It has long been a puzzling macro feature that asset prices and turnover rates are positively correlated. Trading is more frequent and volume is higher when prices go up and vice versa. The most recent and striking experience of an asset boom is in the residential housing market in the early 2000's when both turnover rates and prices were high. This paper introduces agent heterogeneity, shows its implication on the housing market dynamics and proposes a new channel to explain the positive price-volume correlation in the housing market.

Heterogeneity’s role

There is a growing literature that examines how heterogeneous beliefs (combined with short sales constraints) help explain the price and trading dynamics. This paper focuses on the housing market, where the impact of heterogeneous opinions is relatively under-researched. Additional to the agents’ heterogeneity in the private value in the house ownership, this paper introduces an additional layer of investor heterogeneity and shows its implication on the price and transaction activities in the residential housing market.

If there is divergence of valuation, existing theories imply that the market price will reflect higher private valuations in the investor population, given supply constraints. More importantly, buyers with a shorter holding horizon are more sensitive to changes in market conditions. This is because transaction costs at the time of sale will affect their valuations more unfavourably. Furthermore, short horizon buyers' valuations are reduced particularly more when market conditions are expected to worsen. In equilibrium, there will be more short-term investors in the owner population in a booming market and the aggregate transaction activity is high at those times. Therefore, the model suggests that the composition of owners in terms of holding horizons contains valuable information regarding the current and expected price and trading dynamics.

Model implication
     
The equilibrium price is in general higher than the mean valuation of all potential buyers in the economy. In addition, in the presence of a persistent rent growth rate, the transaction cost at the time of sale will have state-dependent and horizon-specific implications for agents' valuations. Short horizon buyers' valuation in general will be affected more since they are subject to a higher transaction cost per unit of time. There are two implications. First, this leads to return predictability by owner's investment horizons since market states persist in the short run. Second, this particular state-dependent ownership structure translates into more trading in the booming markets and vice versa. The intuition is that short horizon owners who buy in the booming market are also very likely to sell in the good state, given persistence in the short run. In the empirical analysis, I focus on the model's novel predictions on the return predictability of the (time-varying) ownership composition.

Data

This analysis uses data from the American Housing Survey (AHS). Starting from the 1970s, the survey is conducted by the Bureau of the Census for the Department of Housing and Urban Development (HUD). It has the longitudinal design since AHS returns to the same housing units year after year to gather data. The date used here is the biennial data from the national sample from 1985 to 2005. Although I do observe realized durations from the AHS survey, there are several problems of using them directly. I model the expected duration for each homeowner as the projection of the observed duration on a vector of demographic and housing unit specific characteristics at the time of the purchase.

Empirical result
     
First, in the constructed panel data of around 120 Metropolitan statistical Areas (MSA) over a 20 year period, using the average expected duration for home owners at the metropolitan level as the test variable, there is a strong negative co-movement between the expected duration series and house transaction prices at the MSA level. This relationship is statistically significant and robust to a vector of control variables including time and location fixed effects.

Summary

This paper introduces agent heterogeneity (in particular their investment horizon differences) and shows its implication on the housing market dynamics. In summary, agents with heterogeneous investment horizons tend to have different demand sensitivities to market conditions. Thus the economy is composed of a different mix of owner population at different market states, which has implications for the joint dynamics of price and trading volume. I show both theoretically and empirically that the owner composition is more likely to be short-term oriented in a rising market. Results suggest that investment horizon heterogeneity is an interesting channel that can improve our understanding of the aggregate housing market.

This paper recently received the Best Paper Award in Joint International Conference of the American Real Estate and Urban Economics Association and Asian Real Estate Society 2009 Conference.