What Caused the Housing Boom?

This is an econometric analysis of the forces driving the housing boom from Jonas D. M. Fisher and Saad Quayyum of the the Chicago Fed. It comes to the conclusion that fundamentals other than low interest rates explain most of the change in homeownership -- speculative bubbles and monetary policy had little to do with it. If that's true, it means that monetary policy has less influence than commonly presumed on the level of economic activity. The body of the paper is a fairly technical econometric investigation, so I've only included the introduction and conclusions and a few extra graphs and tables:

The great turn-of-the-century housing boom, by Jonas D. M. Fisher and Saad Quayyum, Federal Reserve Bank of Chicago: Introduction and summary In the last ten years, residential investment as a share of gross domestic product (GDP) has reached levels not seen since the 1950s. At the same time, the homeownership rate has climbed to levels never before achieved. This article discusses the forces underlying these developments and argues that they are connected.

Figure 1 shows the ratio of nominal residential investment to GDP from 1947 to 2005, with shaded regions indicating years in which the economy was in recession. The spending share of residential investment is clearly highly cyclical, but in the last ten years it has seemed relatively immune to macroeconomic disturbances. Indeed, from a near historic low below 3.5 percent in 1991, residential investment spending has grown rapidly, passing 6 percent of nominal GDP in 2005.

Figure 2 shows the history of the homeownership rate from 1890 to 2004. The homeownership rate equals the number of owner-occupied housing units divided by the number of occupied housing units. Between 1890 and 1940, the homeownership rate varied between 43 percent and 48 percent. After World War II, the homeownership rate rose rapidly, and by the mid-1960s it had surpassed 64 percent. Upward progress in homeownership stalled in the 1970s and even fell in the 1980s. It began growing again in the mid-1990s and by 2005 had reached a new high of 69 percent.

Understanding why residential investment and homeownership have reached such unusually high levels is useful from a policymaking standpoint. For example, monetary policy has been traditionally viewed as having a strong influence over new home construction. Have the high levels of residential investment been driven by unusually loose monetary policy? Another concern of policymakers is that the unusually high level of spending on new housing reflects speculation and is not driven by underlying fundamentals. The increase of rates of homeownership has long been an announced goal of policymakers. ... So, understanding why homeownership rates have risen should help in the development of policies directed at establishing socially and economically desirable levels of homeownership.

Much has been said in the press about high levels of house prices and the possibility of a house price bubble. ... This article does not address house prices directly. Rather, it seeks to understand recent developments by focusing on quantities. To the extent that the quantities can be understood by considering the underlying economic fundamentals, such as productivity growth and the evolution of the mortgage market, then the recent growth in house prices is probably not due to excessive speculation in the housing market, such as occurs in a bubble. We argue that our findings point toward the high prices being driven by fundamentals.

The article begins by describing the evolution of key variables that should influence residential investment. While informative, this discussion has the drawback that it is difficult to distinguish the truly exogenous factors driving the spending. For example, showing that real interest rates have been relatively low as residential investment has surged beyond its trend level does not establish that unusually loose monetary policy is to blame. Consequently, the next phase of the analysis involves an econometric study of the effects of identified exogenous shocks to the economy. This study focuses on the roles of technological change and monetary policy.

We then turn to the homeownership issue. We begin by describing how homeownership rates have changed across various racial, generational, educational, and income categories. Then we address the question of how much of the increase in homeownership can be explained by changes in the distribution of the population across these categories. For example, older people have higher homeownership rates than younger people, so, all else being equal, an aging population tends to increase the homeownership rate. By accounting for all easily measurable factors, this analysis provides a bound on what needs to be explained by other, more difficult to measure factors, such as the increased use of new mortgage products.

The final part of the article connects the overall increase in residential investment with the increase in homeownership. This analysis focuses on the impact of the rapid growth of the subprime mortgage market.

Our main findings are as follows. First, it appears that the housing boom has not been driven by unusually loose monetary policy. This is not to say that monetary policy has not been unusually loose, but that to the extent it has been loose, this is not what has been driving spending on housing. Second, the current levels of spending on housing are largely explained by the wealth created by dramatic technological progress over the previous decade. Third, changes in the demographic, income, educational, and regional structure of the population account for only one-half of the increase in homeownership. ... The last finding is that substitution away from rental housing made possible by technology-driven developments in the mortgage market, such as subprime lending, could account for a significant fraction of the increase in residential investment and homeownership. The current spending boom thus may be a temporary transition toward an era with higher homeownership rates and a share of spending on housing that is nearer historical norms.

Factors affecting residential investment Figure 4 is helpful as a starting point for gauging whether residential investment is currently at unusual levels. The figure displays the log of real residential investment per household from 1947 to 2005 along with a trend line. While certainly subject to large cyclical variations, residential investment seems to follow a linear trend quite closely. So while investment grew rapidly after 1991 until the latter part of the decade, this was largely a return to trend. Only after the 2001 recession has spending grown substantially above trend. By 2005, it was roughly as far above trend as occurred during the boom part of the boom–bust cycle of the 1970s and the early 1980s. The dramatic swings in residential investment in the 1970s and early 1980s contributed significantly to three recessions. From this perspective, the current levels of residential investment may seem alarming. In the remainder of this section, we consider some of the factors that may underlie the current high levels of residential investment.

Household formation Household formation, to the extent that it is governed primarily by long-term social and demographic developments, is the most basic determinant of home building and residential investment ... Figure 5 shows the evolution of new households and housing starts since 1960. The light blue lines indicate the number of new households in a given year, the black lines indicate the average number of new households per year for each decade, and the dark blue line indicates the level of housing starts. In the 1970s, 1980s, and 1990s, there seems to be a close association between home building and household formation. The increases in household formation and home building over the 1970s are an example of the strong influence of demographic factors—this is when the baby boom generation moved out on its own. The close association of home building with household formation is less true of the 1960s, when home building is near its 1980 levels, but new household formation is much lower. Since the 2001 recession, housing starts have also risen to levels that do not seem closely tied to new household formation.

Migration Another factor determining new home building is migration. With migration, the number of households can stay fixed while there is still a demand for new homes. In the region where households migrate from, vacancy rates go up, while in the region where households migrate to, houses need to be built. To assess the possible impact of migration, consider figure 6. Panel A shows the shares of national population increases attributable to the four census regions from 1982 to 2005. Panel B displays the corresponding shares of all housing starts. In each case, a rising or falling share indicates that either population or housing starts are increasing faster in the region than for the nation as a whole. This figure shows that the relative shares of housing starts generally correspond to the relative population shares. Some of the trends seem to correspond as well. For example, the dip in the share of housing starts for the South in the 1980s is associated with a downward trend in the population share as well. Of particular interest is the uptick in the population share of the South from 2000 to 2005. Consistent with a role for migration in the current housing boom, the share of housing starts also picked up, although with a delay.

Interest rates The discussion until now has focused on relatively long-term developments. As is evident from figures 1 and 2, home building historically has been highly cyclical. The conventional wisdom on why this is so is that the demand for housing is sensitive to movements in mortgage rates. If mortgage rates are unusually low, then this could fuel unusually high rates of residential investment. Figure 7 displays a measure of the nominal effective mortgage rate along with an estimate of the corresponding real rate. The term “effective” means that the mortgage rate incorporates the various points, fees, and other closing costs associated with a mortgage. These have generally been declining since the early 1980s. The real rate is equal to the nominal rate less an estimate of the expected rate of inflation. ... Figure 7 shows that both real and nominal mortgage rates were low in the 1990s compared with the 1980s. This presumably contributed to the return to trend of residential investment over this period. Interestingly, the period when home building accelerated beyond its trend level was also a time when nominal and real rates were falling even further. While over the last two years nominal mortgage rates have started creeping up, real rates have continued to fall because inflation expectations have been rising. These considerations suggest that sustained low interest rates, possibly driven by unusually loose monetary policy, could be fueling the housing boom.

Wealth The final factor affecting home building that we consider is household wealth. All else being equal, the richer households are, the more housing they demand. The latter half of the 1990s was a period of rapid wealth accumulation. For example, according to the Survey of Consumer Finances, average family net worth increased by 72 percent between 1995 and 2001. These increases in wealth were primarily due to the large increases in stock values over this period. As a consequence, housing and other nonfinancial assets’ average share of total assets, fell from 63.3 percent in 1995 to 58 percent in 2001. If the share in 1995 was “normal” or close to the “desired” household allocation of nonfinancial assets in households’ portfolios, then it is to be expected that the share would eventually start rising again. Indeed, by 2004, nonfinancial assets’ share of total assets had risen to 64.3 percent, near its 1995 level. While the share of housing in total assets, in the form of primary residences, rose faster over this period, the behavior of nonfinancial assets as a whole suggests that much of the acceleration in residential investment after the 2001 recession might be due to households rebalancing their portfolios. That is, it may be a natural consequence of the stock market boom of the 1990s.

The macroeconomic shocks driving residential investment The foregoing discussion suggests various factors that may be influencing the high levels of residential investment, but the underlying causes remain unclear. Determining the causes of macroeconomic fluctuations is notoriously difficult because essentially all the variables of interest are endogenous—no single variable moves independently and drives movements in other variables. The traditional approach to assessing the causes of fluctuations is to posit that the economy is subject to exogenous random disturbances, which are called shocks. Macroeconomists have formulated methods for identifying three kinds of shocks—two kinds of shocks to technological possibilities and one kind of monetary policy shock. The procedure for identifying these shocks involves specifying a statistical model of the variables of interest and making a series of identification assumptions that make it possible to extract the exogenous shocks from the statistical model. Once the shocks have been identified, it is possible to determine how much of the growth in residential investment from 1995 to 2005 can be attributed to these shocks using what is called a historical decomposition. The strategy for identifying the technology shocks builds on Fisher (2006) and the monetary policy shock identification builds on Christiano, Eichenbaum, and Evans (2005).

Identifying the exogenous shocks

We begin by supposing that the economy evolves according to the following vector autoregression (VAR):

...[discussion gets fairly technical - move to conclusions]...

Conclusions This article has attempted to explain two features of the turn of the twenty-first century U.S. economy: high levels of residential investment and homeownership rates. Our main findings are as follows. First, it appears that the housing boom has not been driven by unusually loose monetary policy. This is not to say the monetary policy has not been unusually loose, but that to the extent it has been loose, this is not what has been driving spending on housing. Second, the current levels of spending on new housing are largely explained by technology-driven wealth creation over the previous decade. Third, changes in the demographic, income, educational, and regional structure of the population account for about one-half of the increase in homeownership. That is, without any other developments, the homeownership rate is likely to have gone up anyway, but not by as much as it has done. The last finding is that substitution away from rental housing made possible by developments in the mortgage market, such as subprime lending, could account for a significant fraction of the increase in residential investment and homeownership.

We view our findings as supporting the view that the current housing boom may be a temporary transition toward an era with higher homeownership rates in which spending is temporarily higher than historical norms but will eventually return to such norms. While we have so far mostly avoided discussing housing prices, our findings do suggest that to the extent that house prices have grown considerably in recent years, this is not due to unusually excessive speculation in the housing market, such as would occur in a bubble. Instead, our findings point toward the high prices being driven by fundamentals.

Here's a two more tables from the paper on the distribution of home ownership by household characteristics (click on tables to enlarge):

Home Ownership Rates and Percent of Population for Various Household Characteristics:

Homeownership Rate by Income Deciles