Price-volume Correlation in the Housing Market: Causality and Co-movements (P.10)

causal relation between prices and trading volume is more likely to be identified in
high frequency data. The highest frequency we are able to obtain is quarterly.
Our analysis uses MSAs that satisfy the following three requirements. First, a
qualifying MSA needs to exist in 1990:2 (1990 Census period) and 2000:2 (2000
Census period), and has single family housing unit data in the two periods. Second, the
MSA also needs to have unchanged definitions and boundaries in the sample period.
To check whether the boundaries have changed, we first check if the MSA name has
changed from 1990:2 to 2000:2. If the name remains the same, we manually check the
BOC historical records of boundary changes to verify if the MSA has unchanged
boundaries. 114 MSAs satisfy the first two requirements. The third requirement is that
MSAs cannot have any of the following variables missing: existing single family
home sales, total nonagricultural employment, the unemployment rate, average
housing income, single family home permits, and the house price index. Using all
three requirements, we end up with 114 MSAs,4 which are listed in the Appendix.
In our analysis, trading volume is measured by turnover, which is defined as the
ratio of existing single family home sales to the stock of existing single family
homes. Since we only observe the number of single family units for the two Census
quarters, we estimate the units in other time periods using the following formula.
In (3), for MSA i in period t, unit i,t is the single family units, completion i,t is
units completed, and demolish i,t is the units demolished.
We estimate completion i,t using permit information as well as the relation among
permits, starts, and completions. According to BOC, actual starts in new home
building are on average 2.5% more than permits and the completion rate is 3.5% less
than starts. Therefore, we assume that 100 issued permits will lead to 100  1:025 
0:965 ¼ 98:9 new units. It is also worth noting that completion of new units may not
be in the same quarter as permit issuances. According to the BOC, almost all (97%
to be exact) constructions start in the same quarter when permits are issued. Further,
20% of starts are completed in the same quarter, 49% in the next quarter, 19% in the
third quarter, 7% in the 4th quarter, and 6% in the 5th quarter or beyond. The above
relations help us estimate completioni,t using the following equation.

We estimate demolitions after estimating completions. We assume the same
number of demolished units per period,5 and calculate the demolished units per
quarter using the following equation.

Next Page 11



Leave a Reply

You must be logged in to post a comment.