Discussion of Randal Verbrugge’s: The Puzzling Divergence of Aggregate Rents and User Costs, Leonard Nakamura Federal Reserve Bank of Philadelphia* * My views, not the Fed’s
Back to basics Paper points out that because interest rates and asset prices are volatile, short-term real rates of return can vary substantially from real rental rates –This is certainly true –Applies to all long term assets, not just residences This paper is a useful reminder of what economists don’t understand –And a valuable challenge to economists I am not so sure that it puts statistical agencies in a quandary: Empirically, they should use rental equivalence –Rents are understood at least somewhat –User costs are (empirically) not
Owner occupied homes: costs and equivalent rents Stock of housing is large, and adjusts slowly Households adjust their use of housing slowly –Implied real stream of housing services must change only slowly User costs are too volatile compared to this shadow price –Not just true for housing – user costs of capital are too volatile (Shiller)
Implication: real quantity of housing services changes slowly At the margin in equilibrium, the rent summarizes the information in the user cost. –Empirical question: are rents or user costs closer to the truth? Good evidence that rents are reasonable: –Hedonic estimates (Crone et al, Kurz and Hoffman) –Rents and housing prices are cointegrated: J. Gallin, BoG, Long horizon of house prices and rents Real rents move with real net stock of residential fixed assets –Ratio is relatively constant over past 60 years –Particularly if we adjust BLS residential rents for response bias (Crone et al, CPI for rents, a case of understated inflation)
Minor problems in Verbrugge measure of user costs To get user cost right, very hard work Risk premiums? Shouldn’t use deterministic model Hard to measure: –Expected appreciation :4 quarter horizon or 40 quarter horizon? –Relevant interest rate: Mortgage rate includes refinance option The exercise cost has changed over time Type of mortgage used varies over time (ARMs when rates are high) –Cost of maintenance and repairs These are minor problems in the sense they are very unlikely to influence the result User costs too volatile to know whether trend is right
Explaining rents Are rents related to theoretical variables? –Another valuable question Rents = β(user cost) –User costs are measured with error – β likely biased downwards Detached units rental rates not very reliable –Too many special cases –Possibly better to use regular rental units Also, adjust rent inflation rates for a la Crone et al
Bottom line: a valuable and difficult line of research This is an important issue for economics: Residential assets very large share of total wealth How useful are rents to measure the shadow price of owner occupied housing services? Why are user costs so volatile? –Why don’t house prices appear to be martingales? What are the returns to owner-occupied housing and are they reasonable? –Is there a equity premium puzzle for housing Need stochastic GE model (Luengo-Prado, Nakajima)