Is There a Bubble in the Housing Market?
Journal article by Karl E. Case, Robert J. Shiller; Brookings Papers on Economic Activity, No. 2, 2003
Journal Article Excerpt
Is There a Bubble in the Housing Market? by Karl E. Case , Robert J. Shiller THE POPULAR PRESS is full of speculation that the United States, as well as other countries, is in a "housing bubble" that is about to burst. Barrons, Money magazine, and The Economist have all run recent feature stories about the irrational run-up in home prices and the potential for a crash. The Economist has published a series of articles with titles like "Castles in Hot Air," "House of Cards," "Bubble Trouble," and "Betting the House." These accounts have necessarily raised concerns among the general public. But how do we know if the housing market is in a bubble? The term "bubble" is widely used but rarely clearly defined. We believe that in its widespread use the term refers to a situation in which excessive public expectations of future price increases cause prices to be temporarily elevated. During a housing price bubble, homebuyers think that a home that they would normally consider too expensive for them is now an acceptable purchase because they will be compensated by significant further price increases. They will not need to save as much as they otherwise might, because they expect the increased value of their home to do the saving for them. First-time homebuyers may also worry during a housing bubble that if they do not buy now, they will not be able to afford a home later. Furthermore, the expectation of large price increases may have a strong impact on demand if people think that home prices are very unlikely to fall, and certainly not likely to fall for long, so that there is little perceived risk associated with an investment in a home. If expectations of rapid and steady future price increases are important motivating factors for buyers, then home prices are inherently unstable. Prices cannot go up rapidly forever, and when people perceive that prices have stopped going up, this support for their acceptance of high home prices could break down. Prices could then fall as a result of diminished demand: the bubble bursts. At least one aspect of a housing bubble--the rapid price increases--has clearly been seen recently. A rapid surge in home prices after 2000, as tabulated, for example, by the Economist Intelligence Service, has been seen in almost all the advanced economies of the world, with the exception of Germany and Japan. In some of these countries, price-to-rental ratios and price-to-average income ratios are at levels not seen since their data begin in 1975. (1) But the mere fact of rapid price increases is not in itself conclusive evidence of a bubble. The basic questions that still must be answered are whether expectations of large future price increases are sustaining the market, whether these expectations are salient enough to generate anxieties among potential homebuyers, and whether there is sufficient confidence in such expectations to motivate action. In addition, changes in fundamentals may explain much of the increase. As we will show, income growth alone explains the pattern of recent home price increases in most states. Falling interest rates clearly explain much of the recent run-up nationally; they can also explain some of the cross-state variation in appreciation because of differences in the elasticities of supply of homes, including land. To shed light on whether the current boom is a bubble and whether it is likely to burst or deflate, we present two pieces of new evidence. First, we analyze U.S. state-level data on home prices and the "fundamentals," including income, over a period of seventy-one quarters from 1985 to 2002. Second, we present the results of a new questionnaire survey conducted in 2003 of people who bought homes in 2002 in four metropolitan areas: Los Angeles, San Francisco, Boston, and Milwaukee. The survey replicates one we did in these same metropolitan areas in 1988, during another purported housing bubble, after which prices did indeed fall sharply in many cities. The results of the new survey thus allow comparison of the present situation with that one. Our survey also allows us to compare metropolitan areas that have reputedly gone through a bubble recently (Los Angeles, San Francisco, and Boston) with one that has not (Milwaukee). The notion of a bubble is really defined in terms of people's thinking: their expectations about future price increases, their theories about the risk of falling prices, and their worries about being priced out of the housing market in the future if they do not buy. Economists rarely ask people what they are thinking when they make economic decisions, and some economists have argued that one should never do so. (2) We disagree. If questions are carefully worded and people are surveyed at a time close to their making an actual economic decision, then by making comparisons across time and economic circumstances, we can learn about how the decisions are made. (3) On the Origin of the Term "Housing Bubble" There is very little agreement about housing bubbles. In fact, the widespread use of the term "housing bubble" is itself quite new. Figure 1 shows a monthly count since 1980 of stories incorporating the words "housing bubble" in major newspapers in the English language around the world, as tabulated using Lexis-Nexis. (The data in years before 2003 are rescaled to account for the smaller coverage of Lexis-Nexis in earlier years.) The term "housing bubble" had virtually no currency until 2002, when its use suddenly increased dramatically, even though the run-up in real estate prices in the 1980s was as big as that since 1995. The peak in usage of "housing bubble" occurred in October 2002. The only real evidence of its currency before 2002 is a few uses of the term just after the stock market crash of 1987, but that usage quickly died out. [FIGURE 1 OMITTED] The term "housing boom" has appeared much more frequently since 1980. As figure 1 also shows, the use of this term was fairly steady from 1980 through 2001, although it, too, took off in 2002, also peaking in October. The term "boom" is much more neutral than "bubble" and suggests that the rise in prices may be an opportunity for investors. In contrast, the term "bubble" connotes a negative judgment on the phenomenon, an opinion that price levels cannot be sustained. Perhaps journalists are shy about using the word "bubble" except after some salient public event that legitimizes the possibility, such as the stock market crash of 1987 or that after 2000. The question is whether such journalistic use of the term also infects the thinking of homebuyers: do homebuyers think that they are in a bubble? The Previous "Housing Bubble" The period of the 1980s and the declines in housing prices in many cities in the early 1990s are now widely looked back upon as an example, even a model, of a boom cycle that led to a bust. A pattern of sharp price increases, with a peak around 1990 followed by a decline in many important cities around the world, including Boston, Los Angeles, London, Sydney, and Tokyo, looks consistent with a bubble. Housing prices began rising rapidly in Boston in 1984. In 1985 alone, home prices in the Boston metropolitan area went up 39 percent. In a 1986 paper, Case constructed repeat-sales indexes to measure the extent of the boom in constant-quality home prices. (4) The same paper reported that a structural supply-and-demand model, which explained home price movements over ten years and across ten cities, failed to explain what was going on in Boston. The model predicted that income growth, employment growth, interest rates, construction costs, and other fundamentals should have pushed Boston housing prices up by about 15 percent. Instead, they went up over 140 percent before topping out in 1988. The paper ended with the conjecture that the boom was at least in part a bubble. The following year we described price changes by constructing a set of repeat-sales indexes from large databases of transactions in Atlanta, Chicago, Dallas, and San Francisco. (5) We used these indexes in a subsequent paper to provide evidence of positive serial correlation in the changes in real home prices. (6) In fact, that paper showed that a change in price observed over one year tends to be followed by a change in the same direction the following year between 25 and 50 percent as large. The paper found evidence of inertia in excess returns as well. This strong serial correlation of price changes is certainly consistent with our expectation of a bubble. (7) During the 1980s, spectacular home price booms in California and the Northeast helped stimulate the underlying economy on the way up, but they ultimately encountered a substantial drop in demand in the late 1980s and contributed significantly to severe regional recessions in the early 1990s. The end of the 1980s boom led to sharp price declines in some, but not all, cities. Since 1995, U.S. housing prices have been rising faster than incomes and faster than other prices in virtually every metropolitan area. Despite the fact that the economy was in recession from March to November of 2001, and despite the loss of nearly 3 million jobs since 2000, prices of single-family homes, the volume of existing home sales, and the number of housing starts in the United States have remained at near-record levels. There can be no doubt that the housing market and spending related to housing sales have kept the U.S. economy growing and have prevented a double-dip recession since 2001. The big question is whether there is reason to think that such a run-up in prices will be followed by a similar or even worse decline than the last time. To answer this question, we need to try to understand better the causes of these large movements in the housing market. Home Prices and the Fundamentals, 1985-2002 A fundamental issue to consider when judging the plausibility of bubble theories is the stability of the relationship between income and other fundamentals and home prices over time and space. Here we look at the relationship between home price and personal income per capita and a number of other variables by state, using quarterly data from 1985:1 to 2002:3. The data contain 3,621 observations covering all fifty states and the District of Columbia. (8) Measures of Home Prices The series of home values was constructed from repeat-sales price indexes applied to the 2000 census median values by state. Case-Shiller (CS) weighted repeat-sales indexes constructed by Fiserv CSW Inc. are available for sixteen states. (9) In addition, the Office of Federal Housing Enterprise Oversight (OFHEO) makes state-level repeat-value indexes produced by Fannie Mae and Freddie Mac available for all states. The Case-Shiller indexes are the best available for our purposes, and wherever possible we use them. Although OFHEO uses a similar index construction methodology (the weighted repeat-sales method of Case and Shiller), (10) their indexes are in part based on appraisals rather than exclusively on arm's-length transactions. CS indexes use controls, to the extent possible, for changes in property characteristics, and it can be shown that they pick up turns in price direction earlier and more accurately than do the OFHEO indexes. Nonetheless, for capturing broad movements over long periods, the indexes tend to track each other quite well, and OFHEO indexes are used in most states to achieve broader coverage. The panel on home prices was constructed as follows for each state: (1) [V.sup.t.sub.i] = [V.sup.1999:1.sub.i] [I.sup.t.sub.i], where [V.sup.t.sub.i] = adjusted median home value in state i at time t [V.sup.1999:1.sub.i] = median value of owner-occupied homes in state i in 1999:1 [I.sup.t.sub.i] = weighted repeat-sales price index for state i at time t, 1999:1 = 1.0. The baseline figures for state-level median home prices are based on owner estimates in the 2000 census. A number of studies have attempted to measure the bias in such estimates. The estimates range from -2 percent to +6 percent. (11) Measures of the Fundamentals Data on personal income per capita by state are available from the Bureau of Economic Analysis website. The series is a consistent time series produced on a timely (monthly) schedule. Population figures by state are not easy to obtain on a quarterly basis. The most carefully constructed series that we could find was put together by Economy.com (formerly Regional Financial Associates). The most stable and reliable measure of employment at the state level is the nonfarm payroll employment series from the Bureau of Labor Statistics (BLS) Establishment Survey, which is available monthly, and which we have converted to quarterly data. The unemployment rate by state is available monthly from the BLS as part of its Household Survey. Data on housing starts are not generally available by state before 1995. The series used here was produced by Economy.com based on the historical relationship between permits and starts and a proprietary data base on permits. Data on average mortgage interest rates on thirty-year fixed rate mortgages, assuming payment of 2 points (2 percent of the loan value) and an 80 percent loan-to-value ratio, are available from Fannie Mae. For each quarter the ratio of income to mortgage payment per $1,000 borrowed was calculated by dividing annual income per capita by twelve (to convert it to monthly) and then dividing by the monthly mortgage payment per $1,000 of loan value for a thirty-year fixed rate with 2 points. Home Prices and Income: A First Look Table 1 presents ratios of home price to annual income per capita for the eight states where prices have been most volatile and the seven states where they have been least volatile. The least volatile states exhibit remarkable stability and very low ratios. The ratio for Wisconsin, for example, a state that we will explore at some length later, remains between 2.1 and 2.4 for the entire eighteen years of our sample. A simple regression of home prices on income per capita in Wisconsin generates an [R.sup.2] of 0.99. On the other hand, the eight most volatile states exhibit equally remarkable instability. Connecticut's ratio, for example, varies between 4.5 and 7.8, and we find that income explains only 45 percent of the variation in home prices. Table 2 shows the variation for all fifty states and the District of Columbia. Glancing down the table reveals that forty-three of the fifty-one observations have a standard deviation below 0.41, whereas only those eight states listed in table 1 as most volatile have standard deviations above 0.41. These calculations reveal that states seem to fall into one of two categories. In the vast majority of states, prices move very much in line with income. But in New England, New York, New Jersey, California, and Hawaii, prices are significantly more volatile. Plots of the ratio of price to income per capita for the states of California, Massachusetts, and Wisconsin (figure 2) show clearly that the pattern of variation is anything but a random walk. In California and Massachusetts the pattern is one of a long inertial upswing followed by a long inertial downturn followed by another rise that has now lasted several years. In Wisconsin the ratio is much smaller and remarkably stable. [FIGURE 2 OMITTED] We conclude that whereas income alone almost completely explains home price increases in the vast majority of states, about eight states are characterized by large swings in home prices that exhibit strong inertia and cannot be well explained by income patterns. Home Prices and Other Fundamentals To explore the relationship between housing prices and other fundamental variables, we performed linear and log-linear reduced-form regressions with three dependent variables: the level of home prices, the quarter-to-quarter change in home prices, and the price-to-income ratio described above. The results for the linear versions of these regressions are given in tables 1 and 3; the results for the log-linear regressions are similar. In those states where income and home prices are very highly correlated, the addition of mortgage rates, housing starts, employment, and unemployment to the regression added little explanatory power. However, for the eight states where income is a less powerful predictor of home prices, the addition of changes in population, changes in employment, the mortgage rate, unemployment, housing starts, and the ratio of income to mortgage payment per $1,000 borrowed added significantly to the [R.sup.2] (table 1). Table 3 reports the pattern of significant coefficients for three sets of regressions on data from the eight states where price-to-income ratios are most volatile. Since the equations are in reduced form, the individual coefficients are plagued by simultaneity. For example, housing starts may proxy for supply restrictions. That is, where supply is restricted, starts may be low, pushing up prices. On the other hand, builders clearly respond to higher prices by building more. Similarly, the change in employment could have a positive impact on home prices as a proxy for demand. On the other hand, rising home prices have been shown to have a negative effect on employment growth in a state by making it difficult to attract employees to a region with high housing costs. (12) In the equations in which the change in price is the dependent variable (top panel of the table), the number of housing starts has a positive and significant coefficient in seven of the eight states. However, in equations in which the price level is the dependent variable (middle panel), which are estimated over a shorter time horizon (1985:2 through 1999:4), housing starts has a significant but negative coefficient in five of the eight states. Income has a significant and positive coefficient in twenty of the twenty-four equations presented. The change in employment had a significant and negative effect in fourteen of the twenty-four equations. Unemployment has a significant and negative coefficient in the price level equations in five of the eight states. Of interest is the fact that the mortgage rate has an insignificant coefficient in all but one of the regressions presented. This again could be the result of simultaneity: low rates stimulate the housing market, but low rates may be caused by Federal Reserve easing in response to a weak economy and housing market. Including the ratio of income to mortgage payment in the regression allows us to take account of the wide swings in interest rates over this period. During 2000-02, the combination of low interest rates and high incomes made housing more affordable. Although this variable had a positive and significant sign in the equations run on all quarters in twenty-one states, it was significant and positive only in New York among the eight states with a high variance of income to home price. To look more closely at the strength of the housing sector since the stock market crash of 2000-01 and the recession of 2001, we used the results from the price level equation estimated with 1985:2-1999:4 data, described above, to forecast the level of home prices for the period from 2000:1 through 2002:3. We did the same exercise with two sets of regressions described in the bottom two panels of table 3. The results from the middle panel of table 3 are presented in figure 3. In all of the eight states except Hawaii, the fundamentals significantly underforecast the actual behavior of home prices since 1999. Diagrams constructed from the results of the bottom panel of table 3 look exactly the same. [FIGURE 3 OMITTED] To conclude this section, we find that income alone explains patterns of home price changes since 1985 in all but eight states. In these states the addition of other fundamental variables adds explanatory power, but the pattern of smoothly rising and falling price-to-income ratios and the consistent pattern of underforecasting of home prices during 2000-02 mean that we cannot reject the hypothesis that a bubble exists in these states. For further evidence we turn to our survey. The 1988 Survey In our 1988 paper we presented the results of a survey of a sample of 2,000 households who bought homes in May 1988 in four markets: Orange County, California (suburban Los Angeles); Alameda County, California (suburban San Francisco); Middlesex County, Massachusetts (suburban Boston); and Milwaukee County, Wisconsin. (13) The four locations were chosen to represent hot (California), cooling (Boston), and steady (Milwaukee) markets. The survey was inspired in part by an article on page 1 of the June 1, 1988, Wall Street Journal, which described the current "frenzy in California's big single family home market" and included colorful stories of angst and activity in the housing market there. (14) We wanted to find out what was going on in California and compare it with other places in a systematic way. The results of that survey provide strong evidence for some parameters of a theory that a housing bubble did exist in 1988: that buyers were influenced by an investment motive, that they had strong expectations about future price changes in their housing markets, and that they perceived little risk. Responses to a number of questions revealed that emotion and casual word of mouth played a significant role in home purchase decisions. In addition, there was no agreement among buyers about the causes of recent home price movements and no cogent analysis of the fundamentals. One additional finding in our 1988 paper lends support to an important stylized fact about the U.S. housing market that has not been well documented in the literature, namely, that home prices are sticky downward. That is, when excess supply occurs, prices do not immediately fall to clear the market. Rather, sellers have reservation prices below which they tend not to sell. This tendency not to accept price declines is connected with a belief that prices never do decline, and with some of the parameters of thinking that underlie a housing bubble. Homebuyer Behavior in Four Metropolitan Areas, 1988 and 2003 Before we present the results of a virtually identical survey done in 2003, we describe home price behavior in the four survey areas. Although the timing was not identical, Los Angeles, San Francisco, and Boston have experienced two boom cycles and a bust in housing prices over the last twenty years. Table 4 describes the timing and the extent of these cycles, which are also shown in nominal terms in figure 4. [FIGURE 4 OMITTED] The first boom in California was similar in Los Angeles and San Francisco. Prices in both metropolitan areas peaked in the second quarter of 1990 after a 125 percent nominal (55 percent real) run-up, which began slowly, gradually accelerated into 1988, and then slowed as it approached the peak. The first boom in Boston was also similar, but it accelerated earlier and actually peaked in the third quarter of 1988 after a 143 percent nominal (more than 100 percent real) increase. The bust that followed was most severe and longest lived in Los Angeles, where prices dropped 29 percent in nominal terms (40 percent in real terms) from the peak to a trough in the first quarter of 1996. Prices in San Francisco dropped only 14 percent (20 percent real) from the 1990 peak and began rising again in the first quarter of 1993, three years earlier than in Los Angeles. Boston was on the mend two years earlier than that. All three metropolitan areas have seen a prolonged boom ever since, although San Francisco has shown some volatility since mid-2002. Home prices during this boom rose 129 percent in nominal terms in San Francisco, 94 percent in Los Angeles, and 126 percent in Boston, despite very low overall inflation. At the time participants in the second survey sample were buying their homes, prices were still rising in all four metropolitan areas. The price index for Milwaukee could not be more different. It shows a very steady climb at a rate of 5.6 percent annually, essentially the same rate of growth as income per capita. Interestingly, over the entire cycle, Milwaukee did about as well as Los Angeles, but not as well as Boston or San Francisco. Home prices in Boston increased more than fivefold in nominal terms over the cycle, while prices in San Francisco quadrupled and prices in both Milwaukee and Los Angeles tripled. Three of the four metropolitan areas--Los Angeles, San Francisco, and Boston--show pronounced cycles. These three might be called glamour cities, in that they are the home of either international celebrities, or the entertainment industry, or world-class universities, or high-technology industries, and the prices of homes in these metropolitan areas are high as well as volatile. (15) Table 5 looks at the latest boom cycle in a bit more detail. Using the state data described in the earlier section, the table makes two points. First, in all three states, home price increases outpaced income growth. (Note that the price increases are not as great as in the metropolitan area data because the indexes are for the entire state.) All three states had increases in their ratios of home price to annual income, but the changes were dramatically larger in the boom-and-bust states. After peaking at nearly 10 percent in early 1995, the thirty-year fixed rate dropped below 7 percent by mid-1999. During 2000 rates spiked back to 8.5 percent but then fell steadily from mid-2000 until 2003, when they briefly went below 5 percent. Table 5 also shows the effect of declining mortgage rates on the cash costs of buying a home. In 1995, at the beginning of the current run-up, the thirty-year fixed rate was 8.8 percent. It had fallen to 6 percent at the time the sample was drawn, keeping the monthly payment required to buy the median home from rising faster than income. The ratio of annual payment to income per capita actually fell in California and Wisconsin and stayed constant in Massachusetts. This fact adds weight to the argument that fundamental factors have an important effect on current home prices. Survey Method A random sample of 500 home sales was drawn from each of the same four counties as in our 1988 survey, and so we can make comparisons with these earlier results. We also used the very same questionnaire as in our 1988 survey, adding only several new questions at the end so that there was no change in the context of any questions. The accompanying letters were essentially similar to those of 1988. Survey methods followed guidelines outlined elsewhere. (16) Ordinary mail was used because we judged that the use of e-mail was still not widespread enough to produce a representative sample. The questionnaire was ten pages long and included questions on a number of topics. The focus was on the homebuyers' expectations, understandings of the market situation, and behavior. The questionnaire encouraged respondents to "write comments anywhere on the questionnaire," and their comments were indeed helpful to us in interpreting the significance of the answers. During the first survey, in 1988, two of the four markets were booming (the California counties), one market was at its peak and showing excess supply (Boston), and one was drifting (Milwaukee). This time three of the four markets were in remarkable booms, and Milwaukee again served as a control city, where no real boom was taking place. The survey was sent to 2,000 persons who had bought homes between March and August 2002. These dates fall just before the peak in media usage of the term "housing bubble" in October 2002. Questionnaires with personalized letters to the respondents were mailed in January 2003, a reminder postcard was sent in February, and replacement questionnaires with personalized letters were again sent to those who had not responded in March. These dates were just after the peak in media use of the term "housing bubble." Thus we managed to get our questionnaire survey out at a time when attention to the possibility of a housing bubble must have been close to its maximum. Our respondents had the opportunity to participate in the real estate market at a time of intense public attention to the possibility of a bubble and had the opportunity to read and think about this experience for some months afterward. This is what we wanted to do, since our purpose is to gauge human behavior during a purported bubble. Just under 700 questionnaires were returned completed and usable in the 2003 survey, for a somewhat lower response rate than in the 1988 survey. Response rates for each county are given in table 6. At the time of the 2003 survey, the economy was recovering from the recession that had ended in November 2001, but the recovery was slow, and the National Bureau of Economic Research had not yet announced that the recession was over. In contrast, at the time of our 1988 survey, there had been no recession for several years. In addition, the Federal Reserve had reduced interest rates to historic lows at the time the buyers in our 2003 survey were signing purchase and sale agreements. In 1988, in contrast, interest rates were on the rise. Table 7 describes the sample. A substantial majority of buyers were buying as a primary residence, and only a small minority were buying to rent. First-time buyers were a majority of the sample in Milwaukee. The lowest percentage of first-time buyers was in Los Angeles. We were surprised to see that, in the 2003 survey, more than 90 percent of the homes purchased in all four markets were single-family homes, a much larger share than in the 1988 survey. We have no explanation as yet for this. Survey Results The results of the 2003 survey, presented in tables 8 through 14, shed light on a number of aspects of homebuying behavior--including investment motivations and the expectation of further price rises, the amount of local excitement and discussion about real estate, the sense of urgency in buying a home, adherence to simplistic theories about housing markets, the occurrence of sales above asking prices, and perceptions of risk--that suggest the presence or absence of a bubble in home prices. Housing as an Investment A tendency to view housing as an investment is a defining characteristic of a housing bubble. Expectations of future appreciation of the home are a motive for buying that deflects consideration from how much one is paying for housing services. That is what a bubble is all about: buying for the future price increases rather than simply for the pleasure of occupying the home. And it is this motive that is thought to lend instability to bubbles, a tendency to crash when the investment motive weakens. Table 8 presents the responses to questions about housing as an investment. For the vast majority of buyers, either investment was "a major consideration" or they at least "in part" thought of their purchase as an investment. In Milwaukee and San Francisco investment was a major consideration for a majority of buyers. This tendency to view housing as an investment is similar to what it was in the boom period that we observed in our 1988 survey, although somewhat weaker. Far fewer of the homebuyers in 2003 said that they were buying "strictly for investment purposes." Thus conditions reported in 2003 would appear to be consistent with a bubble story, although less so than they were in 1988. The apparent attractiveness of housing as an investment is further enhanced if the buyer perceives that the investment entails only very little risk. As table 8 also shows, in all cities in both 1988 and 2003, only a small percentage of buyers thought that housing involved a great deal of risk, although the fraction seeing a great deal of risk rose (perhaps not surprisingly) to a fairly high level (14.8 percent) in San Francisco in 2003. In three of the four cities (Milwaukee being the exception), there was more perception of risk in 2003 than there had been in 1988, which is what one would expect given all the media attention to bubbles in 2003. Even so, the perception of risk of price decline is small: one may say that homebuyers did not perceive themselves to be in a bubble. Exaggerated Expectations, Excitement, and Word of Mouth Table 9 gets to the meat of the housing bubble issue: the role of price expectations, the emotional charge, and the extent of talk about real estate. Expectations about the future price performance of homes were high in both 1988 and 2003. In both of these housing booms, roughly 90 percent or more of respondents expected an increase in home prices over the next several years, and the average expected increase over the next twelve months was very high, even surpassing 9.8 percent in San Francisco in 2003. (17) But it is the long-term (ten-year) expectations that are most striking. When asked what they thought would be the average rate of increase per year over the next ten years, respondents in Los Angeles gave an average reply of 13.1 percent (versus 14.3 percent in 1988); in San Francisco they were even more optimistic, at 15.7 percent (14.8 percent in 1988); in Boston the answer was 14.6 percent (8.7 percent in 1988); and in Milwaukee it was 11.7 percent (7.3 percent in 1988). Note that even a rate of increase of only 11.7 percent a year means a tripling of value in ten years. Thus, although the one-year expectations in the glamour cities were lower in 2003 than they had been in 1988, the ten-year expectations were even higher. (18) Fewer respondents in 2003 said that it was a good time to buy a home because prices may be rising in the future, but at least two-thirds agreed with the statement in all four cities. Many thought not only that now was a good time to buy, but also that there was a risk that delay might mean not being able to afford a home later. The number who admitted to being influenced by "excitement" about home prices was still high, close to 50 percent in Los Angeles, but lower than in 1988. The amount of talk was nearly as high as in 1988, and talk is an important indicator of a bubble, since word-of-mouth transmission of the excitement is a hallmark. We conclude that these general indicators of the defining characteristics of bubbles were fairly strong in 2003. However, they were generally less strong than in 1988 in the glamour cities and stronger than in 1988 in Milwaukee. Simple (or Simplistic) Theories Table 10 shows results on respondents' agreement with a number of simple, popular theories or stories about speculative price movements that might influence how their interpretation of recent events translated into bubble expectations. Our survey results indicate that these simplistic theories are quite a powerful force and, moreover, a bit different in the glamour or bubble cities of Los Angeles, San Francisco, and Boston than in cities generally thought less exciting, like Milwaukee. The most simplistic theory is one that we have often heard expressed in casual conversation: that desirable real estate just naturally appreciates rapidly. The theory expressed seems to confuse the level of prices with the rate of change. The most elementary economic theory would say that properties that people find most attractive will be highly priced, but not necessarily increasing more rapidly in price than other properties. We tried to gauge agreement with this theory by asking whether people agreed with the statement "Housing prices have boomed in [city] because lots of people want to live here." There was overwhelming agreement with this statement in all the glamour cities, but not in Milwaukee. An even more outrageous fallacy that we detect in popular conversation about home prices is that "When there is simply not enough housing available, price becomes unimportant." To our respondents' credit, most did not agree with this statement. But from 20 to 40 percent did agree, particularly in the glamour cities. Another fallacy we think we have detected is in the interpretation of prices closing above asking prices. Homeowners sometimes seem to think that this phenomenon is a sign of a crazy boom that suspends the economic laws of supply and demand. Indeed, most homebuyers in the glamour cities thought that at such a time "there is panic buying and price becomes irrelevant." These results do not firmly prove that people are guilty of economic fallacies, because the questions admit of alternative interpretations, and people were probably not focusing clearly on their exact wording. However, we do believe that the strong agreement with some of these statements is at least suggestive of such fallacies. We believe that there is a sort of knee-jerk reaction to stories about boom markets in real estate that does not accord with economic theory, but that does affect the prices people are willing to pay for their homes. We leave clearer proof that people adhere to such fallacies to further work. A closer study of such popular fallacies is difficult to carry out, for if we draw out the fallacy clearly enough to reveal their belief in it to our satisfaction, respondents may be educated out of the fallacy by the very questioning intended to reveal it. All these theories about panic buying and the irrelevance of price do not, however, indicate that people generally believe that markets are driven by psychology. The results of the last question in table 10 show that people generally do not believe that markets are driven primarily by psychology, even in a booming real estate market. We interpret this as further confirming our general conclusion that most homeowners do not perceive themselves to be in a bubble even at the height of a bubble. Popular Themes in Interpreting Recent Price Movements We have documented that people talked a lot about the housing market both in 1988 and in 2003. What is it that they are likely to have talked about? We need to know the news stories that are on their mind if we are to understand the origins of the purported housing bubble. Table 11 shows some results from two open-ended questions that were put on the questionnaire, along with a space for the respondent to write in answers in his or her own words. Responses to these questions are especially interesting because they elicit themes that are already on the minds of respondents, rather than putting words in their mouths. One would perhaps not expect any one theme to dominate in answers to such questions, since people are so different and such broad questions allow so many different interpretations. But we do see what appears to be a dominating theme both in 1988 and in 2003, namely, interest rates. Clearly, interest rates have fallen substantially and have contributed to the run-up in prices since 1995, at least in the cities where, in our regressions, the interest rate variable was significant. Although, according to basic economic theory, interest rates should be more important in regions where the elasticity of supply of housing is relatively low or the likely growth of future demand relatively high, there is little evidence of this effect in state-by-state regressions. Many of the answers to these questions are disappointing. Typically the answers read like random draws from the business section of the newspaper, or else the respondents refer to casual observations that one might make just driving around town. Respondents presented no quantitative evidence and made no reference to professional forecasts. One should not be surprised at this, however. After all, the single-family home market is a market of amateurs, generally with no economic training. Once more we see evidence that in neither period did many homebuyers perceive themselves to be in a housing bubble. References to market psychology were quite rare. Relation of Investment Demand in 2003 to the Stock Market Boom and Bust The appearance of the real estate bubble right after the stock market drop has lent support to the notion that the two are somehow connected. One popular theory is that the stock market drop was followed by investor disgust with the stock market and a "flight to quality," as people sought safer investments in real assets like homes. There has been a lot of discussion about people shifting their assets toward housing because the stock market has performed so poorly since 2000. On the other hand, a falling stock market could have a negative wealth effect on home buying decisions. (19) Table 12 presents the responses to three questions that we did not ask in 1988 but were added at the end of the questionnaire in 2003. Recall that the survey was virtually finished before the stock market rally (25 percent on the S&P500) of March 11-July 8, 2003, and that the respondents had purchased their homes several months before. The answers to the last question in table 12, about whether the experience with the stock market encouraged purchase of a home, show that for the vast majority of people in all four counties the performance of the stock market "had no effect on my decision to buy my house." However, one should not discard the notion that the stock market's behavior was at least partly responsible for the boom in the real estate market. Judging from their additional comments, it appears that some of the majority who said the stock market had no effect on the decision to buy a home said so only because they would have bought some home in any event, even if perhaps a smaller home. More significantly, many other respondents (roughly between a quarter and a third) said that the stock market's performance "encouraged" them to buy a home, whereas only a small percentage found it discouraging. Immediately after this question we included an open-ended question, "Please explain your thinking here," followed by an open space. Although most left this space blank, the answers we did get were all over the map, as respondents apparently viewed the question as an opportunity to vent on any subject. Some of the answers from those who said they were encouraged by the stock market did refer to the drop in the stock market after 2000 as a reason to buy a home now. Quoting a few of their answers verbatim will illustrate: "Housing costs continue to increase. Value of home investment to increase. Stock market not so promising." "Could be better investment than stock market." "I lost $400,000 in my pension and personal stock portfolio--at least buying this big beautiful home I know it's a hard asset that would hold its value & appreciate while it gives me great enjoyment." "Money that we had saved for a house was starting to become a loss in the market." "I have only made money in real estate and lost a lot in the stock market." "The stock market at my age is not helping me. Short-term real estate is the strongest investment you can make short or long term." "Stock market went down. House market is still going up." "Renting is not cheap, stock is declining, this implies our total assets is [sic] not going anywhere." "The value of my condo had increased significantly compared to the gains to my portfolio. With interest rates low a new home seemed more likely to increase than a comparable investment in the stock market and brings tax & quality of life benefits." Some respondents referred to the increased volatility or other uncertainty in the stock market since 2000, rather than its changed level, as a reason to shift their portfolio: "It seemed that shifting some of our net worth to cash and homeownership was a wise move in the face of the market volatility in 2000-2002." "I'm buying the house for the long term. The house will probably depreciate in the next couple years, but it will certainly appreciate over 10+ years. This is because it is a good house in a good community. This is information that I am confident of. In contrast, there is no confidence that I have full (or even good) information about the stock market (or that even my mutual fund managers have good information about the companies they invest in). So, I buy the house." "A house seems like a more solid investment than stocks. Less volatile." Although this evidence is far from proof of a connection between the stock market and the housing market, we interpret it as confirming the notion that people got fed up with the stock market after the decline and high volatility following the 2000 peak and became more positive about real estate. Excess Demand and Upward Rigidity in Asking Prices In the boom cities, newspaper articles feature stories of homes that sold well above the asking price. We have already noted that it was an article in the Wall Street Journal referring to "frenzy in California's big single family home market" that inspired our original survey. In fact, such frenzy seems to be a fairly common occurrence in boom cities. As table 13 shows, quite a large number of people reported selling above the asking price in both the 1988 and 2003 surveys. An amazing 45 percent of respondents in San Francisco in the 2003 survey reported selling at above the asking price in 2002, well after the sharp decline in employment following the NASDAQ collapse, which began in 2000. Sellers in Los Angeles reported that about 20 percent of properties sold for more than the asking price, as did a slightly smaller share in Milwaukee, which had no boom. Many of those who sold felt that if they had charged 5 or 10 percent more, the property would have sold just as quickly. This was the sense of over 20 percent of sellers in all markets in 2003, a substantially larger fraction than in 1998 except in Los Angeles, where it stayed the same. An amazing number of the 2003 respondents--in fact, a majority in San Francisco and Boston, a near majority in Milwaukee, and 26 percent in Los Angeles--thought that charging more than they did would be unfair. On the other hand, the number who reported that their home was not intrinsically worth more than they were asking dropped in the latest survey compared with that in 1988. Downward Rigidity and Excess Supply An important question on which the survey sheds some light is, What happens in a bust? How do sellers respond to rising inventories and increasing time on the market? It is important first to point out that the housing market is not an auction market. Prices do not fall to clear the market quickly, as one observes in most asset markets. Selling a home requires agreement between buyer and seller. It is a stylized fact about the housing market that bid-ask spreads widen when demand drops, and the number of transactions falls sharply. This must mean that sellers resist cutting prices. Table 14 supports the notion that sellers lower their asking prices only as a last resort. A majority of respondents in all cities and in both years of the survey argue that the best strategy in a slow market is to "hold up until you get what you want." Only a small minority reported that they would have "lowered the price until I found a buyer." In addition, large majorities ranging from 79 percent in San Francisco in 1988 to 93 percent in post-boom Boston reported having reservation prices. There is clear evidence that such resistance prevents home prices from falling at the onset of a down period and that, if the underlying fundamentals come back quickly enough, they can prevent a bubble from bursting. Instead, the danger when demand drops in housing markets is that the volume of sales may drop precipitously. This could do more damage to the U.S. economy today than a modest decline in prices. A Model of Speculative Bubbles in Housing Buyers and sellers in the housing market are overwhelmingly amateurs, who have little experience with trading. High transactions costs, moral hazard problems, and government subsidization of owner-occupied homes have kept professional speculators out of the market. These amateurs are highly involved with the market at the time of home purchase and may overreact at times to price changes and to simple stories, resulting in substantial momentum in housing prices. Shiller argues that speculative bubbles are caused by "precipitating factors" that change public opinion about markets or that have an immediate impact on demand, and by "amplification mechanisms" that take the form of price-to-price feedback. (20) A number of fundamental factors can influence price movements in housing markets. On the demand side, demographics, income growth, employment growth, changes in financing mechanisms or interest rates, as well as changes in locational characteristics such as accessibility, schools, or crime, to name a few, have been shown to have effects. On the supply side, attention has been paid to construction costs, the age of the housing stock, and the industrial organization of the housing market. The elasticity of supply has been shown to be a key factor in the cyclical behavior of home prices. The cyclical process that we observed in the 1980s in those cities experiencing boom-and-bust cycles was that general economic expansion, best proxied by employment gains, drove demand up. In the short run those increases in demand encountered an inelastic supply of housing and developable land, inventories of for-sale properties shrank, and vacancy declined. As a consequence, prices accelerated. This provided the amplification mechanism as it led buyers to anticipate further gains, and the bubble was born. Once prices overshoot or supply catches up, inventories begin to rise, time on the market increases, vacancy rises, and price increases slow, eventually encountering downward stickiness. With housing, a significant precipitating factor may be employment gains, if only because they are highly visible. Employment releases occur on the first Friday of each month, with state data released somewhat later. Both national and state releases by the BLS receive dramatic fanfare in the press. In all three of the cities with volatile prices, substantial employment gains and falling unemployment preceded the upward acceleration of home prices during both booms. The predominant story about home prices is always the prices themselves; the feedback from initial price increases to further price increases is a mechanism that amplifies the effects of the precipitating factors. If prices are going up rapidly, there is much word-of-mouth communication, a hallmark of a bubble. The word of mouth can spread optimistic stories and thus help cause an overreaction to other stories, such as stories about employment. The amplification can also work on the downside as well. Price decreases will generate publicity for negative stories about the city, but downward stickiness is encountered initially. The amplification mechanism appears to be stronger in the glamour cities that were undergoing rapid price change at the time of our surveys than in our control city of Milwaukee. We saw in our survey results that talk about real estate is more frequent in those cities and that excitement is stronger there. Presumably this greater talk and excitement have something to do with the greater price volatility seen historically in the glamour cities, leading to greater public interest and concern with movements in real estate prices. Thus real estate price volatility can be self-perpetuating: once started, it generates more public attention and interest, and thus more volatility in the future. Longer-run forces that come into play tend eventually to reverse the impact of any initial price increases and the public overreaction to them. New construction can bring some new housing online in the space of about a year. The United States now has a highly sophisticated national construction industry, dominated by national firms such as Pulte Homes, Lennar Corporation, and Centex Corporation. These firms are capable of moving their operations into a city quickly if they perceive the ability to build homes for less than the going price. However, there are important barriers to their moving into certain cities, as executives from these firms will animatedly tell you. In many mature cities there is no place to build, and obtaining permits can be long and costly. Case has argued that differences in supply elasticity across cities explained a larger percentage of price changes than do demographics. (21) Clearly, prices of homes can go up more rapidly than building costs only if supply is inelastic at least in the short run. Zoning restrictions are an important barrier to the construction of new homes. These restrictions prevent more intensive use of available land, for example by building more closely spaced houses or taller high-rise apartment buildings. Edward Glaeser and Joseph Gyourko have shown a close correlation across U.S. cities between a measure of zoning strictness derived from the Wharton Land Use Control Survey and the ratio of existing housing prices to the cost of new construction. (22) They found that there is relatively little correlation between population density and home prices, even though economic theory might seem to suggest such a correlation. Thus zoning has been fundamental in limiting the supply of housing. Even if shortages of places to build are long lasting, in the longer run positive impulses to employment can, if there are barriers to the supply response, lead to outflow of industries that have little reason to stay in the city, thereby eventually reversing the high demand for homes. At the height of a boom, both labor supply and labor demand can be negative factors, with high home prices deterring workers from coming to an area and a labor shortage deterring industry from locating there. Moreover, retirees and families with children (who have higher housing demand) will tend eventually to leave high-price cities. Thus cities that have attracted certain industries and have seen a surge in employment eventually become more specialized: Silicon Valley, for example, has become almost exclusively a mecca for people who need to benefit from the synergies of the electronics industry. This process can eventually reverse the price increases. This process of reversal, however, is hardly on the minds of most homebuyers, who, as we have seen, are preoccupied with relatively simplistic stories about housing when they consider their investments. The relatively poor performance of their city after the boom comes as a surprise to them. Over long intervals in most states, the growth rate of home prices has tended to track growth in nominal income per capita. It is not surprising that this should be so, for two reasons. First, land zoned for new construction in scarce or important locations is fixed, and if people target a fraction of their income for the costs of a home, given fixed supply the price of that fixed land should increase with income. Second, construction costs, which are mostly labor costs, tend to track income per capita as well. Thus, over the period from 1980 to 2000, price growth in Los Angeles and price growth in Milwaukee have been about the same. But there is a big difference in the shorter-run behavior of prices in those two cities. The upward trend in home prices that is implied by the growth rate of income per capita, along with the tendency for home price decreases to be slow and sluggish, has meant that relatively few citywide home price declines have been observed in history. More often one sees periods of flat real estate prices, where the ratio of price to income, or the ratio of price to the consumer price index, is falling but nominal prices themselves have not fallen. Outright price declines are much more salient in investor psychology than failures of prices to keep up with income. Thus popular culture has not identified bubbles as a problem in real estate, or did not until last year. The popular impression has been that real estate is an investment that cannot lose money. The declines in prices in the early 1990s in many cities, documented for the first time in history by accurate real estate price indexes developed by us and others, have forever reduced the salience of this public impression, but, as our latest survey documents, the idea still lingers. There is also a popular impression that real estate is a candidate for the "best investment" that can be made (see top panel of table 12). Whether real estate is in fact the best possible investment is not something amenable to economic analysis, since one cannot measure the "dividend" in the form of housing services that homes offer. Presumably there is diminishing marginal utility to owning a bigger and bigger house, and so the psychic dividend declines with the amount of house purchased. The basic question that individuals must resolve is how big a house to buy, and the theory that "housing is always the best investment" is a poor clue to how to answer this question. Yet that theory has a salience that is quite strong in the current market. Is a Housing Bubble about to Burst? Clearly, one can construct an argument that home price increases nationally since 1995 have been driven by fundamentals. For more than forty states, income growth alone explains virtually the entire increase in housing prices, and falling interest rates have reduced financing costs sufficiently to keep the ratio of annual mortgage payments to income from rising even in the boom states of Massachusetts and California. In the vast majority of states, housing is actually more affordable than it was in 1995. Nonetheless, our analysis indicates that elements of a speculative bubble in single-family home prices--the strong investment motive, the high expectations of future price increases, and the strong influence of word-of-mouth discussion--exist in some cities. For the three glamour cities we studied, the indicators of bubble sentiment that we documented in tables 8 and 9 remain, in general, nearly as strong in 2003 as they were in 1988. Some of these are surprisingly high in 2003, notably the ten-year expectations for future price change, where the average expected annual price increase is in the 13 to 15 percent range for all these cities. Even our fourth city, Milwaukee, is perhaps showing some bubble sentiment, for the expected annual price increase for the next ten years there is 11.7 percent. All of the fundamental measures of bubble activity--the expectations, the sense of opportunity and urgency, the excitement and amount of talk--are generally down from their levels in 1988 in the glamour cities, but up from their levels of 1988 in Milwaukee. (Long-run expectations, however, are generally up substantially from 1988. If long-run expectations matter most, one might say that the 2003 exuberance is just as strong as the 1988 one.) Most people do not perceive themselves in 2003 as in the midst of a bubble, despite all the media attention to the possibility. However, neither did people perceive themselves to be in a bubble in 1988, after which real prices fell sharply in many cities. Although these indicators do not suggest such strong evidence of a bubble as was observed in 1988, it is reasonable to suppose that, in the near future, price increases will stall and that prices will even decline in some cities. We have seen that people are not as confident of real estate prices as they were even before the 1980s real estate bubble burst, and this lack of confidence may translate into an amplification of any price declines. Real home prices are already flat in Denver and Detroit, following periods of rapid growth. More declines in real home prices will probably come in cities that have been frothy, notably including some cities on both coasts of the United States, and especially those that have weakening economies. But declines in real estate prices might appear even in cities whose employment holds steady. The consequences of such a fall in home prices would be severe for some homeowners. Given the high average level of personal debt relative to personal income, an increase in bankruptcies is likely. Such an increase could potentially worsen consumer confidence, creating a renewed interest in replenishing savings. Personal consumption expenditure, which has driven the economy so far in the current recovery, may drop, stalling the recovery. However, judging from the historical record, a nationwide drop in real housing prices is unlikely, and the drops in different cities are not likely to be synchronous: some will probably not occur for a number of years. Such a lack of synchrony would blunt the impact on the aggregate economy of the bursting of housing bubbles. Table 1. Ratio of Average Home Price to Personal Income |