of some areas to regulate prices led to controls, the concern over rent control being endogenous is not as great. The results suggest that the price of rental housing in the uncontrolled sector is positively and statistically significantly related to mean income, the presence of a geographic boundary, and construction costs. Surprisingly, the coefficients measuring the effects of other measures of the cost of supplying housing, namely tax and interest rates, were insignificant under all specifications. Demographic characteristics did not show a consistent relationship across model specifications and were rarely significant. To measure the magnitude of price change in uncontrolled rental housing, the increase in rent for a standard unit of housing attributable to the existence of rent control is estimated for each MSA with a controlled sector. The results of the panel regression allow for an estimate of the price of rental housing in the uncontrolled sector if rent controls are removed. Comparing this with the predicted price of controlled rental housing gives a dollar estimate of the magnitude of the effect of rent controls on rents in the uncontrolled sector. The estimated rents in the uncontrolled sector, with and without rent control, and the monthly increases in rents due to controls are presented in table 4. All pecuniary values are in 1996 dollars. The results suggest that, on average, the monthly rent of a typical uncontrolled unit is roughly $85 higher because of the existence of rent controls. Since the price of rental housing is not expected to adjust immediately to the removal of rent controls, these estimates should be considered the long-term pricing change in rental housing if controls are lifted. If prospective investors in rental housing were hesitant to invest in areas where rent controls existed, they may be equally hesitant to invest shortly after their removal, anticipating the possibility of future ordinances. It may take many years for investors to believe that future controls are unlikely and stop demanding a risk premium to invest in certain areas. Conclusion This study uses data from the American Housing Survey from 1984 to 1996 to estimate the effects of rent control on the price of housing in the uncontrolled sector. The results suggest that the introduction of rent controls increases the price per unit of rental housing service in the uncontrolled sector. However, this effect is more prominent shortly after the implementation of controls and may completely disappear once sufficient time has passed without new construction being subject to controls. An analysis of housing markets, roughly a decade after they were established, found a positive and statistically significant relationship be- 278 Dirk W. Early and Jon T. Phelps Table 4. The Effect of Rent Controls on Rental Housing Prices in the Uncontrolled Sector Survey Year Price with Rent Controls Price without Rent Controls Buffalo, NY 1984 454.98 399.16 55.81 San Jose, CA 1984 831.61 729.60 102.01 Boston, MA 1985 622.73 546.34 76.39 Metropolitan Area Monthly Rent Increase Los Angeles–Long Beach, CA 1985 687.40 603.07 84.32 San Francisco, CA 1985 738.10 647.56 90.54 Washington, DC–MD–VA 1985 703.11 616.86 86.25 New York–Nassau–Suffolk, NY 1987 711.13 623.90 87.23 Newark–Northeastern New Jersey, NJ 1987 668.11 586.15 81.96 Paterson–Clifton–Passaic, NJ 1987 698.88 613.15 85.73 Buffalo, NY 1988 493.74 433.18 60.57 San Jose, CA 1988 807.13 708.12 99.01 Boston, MA 1989 696.29 610.87 85.41 Los Angeles–Long Beach, CA 1989 784.41 688.19 96.22 San Francisco, CA 1989 795.30 697.74 97.56 Washington, DC–MD–VA 1989 739.42 648.71 90.70 New York–Nassau–Suffolk, NY 1991 791.32 694.25 97.07 Newark–Northeastern New Jersey, NJ 1991 690.65 605.93 84.72 Paterson–Clifton–Passaic, NJ 1991 716.81 628.88 87.93 Boston, MA 1993* 705.41 618.88 86.53 San Francisco, CA 1993* 759.43 666.27 93.16 San Jose, CA 1993* 790.72 693.73 97.00 Washington, DC–MD–VA 1993* 714.26 626.64 87.62 Buffalo, NY 1994* 537.69 471.73 65.96 Los Angeles–Long Beach, CA 1995* 705.28 618.76 86.52 New York–Nassau–Suffolk, NY 1995* 741.03 650.13 90.90 Note: All dollar values are in 1996 dollars. *The hypothesis that the existence of rent control does not alter the price in the uncontrolled sector cannot be rejected for these years. tween prices in the uncontrolled sector and the existence of controls. Furthermore, the coefficient estimates suggest a substantial increase of nearly 15 percent in prices. However, it is not possible to reject the hypothesis that controls have no effect on prices in the uncontrolled sector, once 20 to 30 years have passed without regulating the rent of new construction. These results suggest that the introduction of new controls would increase the price of uncontrolled housing. However, policy makers concerned with the second-generation controls Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation 279 that are currently in existence can look to these results as an indication that the detrimental effects on the price of uncontrolled housing have passed. According to our findings, the elimination of current controls would not be expected to alter the price of uncontrolled housing. One concern with the above analysis is the possible endogeneity of rent controls in a model explaining variations in prices of uncontrolled rental housing. It could be argued that high growth rates of housing costs led to cities implementing rent controls. If this is true, the coefficients on the existence of controls reported above must be interpreted with care. Future research needs to investigate the direction of causality between higher prices in the uncontrolled sector and the existence of rent controls. A second concern with the finding that rent controls alter rental prices in the uncontrolled sector is the lack of a comparative analysis of the characteristics of controlled and uncontrolled units. If controlled units are substantially different from uncontrolled units, they may represent different submarkets, with different demands for rental housing, reducing the link between the two subsectors. Future research is needed to explore this possibility. Two extensions of this research should be considered. First, the method used to ration controlled housing was not modeled in this study. A careful analysis of the mechanisms employed by landlords to allocate regulated units would increase our understanding of the role played by rent controls in determining prices in the uncontrolled sector. Analysis of the rationing of units would also yield important measures of the distribution of the benefits of rent control. Second, an examination of the specifics of each rent control regulation would allow for a more detailed analysis to determine which restrictions led to the largest price changes in the uncontrolled sector. This type of analysis would help determine which specific restrictions of second-generation controls are less detrimental than the restrictions found in first-generation, or hard, controls. 280 Dirk W. Early and Jon T. Phelps Appendix Table A.1. Explanatory Variables Used in the Hedonic Regression One and one-half bathrooms Two or more full baths No bedrooms One bedroom Three bedrooms 0.25 times the number of bedrooms, if number of bedrooms is greater than 4 One other room Three other rooms 0.25 times the number of other rooms, if number of other rooms is greater than 3 Single-family detached Single-family attached or duplex 3- or 4-unit multifamily 10- to 19-unit multifamily 20-unit or larger multifamily Age of the building Age of the building, squared Age of the building, cubed Built before 1940: 1 4 yes, 0 4 no Central electric heat Built-in electric units Central oil heat Other heating system not specified above No air conditioning At least one room air conditioner but no central air conditioning Building problems: 1 4 the unit has two or more of the following problems: basement leaks, roof leaks, open cracks or holes in walls or ceilings, holes in floor, or broken plaster or peeling paint over an area exceeding one square foot; 0 4 otherwise Public hallway problems: 1 4 unit is in a multifamily building and has at least two of the following problems: absence of light fixtures in public halls, hazardous steps on common stairs, or stair railings not firmly attached; 0 4 otherwise Lack important features: 1 4 unit has any of the following deficiencies: lacks complete plumbing; lacks complete kitchen facilities; sewer system is a chemical toilet, privy, outhouse, facilities in another structure, or some sewage/toilet facilities; wiring in house not concealed; or some rooms lack working electrical outlets; 0 4 otherwise Multiple equipment breakdowns: 1 4 unit has any of the following equipment breakdowns: two or more water breakdowns lasting six hours or more, two or more flush toilet breakdowns lasting six hours or more, two or more public sewer breakdowns lasting six hours or more, or fuses or circuit breakers blew two or more times within the last 90 days; 0 4 otherwise Tenant rates neighborhood quality as excellent: 1 4 yes, 0 4 otherwise Tenant rates neighborhood quality as good: 1 4 yes, 0 4 otherwise Tenant observed signs of rats in last 90 days: 1 4 yes, 0 4 otherwise Census enumerator observed abandoned buildings on the street: 1 4 yes, 0 4 otherwise Tenants are disturbed by trash, litter, or junk in the streets, on empty lots, or on properties in the neighborhood: 1 4 yes, 0 4 otherwise Crime is a problem: 1 4 yes, 0 4 otherwise Noise is a problem: 1 4 yes, 0 4 otherwise Race of the head of the household: 1 4 African American, 0 4 otherwise Ethnicity of the head of the household: 1 4 Hispanic, 0 4 otherwise Number of persons per room Length of tenure in years Length of tenure, squared Electric heat is included in rent: 1 4 payment for electric heat is included, 0 4 otherwise Electricity is included in rent: 1 4 yes, 0 4 otherwise Gas heat is included in rent: 1 4 yes, 0 4 otherwise Oil heat is included in rent: 1 4 yes, 0 4 otherwise Other utilities included in rent: 1 4 yes, 0 4 otherwise Surveyed in August: 1 4 yes, 0 4 otherwise Surveyed in September: 1 4 yes, 0 4 otherwise Surveyed in November: 1 4 yes, 0 4 otherwise Surveyed in December: 1 4 yes, 0 4 otherwise Note: The definitions of the variables used in this study follow Thibodeau’s methodology (1995, table 1, 444–48). Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation 281 Table A.2. Price of Rental Housing Service in the Uncontrolled Sector Metropolitan Statistical Area Survey Substandard Standard New Metropolitan Year Housinga Housingb Constructionc Aread Birmingham, AL Buffalo, NY Cleveland, OH Indianapolis, IN Memphis, TN Milwaukee, WI Newport News–Hampton, VA Norfolk–Virginia Beach– Newport News, VA Oklahoma City, OK Providence, RI Salt Lake City, UT San Jose, CA 1984 1984 1984 1984 1984 1984 1984 1984 165.99 230.05 236.96 191.14 169.79 295.63 250.64 238.08 197.36 289.04 297.14 246.46 215.32 364.15 276.60 287.05 293.02 378.65 400.84 395.60 296.09 539.25 387.74 383.43 196.10 284.58 292.84 243.38 212.04 360.43 276.54 283.95 1984 1984 1984 1984 207.25 291.71 258.23 435.97 257.69 364.19 301.76 536.58 412.18 444.01 428.83 665.82 255.34 358.09 299.96 528.52 Boston, MA Dallas, TX Detroit, MI Fort Worth, TX Los Angeles–Long Beach, CA Minneapolis–St. Paul, MN–WI Philadelphia, PA Phoenix, AZ San Francisco, CA Tampa–St. Petersburg, FL Washington, DC–MD–VA 1985 1985 1985 1985 1985 1985 1985 1985 1985 1985 1985 454.94 297.51 247.90 239.68 369.22 337.21 285.99 220.75 444.63 173.45 363.07 505.46 342.73 310.22 260.14 446.92 380.92 352.22 279.60 506.32 227.76 417.91 594.15 466.53 452.30 382.53 665.52 483.47 492.16 417.06 634.00 323.41 538.11 502.45 340.78 306.45 260.85 443.24 378.73 348.06 276.18 502.77 223.78 414.89 Anaheim–Santa Ana, CA Cincinnati, OH–KY–IN Denver, CO Kansas City, MO–KS Miami–Fort Lauderdale, FL New Orleans, LA Pittsburgh, PA Portland, OR–WA Rochester, NY San Antonio, TX San Bernardino–Riverside–Ontario, CA 1986 1986 1986 1986 1986 1986 1986 1986 1986 1986 1986 484.59 258.51 296.65 176.94 255.52 255.35 175.34 251.86 264.06 223.07 258.43 520.98 310.42 353.85 234.05 317.37 299.49 233.89 329.69 319.28 270.00 320.75 739.30 392.64 507.93 362.01 424.15 390.62 323.48 494.86 405.96 381.78 486.29 522.33 306.69 351.08 230.35 312.96 296.89 229.17 324.47 315.24 267.43 317.53 Atlanta, GA Baltimore, MD Chicago, IL Columbus, OH Hartford, CT Houston, TX New York–Nassau–Suffolk, NJ Newark–Northeastern New Jersey, NJ Paterson–Clifton–Passaic, NJ St. Louis, MO–IL San Diego, CA Seattle–Everett, WA Tacoma, WA 1987 1987 1987 1987 1987 1987 1987 1987 1987 1987 1987 1987 1987 245.76 297.85 290.09 207.31 409.78 201.63 472.57 446.83 439.43 211.69 451.82 375.21 283.20 323.79 362.88 355.34 249.51 452.32 245.66 538.09 491.41 495.90 265.46 499.94 429.86 338.80 465.17 533.24 501.84 382.59 610.11 343.42 735.87 669.84 735.15 414.37 692.98 549.77 430.59 318.01 359.46 351.48 247.79 451.63 243.06 535.76 491.05 495.72 262.79 499.39 426.89 335.11 Birmingham, AL Buffalo, NY 1988 1988 168.23 287.39 233.07 337.32 319.82 456.02 227.48 334.76 282 Dirk W. Early and Jon T. Phelps Table A.2. Price of Rental Housing Service in the Uncontrolled Sector (continued) Metropolitan Statistical Area Survey Substandard Standard New Metropolitan Year Housinga Housingb Constructionc Aread Cleveland, OH Indianapolis, IN Memphis, TN Milwaukee, WI Newport News–Hampton, VA Norfolk–Virginia Beach– Newport News, VA Oklahoma City, OK Providence, RI Salt Lake City, UT San Jose, CA 1988 1988 1988 1988 1988 1988 279.95 213.33 206.96 341.18 304.95 304.58 353.97 273.62 254.05 389.91 347.12 353.38 503.02 407.59 402.62 585.69 465.03 482.68 348.96 269.78 252.03 389.05 345.47 351.10 1988 1988 1988 1988 167.74 392.52 232.21 599.82 204.29 448.78 265.82 677.08 326.92 596.94 374.04 849.32 202.96 446.25 264.72 673.11 Boston, MA Dallas, TX Detroit, MI Fort Worth, TX Los Angeles–Long Beach, CA Minneapolis–St. Paul, MN–WI Philadelphia, PA Phoenix, AZ San Francisco, CA Tampa–St. Petersburg, FL Washington, DC–MD–VA 1989 1989 1989 1989 1989 1989 1989 1989 1989 1989 1989 583.98 275.22 337.85 242.48 557.30 392.46 399.28 270.95 571.47 242.64 420.71 667.73 319.65 405.57 264.97 634.39 427.45 472.20 320.46 665.77 283.02 463.05 676.48 520.04 548.17 418.09 781.76 552.10 655.37 452.90 827.40 412.46 590.01 658.90 319.21 401.47 266.01 629.73 426.85 468.49 318.15 659.60 281.63 461.58 Anaheim–Santa Ana, CA Cincinnati, OH–KY–IN Denver, CO Kansas City, MO–KS Miami–Fort Lauderdale, FL New Orleans, LA Pittsburgh, PA Portland, OR–WA Rochester, NY San Antonio, TX San Bernardino–Riverside–Ontario, CA 1990 1990 1990 1990 1990 1990 1990 1990 1990 1990 1990 648.64 274.08 298.60 197.26 383.56 273.99 232.72 244.54 330.75 233.98 373.53 693.48 317.11 340.77 258.95 448.67 309.11 293.65 302.48 377.95 282.18 429.10 889.72 440.22 511.66 390.84 637.46 403.22 377.58 470.06 433.40 426.01 604.30 693.68 315.32 340.05 254.81 445.90 307.64 288.80 299.67 374.27 280.12 427.15 Atlanta, GA Baltimore, MD Chicago, IL Columbus, OH Hartford, CT Houston, TX New York–Nassau–Suffolk, NJ Newark–Northeastern New Jersey, NJ Paterson–Clifton–Passaic, NJ St. Louis, MO–IL San Diego, CA Seattle–Everett, WA Tacoma, WA 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 1991 271.19 365.89 417.94 250.82 539.35 268.07 610.66 584.96 635.85 251.37 521.21 465.74 421.83 342.55 421.65 500.31 300.66 589.15 317.76 704.58 614.06 730.66 284.24 574.73 541.56 466.83 496.36 590.77 758.18 457.39 723.26 420.77 1004.36 693.85 741.37 427.10 771.23 734.44 723.03 338.05 419.53 497.14 298.68 587.30 314.85 701.37 613.25 721.09 283.98 573.70 537.93 467.94 Birmingham, AL Cleveland, OH Indianapolis, IN Memphis, TN 1992 1992 1992 1992 200.89 293.08 297.54 200.28 248.42 369.31 345.59 253.55 354.11 549.62 524.21 392.57 245.52 364.77 344.38 250.62 Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation 283 Table A.2. Price of Rental Housing Service in the Uncontrolled Sector (continued) Metropolitan Statistical Area Survey Substandard Standard New Metropolitan Year Housinga Housingb Constructionc Aread Newport News–Hampton, VA Norfolk–Virginia Beach– Newport News, VA Oklahoma City, OK Providence, RI Salt Lake City, UT 1992 1992 340.29 300.54 356.22 353.93 522.42 491.43 358.43 351.32 1992 1992 1992 194.53 445.20 300.34 245.65 515.49 339.27 370.44 650.58 487.67 242.65 511.15 338.58 Boston, MA Detroit, MI Minneapolis–St. Paul, MN–WI San Francisco, CA San Jose, CA Tampa–St. Petersburg, FL Washington, DC–MD–VA 1993 1993 1993 1993 1993 1993 1993 637.69 393.97 433.17 550.70 697.87 290.99 550.13 696.34 475.88 492.05 639.73 764.26 323.41 652.33 826.16 698.03 699.65 808.90 958.99 474.49 777.32 693.45 471.85 490.63 634.20 762.05 323.44 644.36 Anaheim–Santa Ana, CA Buffalo, NY Dallas, TX Fort Worth–Arlington, TX Milwaukee, WI Phoenix, AZ San Bernardino–Riverside–Ontario, CA San Diego, CA 1994 1994 1994 1994 1994 1994 1994 1994 647.25 346.01 321.60 275.93 452.58 342.09 410.29 564.03 692.42 399.34 373.03 314.54 530.55 373.83 481.97 622.47 893.94 466.85 593.63 486.96 745.58 573.97 616.79 835.70 692.56 395.30 372.35 314.27 527.01 375.03 477.45 621.42 Charlotte–Gastonia–Rock Hill, NC–SC Columbus, OH Denver, CO Kansas City, MO–KS Miami–Fort Lauderdale, FL New Orleans, LA Pittsburgh, PA Portland, OR–WA San Antonio, TX Chicago, IL Detroit, MI Los Angeles–Long Beach, CA New York–Nassau–Suffolk, NY Philadelphia, PA 1995 1995 1995 1995 1995 1995 1995 1995 1995 1995 1995 1995 1995 1995 274.92 300.43 426.65 252.78 456.80 310.63 287.24 474.72 308.70 439.90 352.49 626.12 743.31 409.88 331.91 368.10 496.01 333.33 533.72 359.10 378.21 554.76 366.65 524.25 444.90 690.23 810.31 511.64 443.91 549.41 832.94 510.50 748.34 531.42 463.46 811.07 571.22 743.12 621.54 850.31 953.25 527.80 328.34 364.66 495.70 328.02 530.31 357.80 369.94 551.91 364.83 520.22 438.53 687.41 807.26 500.87 Atlanta, GA Cleveland, OH Hartford, CT Indianapolis, IN Memphis, TN Oklahoma City, OK Sacramento, CA St. Louis, MO–IL Seattle–Everett, WA 1996 1996 1996 1996 1996 1996 1996 1996 1996 343.66 361.83 536.11 353.13 258.87 235.17 402.19 316.92 546.57 404.23 422.45 588.34 391.05 343.60 281.22 433.25 359.69 637.54 530.28 562.86 732.21 597.72 574.30 427.67 547.30 456.66 895.48 400.68 419.24 586.43 391.63 338.77 279.47 432.88 357.49 633.69 Source: Authors’ calculations using the American Housing Survey. a Substandard units, as defined by the U.S. Department of Housing and Urban Development, include units that are lacking amenities common to most units (e.g., a working toilet, complete kitchen facilities, or heating equipment) or have substantial maintenance problems (e.g., exposed wiring, missing steps in common hallways, holes in floors, or a leaky roof). b Standard housing (housing not less than three years old or substandard). c New construction (housing less than three years old and not substandard). d The dependent variable used in this study, all rental units in the metropolitan area. 284 Dirk W. Early and Jon T. Phelps References American Chamber of Commerce Researchers Association. 1984–1996. Intercity Cost of Living Index. Indianapolis, IN: American Chamber of Commerce Researchers Association. Arnott, Richard. 1995. Time for Revisionism on Rent Control? Journal of Economic Perspectives 9(1):99–120. Ault, Richard, and Richard Saba. 1990. The Economic Effects of Long-Term Rent Control: The Case of New York City. Journal of Real Estate Finance and Economics 3(1):25–41. Bureau of Labor Statistics. 1998. Consumer Price Index—All Urban Consumers, All Items Less Shelter. World Wide Web page <http://146.142.24/cgi-bin/dsrv?cu> (last modified May 26, 1998). Caudill, Steven. 1993. Estimating the Costs of Partial Coverage Rent Controls. Review of Economics and Statistics 75(4):727–31. Early, Dirk W., and Edgar O. Olsen. 1998. Rent Control and Homelessness. Regional Science and Urban Economics 28(6):797–816. E. H. Boeckh. 1984 to 1996. Boeckh Building Cost Index Numbers. Milwaukee, WI: Boeckh. Fallis, George, and Lawrence B. Smith. 1984. Uncontrolled Prices in a Controlled Market: The Case of Rent Controls. American Economic Review 74(1):193–200. Gyourko, Joseph, and Peter Linneman. 1989. Equity and Efficiency Aspects of Rent Control: An Empirical Study of New York City. Journal of Urban Economics 26(1):54–74. Gyourko, Joseph, and Peter Linneman. 1990. Rent Controls and Rental Housing Quality: A Note on the Effects of New York City’s Old Controls. Journal of Urban Economics 27(2):398–409. Hohm, Charles F. 1983. The Reaction of Landlords to Rent Control. American Real Estate and Urban Economics Association Journal 11(4):504–20. Hubert, Franz. 1993. The Impact of Rent Control on Rents in the Free Sector. Urban Studies 30(1): 51–61. Kutty, Nandinee K. 1996. The Impact of Rent Control on Housing Maintenance: A Dynamic Analysis Incorporating European and North American Rent Regulations. Housing Studies 11(1):69–88. Linneman, Peter. 1987. The Effect of Rent Control on the Distribution of Income among New York City Renters. Journal of Urban Economics 22(1):14–34. Malpezzi, Stephen. 1996. Housing Prices, Externalities, and Regulation in U.S. Metropolitan Areas. Journal of Housing Research 7(2):209–41. Malpezzi, Stephen, Gregory Chun, and Richard Green. 1998. New Place-to-Place Housing Price Indices for U.S. Metropolitan Areas, and their Determinants. Real Estate Economics 26(2):235–74. Moon, Choon-Geol, and Janet Stotsky. 1993. The Effects of Rent Control on Housing Quality Change: A Longitudinal Analysis. Journal of Political Economy 101(6):1114–48. National Bureau of Economic Research. 1999. US Business Cycle Expansions and Contractions. World Wide Web page <http://www.nber.org/cycles.html> (accessed July 25). Rent Regulations’ Pricing Effect in the Uncontrolled Sector: An Empirical Investigation 285 New York State Division of Housing and Community Renewal. 1997. Rent Administrations Fact Sheet: Rent Control and Rent Stabilization. World Wide Web page <http//www.dhcr.state.ny.us/ora/pubs/html/ orafac1.htm> (last modified September 30). Olsen, Edgar. 1972. An Econometric Analysis of Rent Control. Journal of Political Economy 80(6):1081–100. Olsen, Edgar. 1988. What Do Economists Know About the Effect of Rent Control on Housing Maintenance? Journal of Real Estate Finance and Economics 1(3): 295–307. Ozanne, Larry, and Thomas Thibodeau. 1983. Explaining Metropolitan Housing Price Differences. Journal of Urban Economics 13(1):51–66. Thibodeau, Thomas G. 1992. Residential Real Estate Prices: 1974–1983, From the Standard Metropolitan Statistical Area Annual Housing Surveys. In Studies in Urban and Resource Economics, ed. Richard B. Clemmer. Mount Pleasant, MI: Blackstone Company. Thibodeau, Thomas G. 1995. House Price Indices from the 1984–1992 MSA American Housing Surveys. Journal of Housing Research 6(3):439–81. U.S. Department of Commerce. 1985–1997. American Housing Survey, 1984–1996: Metropolitan Statistical Area File [Computer file]. Washington, DC: Bureau of the Census [producer]. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor]. U.S. Department of Commerce. 1996. American Housing Survey, 1995: National File [Computer file]. Washington, DC: Bureau of the Census [producer]. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor]. U.S. Department of Housing and Urban Development. 1991. Report to Congress on Rent Control. Washington, DC: Office of Policy Development and Research.