Housing supply.

• Author: Glaeser, Edward L.
• Position: Research Summaries - Growth in housing rates

In 1981, Lawrence Summers noted the 35 percent increase in real housing prices between 1965 and 1980 and argued that this increase could be explained by inflation. Summers (1) and Poterba (2) persuasively showed that higher levels of inflation increase the interest rate subsidy on home mortgages and essentially shift out the demand curve for housing. Ten years later, Mankiw and Weil (3) argued that demographics drive housing demand and, because of failing demand, housing prices will experience painfully slow growth by the year 2000.

The appropriate renown of these papers indicates the degree to which demand-side analysis has dominated the housing literature, but an increasing body of facts is beginning to challenge this orientation. It is becoming increasingly obvious that we must understand housing supply if we are to understand booms and busts in housing prices. Over the past five years (1998-2003), despite low inflation and the baby bust, real housing prices increased by 25 percent, according to the Freddie Mac Repeat Sales Index. During the 1975 to 1980 period, when inflation was soaring and baby boom children were moving out of their parents' homes, the same index showed real housing price increases of less than 20 percent.

Rising housing prices over the past ten years can always be explained by another omitted shifter of demand. However, evidence on construction suggests that demand alone cannot provide the answer. For example, in Manhattan, before 1975, housing price growth was modest, and there was abundant new construction. Since 1980, housing prices have soared and there have been few new units. (4) The physical character of Manhattan has not changed between 1960 and today: If the rise in housing prices during the 1990s were the result of demand pushing along a stable supply curve, then surely we would see an explosion in new construction as we did in the past. The increasingly common combination of rising prices and tiny amounts of construction pushes us to focus on housing supply.

Differences across regions confirm the need for supply-side analysis. High housing prices are not ubiquitous. The median housing value in the median county in America in the 2000 census is $75,300. More than 95 percent of countries have median housing values below $160,000. Soaring home prices are primarily coastal phenomena that have left the growing states of the American interior untouched.

If the heterogeneity in price growth with the United States were the result of different patterns of demand, then we would expect to see quantities and prices move together. Places with high price growth would be places with new construction. Figure 1 graphs the rate of housing price growth (again using to the Freddie Mac Repeat Sales Index) and permits for new housing units (divided by the stock of housing units in 1990) between 1998 and 2003 across census divisions. There is a negative 50 percent correlation between price growth and new construction.

[FIGURE 1 OMITTED]

The places that are building have little housing price appreciation and the places that have housing price appreciation are not building. Demand alone can't explain the difference in housing price...