Price-volume Correlation in the Housing Market: Causality and Co-movements (P.3)

Hort (2000), and no significant relation is found in commercial real estate by Leung
and Feng (2005).
Second, does the causal relation from prices to trading volume necessarily explain
the contemporaneous price–volume correlation? The causal relation, though strongly
supported by empirical evidence, more naturally implies a lead–lag relation instead
of a positive correlation. While it is possible that a lead–lag relation at high
frequency helps generate a contemporaneous correlation at low frequency, or a
correlation at the same frequency due to possible positive autocorrelations of prices,
to date no empirical study has been conducted to assess the extent to which the
causal relation from house prices to trading volume helps explain the price–volume
Third, is the price–volume correlation necessarily, or solely, due to the causal
relation between prices and trading volume? Houses are not only assets, but also
consumption goods. While the supply of many assets such as common stocks may
be fixed in the short term, the aggregate demand and supply for housing in a market
is often elastic. In fact, Smith (1976), Hanushek and Quigley (1980), DiPasquale and
Wheaton (1994), and Malpezzi and Maclennan (2001) among others, provide
evidence of negative price elasticity of housing demand and positive price elasticity
of housing supply. Therefore, shocks to the housing market may affect both home
prices and trading volume, and thus cause co-movements of them, which may lead
to a price–volume correlation. Further, there is a solid theoretic foundation for the
co-movements of prices and trading volume in housing markets. Wheaton (1990)
provides theoretical evidence that exogenous variables such as demand shocks can
affect both vacancy and sales time, which usually relate negatively to turnover, and
prices in housing markets. More theories along this line are proposed by Krainer
(2001), Ortalo-Magné and Rady (2006), and Novy-Marx (2007). While the theories
suggest that the price–volume correlation could simply be co-movements, there is no
empirical study for such co-movements and the extent to which they help explain the
price–volume correlation.2 Moreover, Wheaton (1990) also predicts that trading
volume itself can affect house prices: a higher rate of successful matching between
buyers and sellers reduces the supply of for-sale units; therefore, sellers adjust their
reservation prices upward.3 This causal relation from trading volume to prices might
also help explain the price–volume correlation in the housing market, although there
is no empirical study on this possibility at this moment.
This paper aims to shed light on these three questions using an unusually large
panel dataset which comprises 114 metropolitan statistic areas (MSAs) in the U.S.
and covers a sample period from 1990:2 to 2002:2. First, we fit to the data a
bivariate VAR model with both prices and volume (measured with turnover) being
endogenous, and estimate how exogenous variables, such as conditions in the labor
market, the mortgage market, and the financial market, and lagged endogenous
2 While there is a large literature in finance about the determinants of trading volume as well as returnvolume
relations in the stock market, this paper focuses on the theories that are specifically developed for
housing markets and motivated by the fact that houses are both consumption goods and investments.
3 It is worth noting that many housing analysts presume that volume increases lead price increases. For
example, Miller and Sklarz (1986) show that changes in sales volume in Honolulu and Salt Lake lead
price changes by one or two years.

Next Page 4

Leave a Reply

You must be logged in to post a comment.