Saturday, January 9, 2010

Household Survey Shows No "Deceleration"

The chart below graphs monthly employment as measured by the household survey (red) and payroll employees (green). I have subtracted 7000 from the household survey in order to put them on the same scale.

The household survey shows little, if any, reduction in the rate of job loss.

Which survey should we trust? The household survey is based on a smaller sample, so for that reason economists tend to discount its monthly fluctuations. For a given sample size, economists think that the household survey is more reliable for most purposes.

Thus, if a pattern in that data is repeated a couple of months in the household survey, then sampling error cannot be blamed for creating that pattern. Although some observers look at the green series and expect an upturn soon, as I posted yesterday, there is at least as much room for employment to fall over the 4-5 months as to increase, even on the basis of pure statistics.

[My forecasting model gives another reason, separate from the household survey, that employment could fall: consumption behavior seems to indicate that consumers have long expected the recession to get a bit worse than it had gotten through Nov 2009].

No comments: