2011/04/28

5 Government Statistics You Can't Trust


Widely watched economic gauges have some serious flaws

As pattern-seeking creatures, statistics have a peculiar hold on our minds. Quite a large number of decisions are undertaken on the basis of what statistics tell us. That is certainly true when it comes to government economic data and the blizzard of stats that come out every month.

Billions of dollars’ worth of value appears or vanishes on the basis of what these numbers say about the health, growth and direction of the economy, and the implications for company profits, interest rates and so on.
Some of that faith seems to be misplaced. While great reliance is placed on government economic numbers and the financial media report on them at length, in-depth discussions of how the numbers are created — and where the weaknesses may lie — is relatively rare.

Unfortunately, those gaps are significant.

Unemployment

Two surveys examine employment — the household survey and the payroll survey. While many seem to think that the larger sample size of the payroll survey makes it more accurate and reliable, from a statistical standpoint the household survey's design is more sound, and the margin of error is usually better.

That said, the unemployment numbers give a few examples of the problems of government statistics. Starting during the 1960s, the methodology was changed to exclude discouraged workers — people who are out of work and have met with so little success in finding a new job that they have quit trying. This had the instant effect of lowering the unemployment number.