Price-volume Correlation in the Housing Market: Causality and Co-movements (P.17)

Table 2 also suggests that lagged prices and trading volume significantly affect
home prices and trading volume. The first-order autoregressive coefficients are
significantly negative for both prices and trading volume, therefore, prices and
volume tend to reverse in the next quarter, which may indicate the adjustments of the
housing market to exogenous shocks. An adjustment of market supply to a demand
shock appears consistent with the feedback effects predicted by Novy-Marx (2007).
The negative coefficients of lagged prices and trading volume are also consistent
with the overshooting of home prices predicted by Ortalo-Magné and Rady (2006).
Table 2 also reports the tests of Granger causality between prices and turnover.
We find strong evidence that prices Granger cause turnover, with the F statistic being
4.811, and the P value being 0.002. This directly supports the theory by Stein
(1995), and is consistent with empirical evidence provided by Chan (2001),
Engelhardt (2003), Genesove and Mayer (1997), Genesove and Mayer (2001), etc.
At the same time, we find weak evidence that turnover Granger causes prices, with
the F statistic being 2.436, and the P value being 0.063, which provides some
evidence for the part of the theory in Wheaton (1990) that suggests turnover might
affect prices, particularly in tight markets.
Table 3 reports the results of the second specification, which separates positive values
from negative values for lagged log differences of house prices and thus accommodates
asymmetric effects of house prices on turnover. While almost all results in Table 2
remain, we find that, in the equation with turnover being the dependant variable, the
negative value of one-quarter lag log difference of house prices is significantly positive
at the 1% level, while the positive value and all other lagged house prices are
insignificant. This indicates that decreases in house prices reduce market turnover, but
increases in house prices do not have significant effects. The result is consistent with
theories in Stein (1995) etc., which suggest that equity constraints or loss aversion due
to decreasing house prices reduce market trading volume.
Tables 4 and 5 report the results for MSAs with high and low supply elasticity
respectively. While our early results remain, these tables reveal interesting differences
across markets with different supply elasticity. First, turnover Granger causes
prices in tight markets (MSAs with low supply elasticity) but not in loose markets
(MSAs with high supply elasticity). Relating to Wheaton (1990), this seems to
indicate that sellers more likely raise their reservations as a reaction to increasing
trading volume in markets with an inelastic supply of housing. Therefore, in tighter
markets, due to the lack of new homes, sellers are able to profit more from
increasing housing demand. Second, the results seem to suggest that homebuyers in
tight markets are less financially constrained. The first piece of evidence for this is
that house prices in tight markets are less sensitive to mortgage interest rate levels
and trends, possibly due to the fact that homebuyers are less financially constrained.
This may have interesting implications on the risk of home equity: although houses
tend to be more expensive in tight markets, they are less risky in the sense that their
prices are less sensitive to mortgage interest rates. The second piece of evidence is
that growth in average household income has weaker effects on house prices in tight
markets than in loose markets, which seems to suggest that income is less likely a
financial constraint for homebuyers in tight markets with high house prices. Third,
growth in employment has stronger effects on house prices in tight markets than in
loose markets, which is sensible given the low supply elasticity in tight markets.

Next Page 18



Leave a Reply

You must be logged in to post a comment.