Does It Matter How Aggregate Activity Is Measured?

When the latest data on unemployment, GDP, industrial production, and so on are released lots of attention is generated in the press, the blogosphere, and elsewhere. With the release of each new piece of information we hear the economy is headed towards recession or moving towards full employment depending upon who is doing the reporting and what piece of information is being reported. When measures of aggregate activity disagree, which measure should be given more weight? Does it matter which measure of aggregate activity is used to measure the state of the economy? Let’s start by looking at graphs of commonly used measures of aggregate activity.

Here are three measures of aggregate economic activity, the deviation of unemployment from its natural rate, the deviation of GDP from its natural rate, and the deviation of industrial production for its natural rate. Thus, there are three gaps, an unemployment gap, an output gap, and an industrial production gap. All are measured in real terms. The first two are created using the estimates of the natural rate of unemployment and output from the Congressional Budget Office, and the last is derived from a measure of the natural rate of industrial production I created using the natural rate of GDP from the CBO (I was curious to see how well it would correspond to the other two more commonly used measures). The measures are rescaled to make it easy to graph them on the same vertical scale. The first graph shows the unemployment and output gaps, and the second shows the unemployment and industrial production gaps:

What do we learn?  Here are some broad conclusions:

1. Over the long-run, there is very little difference in the underlying trend movements in the three measures. This shows that general trends in aggregate activity are evidenced fairly well by examining general trends in all these series, i.e. by examining the average movement across a variety of measures of activity over many quarters.
2. But what if recent data for the last quarter or two differ? For example, suppose the output gap is falling but the unemployment gap is not. In this case, the output gap movement might command more attention than the stagnant unemployment gap, though not too much should be made of a single observation in any case. A close examination of the first graph (and something more sophisticated empirical investigation verifies) shows that turning points in the output gap tend to lead turning points in the unemployment gap. Thus, data indicating a turning point in output without a corresponding turning point in employment are worth more attention than the opposite situation, a change in the unemployment gap not evidenced in the output gap measures. There also appears to be a similar lead-lag relationship between the industrial production and unemployment gaps, but it is not as evident as with the output gap. In addition, it is interesting to note that the lag in the unemployment gap relative to the output gap seems more pronounced since 1990.
3. While the three series move closely in the long-run, there is considerable variation across the three measures in the short-run. Thus, the current situation we are seeing in the data – monthly or quarterly data that appear confusing – is normal. Averaging the information content of the three measures both in the current time period and over time (and, of course, taking account of lags in relationships as well as other available data) will improve the picture when recent measures disagree as to the direction the econoomy is headed.

Comments from old site