No. 36 May 2003
Rent Control and Housing Investment: Evidence from Deregulation in Cambridge, Massachusetts
Cambridge Neighborhoods and Housing Stock
It is important to examine Cambridge neighborhoods and the housing stock to understand patterns of affluence, mid-size and large rental buildings, and rent control. Figure A-1 shows the physical boundaries of neighborhoods within Cambridge. While the city’s population is diverse in terms of demographics and income levels, its thirteen neighborhoods can be broadly divided into two groups according to level of affluence. The neighborhoods surrounding Harvard Square are generally wealthier, although they vary from having a significant number of single-family homes to being dominated by large apartment buildings. On the other hand, the neighborhoods to the east of Central Square plus an outlying neighborhood are more modest income with a mix of all property types except single-family and a larger percentage of rental housing. Table A-1 presents the distribution of rental structures by rent control status, number of dwelling units, and neighborhood. Figure A-2 presents 1990 median household income for each neighborhood.
The more affluent group is comprised of the northern neighborhoods of Agassiz, Neighborhood 9, and Neighborhood 10, as well as Mid-Cambridge, all with 1990 household income levels above the citywide median of $33,140. Agassiz is a small but dense neighborhood, largely populated by Harvard employees and students as well as professionals, with the highest levels of educational attainment in the city. Over 80 percent of the rental housing in this neighborhood was under rent control, one of the two highest percentages in the city. Neighborhood 9 is one of the city’s largest neighborhoods, with all building types represented and an above-average incidence of rent control. Neighborhood 10 includes the well-known affluent Brattle Street area, which is characterized by substantial houses and sizable lots. Population density is relatively low compared to the other neighborhoods. This is the only neighborhood where owner-occupied units outnumber renter-occupied units. Housing over 13,000 residents, Mid-Cambridge is the one of the most populous and highly-educated neighborhoods in the city. Unlike the above neighborhoods, it consists of large apartment buildings, with more than 80 percent of the rental units formerly under rent control. As shown in Figure A-3, it had the largest number of rent-controlled buildings in the city.
The less affluent group of neighborhoods is comprised of East Cambridge, Wellington-Harrington, Area IV, Cambridgeport, Riverside, and North Cambridge. These neighborhoods are located in the south and east of the city, with the exception of North Cambridge. East Cambridge has traditionally been a blue-collar neighborhood with a sizable immigrant population since the mid-nineteenth century. Here and in nearby Wellington-Harrington multi-family structures are smaller than elsewhere, and the percentage formerly under rent control is below the city average. The median household income of $24,665 in Area IV is the lowest in Cambridge, about 75 percent of the citywide median. Mid-size buildings dominate, with the percentage formerly under rent control about average for the city. Cambridgeport is a diverse neighborhood that is larger than Area IV, with a somewhat higher median income. It contains a large number of duplexes in addition to a considerable number of midsize buildings. While its percentage of formerly rent-controlled duplexes and “triple-deckers” is somewhat larger than average, its percentage of formerly rent-controlled buildings with four or more units is slightly below average. Riverside is a diverse neighborhood, running from Harvard in the west to a more modest income area in the east, and bordering higher density, more affluent Mid-Cambridge to the north. Excluding Harvard dormitories, it is smaller than Cambridgeport. It has the highest percentage of rental housing (90 percent) in the city. Finally, North Cambridge is predominantly professional and middle-class. However, a substantial blue-collar population remains in this former industrial area. The percentage of formerly rent-controlled buildings is less than average.
Throughout the City, renter-occupied units as a percentage of total units remained fairly steady throughout the 1990’s, at about 65 percent, but decreased considerably from nearly 80 percent in the 1970’s and 1980’s. Across the city’s neighborhoods, the percentage of units that are renter-occupied varies from 48 percent in Neighborhood 10 to over 90 percent in Riverside.
At the end of 1994, over 13,400 units in the city of Cambridge were under rent control. This represented slightly less than half of the city’s rental housing stock. More than two-thirds of units in rental buildings with four or more units were under rent control. As with the composition of the housing stock, the intensity of rent control varied across neighborhoods, with the percent of rental dwellings under rent control somewhat higher in the more affluent neighborhoods.
Table A-1 presents the distribution of buildings throughout the city according to category of structure. The city Assessor’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 unit. These categories differ slightly from the US Census, which groups 3- and 4- unit structures together and classifies 5 to 9 unit structures as a single group. Single-family homes make up 37 percent of residential buildings, but only 12 percent of the city’s housing units. Duplexes and especially “triple-deckers” are more prevalent in the lower and eastern neighborhoods, such as Wellington-Harrington and East Cambridge. Larger residential buildings, those of 10 or more units, are most prevalent near the city’s two largest universities, Harvard and MIT. In neighborhoods such as Mid-Cambridge, Riverside, and Agassiz, more than half of all dwelling units are located in apartment buildings.
Very Large Renovation/Reconstruction Projects
As discussed in the presentation of the building-level renovation and repair simulation results, a small number of projects were not included in the simulation model. The typical cost of one of these projects was several orders of magnitude greater than most of the renovations considered in the simulation model. They are typically qualitatively different, in that they may involve complete gutting and reconstruction of a building. Also, from a statistical point of view, the type of simulation model used here cannot encompass such disparate cases. In some cases it would be the equivalent to comparing construction of a new multifamily building with remodeling a kitchen. It is important to have a rule to follow in segmenting a sample in this way. The rule used here is for each neighborhood to remove the largest five percent of the projects that occurred (note that this is not five percent of the total sample, since many building/year observations showed zero investment). Thus the four model variations were estimated using all observations of zero investment and all observations of positive investment that met the criterion of having cost per unit less than the 95th percentile for the relevant neighborhood. The 35 very large investments not included in the four model variations are depicted in Figure B-1 by regulation status and time period. It is important to emphasize that ideally such sample segmentation should be done in terms of characteristics other than the value of the model’s dependent variable (such as gutted property). Unfortunately such characteristics were not available.