Price-volume Correlation in the Housing Market: Causality and Co-movements (P.5)

Specifically, we find the positive correlation is almost completely explained by the
home prices and trading volume fitted by our panel VAR model. In addition, we find
that the price-caused components of prices and volume are negatively correlated,
which indicates that the Granger causality from prices to trading volume does not
appear to help explain the positive price–volume correlation. We also find that the
trading volume-caused components of prices and volume are positively correlated in
markets with high supply elasticity, but negatively correlated in markets with low
supply elasticity. Therefore, the Granger causality from trading volume to prices
does not seem to provide a good explanation for the average price–volume
correlation that is positive. Finally, the co-movement components of prices and
volume are significantly positively correlated for markets with both high and low
supply elasticity; therefore exogenous shocks seem to explain the positive price–
volume correlation well. Overall, our empirical evidence suggests that home prices
and trading volume indeed Granger cause each other, but the causal relations do not
appear to be driving the positive price–volume correlation, at least not at quarterly
frequency.
This paper is original in four aspects. This is the first study that investigates the
contemporaneous price–volume correlation in the housing market using a large panel
data set comprising a large number of markets (114 MSAs) that are arguably distinct
from each other. Second, this paper is the first to test the Granger causality between
house prices and trading volume and compare and contrast Stein’s (1995) and
Wheaton’s (1990) theories. Third, this paper is the first to empirically study the comovements
of prices and trading volume caused by exogenous economic/
demographic shocks. Finally, this paper is the first to assess the extent to which
the co-movements of prices and volume and causality between them, respectively,
help explain the price–volume correlation.
The paper proceeds as follows. The next section presents the econometric model.
“Model Specifications and Data” discusses the specification of the model and the
data. Empirical evidence is presented in Section “Empirical Evidence”. Section
“Conclusions” provides conclusions.
Econometric Model
The Model
We use a bivariate VAR model to analyze the determinants of both house prices and
trading volume. This approach has a few important merits. First, it allows us to
directly test the Granger causality between prices and trading volume. Wheaton
(1990) suggests not only that turnover and house prices are jointly determined, but
also that greater market turnover itself can generate higher house prices by reducing
sales time and increasing seller reservations, which predicts that turnover Granger
causes house price changes. On the other hand, Stein (1995) and others suggest that
house price changes should Granger cause turnover, due to equity constraints, loss
aversion, or the option value of homeowners. While all the above predictions
provide important theoretical insights, they have not been tested using large crosssectional
time series data in the literature.

Next Page 6



Leave a Reply

You must be logged in to post a comment.