Tuesday, October 12, 2010
Don't sweat the monthly stuff. . . at least too much. . .
Although most economists are now discounting the threat of a "double-dip" recession (which in my opinion was never much of a threat to begin with), there still seems to be a lot of hand wringing every time a new monthly economic indicator is released. One month the indicator is up, then the next month the indicator is down, during the third month, it is up again. Everyone is analyzing and spinning every little monthly movement. The result? No one seems to have a clear picture of what is going on in the economy.
Our obsession with having to know the most current monthly change obscures our true understanding of where we are in the business cycle.
Why don't I like relying on month-to-month changes for my analysis of the economy? Well, first of all, those monthly changes are ALWAYS erratic--in recession and in booming expansion. It's just their nature. With that kind of movement, it's just plain difficult to extract any kind of trend. You'll notice in the first graph that charting the monthly changes in U.S. nonfarm employment shows that the gains/losses are wildly irregular. You can kind-of see a trend, but if you just focus on one month's change, you really don't have a clue what is going on. Some expansionary changes aren't much different than some recessionary changes.
Second, most monthly data is seasonal in nature. For example, employment levels are typically very high in December because there's a lot of temporary hiring for the holiday season. In cold-winter areas, construction employment tends to ebb and flow with the weather. As students flood the labor market during the summer, unemployment rates tend to rise.
Because of that seasonality, if you want to extract the business cycle trend from what just happens as a normal part of the a seasonal pattern, you need to "seasonally adjust" the data. Seasonal adjustment is a statistical procedure that uses past history to smooth out the seasonal changes in data to reveal the underlying trend. Unfortunately, there are times when seasonal adjustment doesn't work all that well. Past history isn't always the best predictor of the current situation. What if the snow comes early in the year? What if school schedules change? What if Thanksgiving is a week later than usual? Then, seasonal adjustment process can distort the current seasonal pattern.
To me, the real question needs to be: Is the economy doing better this year than it did last year? When it comes to jobs, do we have more jobs this year than we did last year? If yes (and the answer is yes), then the economy is improving. Are consumers buying more this year than last year? If yes (and the answer is yes), then the economy is improving.
Examining the year-over change (for example the September 2010 figure compared to the September 2009 number), eliminates most of the seasonality, tells me how the economy has done over the past year, and is just as current as the month-to-month change. Plus, as you can see from the second chart, I can easily decipher the trend.
Labels: Economist--Lecia Langston