This figure shows those series (highlighted in red) which had coefficients on the region dummy that were statistically different from zero at 5%. The left side of the figure identifies the product and the top row identifies the new Timber Mart-South region. The Timber Mart-South regions that had no boundary changes are darkened (FL1, LA2, MS2, SC1, TX1, TX2, VA1).
The region with the most series with significant dummies was South Carolina Region 1, which saw no change in its borders. Indeed, it seems that the likelihood of a significant shift is unrelated to whether or not a boundary change occurred. Six significances out of the 42 series in the seven regions that were not reconfigured is roughly the same percentage as the eleven out of 84 significances in the regions that were reconfigured.
Further, there appears to be some spatial correlation in the series which had statistically significant dummy coefficients. Almost all dummy coefficient estimates in these series were positive. Many of the series were in regions surrounding the southern Appalachians and other interior portions of the South. This spatial pattern and similar direction of shift could point to some weather-related or economic phenomenon common to those regions with significant shifts.
Finally, we compared the shift seen in the TMS Mississippi pine sawtimber series with that from a similar but independently-collected price series from the State of Mississippi and found the shifts to be identical, arguing that the shift is more than a methodological artifact.
