ArticlePDF Available

Abstract and Figures

This article examines changes between 1980 and 1990 in the number of rental units and the demographic composition of tenants in four California cities that adopted rent control with vacancy control provisions. Six border areas within the four cities were compared to border areas of adjoining cities that did not have vacancy control. A spatial lag regression model was constructed to estimate the changes in regional and neighborhood components in addition to vacancy control policies. Vacancy control contributed to lower rents and longer tenure by tenants compared to non-vacancy-controlled areas. There were also fewer rental units in part because of a shift from rental housing to owner-occupied housing.
Content may be subject to copyright.
1. Allan Heskin is Professor of urban planning at the University of California, Los Angeles.
He is the author of Tenants and the American Dream (Praeger, 1983), a book about the
tenant movement in Los Angeles and Santa Monica. Ned Levine is director of Ned
Levine & Associates of Annandale, VA. He is the author of studies on rent control,
growth control, community security, and spatial statistics and has published numerous
times in the Journal of the American Planning Association. Mark Garrett is an attorney
and Ph.D. student in urban planning at the University of California, Los Angeles.
RENT CONTROL AND VACANCY CONTROL:
A SPATIAL ANALYSIS OF FOUR CALIFORNIA CITIES
by
Allan D. Heskin, Ned Levine and Mark Garrett
1
RENT CONTROL AND VACANCY CONTROL:
A SPATIAL ANALYSIS OF FOUR CALIFORNIA CITIES
ABSTRACT
Rent control is always a controversial policy. Supporters frequently claim exceptionally
beneficial effects to support the adoption of the policy, while opponents assert a myriad of
negative effects resulting from the policy's adoption. In this study of rent control, four California
cities (Santa Monica, Berkeley, West Hollywood and East Palo Alto) which had vacancy control
on rental units were examined for changes between 1980 and 1990 to assess the effect on rental
unit creation and the composition of tenants. Comparison of demographic changes in six border
areas within the four cities were made with the border areas of the adjoining cities which did not
have vacancy control. A spatial lag model was constructed to estimate the effects on change of
regional, neighborhood, and vacancy control components.
The results show that vacancy control regulation contributed to lower rents and longer
tenure by tenants. However, there were also fewer new rental units created in these areas than in
the comparison areas and apparent conversion of a portion of the housing stock from rent to
ownership. Few other variables were different in the two areas although there was an increase in
Latino renters and children under age 18 in the vacancy controlled areas. There were no
significant differences in the change in the distribution of Whites, Blacks, Asians or seniors.
Also, there were no differences in poverty levels, female-headed families, and the disabled
population.
Within these findings, there is great variance from area to area, suggesting important
neighborhood effects. A lesson of the study is that it is a mistake to examine rental and
ownership housing in isolation. Studies need to look at the interactive effects of change of
rentership on ownership, and vice versa.
INTRODUCTION
In the California study reported upon in this paper, we set out to examine how rent
control with vacancy control effects renter demographics. Vacancy control policies have become
the center of the rent control debate in California with opposition to rent regulation focusing on
vacancy control. Rent regulation, while not favored by many in the rental industry, is
significantly more tolerable without this provision. This position has been adopted by the
California Legislature in the Costa/Hawkins bill of 1997 which authorizes local government to
adopt rent control, but requires the phasing out of vacancy control through 1999 (California
Assembly Bill No. 1164). This makes the results of this study a particularly important baseline
for future research efforts.
There has been a long literature on rent control, much of which consists of debates about
the existence of inequities and market dysfunctions. On the one hand are arguments claiming
that rent control discourages investors from purchasing existing rent controlled properties,
reduces the level of rental housing creation, and creates disincentives for rental maintenance
(Friedman and Stigler, 1981; Grampp, 1950; Johnson, 1951). On the other hand are arguments
claiming that rent control protects low income households, acts as a barrier to housing
discrimination against minorities, female-headed households and the elderly, and in general
promotes tenure stability in communities (Appelbaum and Gilderbloom, 1990).
1
The issue of the effect of rent control on tenant demographics also has a long history. For
example, it has been argued that rent-control creates inequalities among tenants themselves by
encouraging landlords to selectively choose more affluent or stable tenants (Grebler, 1952). The
effect of this landlord behavior is, as argued by Kristoff (1970), that middle class renters were
able to gain a disproportionate share of the benefits of rent control by moving less frequently.
This view was supported by De Salvo (1971), Olsen (1972), Roistacher (1972) and Devine
(1985). There is a counter view that claims that rent control benefits low income tenants, rather
then favoring the middle class, or is at least economically neutral, benefitting all classes of
tenants irrespective of need (Gilderbloom, 1978; Clark and Heskin, 1982; Linneman, 1980, 1987;
Gyourko and Linneman, 1985; Levine and Grigsby, 1985; Levine, Grigsby and Heskin, 1990).
The years of analysis and reform have taught us that it is very difficult to generalize
about rent control. We now have a second generation of rent control laws with varying
provisions, making comparison with first generation laws difficult (Arnott, 1995). Rent control
varies in the allowable rate of rent increases, in the duration of the controls, in the rate of
passable improvement costs, in the amortization rate of improvements, and in the degree of
enforcement. By focusing on cities with vacancy control provisions, we seek to limit variability
although we acknowledge that complete experimental comparisons are not possible.
In the course of the study, data relevant to another element of the rent control debate were
encountered, the arguments claiming that rent control discourages investors from purchasing
existing rent controlled properties and reduces the level of rental housing creation (Gilderbloom,
2
1983; Friedman and Stigler, 1981; Grampp, 1950; Johnson, 1951). Within the confines of the
data that were available, that question is addressed as well.
Vacancy Control Conditions
To date, approximately 13 cities in California have some form of rent regulation.
However, of all the rent control cities, only five - Berkeley, Cotati, East Palo Alto, Santa Monica
and West Hollywood, had vacancy control provisions (i.e., units are permanently controlled
whether tenants stay or not). This study concentrated on the four vacancy control cities that are
in metropolitan areas; Cotati is a small, nearly rural town with a very low population density.
METHODOLOGY
Four California Cities
The four cities have different histories with respect to rent control. Both Berkeley and
Santa Monica adopted rent control in the late 1970s in response to escalating housing values
(Barton, 1998; Capek and Gilderbloom 1992; Baar and Squier, 1987; Heskin, 1983). While there
have been some modifications over time, these laws permanently control units and set an annual
rate of allowable increase. Both cities also have exclusions from the rent control law including
newly constructed rental units, single family homes, and buildings with three or fewer units
where the owner is an occupant. In 1982 Berkeley exempted owner-occupied duplexes and
3
shared housing. Also, both cities have attempted to regulate conversion of rental housing to
ownership. West Hollywood, however, has experienced two different rent control laws. Prior to
becoming a city in 1984, it was governed by a Los Angeles County rent control law that had been
in place since 1979; that law controlled rents only as long as the 1979 tenant remained in place.
In 1985, the city adopted its own rent control law, which established a modified form of vacancy
control over rental units permitting up to 15% increases under certain conditions upon vacancy
and, like Berkeley and Santa Monica, allowing exclusions. East Palo Alto adopted rent control
in 1984 with a rent rollback to April 1983 (Barton, Breslin, Hicks and Tesh, 1994).
Demographic Changes between 1980 and 1990
We examined changes in these four cities between 1980 and 1990. With the exception
perhaps of East Palo Alto which adopted rent control in 1984, the ten year comparison should be
sufficient to demonstrate the consequences of vacancy control on the number of housing units
and the tenant composition. There were a few commonalities. The average rent and median
household income went up in all four cities over the decade. All four cities had a decrease in the
number of tenants who moved into their units within the past 5 years. All four cities also had a
decrease in the percentage of their rental households occupied by white renters, although by a
smaller amount in Santa Monica and a much larger amount in East Palo Alto. All four cities also
had an increase in the percentage of the renter households occupied by Latino (Hispanic) and
Asian renters, particularly in East Palo Alto.
4
On the other hand, there were a number of discrepancies in other variables. Berkeley
showed a slight increase in the percentage of households rented by seniors (age 65+) while East
Palo Alto, Santa Monica and West Hollywood showed a slight decrease. East Palo Alto and West
Hollywood showed increases in the percentage of children under age 18 (for all households)
while Berkeley and Santa Monica showed decreases. Berkeley showed an increase in the
percentage of the population, age 16-64, who were disabled, while Santa Monica and West
Hollywood showed decreases; East Palo Alto showed no change.
In short, a comparison of these cities between 1980 and 1990 reveal few consistencies in
demographic characteristics. Those that were found (decreasing tenant mobility, increasing
ethnic diversity) could be caused by parallel changes in the larger regional environment.
Comparison of Border Areas of Cities
While these four cities have differing conditions in their rent control law, they all had
permanent control over units (vacancy control). Consequently, it is meaningful to compare these
cities with adjacent cities which have not had this condition. Berkeley is adjacent to Oakland,
Emeryville, Albany, and Kensington, none of which have rent control. East Palo Alto is adjacent
to Palo Alto and Menlo Park, neither of which has rent control. Santa Monica, on the other hand,
is adjacent to Los Angeles which does have rent control, though not vacancy control. Finally,
West Hollywood is adjacent to both Los Angeles and Beverly Hills, both of which have rent
control laws, though not vacancy control conditions.
5
We compared the border areas of each of the four cities with the border areas in the
adjacent cities using as small a geographical unit as possible. The advantage of comparing the
1
border areas of cities is that differences in socio-economic characteristics, which tend to cluster
spatially, are minimized. Similarly, land use and building characteristics tend to be more similar
on both sides of a border than, for example, two sub-areas randomly selected from within each of
two adjacent cities. Most of the strong rent control cities are essentially built-out and gross
comparisons with nearby cities, which may not be as developed, are misleading. By comparing a
small geographical area on both sides of a border, where population, housing and land use
characteristics tend to be similar allows a more precise examination of the effects of the different
types of regulation that tenants experience.
The geography was that used by the U.S. Census Bureau to define geographical
enumeration areas for their decennial census. A census block typically corresponds to a normal
city block, though there are exceptions (particularly in rural areas). A block group is typically 7-
12 blocks. A census tract is typically 3-5 block groups. Many of the housing tenure and
demographic variables were available at the individual block level. However, the Census Bureau
is required by law to protect the identity of individual households and persons. To do this, they
develop procedures to ensure confidentiality. Unfortunately these procedures changed between
1980 and 1990 making comparisons difficult at the individual block level. Instead, we
2
conducted the analysis at the block group level where there was very little data suppression in
1980.
6
We examined each side of the border for these four cities and removed from the analysis
areas where at least one side of the border was not populated or where one or both sides of the
border had too few renters to make a comparison meaningful. The result was 100 individual
block groups from 10 individual cities and one unincorporated area that is a census designated
place (Kensington CDP). These block groups are equally divided between vacancy control and
non-vacancy control cities. The comparisons and the number of block groups are:
1. South Berkeley. Berkeley border with Oakland and Emeryville (n=16)
2. North Berkeley. Berkeley border with Albany and Kensington CDP (n=16)
3. East Palo Alto. East Palo Alto border with portions of Menlo Park and Palo Alto
(n=11)
4. West Los Angeles. Eastern Santa Monica border with Los Angeles (West Los
Angeles district) (n=13)
5. Venice. Southern Santa Monica border with Los Angeles (Venice district) (n=8)
6. West Hollywood. Eastern West Hollywood border with Los Angeles (Fairfax/La
Brea district) (n=36)
Figure 1 shows the block groups of the six study areas and the corresponding control area
block groups. The comparisons are not perfect. Two of the comparisons are between vacancy
control cities and those without rent control while the other four are between vacancy control
cities and rent controlled cities without vacancy control.
7
Census Variables
Housing and population data were obtained from the 1980 and 1990 census for each of
the block groups making up the six study areas. Where possible, variables relating to rental
housing were selected, though several general population characteristics were also used. The
variables included were:
1. Total population
2. Number of families
3. Number of occupied housing units
4. Number of vacant rental units
5. Number of occupied rental units
6. Number of owner-occupied units
7. Percent of all units which are rented
8. Median rent level
9. Median household income
10. Percent of renter households who moved in within previous five years
11. Percent of renter households headed by a person of non-Hispanic White ethnicity
12. Percent of renter households headed by a person of Black ethnicity
13. Percent of renter households headed by a person of Hispanic ethnicity
14. Percent of renter households headed by a person of Asian ethnicity
8
15. Percent of renter households headed by a person age 65 or older
16. Percent of families headed by a single female
17. Percent of families below the poverty line
18. Percent of population under age 18
19. Percent of population who are disabled
To measure direct changes of the vacancy control policies, totals were used. To gauge
proportional shifts, percentages were used by dividing the totals by the appropriate denominator.
These variables were selected because they are central to arguments about the assumed
consequences of rent control. For example, if vacancy control reduces the incentives to build
new rental housing, then this would be seen in a net reduction in new rental units or, even,
owner-occupied units. Similarly, if vacancy control favors higher income families (or,
alternatively, if rent control favors lower income households), then this would be seen in relative
increases (or decreases) in median household incomes, median rent levels, and median home
values. Finally, if rent control favors persons who are economically vulnerable, as some
supporters have claimed, then this would be seen in proportional shifts towards minorities, the
elderly and disabled persons. In short, implicit in most of the rent control arguments are
assumptions about changes in various housing and socio-economic variables.
For the neighborhoods on each side of the comparison border, the individual block groups
were summed and indices calculated. To estimate the study area median rent level and median
9
household income, we weighted the block group medians in each study area by the number of
households in each block group.
RESULTS
Table 1 presents a summary of the changes that occurred between 1980 and 1990 for all
the vacancy control and comparison block groups while Appendix A presents the summaries for
the individual study areas. For each variable, the absolute change is presented. There are equal
numbers of block groups in the vacancy control and comparison areas (50 each) and a t-test of
the differences between the mean change between 1980 and 1990 was conducted.
An examination of table 1 reveals some similarities and differences between the vacancy
control and the comparison areas (the border in the adjacent cities). Both types of areas showed
increases in population, though slightly more in the comparison areas; the difference is not
significant, however. Similarly, the total number of units and number of occupied units
increased more in the comparison areas; but, again, these differences are not significant. More
dramatically, the number of rental units decreased and owner occupied units increased in the
vacancy controlled areas whereas there was a large increase in rental units but only a small
increase in owner-occupied units in the comparison areas; these differences are statistically
significant. There were also significantly more vacant rental units in the comparison areas than
in the vacancy control areas.
10
As expected, median rent levels showed a smaller increase in the vacancy control areas
than in the comparison areas, a difference which is significant. The median rent level increased
in the vacancy control areas only about two-thirds that of the comparison areas over the decade
(an average increase of $279 compared to $422). Further, in every one of the six study areas, the
vacancy control block groups showed smaller increases in median rent levels than the non-
vacancy control block groups (Appendix A).
One of the primary objectives of the rent control law in these cities was to increase
community stability by reducing tenant turnover (Heskin, 1983). These results suggest that the
goal was achieved, at least for these block groups. This can be seen by the variable measuring
the percentage of renter households who moved into their units within five years prior to the
census. In both types of area, there was a decline in the percentage who moved in during the
previous five years, a result partly of an aging population (Levine, Grigsby, and Heskin, 1990).
However, the reduction was more dramatic for the vacancy control block groups than for the
comparison areas; the difference is significant. Further, this is seen is all six study areas
(Appendix A), though only slightly in East Palo Alto.
When the other variables are examined, there are less clear differences. There was a
greater decline in rental units headed by non-Hispanic White persons and a greater increase in
rental units headed by African-Americans and by Latinos (Hispanics) in the vacancy control
areas. There was an almost identical change for rental units headed by Asians. However, these
are not very strong differences and only the change in Latinos is statistically significant.
11
Households with senior tenants declined as a percentage of renter households in both
types of area. Families also declined in both types of areas. Female-headed families increased
slightly in the vacancy control areas and decreased slightly in the comparison areas; the
difference is not statistically significant, however. Families below the poverty line increased the
same in both types of areas. Children increased as a percentage of the population in the vacancy
controlled areas and decreased slightly in the comparison areas; this difference was statistically
significant. Finally, the percentage of the population that was disabled decreased about the same
in both groups; this was not significant.
To summarize, it appears that the areas within the vacancy control cities had lower rates
of increase in rent levels, lower renter turnover, and a higher percentage of households with
children than the comparison areas. On the other hand, the number of rental units declined in the
vacancy control areas whereas they increased in the comparison areas, but there was also a
greater increase in the number of owned units created in the vacancy control areas.
However, these are summaries over all 50 block groups of each type. There are
exceptions for different borders and there is considerable variability among individual block
groups. Further, the comparisons are crude and reflect the interaction of many different
variables. To understand the effect of vacancy control policies on rental unit formation and
tenant composition, it is necessary to examine how the policies affect each individual data unit,
which in this case are census block groups.
12
A Model of Neighborhood Housing and Population Change
In order to examine this issue more precisely, we constructed a spatial decomposition
model of housing and population change between 1980 and 1990 and applied it to individual
block groups. The conceptual form of the model is
1990 1980 Change between
Characteristic = Characteristic + 1980-1990 (1)
In turn, the change between 1980 and 1990 is hypothesized to be a function of three components:
1. Regional change
2. Local change ('neighborhood effect')
3. Vacancy control policy ('vacancy control' or not)
The formal model is
1990 1 1980 2 regional 3 local 4 policy
Y = á + â Y + â X + â X + â X + å (2)
1990 1980 regional
where Y is the 1990 variable, Y is the 1980 value for the same variable, X is an
local
indicator of change between 1980 and 1990 at the regional level, X is an indicator of the
policy
change between 1980 and 1990 at the local (neighborhood) level, X is a dummy variable
13
1 4
indicating whether the area is under vacancy control conditions or not, á is a constant, â ....â are
coefficients, and å are the residual errors (assumed to be normally distributed, uncorrelated with
the predicted variable, and uncorrelated with the other independent variables).
The model is applied to the 100 block groups which span the six comparison areas. The
1990 and 1980 variable values are those defined above. If there was no change between 1980
1980 1980
and 1990, then the coefficient for Y would, of course, be 1.00. If the coefficient for Y is
less than 1.00, this indicates that there was either a decrease in the variable between 1980 and
1980
1990 or that other variables account for some its variance; conversely, a coefficient for Y
greater than 1.00 indicates an increase in the variable not associated with the other variables in
the model. In any case, the correlation between the 1980 and 1990 value would be expected to
be very high. Thus, the model is actually testing the effects of the additional variables on the
change component.
In terms of the other variables, region is subject to different interpretations. For a rental
market, it could be defined as the city, a subarea within the county, the county, or a multi-county
area (e.g., the Bay Area, Southern California); many sources (e.g., newspaper like the Los
Angeles Times real estate section) use the county as a reference for housing markets, though the
Census Bureau uses the larger metropolitan area. We arbitrarily took the county to represent the
regional effect and measured county change between 1980 and 1990 for each variable. For the
policy variable, we assigned '1' to each block group within the four vacancy control cities and ‘0'
to each block group within the comparison areas.
14
For a neighborhood effect, we used a particular form of regression model called a spatial
lag regression (Anselin, 1988; Levine, Kim and Nitz, 1995; Levine, 1996). The explicit location
of each block group in relation to all other block groups in the data set is taken into account by
incorporating the value of the dependent variable at all other locations and weighting these by the
inverse square of its distance from the particular block group (spatial lag). This weighting uses a
3
spatial weights matrix in order to examine the effects of concentration or dispersion on the
dependent variable. It is a measure of spatial autocorrelation. It is a weighted average and
4
indicates the extent to which the value of a variable for a block group is influenced by the values
of the same variable for nearby block groups. Using a spatial statistics software package,
SpaceStat (Anselin, 1992), the spatial lag is estimated using a maximum likelihood estimator.
Once the coefficient of the spatial lag is estimated, it is subtracted ('filtered') from the dependent
variable and the remainder is then regressed against the independent variables using a least
squares estimator; the process is repeated until the solution optimizes a non-linear likelihood
function. This method, therefore, allows for an estimate of a neighborhood effect independent of
the 1980 value, regional change or the vacancy control policy variable.
This model was used to estimates volumes, the number of persons or households with
various characteristics. To estimate percentages, we added an additional variable to control for
the considerable differences in size among the block groups, either the total number of units in
1990 (for the renter variables), the total number of families in 1990 (for the two family
variables), or the total population in 1990 (for the two population variables). Without this
5
statistical control, larger block groups will have a greater impact on the model than smaller ones.
15
To illustrate the method, table 2 presents the spatial lag result for changes in the number
of rental units between 1980 and 1990. The R for the equation is highly significant (0.95). The
2
maximum likelihood function reported is the logarithm of the likelihood obtained from the
maximum likelihood estimate; the model with the highest log likelihood is the one that achieves
the best fit. The Akaike Information Criterion (AIC) corrects the log likelihood for the degrees
of freedom; the best model is the one with the lowest AIC.
The coefficient of the number of 1980 rental units is highly significant, as would be
expected, and is 1.0440; this indicates that there was a 4.4% increase in rental units over the
period that can't be explained by the other variables in the equation. The regional change
variable is slightly negative but is not significant. All other things being equal, the effect of
regional change is to decrease the number of rental housing units added; the effect was
insignificant, however, and must be considered as random. The spatial lag variable is also not
significant, indicating that there was no apparent clustering in created rental housing units. Since
the spatial lag variable is a weighted index, the coefficient does not have an intuitive meaning.
Finally, the policy variable is significant and negative. There were approximately 60 fewer rental
units in block groups within vacancy control cities than in block groups within non-vacancy
control cities.
Table 3 presents a summary of 19 spatial lag models that were constructed. The R are
2
generally high though there are exceptions. The model has a good fit for the population, the
number of occupied units, the number of rental units, the rental percentage, the percentage of
16
White renter households, the percentage of Black renter households, the percentage of Latino
renter households, the percentage of female-headed families, and the percentage of the
population under age 18. Moderate results are found for the number of families, the number of
owned units, the median rent level, the median household income, and the percentage of senior
renter households, while poor results are found for the remaining variables.
Intercept and 1980 Characteristics
In all 19 models, the constant is significant in eight, two of which are negative. For the
13 models examining changes in percentages, the control variable (either 1990 housing units,
families or population) is significant in four. In terms of the conceptual model, the 1980 value is
positive and highly significant in 18 of the 19 models, as would be expected; block groups which
had relatively high values in 1980 also had high values in 1990, and vice versa. The one
exception is changes in the percentage of families below the poverty line. Other than the one
exception, there was a high degree of consistency for these block groups over the decade.
Regional Effects
Five of the models show significant regional effects, four of which are negative. Positive
regional change is associated with increases in the percentage of Latino renter households while
negative changes are associated with decreases in population, numbers of families, the
percentage of White renter households, and the percentage of the population under age 18. In the
17
five counties covering these areas, all showed decreases in the percentage of White renter
households and increases in the percentage of Latino households. On the other hand, the
negative changes for population, families and the percentage of the population under age 18 may
reflect sub-regional changes for both the vacancy control and comparison block groups since
1980-90 population growth in both the vacancy control and comparison areas was smaller than in
the five counties in which they are found. The number of families decreased in both types of
areas compared to an increase in all five counties. The percentage of the population under age 18
increased slightly in the vacancy areas whereas they decreased in the comparison areas; in all five
counties, they increased substantially.
Local Effects
Seven of the models show significant local effects, six of which indicate positive spatial
autocorrelation (spatial concentration). There are positive neighborhood effects (spatially
concentrated) for the number of vacant rental units, the percentage of all units which are rented,
the percentage of renter households headed by Whites, the percentage of renter households
headed by Blacks, the percentage of families headed by single women, and the percentage of
families below the poverty line, while there is a negative local effect (spatial dispersion) for the
number of families. The significant relationship for White and Black renter households suggests
the continuance of spatial separation for these ethnic groups, either through choice or through
discrimination.
18
Effects of Vacancy Control
Finally, eight of the models show significant effects for the vacancy control policy
variable. Compared to the non-vacancy control block groups, those in vacancy control cities had
a smaller number of vacant rental units (about 13 fewer), a decline in the number of occupied
rental units (about 60 fewer), a greater increase in the number of owner-occupied units (about 34
more), a proportional decline in the percentage of units which are rented (about 7% lower), lower
median rent levels (about $117 a month lower), a lower percentage of renter households who
moved in between 1985-90 (renter 'turnover', about 10.1% lower), a higher percentage of Latino
renter households (about 1.4% higher) and a higher percentage of the population under age 18
(about 1.4% higher). No effects were seen for median household income, the percentage of
Black renter households, the percentage of Asian renter households, the percentage of senior
renter households, the percentage of female-headed families, the percentage of families below the
poverty line, and the percentage of households with disabled persons.
In other words, these models suggest five main consequences associated with vacancy
control conditions. First, rent levels are lower, which would be expected. Second, rental tenure
is longer. Third, a shift took place from rentership to ownership of housing units. Fourth, there
was increased diversity, at least with respect to Latino renters. Fifth, a higher percentage of the
population was under age 18; this latter result, however, is mostly due to changes in East Palo
Alto where there has been a rapidly increasing Hispanic population.
19
Community Stability and Increased Ownership
The overall thrust of the findings involves increased community stability in vacancy
control areas, both in longer stays for tenants and increased ownership. The longer stays by
tenants are undoubtedly encouraged by the more moderate increases in rents. The shift away
from rental units and the increase in home ownership was found in three of the four cities:
Berkeley, Santa Monica, and West Hollywood. Also, though it was not uniform across the block
groups, the shift was remarkably consistent in those cities. Thirty-five of the block groups (or
68.6%) in the vacancy control areas showed a decrease in rental housing units during the 1980s
while only 21 (or 41.2%) of the comparison block groups showed a decrease in rental housing
units. Conversely, 40 of the block groups (or 78%) in the vacancy control areas showed
increases in the number of owner-occupied units compared to 18 (or 35%) of the block groups in
the comparison areas. In terms of net increase in occupied units (whether rented or owned) 22 of
the vacancy control block groups and 26 of the comparison block groups showed a net increase, a
difference which was not significant.
Possible Explanations for the Shift to Home Ownership
There are two ways that the shift from rentership to ownership could have occurred in the
vacancy control block groups. First, a shift could occur through differential demolition and
construction, such that some rental units were demolished and an approximately equal number of
new ownership units were constructed. Second, units could be converted from rental to owner-
20
occupied, without any demolition or construction. However, the rent control ordinances in effect
throughout most of the decade in Berkeley, Santa Monica, and West Hollywood all placed
obstacles in the way of the loss of rental property through demolition of rental property or the
conversion of rental property to ownership. While we don't have detailed demolition and
construction data for these cities, from all informal accounts there were few demolitions. The
restrictions, in general terms, required one-for-one replacement of affordable rental housing units
removed from the stock. In the later part of the 1980s the rules were changed to make demolition
easier and, in the case of Santa Monica, the law was changed by the voters to permit conversions
if a large majority of the tenants voted for the change and agreed to purchase. We have to
assume, therefore, that the shift was primarily that of conversion, rather than differential
demolition and construction. Further, exceptions to the rent control laws, which eliminate
controls for smaller buildings in which the owner is living in the building, may have encouraged
the conversion of buildings from rental to owner-occupied. Smaller rental properties, duplexes
and triplexes might have been sold by investor owners to occupants either as tenants-in-common
occupying all units or as a single owner occupier who rented out the adjoining units.
Other Factors Encouraging Shift from Rental to Owner Units
The presence of rent control was probably a factor in the shift, but there are other factors
that could have contributed to the decrease of rental units and an increase in ownership units.
First, increasing property values in these cities during the 1980s may have encouraged
ownership. The three cities where the shift took place are all highly desirable locations where
21
home ownership is valued. In such markets it may be more advantageous to sell rather than to
rent out a single family home. Second, the difficulty of converting units to condominium
ownership in these cities may have encouraged the construction of condominiums in the first
place. In the comparison areas, while there was greater net new construction (the excess of new
units added over demolitions), it appears that much of the new construction was of rental housing
rather than condominiums.
Third, there may be differences in political climate between the two types of study areas.
Even though new rental housing units are excluded from rent control in all four of the vacancy
control cities, the intense politicalization of the issue may have created a perception by
developers to not build new rental housing for fear of future restrictions. Fourth, these vacancy
control cities also have inclusionary policies not present in the surrounding cities. The
inclusionary policies in West Hollywood, for example, require either the construction of
affordable units by the developer or a contribution to the cities housing trust fund that local non-
profit groups use to construct affordable housing. This added cost would drive up the cost of
construction and may push developers toward the higher return at the ownership end of the
market. A fifth factor may have been the high degree of growth management involved in
virtually all the cities in the study (with the exception of East Palo Alto). Restrictive building
conditions could have been a more significant factor than vacancy control, affecting both the
vacancy control cities as well as the comparison cities.
6
22
The question of what contributed to the shift from renting to ownership housing and who
benefitted clearly requires more research that is beyond the scope of this study which set out to
examine the question of demographic effects of vacancy control. All we can say at this point is
that the net effect of vacancy control and the shift from rental to owner occupied units did not
appear to have an inequitable demographic consequence on the mix of tenants that remained.
SUMMARY AND CONCLUSION
This study took on added importance on August 4, 1996 when the Governor signed a bill
creating a three year phase out of vacancy control in all California cities. Thus, the study, in
addition to contributing a new approach to analyzing the effects of rent regulation and providing
data rather than just theory on the consequences of rent control, provides base line data for
examining longer term effects when the data from the year 2000 census is collected.
The data show that vacancy control conditions have created several consequences for
tenants. First, vacancy control has reduced the rate of increase in rents for tenants. In this sense,
it has fulfilled what its designers initially sought, which was to prevent escalating land values
from pushing rent levels beyond most tenants' means. Second, vacancy control appears to have
increased residential stability for tenants. Third, it appears to have encouraged more ethnic
diversity, at least with respect to the growing Latino population. Fourth, vacancy control
conditions are uncorrelated with changes in income levels, ethnic distribution, the elderly
population, female-headed families, families living below the poverty line, and the disabled
23
population. There is no evidence of 'gentrification' as a result of the vacancy control. Fifth, and
finally, vacancy control conditions appear to have contributed to a shift toward ownership in the
composition of the tenure mix in the housing stock.
The rent control debate is a highly politicized, emotional arena with an enormous amount
of hyperbole on both sides. The data we have shown do not fully support any of the positions
and, in fact, show how complex rental housing markets can be. What policy implications are
drawn from our findings depends usually on, first, how the market is valued and, second, on how
neighborhood stability is valued. People who believe in the wisdom of markets and value
mobility usually don't like rent control. People who worry about the potential impact of market
failures and consider neighborhood stability to be of prime importance usually like rent control.
While rent control does not destroy long term underlying land values (Baar and Squier, 1987), it
can take the edge off of prices. Lowering the 'up' side of a market certainly is seen as painful by
owners. However, in Southern California's recent past there was a significant downward spiral in
apartment house prices. People who either bought or refinanced to liquefy equity on the high
side were in serious potential trouble in that downward period. The 'down-side’ pressure is
greatly relieved by being denied 'upside' potential. The owners may not appreciate being
protected from this risk, but if a down-side collapse takes place, public intervention may be
called for, even demanded, if the foreclosure rate in housing accelerates.
A major caveat, however, has to be made regarding both why there was a significant shift
in tenure in Berkeley, Santa Monica and West Hollywood and who benefitted from that shift. If
24
vacancy control, as opposed to changes in the home ownership market, anti-demolition policies,
anti-conversion policies, inclusionary requirements or growth control is found to be a dominant
factor that increased the rate of home ownership, one would have to consider the policy
implications of such an impact. Is this positive or negative? The argument could be made that
such a conversion hurt lower income people because rental units tend to be more affordable than
ownership units or, alternatively, that it was positive because rent control had the effect of
making home ownership more affordable in these cities. These or other arguments that might be
constructed are, at this point, clearly hypotheses. One would have to examine which units where
converted, what was the sales price, and who was the purchaser. More research is clearly
needed.
Finally, we were struck by the complexity of what we found. When data is disagregated
in examining block groups and border areas, we found particular comparisons that could support
virtually any position on rent control. The models represent the commonalities over all block
groups. However, in doing so, they ignore the substantial differences. Further, they show that
the relationship between rentership and ownership is complex. One of the most important
implications is that one should not attempt to isolate rental or home ownership data when doing
housing analysis, but rather see housing markets as a whole.
25
1. We chose not to compare each of the four cities broadly to the adjacent cities for a
number of reasons. First, they differ in size. Los Angeles, for example, is huge
compared to Santa Monica and West Hollywood. It would be misleading to compare a
very large city, with numerous sub-regions and differing populations, with a small one
and then draw conclusions about the effect of a particular policy condition. Second, the
housing situations in each of these cities differ. For Berkeley, the western and
northwestern borders are predominately single family homes or vacant. The same is true
for East Palo Alto’s border with Menlo Park, Santa Monica’s northern border with Los
Angeles, and West Hollywood's western border with Beverly Hills.
2. In 1980, if the number of cases for a given variable was too small, the Census Bureau
suppressed the data. This meant that the sum of the data at the block level did not equal
the total for the block group level. In 1990, rather than suppress the data, the Census
Bureau added in randomly assigned persons and households in such a way that the sums
of the blocks added up to the total for the city.
3. Even though this theoretically allows interaction between block groups which are widely
disparate (e.g., between Berkeley and Santa Monica), the practical effects are negligible
due to a rapid deterioration in the distance effect (i.e., 1/d ). We ran the subsequent
2
models with interaction cut-offs of 1 and 3 miles respectively, and found that there was
virtually no change in the results.
4. The spatial lag model is defined as:
i j
Y = ñW(Y ) + âX + å
i
where Y is an N by 1 vector of observations on the dependent variable for all locations, I,
j
W(Y ) is a weighted matrix of N by 1 vector of values for the dependent variable summed
over all locations, j, where i =/ j (the 'spatial lag'), ñ is the coefficient of the spatial lag (the
spatial autoregressive term), X is an N by K matrix of observations on the explanatory
variables, â is a K by 1 vector of regression coefficients, and å is an N by 1 vector of
normally distributed random error terms, with mean 0 and constant variance, ó (Anselin,
2
1992). The weights are typically inverse distances (e.g., 1/d, 1/d , 1/d ). We used an
2 3
inverse squared distance in our models (1/d ).
2
5. While, in theory, census block groups should be of similar size and, hence, not require a
size adjustment, in practice they differ considerably. For example, the total population in
1990 for the 100 block groups varied between 179 and 4,091!
6. During the 1980s, California cities experienced a huge increase in local growth
management and control legislation designed to contain the rate of new development as
well as maintain a balance between development and infrastructure creation (Glickfeld
Endnotes
and Levine, 1992; Levine, 1997). We have data from two surveys of California cities and
counties in 1988 and 1992 which documented residential, commercial and other types of
local growth management legislation (Glickfeld and Levine, 1992; Levine, Glickfeld, and
Fulton, 1996; Levine, 1999). The 10 cities and one unincorporated area had a total of 42
separate growth measures enacted as of 1992. The average number of measures for the
four vacancy control cities was 2.8 while the average number for the comparison cities
(and one unincorporated area) was even higher, 4.4. Of all the cities involved, only East
Palo Alto did not have any growth management legislation. We don’t know how
stringently these restrictions are enforced or how effective they are in reducing rental
housing construction though at a regional level, growth controls have significantly
displaced rental housing (Levine, 1999).
References
Anselin, Luc. 1988. Spatial Econometrics, Methods and Models. Kluwer Academic: Dordrecht.
Anselin, Luc. SpaceStat: A Program for the Statistical Analysis of Spatial Data. Santa Barbara,
CA: National Center for Geographic Information and Analysis; 1992.
Appelbaum, Richard P., and John I. Gilderbloom. 1990. The redistributional impact of modern
rent control. Environment and Planning A. 22: 601-614.
Arnott, Richard. 1995. Time for Revisionism on Rent Control. Journal of Economic
Perspectives. 9, 1: 99-120
Baar, Kenneth, and Gary Squier. 1987. Perspectives on the Rental Housing Market in the Santa
Monica Area. Report prepared for the Rent Control Board of the City of Santa Monica. August.
Barton, Stephen. 1998. The Success and Failure of Strong Rent Control in Berkeley in W. Dennis
Keating, Michael B. Teitz, and Andrejs Skaburskis, Rent Control: Regulation and Rental
Housing Market. New Brunswick, N.J.: Center For Urban Policy Research .
Barton, Stephen, Kate Breslin, Alison Hicks and Carolyn Tesh. 1994. Rent Control in East Palo
Alto,1983-1994: An Evaluation, prepared for Lenny Goldberg and Associates. Sacramento,
California.
Capek, Stella and John Gilderbloom. 1992. Community Versus Commodity: Tenants and the
American City. Albany: State University of New York Press.
Clark, W. A. V., and Allan David Heskin. 1982. The Impact of Rent Control on Tenure
Discounts and Residential Mobility. Land Economics 58, 1: 109-117.
De Salvo, J. S. 1971. Reforming Rent Controls in New York City: Analysis of Housing
Expenditures and Market Rents. Regional Science Association Paper, 195-227.
Devine (1985).
Friedman, Milton, and George Stigler. 1981. Roofs or Ceilings? The Current Housing Problem.
In Rent Control: Myths and Realities, edited by W. Block and E. Olsen. Vancouver, British
Columbia: Fraser Institute.
Gilderbloom, John I. 1978. The Impact of Moderate Rent Control in the United States: A Review
and Critique of Existing Literature. Sacramento: California State Department of Housing and
Community Development.
Gilderbloom, John. 1983. The Impact of Moderate Rent Control in New Jersey: An impact
Study of 26 Rent Controlled Cities. Journal of Urban Analysis 7: 135-154.
References (continued)
Glickfeld, Madelyn , and Ned Levine, 1992. Regional Growth...Local Reaction: The Enactment
and Effects of Local Growth Control and Management Measures in California. Cambridge,
MA: The Lincoln Institute of Land Policy.
Grampp, William D. 1950. Some Effects of Rent Control. Southern Economic Journal 16, 4:
425-47.
Grebler, Leo. 1952. Implications of Rent Control: Experience in the United States. International
Labour Review 65, 4: 462-85.
.
Gyourko, Joseph, and Peter Linneman. 1985. Equity and Efficiency Aspects of Rent Control: An
Empirical Study of New York City. Wharton School, University of Pennsylvania, Philadelphia.
Typescript.
Heskin, Allan David. 1983. Tenants and the American Dream: Ideology and the Tenant
Movement. New York: Praeger.
Johnson, D. Gale. 1951. Rent Control and the Distribution of Income. American Economic
Review 41, 2: 569-85.
Kristof, Frank B. 1970. Housing: Economic Facets of New York City’s Problems. In Agenda for
a City. Issues Confronting New York, edited by L. C. Fitch and A. H. Walsh. Beverly Hills:
Sage.
Levine, Ned. 1999. The effects of local growth management on regional housing production and
population redistribution in California, In press, Urban Studies. November.
Levine, Ned. 1997. A note on urban sprawl. Journal of the American Planning Association.
Spring 1997. 63 (2), 279-282.
Levine, Ned. 1996. Spatial statistics and GIS: software tools to quantify spatial patterns. Journal
of the American Planning Association. 62, 3: 381-392.
Levine, Ned, Madelyn Glickfeld, and William Fulton (1996). Home Rule: Local Growth
Control...Regional Consequences. Report to the Metropolitan Water District of Southern
California and the Southern California Association of Governments. Los Angeles.
Levine, Ned , and Gene Grigsby. 1985. A Survey of Tenants and Apartment Owners in West
Hollywood Conducted for the City of West Hollywood. The Planning Group, Los Angeles, April.
Levine, Ned, J. Eugene Grigsby and Allan Heskin. 1990. Who Benefits from Rent Control?
Effects on Tenants in Santa Monica, California. Journal of the American Planning Association.
56, 2: 140-52.
References (continued)
Levine, Ned, Karl E. Kim, and Lawrence H. Nitz. 1995. Spatial analysis of Honolulu motor
vehicle crashes: II. Generators of crashes. Accident Analysis & Prevention, 27, 5: 675-685.
Linneman, Peter. 1980. Some Evidence on the Functional Form of the Hedonic Price Function
for Urban Housing Markets. Journal of Urban Economy 15: 129-48.
Linneman, Peter. 1987. The Effect of Rent Control on the Distribution of Income among New
York City Renters. Journal of Urban Economy 22: 14-34.
Olsen, E. A. 1972. An Econometric Analysis of Rent Control: An Empirical Analysis of New
York’s Experience. Journal of Political Economy 82: 1081-1110.
Roistacher, E. A. 1972. The Distribution of Tenant Benefits under Rent Control. Doctoral
dissertation, University of Pennsylvania.
Table 1
SUMMARY CHARACTERISTICS OF SIX COMPARISON AREAS
Aggregated Over Border Block Groups
Not
Vacancy Vacancy
Controlled Controlled
(n = 50) (n = 50)
1980 1990 Change 1980 1990 Change p
*
Population 53426 56867 +3441 50680 54533
+3853 n.s.
Families 11699 11482 -217 11426 10948 -478
n.s.
Occupied units 27628 28806 +1178 24563 26527 +1964
n.s.
Vacant rental units 680 684 +4 505 1162
+657
**
Occupied rental units 21226 20420 -806 16552 18375
+1823 **
Owner-occupied units 6415 8386 +1971 8011
8152
+141
**
% Renter Units 76.8% 70.9% -5.9% 67.4%
69.3%
+1.9% ***
Median Rent Level $276 $555 +$279 $285 $707 +$422
***
Median Household Income $14102 $30260 +$16158 $15649 $32922
+$17273
n.s.
% renter households who
moved in within 5 years 72.1% 59.2% -12.9% 72.3% 70.5%
-1.8% **
% of renter households
headed by person of non-
Hispanic White ethnicity 80.4% 74.7% -5.7% 77.9% 75.3%
-2.6% n.s.
% of renter households
headed by person of
Black/African-American
ethnicity 9.7% 11.2% +1.5% 9.9% 9.8% -
0.1%
n.s.
% of renter households
headed by person of
Hispanic ethnicity 5.7% 9.5% +3.8% 8.1% 9.9%
+1.8
% *
% of renter households
headed by person of
Asian ethnicity 2.7% 4.6% +1.9% 2.9% 5.0%
+2.1
%
n.s.
% of renter households
headed by persons
age 65 or older 20.9% 18.7% -2.2% 20.3% 15.3%
-5.0% n.s.
% of families
headed by females 13.1% 13.8% +0.7% 11.7% 11.5% -
0.2%
n.s.
% of families
below poverty 6.9% 7.9% +1.0% 6.1% 7.1%
+1.0% n.s.
% of population
under age 18 14.2% 15.7% +1.5% 15.8% 14.4%
-1.4% *
% of population who
are disabled 6.2% 5.5% -0.7% 5.8%
5.2% -
0.6%
n.s.
-------------------------------------------------------------------------------------------------------------------------
-----------------------------
t-test of difference in mean change per block group
a
n.s. Not significant
* p<.05
** p<.01
*** p<.001
Table 2
SPATIAL LAG MODEL OF
RENTAL HOUSING CHANGE
Dependent
Variable: 1990 Rental Housing Units
N = 100 block groups
R = 0.955
2
Maximum Log Likelihood Function = -578.34
AIC = 1166.67
Independent
Variables Coefficient Z p
Constant 52.7133 0.98 n.s.
1980 Rental
Housing Units 1.0440 40.92 ****
1980-90 Change
in County Rental
Housing Units -3.5809 -.77 n.s.
(regional change)
Spatial Lag 0.0184 0.94 n.s.
(local change)
Vacancy
Control -60.3492 -3.77 ***
__________________________________________________
n.s. Not significant
* p<.05
** p<.01
*** p<.001
**** p<.0001
Table 3
SPATIAL LAG MODELS OF
HOUSING & POPULATION CHANGE
Coefficients and Significance Levels of Models
1980-1990
Dependent Control 1980 Regional Local Vacancy
Variable Constant Variable Value Change Effect ControlR
2
Population 293.5820 - 1.1761 -254.4660 0.0041 -22.8981
0.896
n.s. **** * n.s. n.s.
Families 146.8470 - 0.9095 -104.4610 -.1098 9.1951
0.798
*** **** *** * n.s.
Occupied Units 30.8382 - 1.1120 -47.7796 0.0034 -
24.9
4720
.950
n.s. **** n.s. n.s. n.s.
Vacant rental units 12.1787 - 0.8602 -.2082 0.1866 -13.2544
0.331
n.s. **** n.s. * **
Occupied rental Units 52.7133 - 1.0440 -3.5809 0.0184 -
60.3
492
0.95
5
n.s. **** n.s. n.s. ***
Owner-occupied Units 59.5274 - 0.9182 -33.9527 -0.0641
34.0
7640
.674
* **** n.s. n.s. **
% renter units 3.3332 0.0032 0.8774 -28.0363 0.0464 -6.9396
0.883
n.s. n.s. **** n.s. * ***
Median rent level -141.6060 -0.0210 2.3888 1.1224 0.0423 -
116.9070
0.541
n.s. n.s. **** n.s. n.s. **
Median household
income 115655 -2.9854 1.8323 -1085.05
0.02
06 -
2362
.17
0.73
9
n.s. n.s. **** n.s. n.s. n.s.
% renter households
who moved in
within 5 years 51.8110 -0.0011 0.2842 -.8116
0.03
21
-
10.1
1500
.164
**** n.s. ** n.s. n.s. **
% White renter
households 13.3465 -0.0071 0.7104 -0.9002 0.05
77
-
3.76
300.
793
*** * **** ** * n.s.
% Black renter
households 2.0939 0.0013 0.7874 -0.1223 0.08
54
-
.151
00.9
75
n.s. n.s. **** n.s. *** n.s.
% Latino renter
households -6.5423 0.0002 0.8621 0.2194 -
.111
8
1.42
380.
846
*** n.s. **** **** n.s. *
Table 3 (continued)
1980-1990
Dependent Control 1980 Regional Local Vacancy
Variable Constant Variable Value Change Effect
Cont
rolR
2
% Asian renter
households 1.9957 -0.0013 1.0404 0.0042
0.11
57
0.50
730.
410
n.s. n.s. **** n.s. n.s. n.s.
% Senior renter
households 3.8517 -0.0005 0.4604 -0.1894 0.07
20
1.55
860.
514
** n.s. **** n.s. n.s.
n.s.
% female-headed
families -2.2191 0.0100 0.8115 0.0454 0.0682
0.18
930.
883
*** **** **** n.s. * n.s.
% families below
poverty line 1.9924 0.0187 -.6958 -.1097
0.35
25
-
.816
10.2
15
n.s. ** n.s. n.s. **** n.s.
% under age 18 9.7712 0.0015 0.6164 -0.6502
0.05
26
1.36
510.
859
**** ** **** **** n.s. *
% who are disabled 2.9375 -0.0003 0.5016 0.3286
0.00
48
1.08
500.
336
n.s. n.s. **** n.s. n.s. n.s.
__________________________________________________________________________________
__________________________
n.s. Not significant
* p<.05
** p<.01
*** p<.001
**** p<.0001
FIGURE 1 - MAP OF BLOCK GROUPS OF FOUR CITIES
Appendix A
SUMMARY CHARACTERISTICS OF SIX COMPARISON AREAS
Selected Block Groups
South Berkeley North Berkeley
Albany/
Berkeley Oakland Berkeley Kensington
(Vacancy (Comparison (Vacancy (Comp
arison
Control) Group) Control) Group)
1980 1990 1980 1990 1980 1990 1980 1990
Population 6552 6378 5612 5921 6780 6592 5631 5643
Families 1363 1363 1283 1251 1722 1677 1525 1470
Occupied units 2978 2874 2527 2594 2975 2908 2475
2419
Vacant rental units 74 80 43 47 25 12 15 16
Occupied rental units 1955 1734 1578 1714 940 730 640 657
Owner-occupied units 1023 1140 949 880 2048 2178 1835
1762
% Renter Units 65.7% 60.3% 62.5% 66.1% 31.6%
25.1% 25.9%
27.2%
Median Rent Level $204 $384 $191 $482 $252 $466 $304 $817
Median Household Income $11455 $25577 $11354 $23328 $19745
$41232 $19629
$45499
% renter households who
moved in within 5 years 74.5% 58.2% 70.2% 65.9% 73.7% 62.3% 70.0% 71.1%
% of renter households
headed by person of non-
Hispanic White ethnicity 48.4% 43.4% 41.4% 42.3% 74.5% 66.3% 86.1% 80.4%
% of renter households
headed by person of
Black/African-American
ethnicity 44.2% 44.6% 51.2% 44.9% 11.6% 11.8% 3.3% 4.6%
% of renter households
headed by person of
Hispanic ethnicity 3.0% 5.4% 3.9% 6.5% 3.8% 5.6% 4.4% 5.6%
% of renter households
headed by person of
Asian ethnicity 2.9% 7.4% 2.5% 6.5% 7.6% 16.7% 5.6%
9.0%
% of renter households
headed by persons
age 65 or older 11.2% 10.9% 9.3% 10.6% 12.6% 10.7% 14.7%
11.9%
% of families
headed by females 28.5% 26.3% 27.1% 24.8% 11.7% 9.7% 10.4% 10.2%
% of families
below poverty 14.6% 14.7% 11.6 14.4% 4.1% 1.7% 1.6%
2.2%
% of population
under age 18 20.3% 19.8% 21.9% 18.1% 19.1% 19.0% 19.6%
21.3%
% of population who
are disabled 9.9% 10.6% 10.2% 7.8% 4.5% 4.9% 4.0%
3.9%
Appendix A (continued)
East Palo Alto West Los Angeles
Menlo Park/ East West
East Palo Alto Palo Alto Santa Monica Los
Angeles
(Vacancy (Comparison (Vacancy (Comparison
Control) Group) Control) Group)
1980 1990 1980 1990 1980 1990 1980 1990
Population 7232 10095 6505 6562 6522 6603 9261 10122
Families 1416 2049 1618 1569 1506 1422 1972 2036
Occupied units 3358 3656 2321 2226 3164 3331 4555 496
1
Vacant rental units 78 172 20 29 43 53 84 221
Occupied rental units 2456 2661 726 679 2207 2082 3518 3695
Owner-occupied units 902 995 1595 1547 957 1249 1037 126
6
% Renter Units 73.1% 72.8% 31.3% 30.5% 69.8% 62.5% 77.2% 74.
5%
Median Rent Level $266 $548 $279 $584 $291 $502 $331 $779
Median Household Income $13536 $25534 $24815 $49484 $17514 $40446
$16227 $36095
% renter households who
moved in within 5 years 85.1% 82.0% 80.7% 81.2% 75.5% 50.1% 78.4% 76.9%
% of renter households
headed by person of non-
Hispanic White ethnicity 62.6% 35.3% 41.6% 37.4% 81.1% 77.6% 80.6% 76.4%
% of renter households
headed by person of
Black/African-American
ethnicity 25.0% 36.1% 41.9% 45.5% 1.,7% 2.7% 2.8%
3.2
%
% of renter households
headed by person of
Hispanic ethnicity 7.0% 22.3% 12.1% 13.0% 11.3% 12.4% 9.6% 11.2%
% of renter households
headed by person of
Asian ethnicity 4.5% 6.7% 2.8% 4.9% 4.1% 7.0% 6.5% 9.0%
% of renter households
headed by persons
age 65 or older 6.1% 5.4% 10.1% 10.0% 14.9% 15.8% 12.1% 10.9%
% of families
headed by females 19.3% 25.5% 15.8% 16.6% 11.2% 9.4% 7.8% 7.2%
% of families
below poverty 11.2% 16.0% 6.5% 6.3% 3.5% 5.6% 4.6%
4.7
%
% of population
under age 18 20.5% 29.7% 27.2% 27.9% 15.6% 14.2% 12.8% 10.
6%
% of population who
are disabled 7.4% 5.8% 7.0% 6.7% 4.4% 3.4% 4.5%
4.5
%
Appendix A (continued)/
Venice West Hollywood
South Venice West Fairfax/La
Brea
Santa Monica Los Angeles Hollywood Los Angeles
(Vacancy (Comparison (Vacancy (Comparison
Control) Group) Control) Group)
1980 1990 1980 1990 1980 1990 1980 1990
Population 4784 5064 3657 3730 21556 22135 20014 22555
Families 891 949 640 571 4801 4022 4388 4051
Occupied units 2628 3054 1695 1700 12525 12983 10990 126
27
Vacant rental units 140 46 24 75 320 321 319 774
Occupied rental units 2321 2376 1426 1410 11347 10837 8664 10220
Owner-occupied units 307 678 269 290 1178 2146 2326
240
7
% Renter Units 88.3% 77.8% 84.1% 82.9% 90.6% 83.5%
78.
8%
80.
9%
Median Rent Level $310 $521 $255 $619 $283 $608 $287
$73
2
Median Household Income $17033 $34765 $13331 $25312 $12184 $26508
$13935 $29224
% renter households who
moved in within 5 years 74.5% 55.4% 77.1% 70.9% 67.7% 56.2% 68.9%
68.1%
% of renter households
headed by person of non-
Hispanic White ethnicity 86.47% 85.4% 67.5% 69.4% 88.8% 87.1%
87.6%
83.
5%
% of renter households
headed by person of
Black/African-American
ethnicity 3.4% 3.2% 7.4% 6.6% 3.1% 3.2%
3.4
%
4.7
%
% of renter households
headed by person of
Hispanic ethnicity 6.7% 7.9% 21.5% 21.8% 4.8% 7.1%
6.0
%
8.4
%
% of renter households
headed by person of
Asian ethnicity 3.2% 3.3% 1.9% 2.0% 1.6% 2.6%
1.5
%
3.4
%
% of renter households
headed by persons
age 65 or older 13.7% 16.0% 13.5% 8.0% 29.2% 24.9%
28.
0%
19.
3%
% of families
headed by females 15.2% 9.3% 16.1% 14.4% 7.6% 7.9%
7.3%
7.5
%
% of families
below poverty 2.7% 0.0% 12.0% 15.4% 6.4% 6.8%
5.7
%
7.0
%
% of population
under age 18 13.1% 9.5% 21.5% 16.8% 8.6% 9.1%
9.6%
9.0
%
% of population who
are disabled 6.9% 3.3% 4.0% 5.0% 5.5% 5.1%
5.5%
4.8
%
Article
Full-text available
In the context of chronic problems with high housing costs, rent regulation has returned to the forefront of policy debate in several countries. This paper addresses three distinct questions related to rent regulation and the role of evidence. First, what are the drivers of policy change on rent regulation and what role does evidence play in shaping change? Second, what is the nature of the evidence base on rent regulation and the key messages that emerge from it? Third, how is this evidence base transmitted into policy debate? We take the example of the recent UK policy debate to examine this issue. The paper discusses the case of current policy development in Scotland to reflect upon policy drivers and the role of evidence. The paper combines insights from a recent evidence review and a decade-long policy ethnography. Not only does research indicate that evidence has played a limited role as a driver for policy change on rent regulation but the nature of the evidence base is such that there are limits on the guidance it can offer and the extent to which policy can in principle be rooted in evidence. The case of Scotland illustrates the forces at play in shaping rent regulation policy.
Article
Private rental markets have become increasingly important since the Global Financial Crisis 2008–2009 and rent controls are back on the political agenda. Yet, they have received less attention from housing scholars than homeownership and public housing. This paper presents new data on the development of private tenancy legislation based on a content-coding of rent control, protection of tenants from eviction, and rental housing rationing laws across more than 15 countries and 100 years. This long-run perspective allows for inquiring about the dynamic effects of rent control on the rise of homeownership as the dominant tenure during the twentieth century. We find that both rent regulation and rationing measures were followed by increases of homeownership and decreases of private rentals. We suggest that homeownership was not just produced by generous subsidies or the homeownership dream, but also through the push-effect of regulation crowding out rental units.
Article
Based on two surveys of 490 Californian cities and counties, the study examines the effects of local growth-control enactment between 1979 and 1988 on net housing construction between 1980 and 1990. It is shown that local growth-management measures significantly displaced new construction, particularly rental housing, possibly exacerbating the expansion of the metropolitan areas into the interiors of the state. Further, the measures impacted low-income households and minorities particularly. Not all growth-control measures were associated with this change. Measures which limited available land or which downsized existing zoning had stronger effects.
Article
Full-text available
 This study reexamines the conventional claims made by economists and policy analysts concerning modern rent control. We look at 76 New Jersey rent-controlled cities over a 30-year period and pose questions about the impact of rent control on rents, number of rooms, quality of units, and new rental construction using the latest Census data on cities for 2000. Our study is a comprehensive empirical study of rent control using multiple regression as the primary form of analysis. We find the intended impacts of New Jersey rent control over a 30-year period seem minimal when you compare cities with and without regulations. Housing activists and policymakers need to look at new kinds of approaches to address rental affordability problems.
Article
Full-text available
Examines the argument that short term, moderate rent controls cause a rapid decline in construction and taxable valuation of multifamily housing and an increase in demolitions. Argues that previous studies which support these conclusions used faulty methodological and statistical procedures. Attempts to avoid these problems by using regression techniques to study 26 rent controlled and 37 non-rent controlled cities in New Jersey. The results indicate that in the first five years following its establishment, moderate rent control had no significant effect on the amount of construction, the number of demolitions, or the total taxable value of apartments. -Author
Book
Reports on a survey to tenants in Los Angeles and Santa Monica looking at the tenant level of tenant consciousness.
Book
1: Introduction.- 2: The Scope of Spatial Econometrics.- 3: The Formal Expression of Spatial Effects.- 4: A Typology of Spatial Econometric Models.- 5: Spatial Stochastic Processes: Terminology and General Properties.- 6: The Maximum Likelihood Approach to Spatial Process Models.- 7: Alternative Approaches to Inference in Spatial Process Models.- 8: Spatial Dependence in Regression Error Terms.- 9: Spatial Heterogeneity.- 10: Models in Space and Time.- 11: Problem Areas in Estimation and Testing for Spatial Process Models.- 12: Operational Issues and Empirical Applications.- 13: Model Validation and Specification Tests in Spatial Econometric Models.- 14: Model Selection in Spatial Econometric Models.- 15: Conclusions.- References.
Article
With the increasing use of GIS by planners, statistics routines are needed that quantify relationships taking into account spatial location. Spatial statistics is a branch of statistics that includes measures of spatial distribution, spatial autocorrelation, and spatial association. Five statistics packages are described here that calculate various spatial indices useful for planners. In the future, planners will use these methods along with GIS programs to draw more rigorous and quantifiable deductions from their data.