Yesterday, the Commerce Department released its New Residential Sales report, which seemed to show a surprising gain in new home sales -- but only after a surprisingly large downward revision to previous sales numbers. From the report:
Sales of new one-family houses in October 2007 were at a seasonally adjusted annual rate of 728,000, according to estimates released jointly today by the U.S. Census Bureau and the Department of Housing and Urban Development. This is 1.7 percent (±11.0%) above the revised September rate of 716,000 and is 23.5 percent (±10.3%) below the October 2006 estimate of 952,000.
The previously reported September rate was 770,000 -- so the revision in September's new home sales number to 716,000 is HUGE, for two reasons. For one, it means that the new home sales numbers reported for September -- which at the time seemed to show a gain of 4.8 percent -- really landed in negative territory, losing 2.6 percent. For another, it means the phantom gain now appearing for October is most likely the result of comparing apples to oranges, taking an adjusted number and comparing it to an unadjusted number that will eventually be adjusted (and probably downward). And that doesn't mention the actual magnitude of adjustment for September; going from 770,000 to 716,000 is a very large revision. Calculated Risk has some good discussion about this and the estimating process, worth your read. It's worth noting the confidence intervals given above for each estimate; note that the 23.5 percent drop is in negative territory throughout the error margin of error provided.