The natural rate of unemployment varies through time with structural change, demographic change, and so on. This report released today on existing home sales and recent reports on inventories are a reminder that a similar concept exists for homes and machinery.
The vacancy rate in housing is similar to the unemployment rate in labor markets. Why isn't 0% unemployment for labor optimal? Some unemployment is optimal because it allows people to change jobs and allows new entrants to enter the labor market without a "double coincidence of wants." Without vacancies, to change jobs you would need to find someone who has the job you want and wants the job you have, and then trade. Those entering the labor market would have to find someone who is leaving the labor market and has an acceptable job, and they in turn must be acceptable to the employer. The matching costs are high with such an arrangement. With some unemployment, costs fall since finding an offsetting match is unnecessary.
Housing is no different. Without vacancies, to move from New York to Los Angeles would require finding someone moving in the other direction who has a house you are willing to buy and is also willing to buy your house, a difficult task (rentals would be similar). But with vacancies, the task is much easier.
But what should the optimal vacancy rate be and what causes it to change? If the actual vacancy rate is below the natural rate, prices should rise. If the vacancy rate is above the natural rate, prices should fall, so knowledge of the natural vacancy rate is helpful in understanding markets. The natural rate of vacancy is not a constant. It varies over time and there is no reason to think that one natural rate exists for all areas, for all housing types, and so on, so what causes the natural vacancy rate to change? One approach is to break it down much like unemployment. There are cyclical factors from the macroeconomy and local economy that make the vacancy rate rise and fall over time, but these aren't changes in the natural rate, these are variations around the natural rate. There are structural factors that cause increased job turnover that would presumably change the optimal vacancy rate, i.e. the more dynamic the labor market, particularly if labor is coming from outside or moving away from the local area, the higher the natural rate. And there are frictional factors as people change houses and move up or down depending upon their stage in the life-cycle, and as with unemployment, some of this is driven by demographics. I just started thinking about this, so I have surely overlooked important reasons for variation in natural vacancy rates. What other factors are important?
Will the data identify a natural rate? These graphs (Census data here) show housing and rental vacancy rates since 1956. The rates vary considerably through time:
A few observations. The turnover rate is higher for rentals and this is clearly reflected in the difference in mean vacancy rates. The vacancy rate averages 7.0% for rentals which is quite a bit higher than the 1.4% rate for owner-occupied housing. Second, there are both high frequency and low frequency cycles in the data and the low frequency cycles are, to me, surprisingly long. Third, for both series there has been an upward trend in vacancies over the last 25 years with the most pronounced increases occurring from around 1980-1990 and after 2000. This is most evident in rentals, but owner-occupied shows similar, but less discernable trends (the top graph appears noisier because it is on a smaller scale). This is not what I expected. I anticipated that that factors such as better search technology and more creative financial instruments would have lowered average vacancy rates. Anyone have ideas why the natural vacancy rate appears to be drifting upward over the last 25 years? Were their tax changes, demographic changes, changes in second homes purchases and speculation, or is it just weak demand in some time periods? There is a caution in the footnote about comparing pre and post 1986 rates, and the data do change at this point, but the upward trend exists both before and after this breakpoint. Finally, this leads me to wonder how much variation there is in the natural rate of unemployment over subgroups in labor markets. For example, is the unemployment rate for temporary workers (corresponding to rentals) substantially different from full time employees (corresponding to owner-occupied)? Are we missing anything important by focusing on a single national rate in policy applications such as the Taylor rule, or does that capture the national trend adequately?