Chapter 6: Past and Future Housing Shortfalls
California builders and developers produced 2.8 million new housing units between 1980 and 1997, one for every three new Californians. This was an amazing amount of production. Even so, it was not enough to meet the housing needs of the State's growing population. We estimate that between 1980 and 1989, statewide housing demand exceeded supply by more than 660,000 units—an amount equivalent to 6 percent of the State's housing inventory. The housing production gap disappeared between 1990 and 1994 as the recession depressed demand, but then re-appeared in 1995. Between 1995 and 1997, housing production in California lagged demand by 146,000 units.
California, it seems, chronically under-produces housing, especially in coastal markets. This chapter explores the nature and effects of California's inability to house itself. We begin by analyzing the spatial pattern of past production shortfalls. Next, we consider the effects of past production gaps on housing prices, rents, and housing cost burdens. Logic and economics suggest that when housing grows scarce and less affordable, households will respond by deferring homeownership, by commuting longer distances to access more affordable housing, and by doubling and tripling-up. To what extent have these housing problems actually occurred? And if they have, is there any real evidence that building more housing would help ameliorate them? We conclude prospectively, by comparing projected future housing needs with past production levels to see where and when California's future housing shortages are likely to occur.
Measuring Housing Shortfalls
The traditional tool for measuring housing shortages is vacancy rates. Low vacancy rates indicate a shortage of housing; high vacancy rates indicate a surplus. Vacancy rates are most useful for measuring the existing match between households and housing units. They are less useful for measuring unmet housing demands—that is, for measuring how many new housing units should have been constructed but were not. Nor do vacancy rates account for households who are unable to find housing in the neighborhood or community of their choice, and are displaced to other communities.
To account for these more basic factors, we developed a simple model of county-level housing demand for the 1980-89, 1990-94, and 1995-97 periods. Changes in household demand during a given period are calculated as the number of additional households attributable to job or population growth, plus the number of additional units required to replace any demolished units, plus the number of units required to achieve a "normal" 5 percent vacancy rate.1These demand estimates were then compared with total production, as measured from local building permits.
As an example, consider the Alameda County case shown in Exhibit 39. According to the California Employment Development Department, the Alameda County economy grew by 147,000 jobs between 1980 and 1989. To keep pace with job growth, the number of housing units in Alameda County should have expanded by 116,000 (assuming 1.26 jobs per house hold, the Alameda County average for the 1980s). An additional 3,300 housing units should have been constructed to replace units demolished between 1980 and 1989. Lastly, 3,800 additional units should have been constructed to bring the county's vacancy rate up to 5 percent from its 1980 level of 4.1 percent. Adding these three items together, Alameda County's housing stock should have increased by 123,400 units between 1980 and 1989.
On the supply side, Alameda city and county governments issued a total of 66,000 (new) residential building permits between 1980 and 1989. The difference between these numbers, 57,800 units, is the 10-year cumulative housing production shortfall. This shortfall was equivalent to 46.8 percent of total 1980-89 demand, or 11.5 percent of Alameda County's 1990 housing inventory. Either way, it was one of the largest shortfalls in the state.
Continuing into the 1990s, Alameda County lost 14,300 jobs between 1990 and 1994 as a result of the California recession.2 These losses notwithstanding, Alameda County's population continued to grow. Between 1990 and 1994, the county's population increased by 67,500 persons. At an average of 2.93 persons-per-household, an additional 23,000 housing units would have been needed to accommodate this increment of population growth. Adding in replacements for 641 demolished units and 614 units to bring the county vacancy rate up to 5 percent, housing demand in Alameda County increased by 24,270 units between 1990 and 1994. Note that this increase occurred despite a significant loss of jobs. On the supply side, a total of 13,238 new residential building permits were issued in Alameda County between 1990 and 1994, yielding a cumulative shortfall of 11,030 units. Using the same method, we estimated Alameda County's cumulative housing shortfall for the 1995-97 period at 10,720 housing units.
As of January 1998, Alameda County's cumulative production shortfall for the 1990-97 period stood at 21,750 units. This amount was equivalent to 45.5 percent of household growth and 4.1 percent of January 1998 inventory. If only in Alameda County, the net effect of California's recession was to slow the accrual of additional housing shortfalls. It did not turn a situation of shortages into one of surpluses.
The Geography of Housing Shortfalls: 1980-1997
We duplicated these shortfall calculations for all 58 California counties. The results are presented by county and urban area in Exhibit 40 and by county in Maps 10 and 11. For the state as a whole, we estimate that housing production lagged demand growth by 660,000 units between 1980 and 1989 and by 146,000 units between 1995 and 1997. From 1990 through 1994, thanks to the demand-depressing effects of the recession, statewide housing production actually exceeded demand growth by 275,000 units. Large in absolute terms, these estimates correspond to a percentage shortage of 5.9 percent between 1980 and 1989, and 1.5 percent between 1994 and 1997. The next three sections take a closer look at the pattern of shortfalls and surpluses.
- The 1980s: The 1980s was a period of large production shortfalls in the state's coastal housing markets and of surpluses in its inland ones. Each of the state's coastal metropolitan areas—including Los Angeles, the San Francisco Bay Area, and San Diego—suffered large housing shortfalls during the 1980s. In Los Angeles County alone, supply lagged demand by more than 325,000 units, an amount equivalent to 10 percent of LA County's 1990 housing stock. Elsewhere in Southern California, production lagged demand growth by 79,700 units in San Diego County, by 43,000 units in Orange County, and by 30,000 units in Ventura County.
Among Bay Area housing markets, production lagged demand growth by 75,000 units in the Oakland/East Bay MSA (Alameda and Contra Costa counties), by 40,000 units in the San Jose MSA (Santa Clara County), by 28,000 units in the San Francisco MSA (Marin, San Mateo, and San Francisco counties), by 16,600 units in the Vallejo-Fairfield MSA (Napa and Solano counties), and by 4,000 units in Sonoma County. Among counties along California's central coast, supply lagged demand by 15,300 units in Monterey County, by 11,639 units in Santa Barbara County, and by 5,500 units in Santa Cruz County. Production exceeded demand in San Luis Obispo and San Benito counties.
The shortfalls were all large in percentage as well as absolute terms. They amounted to 7.3 percent of the 1990 housing supply of the Greater Los Angeles Urban Area, to 7.6 percent of Bay Area housing units, to 8.4 percent of San Diego's housing stock, and to 6.2 percent of Central Coast housing units.
The sources of these shortfalls were different in different places. In Los Angeles, San Diego, Alameda, Contra Costa, and Sonoma counties, population growth simply overwhelmed the housing production system's ability to keep up. Limited land supplies and the difficulties associated with redevelopment constrained new home production in Monterey, Orange, San Francisco, and San Mateo counties. In Marin, Napa, Santa Barbara, Santa Cruz, and Ventura counties, housing production was limited by locally imposed development controls.
Inland, where potential development sites were more abundant, and where there were fewer development controls, the situation was far different. In the San Bernardino/Riverside MSA, for example, permits exceeded demand growth during the 1980s by 15,000 units. Permits also exceeded demand growth in the Sacramento (+3,000) and Fresno (+2,900) MSAs. Measured in percentage terms, these surpluses were fairly small: they amounted to 1.8 percent of 1990 housing stock in Riverside and San Bernardino counties, to .5 percent of the supply of housing in the Sacramento MSA, and to 1.1 percent of Fresno County housing units.
Not all inland markets were as elastic as these three. Because of "spill-over" demand from the Bay Area, production lagged demand by 9,200 units in San Joaquin County, and 4,100 units in Merced County.3 Production also lagged demand in Bakersfield, Visalia, Yuba, and Chico MSAs. Measured as a percentage of the housing stock, these shortfalls ranged from a high of 7.9 percent in the Yuba MSA, to 3.2 percent in Bakersfield-Kern County.
- 1990-1994: Except in Los Angeles and Orange counties, the incidence of cumulative shortfalls and surpluses between 1990 and 1994 mirrored the severity of the recession. Because of continued population growth—Southern California's dire economic straits notwithstanding4 —housing production in Los Angeles County between 1990 and 1994 actually lagged household demand by 78,700 units. In Orange County, 1990-94 housing production lagged demand by 13,600 units.
Elsewhere in Southern California, supply led demand growth. In San Diego County, 1990-94 housing production exceeded demand growth by 48,300 units. Even in perennially slow-growth, Ventura County, 1990-94 production exceeded demand growth by nearly 6,000 units.
The recession affected Bay Area housing markets less than those in Southern California. The hardest-hit Bay Area housing markets were those in Contra Costa, Sonoma, and Solano counties. Between 1990 and 1994, housing production exceeded demand growth by 33,400 units in Contra Costa County, by 7,700 units in Sonoma County, and by 7,200 units in Solano County. Elsewhere in the Bay Area, demand ran ahead of production.
If Bay Area housing markets were spared the worst of the recession's effects, those in the Central Valley and Inland Empire were not. Following the economy, housing demand throughout the Central Valley collapsed, even as projects already "in-the-pipeline" continued coming on-line. In Fresno County, 23,000 residential building permits were "pulled" between 1990 and 1994, despite an overall demand decline of 20,000 households. Elsewhere in the Central Valley, the demand fall-off was not so precipitous. The problem of not being able to turn off the production tap was even more severe in Riverside and San Bernardino counties. Nearly 85,000 new housing units became available in the two-county region between 1990 and 1994, even as demand was declining by the equivalent of 86,000 households.
- 1995-1997: As the recession's effects subsided and California's housing markets returned to health, production shortfalls again started accumulating—first in the San Francisco Bay Area, and then in southern California. Between 1995 and 1997, housing production lagged demand growth by 10,700 units in Santa Clara County, by 10,601 units in Alameda County, by 7,900 units in San Mateo County, and by 7,000 units in San Francisco County. (Although it would have to wait until 1998, Contra Costa County's housing recovery has been equally strong.) Measured as a share of the housing stock, Santa Clara County builders under-produced housing between 1995 and 1997 by 5.3 percent. Over the same period, builders in the San Francisco-San Mateo-Marin market under-produced new housing by 4.3 percent.
In Los Angeles County, 1995-97 new home production fell short of demand by 106,000 units, an amount equivalent to 5.5 percent of the county's housing inventory. In Orange County over the same period, home construction lagged demand growth by nearly 25,000 units, or by 4.1 percent as compared to inventory. Elsewhere in Southern California, excess supply would not be fully "worked-off" until 1998.
Housing demand continued to lag supply throughout the Central Valley until 1997, except in Kern, Sacramento and Yolo counties, which all began their housing recoveries in 1996. By the end of 1998, housing price levels throughout the Central Valley had returned to pre-recession levels. Led by Santa Barbara and Santa Cruz counties in 1995, and San Benito County in 1996, housing production shortfalls also started accruing among California's Central Coast markets.
Housing Shortfalls, Prices, And Rents
Recent Housing Price and Rent Trends
California housing prices and rents have long been among the nation's highest (see Exhibit 41). Common sense suggests that this is partly due to the state's inability to produce enough housing. Does the available evidence support this contention? To find out, we compared the percentage shortfalls listed in Exhibit 40 with various housing price and rent measures in 23 California counties:
- Housing Price Trends: There is, regrettably, no single reliable source of multi-year housing price data. The U.S. Census includes estimates of median home value (by state, county, jurisdiction, and tract); however, since this information is self-reported, it is widely regarded as unreliable. (It is nonetheless reported for comparison purposes.) More reliable and recent transaction-based estimates of average home prices 5 are available from the National Association of Homebuilders (new homes only), the California Association of Realtors (existing homes only), and a private firm, DataQuick. NAHB and CAR report their information by market area. DataQuick reports its information by county. Exhibit 42 lists DataQuick's estimates of average existing home prices for 23 California counties for 1990, 1995, and 1997. Adjusted for inflation, average existing home prices fell 21 percent between 1990 and 1995, but then increased 4 percent between 1995 and 1997.
Different regions and counties performed differently during different periods. Home prices increased much faster during the 1980s among coastal counties—especially Ventura, San Francisco, Alameda, Contra Costa, Santa Clara, and Napa—than among inland ones. Bay Area home prices declined less between 1990 and 1995 than elsewhere in the state, while home prices in the Los Angeles region and Sacramento County declined more. Home prices also recovered first in the Bay Area, led by fast-growing Santa Clara County. In southern California, the recovery in home prices started in Orange and Ventura counties in 1995 and San Diego in 1996, and then gradually moved eastward to Los Angeles, Riverside, and San Bernardino counties. Central Valley home prices didn't start recovering until 1998.
- Rent Levels: Apartment rents are even more difficult to track than home prices. Median rents are more indicative of general market conditions, whereas average rents tend to be more indicative of conditions facing tenants currently searching for units.6 The Census Bureau tracks median rents, but only every ten years. Several private and industry groups track average rents on an annual basis. For 1980 and 1990, we compared Census median rents. For 1995 and 1997, we used an average rent series published by RealFacts, a private real estate market research company.
Apartment rents tend to be much higher in California's coastal markets than in its inland ones. This was just as true in 1970 as it is today. Apartment rents also tend to be lower in markets with more apartment units, as competition between landlords serves to holds rents in check.
The 1991-94 recession altered the dynamics of the California apartment market, perhaps permanently. Although the recession's effects were much more moderate in the Bay Area than elsewhere, very few new apartment units were built in the Bay Area between 1990 and 1995. As a result, steady job growth in Silicon Valley quickly translated into a spreading wave of rising rents, first in Santa Clara County, then in San Francisco and San Mateo counties, and more recently in Alameda and Contra Costa counties.
Because southern California went into the recession with a much larger overhang of available apartment units, rents in southern California have not increased as much or as fast as in the Bay Area. In the Central Valley and Inland Empire, where apartment construction costs have always been much lower than in coastal areas, ample supplies and a weaker economic recovery have moderated recent rent increases.
Production Shortfalls, Housing Prices, and Rents: A Two-Test Statistical Analysis
We used two types of statistical tests to see whether housing production shortfalls track with price and rent levels: correlation analysis, and multivariate regression. Correlation analysis is used to compare whether two variables track together; multivariate regression can compare three or more variables.
- Correlation Analysis: Correlation coefficients vary between -1 and +1. A correlation coefficient value of -1 indicates a perfect negative correlation, 0 indicates no correlation, and +1 indicates a perfect positive correlation. All else being equal we would expect housing prices and rent levels to be strongly and positively correlated with production shortfalls—the bigger the shortfall, the higher the price or rent. Following the conventions established previously, production shortfalls were measured both as a percent of demand growth and as a percent of the housing inventory. The results of the correlation analysis are shown in Exhibit 43.
Among the 23 California counties analyzed, average 1990 existing home prices were strongly correlated with production shortfalls measured as a share of demand growth. They were less strongly correlated with shortfalls measured as a share of inventory. Average home prices in 1995 and 1997 were only slightly correlated with production shortfalls measured as a share of demand. They were more strongly correlated with shortfalls measured as a share of inventory. Housing price changes between 1990 and 1995 were unevenly correlated with supply shortfalls. The correlation between supply and price changes between 1995 and 1997 was more consistent, but not strong.
Median and average rents were moderately correlated with both measures of supply shortfall in 1990. Average asking rents in 1995 and 1997 were weakly correlated with shortfalls as a share of demand, but more strongly correlated with shortfalls as a share of inventory. Increases in median and average rents were moderately correlated with both supply shortfall measures for the 1980-89 and 1995-97 periods.
- Regression Analysis: Housing price and rent levels typically reflect other factors besides supply—income among them7 . A second technique, regression analysis, makes it possible to isolate the effects of production shortages on prices and rents by statistically controlling for the effects of income. Household income was measured using Census estimates of median household income at just one point in time, 1990. Regression equations were developed for housing price levels in 1990, 1995, and 1997; and for average and median apartment rents in 1990, 1995, and 1997. Additional models were developed to explain percentage changes in home prices and rents between 1980 and 1990, 1990 and 1995, and 1995 and 1997. Production shortfalls were measured as a percent of both demand growth and total inventory: All measurements were made at the county level. The regression results are presented in Appendix M.
We first consider home prices, then rents. Supply constraints affect median and average housing prices differently in different periods. Holding constant household income, average housing prices in 1990 were consistently higher in counties in which new supply most lagged demand growth. These included Los Angeles and Ventura in southern California, and Alameda, San Francisco, San Mateo, and Santa Clara in northern California. Average home prices in 1995 and 1997 were higher in counties in which production shortfalls were a larger share of inventory. Or, to put it more accurately, home prices in 1995 and 1997 were lower in counties with large production overhangs. These included Riverside and San Bernardino counties in Southern California, Contra Costa and Solano counties in northern California; and Fresno County in the Central Valley. Altogether, income and supply consistently explained 60 to 75 percent of average home prices, although the income effect was far stronger.
Between 1990 and 1995, housing prices decreased somewhat more in places with large housing surpluses, and somewhat less in places with smaller surpluses. Because of inconsistencies in the data, we were unable to evaluate these relationships statistically. Between 1995 and 1997, home price changes were insensitive to supply shortfalls but highly sensitive to income levels, as prices rose more in higher-income counties. Depending on the analysis period, income and supply shortfalls explained anywhere between 8 and 56 percent of county-level housing price changes from 1980 through 1997.
The effects of incomes and supply shortfalls on rent levels also vary over time. Rent levels in 1990, for example, were closely correlated with income levels, but not with supply conditions. As the apartment market started softening in 1990, the correlation between income levels and rents also softened: by 1995, income levels explained less than 50 percent of the variation in rents. Since 1995, average rents have risen disproportionately more in tighter, more supply-constrained markets, especially San Francisco, San Mateo and Santa Clara.
Changes in rent levels, like changes in home prices, are difficult to model at the county level. Between 1980 and 1990, rent levels increased more in upper-income counties than in lower ones, but the effect was inconsistent. The housing supply situation played almost no role. Since 1995, rent level changes have more closely reflected housing market conditions.
- Summary: Unfortunately but not unexpectedly, the results of the statistical models presented in Exhibit 43 and Appendix M say more about the inconsistencies of available data than they do about the consistencies of the housing market. Still, when all the results are deciphered, they suggest that supply shortfalls alone, and also in concert with incomes, are a significant determinant of housing price levels; and that supply-constrained housing markets are more expensive housing markets. This is especially true for owner-occupied housing. A more precise identification of the relationship between housing supply and prices will have to await further analysis.
Housing Prices And Cost Burdens
Housing Cost Burdens
Californians not only pay more for housing than residents of other states, housing costs also account for a greater share of Californian's incomes. The ratio of monthly housing cost to household income is known as cost burden. For renters, cost burden typically includes monthly contract rent plus utility payments. For owners, cost burden is typically calculated on the basis of monthly mortgage payments, property taxes and insurance, and utilities. Estimates of rent and ownership cost burden can be calculated from the American Housing Survey.
Nationwide, homeownership cost burdens have been fairly consistent during the last 20 years (Exhibit 44). Among all U.S. homeowners, the median burden rose from 15 percent in 1975 to 17 percent in 1985, mostly because of higher mortgage interest rates. Despite later interest rate declines, homeownership cost burdens have remained in the 16 to 17 percent range. Among recent movers8 —who are more affected by short-term housing prices and mortgage rate trends than non-overs—the median cost burden rose from 21 percent in 1975 to 23 percent in 1985 and 1989; and then declined to 21 percent in 1995. Among first-time homebuyers-recent movers, the median cost burden rose from rose from 22 percent in 1975 to 25 percent in 1985, and then declined to 23 percent in 1989, and to 21 percent in 1995.9 Continuing a long-term trend, renter cost burdens have increased significantly during the last 20 years. Nationally, renter cost burdens increased from 22 percent in 1975 to 27 percent in 1985 to 28 percent in 1995.
Housing cost burdens tend to be slightly higher in Sunbelt metropolitan areas than elsewhere. This is partly because the housing stock is newer and more expensive, and partly because housing turnover rates are higher. Excluding California, homeowner cost burdens among large southern and western metropolitan areas have consistently been 2 to 3 percentage points higher than in the rest of the country. Rent burdens among the same metropolitan areas have exactly matched the rest of the country.
Among California metropolitan areas ownership cost burdens are consistently two to five percentage points higher than for other large Sunbelt metropolitan areas and for the nation as a whole. The burden gaps, moreover, have been increasing over time. In 1975, for example, the median California homeowner paid 16 percent of their monthly income on housing costs, the same percentage as in the rest of the U.S. By 1995, the median Californian homeowner paid 22 percent of their monthly income on housing costs, compared to 17 percent for homeowners elsewhere in the United States.
Among owners-recent movers, the burden gap between California and the rest of the U.S. rose from two percent in 1975 to six percent in 1995. Among first-time-owners, recent movers, it rose from four percent in 1975 to ten percent in 1985, and then declined to four percent in 1995.
Among renters, the burden gap rose from two percent in 1975, to five percent in 1995. The median California renter living in a metropolitan area paid one-third of their income for rent in 1995; for the U.S. as a whole, the comparable figure was only 28 percent. Within California, sufficient observations are available to construct measures of housing cost burden for Los Angeles County in 1975, 1985, and 1995; and for the 1985-1995 period for Orange, San Bernardino/ Riverside, San Diego, and Santa Clara counties, as well as for the San Francisco-Oakland MSA.10 In all of these markets, homeownership cost burdens were consistently five to nine percentage points above national levels.
Among Los Angeles County homeowners, cost burdens rose from 16 percent in 1975 to 22 percent in 1995. Among the other California markets listed, homeowner burdens fluctuated by one to two percentage points between 1989 and 1995, with the largest gains occurring in Santa Clara County and in the San Bernardino/Riverside MSA.
Among owners/recent movers, homeownership cost burdens declined just about everywhere in California between 1989 and 1995, both as a result of lower interest rates, and of the weakness of the state's economy. The biggest decreases occurred in those markets which were most overheated in 1989, and/or in which the recession allowed supply to begin to catch up to demand: Orange County, San Diego, and San Francisco.
Among first-time owners/recent movers, homeownership burdens increased in the Los Angeles and San Bernardino/Riverside markets, probably as a result of declining incomes. Elsewhere, they either stayed constant or declined.
Renter cost burdens are also much higher in California. In Los Angeles County, for example, the median rent burden increased from 25 percent in 1975 to 39 percent in 1989, before declining to 33 percent in 1995. Among the other five California markets listed in Exhibit 44, median rent burdens ranged between 29 percent and 32 percent in 1989, and between 28 percent and 33 percent in 1995.
Housing Cost Burdens Among Low-Income Households
The problem of high housing costs falls disproportionately on low and very-low income renters. Low-income households are defined as those with incomes less than 80 percent of the MSA or county median-household income. Very-low-income households are those with incomes less than 50 percent of area median income.
In 1995, according to the American Housing Survey, 52 percent of California's 2.5 million low-income renter households paid more than half of their income for rent (Exhibit 45). Of the state's 1.6 million very-low-income renters, 72 percent paid more than half of their income for rent in 1995. Among all renters, the proportion that paid more than half of their income on rent in 1995 was 35 percent. Thirty-six percent of low-income homeowners devoted more than half of their incomes to housing costs.
Among very-low-income renters—the poorest of the poor—the problem of extreme burden worsened in all California's major metropolitan areas during the early 1990s. In Los Angeles County, for example, the number of very-low-income renters paying more than half of their income for rent increased from 312,000 in 1990 to more than 436,000 in 1995. In the San Francisco MSA (including San Francisco, San Mateo, and Marin counties), the number of very-low-income renters with extreme rent burdens nearly tripled between 1989 and 1993, rising from 55,349 to 149,712. In the San Jose MSA, the number of very-low-income renters with extreme rent burdens more than doubled, between 1988 and 1993, rising from 36,621 to 78,137.
Indeed, except for San Diego County, the number and/or percentage of low-income renters who paid more than half of their income for rent increased in all of California's major metropolitan areas during the early 1990s. Depending on the metropolitan area, between 40 and 60 percent of low-income renters faced an extreme rent burden problem in the early 1990s.
The incidence of extreme rent burdens is not just a function of high housing prices and rents. Measured in percentage terms, the problem of extreme rent burden was actually more severe in low-priced San Bernardino and Riverside counties than in high-priced Los Angeles and Orange counties.
The extent to which the problem of high rent burdens is one of too-little income versus not-enough supply varies by market and income range. Very-low-income renters have so little income that any increase in the supply of market-rate rental or ownership housing is unlikely to bring the cost of renting within reach.11 In moderately priced ownership-oriented markets, by contrast, large amounts of new construction may push the price of housing downward, thereby relieving ownership cost burdens.
Lack of supply clearly exacerbates rental burdens. As Exhibit 45 reveals, this was precisely what happened in San Francisco and San Jose MSAs during the early 1990s as low-income renters found themselves hard pressed to keep pace with rising rents.
Nor was the problem of rising cost burdens limited to renters. Among low-income homeowners, the number and/or percentage of households with extreme cost burdens increased significantly during the early 1990s in all of California's major metropolitan areas except Orange County.
The Costs Of Housing Scarcity
Lower Homeownership Rates
California's homeownership rate has always been below that of the nation; however, in recent years, the gap has widened (Exhibit 46). In 1975, the homeownership rate for California's major metropolitan areas stood at 57 percent, eight percentage points below the nation as a whole.12 By 1985, the gap between California metropolitan areas and the U.S. had widened to 13 percent. By 1995, the homeownership rate among major California metropolitan areas had further declined to 51 percent, while the national homeownership rate had increased to 65 percent.13
Such was not the case for individual counties or metropolitan areas. Between 1989 and 1995, homeownership rates declined marginally in Los Angeles and Santa Clara counties; increased marginally in San Diego County; and stayed about the same in Orange County. Among metropolitan areas, homeownership rates remained constant in the San Francisco-Oakland market between 1989 and 1993, and increased by three percentage points in the San Bernardino/Riverside market. Except for San Bernardino/Riverside, homeownership rates in individual California markets were well below those of the nation as whole.
Age of First Home Purchase
One response to rising home prices and declining affordability is to delay buying a home. Statistically, this response reveals itself as an increase in the age of first home purchase. Between 1985 and 1995, the national median age of first time homeownership increased from 29 to 31 years of age (Exhibit 47). Over the same period, and among large Sunbelt and western metropolitan areas, it rose from 29 to 33 years of age. Among California counties and metropolitan areas, the median age of first time homeownership increased slightly from a higher-than-national level of 31 years in 1985 to a near-national level of 32 years.
How much of these changes were due to rising housing costs and burdens? Probably very little, as it turns out. Between 1985 and 1995, the median age of first time homeownership rose just about everywhere, even in those markets in which housing cost burdens declined. Among the large California markets, housing cost burdens and median age of initial home purchase both increased only in Orange County and the San Bernardino/Riverside MSA. This suggests that the general aging of the population, not the cost of homeownership, accounts for recent increases in the median age of first time homeownership.
Housing prices in most American metropolitan areas conform to some form of declining price gradient. Prices are typically highest near job centers where land is expensive and homebuilding is difficult, and lowest at the urban fringe, where land is less expensive. Households unable to afford higher-priced homes must therefore commute longer distances to work, generating what has come to be called the problem of "jobs-housing balance."
Nationwide, median commute distances among homeowners increased from 7 miles in 1975 to 9 miles in 1985 to 11 miles in 1995 (Exhibit 48). At the same time, because of rising travel speeds, median commute times among homeowners have actually declined, from 22 minutes in 1975 to 17.5 minutes in 1995. (Because the distribution of commuting times and distances is typically skewed toward lengthier commutes, average commute distances and times are generally higher than median commute distances and times.) Recent movers and first time homeowners-recent movers have also benefited from the decline in commuting times, although not quite to the same extent as all homeowners.
Inside California, the story is far different. Among residents of metropolitan areas who recently purchased a home, median commute times increased from 20 minutes in 1985 to 25 minutes in 1995. Median commute distances increased from 12.5 to 17 miles during the same period. Among first-time owners/recent movers, median commute times increased from 20 minutes in 1985 to 31 minutes in 1995, while median commute distances increased from 9.5 to 17 miles. The problem of increased commute times is most burdensome for recent homebuyers. Among urban California renters, the median commute time did not change between 1985 and 1995. Renters, moreover, now take less time than recent homebuyers to commute to work.
Among residents of Los Angeles County—the only California county or metropolitan area with a large enough number of survey observations—median commute times increased slightly for recent movers between 1985 and 1995. For renters over the same period, they remained constant. Contrary to expectations, median commute times for Los Angeles County first time homebuyers actually declined between 1985 and 1995.
Exhibits 49a,b,c,d adds the dimension of housing cost burden to this analysis. It summarizes changes in commute times between 1985 and 1995 compared with changes in cost burden.
- Among all U.S. and California homeowners, median commutes were unchanged between 1985 and 1995, even as cost burdens declined (Exhibit 49a).
- Among U.S. homeowners who had moved within the previous year, median housing cost burdens and commute times both declined between 1985 and 1995 (Exhibit 49b). For their California counterparts, cost burdens declined, but commute times increasedùfrom about 20 minutes in 1985, to about 25 minutes in 1995.
- Among California first time homebuyers the increase in median commute times was even more dramatic: their average commute time rose from 20 minutes in 1985 to 31 minutes in 1995 (Exhibit 49c).
- Among California renters, median rent burdens increased from 29 percent in 1985 to 33 percent in 1995. Their average commute times, however, remained unchanged (Exhibit 49d).
These comparisons tell only part of the story. They do not, for instance, account for increases in the number of commuters. Because they (appropriately) focus on rent burdens, they also fail to capture the relationship between household income and commute times. These caveats aside, the results are still dramatic, especially for first time homebuyers. While housing cost burdens in 1995 were down slightly compared with 1985—mostly as a result of lower mortgage interest ratesùcommute times were way, way up.
Overcrowded housing units are those in which the ratio of persons-to-rooms exceeds 1.0. Severely overcrowded units are those in which the persons-per-room ratio exceeds 1.5. According to the Census Bureau, about a half-million California households lived in overcrowded conditions in 1980. By 1990, the number of overcrowded households had increased to 1.2 million, more than half of which were severely overcrowded. (see California's Housing Markets 1990-1997, Statewide Housing Plan Update, Department of Housing and Community Development, January 1999, for a much more detailed discussion of the severity and extent of overcrowding in California.)
While overcrowding increased for both owners and renters during the 1980s, renters were much more likely to live in overcrowded housing than owners. Overcrowding levels generally declined as renter incomes rose, particularly among family renters. The rate of overcrowding for very-low-income households (50 percent of median income) was nearly three times greater than for households with incomes over 95 percent of median. High overcrowding levels were geographically disbursed, including both metropolitan and non-metropolitan areas. More than four-fifths of extremely low-income residents of 17 counties lived in overcrowded conditions in 1990, including such wealthy counties as Santa Clara, San Mateo, and Orange. Overcrowding is particularly exacerbated where there is a mismatch between the number of large family households and the number of available family-sized housing units.
More recent and specific estimates of overcrowding are available in the American Housing Survey. Overall, overcrowding in California's major metropolitan areas increased by about 13 percent from 1989 to 1995, while severe overcrowding decreased modestly (-.7 percent). These overall figures mask important differences between owners and renters. Renter overcrowding increased by over 20 percent between 1989 and 1996, while severe overcrowding of renters increased by 7.2 percent. Among owners, overcrowding decreased by 6.7 percent and severe overcrowding decreased six percent.
Within California, overcrowding and changes in overcrowding vary widely by metropolitan area (see Exhibit 50). Particularly among renters, overcrowding and severe overcrowding was much more widespread—as well as increased more—in Los Angeles, Orange, and Riverside/San Bernardino counties. Compared to southern California, renter overcrowding was less severe in the San Francisco/Oakland and San Jose MSAs; although as Exhibit 50 shows, the San Jose MSA experienced a very large increase in overcrowding between 1988 and 1993. Only in San Diego County did overcrowding not increase significantly after 1990.
In Southern California, increases in overcrowding occurred mostly in suburban areas. In Northern California, by contrast, increases in overcrowding were concentrated in central cities. Regardless where they occurred, increases in overcrowding were accompanied by higher housing cost burdens, especially among renters. Throughout California, an increasing proportion and number of low- and moderate-income households paid more for the opportunity to live in more and more crowded conditions.
Would More Production Make California Housing More Affordable?
Housing prices and rents rise when there is too little construction, but is the opposite also true? Would building more homes and apartments really help restrain housing prices and rent increases? Would ramping up housing production in urban California really do much to address the state's pressing housing problems?
To answer these questions, we analyzed housing price and rent trends during the 1980s across 34 large metropolitan areas, including 20 California metropolitan areas. Specifically, we used multiple regression analysis to compare numerical and percentage changes in housing prices and rents with changes in construction activity, vacancy rates, and per capita incomes. The results of the various regression models are explained below and summarized in Appendix O. (The data are summarized in Appendix N.) For reasons of data availability, all of the models make use of metropolitan-scale measures; none consider the behavior of more-localized housing markets.
Estimating the Effects of Apartment Construction Activity on Rent Levels
Nationwide, inflation-adjusted (median) apartment contract rents rose from $304 per month in 1980 to $350 per month in 1990 (1989 dollars). Among a sample of 34 of the nation's largest metropolitan areas, inflation-adjusted median rents rose slightly more, from $357 in 1980 to $417 in 1990. Among the six California metropolitan areas for which data are available, inflation-adjusted median rents rose from $450 in 1980 to $615 in 1990. To measure the effects of apartment construction activity on rent increases, we divided the percentage change in the number of rental units (in each metropolitan area) between 1980 and 1990 by the percentage change in the number of households during the same period. To simplify things, we will refer to this ratio as the rental housing construction index. Among the same sample of 34 metropolitan areas, the average value of the rental housing construction ratio during the 1980s was 1.12. What this means is that the number of rental units in the sample MSAs grew twelve percent faster during the 1980s than the number of households. Among California MSAs, the average value of the 1980-90 rental housing construction ratio was .98
If economic theory is correct—that is, if new construction really serves to moderate rent increases—then we would expect the relationship between the rental housing construction index and rent growth to be negative; that is, for rents to fall in metropolitan areas with lots of rental housing construction. This was indeed the case for the 1980-90 period, as the results of Model 1 in Appendix O demonstrate. After statistically controlling for initial rents and vacancy rates, apartment rents declined by an average of $30 (per month) per 100% increase in the rental housing construction index. Altogether, these three factors—initial rent levels, initial vacancy rates, and the rental housing construction ratio—explained 62 percent of the inflation-adjusted change in median apartment rents.
New rental supply was important, but not that important. Of the three independent variables included in Model 1, vacancy rates were the most important (as expected, the lower the vacancy rate, the larger the rent increase), followed by initial rents (rents increased more in markets with initially higher rents), followed by the rental housing construction ratio.
Supply effects were comparatively more important among California rental markets (Model 2 in Appendix O). Altogether, initial rent levels and the apartment construction-to-household percentage change ratio explained 79 percent of the inflation-adjusted 1980-90 change in median apartment rents among a sample of 22 California counties. (The coefficient associated with initial vacancy rates was not statistically significant.). For every doubling of the rental housing construction ratio, inflation-adjusted apartment rents fell $122.
Estimating the Effects of Housing Construction on Price Levels
We begin our discussion of the relationship between housing construction and price levels with two caveats. The first is that the supply-price relationship is inherently more complicated for ownership housing than for rental housing. Compared to the rental housing market, the homeownership market is comprised of many more sub-markets, each with its own dynamics. Metropolitan scale measures such as median home prices provide only a cursory view of the homeownership market. Second, accurate, multi-year housing price data are much more difficult to come by than historical rent rates. Home value estimates as reported in Census documents are based on the owner's current estimate of value, not on actual transactions, and therefore tend to be biased. Instead of census data, we relied upon estimates of median existing home prices as reported by the National Association of Realtors.
As with apartments, we would expect increased rates of housing construction to moderate, and perhaps even reverse, housing price inflation. To see whether in fact this is the case, we used regression analysis to compare inflation-adjusted changes in (MSA-median) existing home prices between 1980 and 1990, and between 1990 and 1995 with: (i) initial housing price levels; (ii) initial ownership vacancy rates; and (iii) real changes in per capita income. To measure the effects of housing construction activity on price changes, we divided the percentage change in the total number of housing units (in each metropolitan area) between 1980 and 1990 by the percentage change in the number of households during the same period. To simplify things, we will refer to this ratio as the housing construction index. For reasons of data availability, we included a different measure of relative supply for the 1990-95 period: the ratio of the 1995 (ownership housing) vacancy rates to 1990 vacancy rates. As in the rental housing case above, we would expect a greater supply response—measured as an increase in housing units relative to households, or as an increase in vacancy rates—to be accompanied by lower housing prices.
The 1980-90 comparisons are presented in Appendix O as Model #3. Altogether, the three statistically significant variables—initial prices, per capita income change, and the housing construction index—explain 65 percent of the increase in existing home prices between 1980 and 1990. Based on the calculated "beta weights," initial prices were the most important of the three, followed by the housing construction index, and per capita income growth. As expected, housing prices rose more in markets with greater income growth and a lower supply response. For every doubling of the ownership housing construction index, median existing home prices fell by $40,808. Housing price inflation was not correlated with initial vacancy rates, at least among the MSA sample used to test Model #3.
Among the 32 metropolitan areas for which data were available, the four independent variables included in Model #4 (also presented in Appendix O), explain 87 percent of the change in housing prices between 1990 and 1995.
Its "goodness-of-fit" notwithstanding, the results of Model #4 are ambiguous. Three of the four independent variables—initial housing prices, initial vacancy rates, and changes in real per capita income—were found to be statistically significant; the vacancy rate change ratio was not. Of the three significant variables, initial housing prices were the most important, followed by the percentage change in per capita income, followed by initial vacancy rates. According to the results of Model #4, between 1990 and 1995, real housing prices declined in inexpensive markets, and rose in markets with low vacancy rates and rising per capita incomes. Supply responsiveness as measured as the change in vacancy rates, did not make a difference.
Simulating the Price and Rent Effects of Increased Construction Activity
Regression results can be difficult for non-statisticians to interpret. To gain a clearer sense of the effects of new supply on rents, we used the regression results to simulate the possible effects of a twenty percent increase in rental and ownership housing construction activity on median rent and price levels.
Among the national sample of 34 metro areas, increasing rental construction rates by 20 percent (over their actual 1980-90 levels) would have dampened median rent growth by an average of eleven percent. The more limited the initial supply (as measured by vacancy rates), the greater the dampening effect of additional supply on rents. The same relationship between increased rental supply and reduced rent is especially evident among a number of California markets, including Orange, Los Angeles, Ventura, Santa Clara, and Santa Barbara counties.
Why is the effect of additional new construction on rent levels not greater? Real world housing markets are much more diverse, and operate in much more complex ways than simple models like these are able to capture. This caveat notwithstanding, the model results confirm that rental housing in California in 1990 would have been considerably more affordable had rental construction levels more closely approximated the growth of renter households.
What of housing prices? Among California markets, a twenty percent increase in housing construction relative to household growth between 1980 and 1990 would have dampened median (existing home) price increases by between 20 and 50 percent, depending on the market.
Sometimes, statistical analysis does validate common sense. The considerable difficulties of developing statistical models of housing markets aside, these results suggest that California housing prices and rents do respond to increases in supply. Had housing production in California during the 1980s kept pace with household growth, housing prices and rents in 1990 would have been considerably lower, particularly in coastal markets. The effect of supply on price is especially strong during growth periods like the late-1980s and 1990s. During recessionary periods like the early 1990s, housing prices and rents track more closely with incomes than supply.
The Coming Housing Crunch?
So far in this chapter we have considered the incidence and effects of past housing shortages. What of the future? As difficult as it is to predict the future with precision, it is certainly possible to compare projections of the future with past experiences. Toward this end, Exhibit 51together with Maps 12 and 13 compare projected future household growth with past housing production levels. Results are reported by county, metropolitan area, and urban region. Household growth is projected to the year 2010; past production is estimated from permits and spans the 1987-1997 period. All estimates are annualized to facilitate comparisons.
Between 1987 and 1997, California builders produced an average of 138,761 new housing units each year. In order to meet projected 1997-2010 household demand growth, they would need to produce an average of 220,000 units year. The difference between these two estimates—or, the projected yearly deficit should past production trends continue into the future—is 94,300 units.
There is no reason to assume that the future will perfectly follow the past, but if it does, the counties which will experience the largest absolute housing shortages—calculated as the difference between 1997-2010 average annual projected household growth, and 1987-97 average annual housing production—are mostly in Southern California. If future housing production levels mirror those of the past, then Los Angeles County would suffer a net annual housing deficit of 28,000 units.14
Elsewhere, San Diego County's net annual housing deficit would exceed 7,640 units per year, and San Bernardino County's would exceed 6,600. Riverside County's net annual housing deficit would exceed 5,500 units and Orange County's would be nearly 4,800 units.
The two Bay Area counties with the largest prospective housing deficits are Santa Clara (5,900) and Alameda (2,800). The Central Valley counties with the largest projected deficits are Kern (3,900), Fresno (3,000), San Joaquin (2,800), and Stanislaus (2,200).15
Comparing metropolitan regions, the Los Angeles Metropolitan Region would suffer an average yearly deficit of 48,400 units. Annual production deficits in the San Joaquin Metropolitan Region would approach 15,000 units a year. In the Bay Area, production deficits could exceed 12,000 units per year (see Exhibit 51).
The forces and factors behind these deficits will be different in different places. The challenge in Southern California will be for developers to keep pace with the region's potentially explosive population growth. For that to happen, there will need to be significant levels of private redevelopment in Los Angeles and Orange counties. Moreover, given the region's rapidly changing demographics, Southern California builders will need to become more adept at delivering a broader array of product types.
In the Bay Area, where growth pressures will be more moderate, the challenge facing politicians in fast-growing counties like Santa Clara, Alameda, Contra Costa, and Sonoma will be to zone or otherwise entitle sufficient land to accommodate projected household growth, while also responding to rising demands for growth containment. In order to accommodate projected household growth, planners, elected officials, and builders will have to figure out how to promote neighborhood-friendly redevelopment.
In the Bay Area, where growth pressures will be more moderate, the challenge facing politicians in fast-growing counties like Santa Clara, Alameda, Contra Costa, and Sonoma will be to zone or otherwise entitle sufficient land to accommodate projected household growth, while also responding to rising demands for growth containment.
In the San Joaquin Valley, agriculture will likely be the limiting factor. Cities like Stockton, Modesto, Merced, and Fresno are expanding and will continue to expand onto prime agricultural lands. In the right forms, agricultural production and residential development should certainly be able to co-exist. The question for San Joaquin Valley planners, developers, and elected officials, therefore, is how the planning system should be changed to promote those forms.
Elsewhere, the ability to accommodate projected population and household growth will turn on the ability of the residential approvals process to respond to individual project proposals in ways which are consistent with local environmental and fiscal preferences, and which are saleable (or rentable) in the private housing market.
- Hard-Pressed Developers, Builders: Judging from recent experience, California's residential builders and developers will be hard-pressed to meet the State's projected housing growth needs. California chronically under-produces housing, particularly in coastal markets. As of 1990, the last year before the recession, statewide housing demand exceeded supply by more than 660,000 units, an amount equivalent to six percent of the state's housing inventory. Among large urban counties, production shortfalls (as a share of inventory) ranged between 10 percent (Los Angeles) and -2 percent (Stanislaus), depending on demand and supply constraints. The shortfall gap disappeared during the first-half of the 1990s, but then reappeared in 1995.
- Housing Supply vs. Price Complicated: In theory, housing prices and apartment rents should be higher (and increase faster) in supply-constrained markets. While this is generally true, in practice, the link between housing supply and price is much more complicated. Measured in real dollar terms, housing prices increased more and more rapidly between 1980 and 1990 in supply-constrained coastal markets than in growth-oriented inland markets. Median apartment rents increased more between 1980 and 1990 in smaller and supply-constrained markets than in larger markets, although the difference was not as large as for home prices.
- Housing Price Change: Housing price changes during the 1991-94 recession were more related to employment changes and income levels than housing supply. Since 1995, housing prices and rents have risen more and faster in supply-constrained markets. Housing price and rent increases have gradually emanated outward from supply-constrained/high-growth areas like Santa Clara, San Francisco, and Orange counties to other parts of their respective regions.
- Homeownership Cost Burdens: Californians not only pay more for housing than residents of other states, housing costs also account for a greater share of Californians' incomes. Depending on the market, homeownership cost burdens among California households are two to five percentage points higher than for the nation as a whole. Nationwide, homeownership cost burdens have been fairly consistent over the last 20 years; among California households, particularly recent movers and first time homebuyers, they have consistently risen. Housing cost burdens for California renters are also two to five percentage points higher than for renters elsewhere.
- Gap Has Widened: California's homeownership rate has always been below that of the nation; however, in recent years the gap has widened.
- Reactions to Rising Costs: The usual way that renters and prospective homebuyers respond to rising housing costs is to move to a less expensive location. As a result, homeownership rates are not significantly lower, and housing price and rental cost burdens are not significantly higher in supply-constrained markets than in markets where housing is more plentiful.
- Longer Commutes: The flip side of this bargain is longer commutes. While median and average commuting times for homeowners outside of California have stayed constant or actually declined, inside California, commute times have increased significantly. Among first time owners-recent movers (the group most burdened by rising housing costs), median commute times in California increased from 20 minutes in 1985 to 31 minutes in 1995. Renters have fared far better than homebuyers: their median commute time did not change between 1985 and 1995.
- Housing Overcrowding Increasing: Housing overcrowding is an increasing problem, particularly in Los Angeles, Orange, San Francisco, and Santa Clara counties. (In relative terms, it is an even bigger problem in many rural agricultural communities). Overcrowding is not due to supply constraints, at least not directly. Overcrowding in California is most severe in rental markets with large (and growing) moderate-income and working populations, where high development costs make it impractical to build moderate-income housing.
- More Housing a Solution? Would more housing supply help solve these problems? Based on an analysis of housing price and rent trends in large metropolitan areas around the country, the answer to this question is yes. Nationally as well as in California, housing prices and rents have increased less in markets where housing supplies have kept pace with population and household growth. Among 34 large U.S. metropolitan areas, for every doubling of the ratio of new home construction to household growth between 1980 and 1990, median home prices fell by $40,808.
- Under-production Likely: Unless things change, California will continue to under-produce new housing, further exacerbating many of the problems discussed above. Each year between 1987 and 1997—a not atypical boom-bust cycle—California builders and developers produced an average of 141,000 single-family homes and apartments. In order to meet projected household demand through the year 2020, they will have to increase average annual production levels to more than 220,000 units.
California Employment Development Department. 1998. County Employment Estimates. Sacramento.
California Department of Finance. 1999. Estimates of County Population Change: 1970-1998. Sacramento.
Construction Industry Research Board. 1998. Building Permit Tabulations by City and County. Burbank.
DataQuick. 1998. California Home Price Statistics. La Jolla, California.
National Association of Realtors. Various years. Monthly Reports. Washington, D.C.
Michael Smith-Heimer. 1989. The Potential for Filtering as Public Policy. Berkeley Planning Journal. University of California at Berkeley. 94-104.
U.S. Census Bureau. Various years. C-30: Housing Construction and Demolition Reports. Washington, D.C.
U.S. Census Bureau. 1982, 1992. Census of Population and Housing. Washington, D.C.
U.S. Department of Housing and Urban Development/U.S. Census Bureau. Various years. American Housing Survey. Washington, D.C.
- If the initial vacancy rate is below the normal level, additional units are required; if it is above the normal level, fewer units are required.
- Los Angeles County, by contrast, lost nearly 300,000 jobs during the same period.
- Spillover demand refers to population and household growth which spills-over from one economic region (e.g., the San Francisco Bay Area) to another (e.g., the Central Valley). Spillover demand typically results from an inter-regional imbalance between job growth and affordable housing opportunities.
- The southern California economy lost more than a quarter-million jobs between 1990 and 1993.
- Average home sales prices tend to be much higher than median home sales prices.
- As is the case for home prices, average rents tend to be considerably higher than median rents.
- The relationship between income and housing price levels is two-way: the higher a community's income level, the more expensive the housing its residents can afford. Conversely, the more expensive the housing in a community, the less affordable it is to households with lower incomes.
- Recent movers are households that moved within the prior year.
- Recent mover-first-time homebuyers tend to be younger and have lower incomes than households who have previously owned a home. This serves to make their cost burdens higher.
- The American Housing Survey collects metropolitan area information on a staggered schedule. In Orange County, for example, the AHS was administered in 1988 and 1994; in San Jose, it was administered in 1988 and 1993. For comparison purposes, we report 1987 through 1990 survey results for 1989, and 1993-94 survey results for 1995.
- The theory of filtering suggests that an increase in supply should, sooner or later, result in a reduction in prices or rents at the bottom of the income spectrum. Smith-Heimer's analysis of filtering (1990) suggests that this is rarely the case in growing markets.
- Homeownership rates in California counties and metropolitan areas are generally, although not uniformly, related to housing prices and rents. All else being equal, homeownership rates are generally lower in more expensive markets (e.g., San Francisco-Oakland, San Diego, and Los Angeles) and higher in less expensive markets (e.g., Riverside San Bernardino, Fresno, Sacramento).
- These comparisons are based on American Housing Survey (AHS) sample data. Because the AHS over-samples large metropolitan areas, and under-samples smaller ones and rural areas, it tends to underestimate homeownership rates. The 1990 Census, by comparison, reported that the homeownership rate for the state was 57%.
- Note that these estimates do not include current shortfalls or surpluses.
- The counties likely to experience the largest percentage housing production deficitsùcalculated as the absolute deficit divided by average annual household growthùare mostly rural; they are: Colusa (-78.5 percent), Glenn (-75.7 percent), Santa Cruz (-70.9 percent), Imperial (-68.0 percent), and Modoc (-62.8%). Large percentage shortfalls in small and rural housing markets are mostly an artifact of the method used to calculate shortfalls, and are not generally worrisome.
Copyright © 2000. All rights reserved.