Employment During Recoveries from Recessions – Long Term Trends

by Mike Kimel
Cross posted at the Presimetrics blog

Employment During Recoveries from Recessions – Long Term Trends

The following graph of the employment to population ratio was obtained from the Federal Reserve Economic Database (FRED). The employment to population ratio shows the percentage of American civilians age 16 and over that happen to be employed. Recessions are shaded gray.


Figure 1.

The graph shows that between about 1960 and 2000, the percentage of civilians that were employed tended to rise over time, though those increases were interrupted with big drops in the ratio during recessions. That increase has a number of explanations, among them that women in the workplace is no longer a scandalous thing, that a single paycheck is often no longer enough to sustain the typical household, and that as manual labor’s share of the workforce has fallen retiring later has become more and more feasible.

Since 2000, the picture is more nuanced – there has still been an increase in the percentage of American civilians that are employed during non-recessionary periods, but that increase has not been enough for many people who lost their jobs during recessions find new employment. In fact, the recent drop in the employment to population ratio has brought the percentage of civilians with jobs down to levels last seen in 1983.

So what is going on here?

One clue can be seen by looking at how quickly the economy has regenerated lost jobs after each recessions. The following graph shows the employment to population ratio in months following recessions relative to the employment to population ratio in the month that a recession ends. Thus, if the employment to population ratio has risen since the end of a recession, that number should be above 100%, and if it has fallen, that number should be below 100%. Every recession since 1948 – the first year for which employment to population data is available – is shown on the graph. Because, for reasons that will be evident in a moment, I had to label each recession (or rather, post-recession period), the graph is a little busy. Note also that I am assuming that the most recession ended in June of 2009, even if the recovery has been very weak and unpleasant since.


Figure 2.

Now, assuming that the latest recession ended in June of 2009, we’re about 14 months out from last recession. The first thing the graph shows us is that for most of the recessions in our sample, the employment to population ratio continued decreasing after the recession ended; that is to say, job losses continued after the recession ended. And for four of the eleven recessions (including the latest one), the employment to population ratio 14 months after the end of the recession is still below where it was when the recession ended.

But look closer, and you see something else, something more disconcerting. It seems that the more recent the recession, the poorer the job recovery. In fact, three worst job recoveries came after the last three recessions, and the 1991 and 2001 recessions were pretty mild. (And, to repeat a point from above… the job market never quite recovered after the 2001 recession.)

Possible reasons for this long term trend in the worsening of job recoveries that come to mind include:

1. simple mathematics – as the employment to population ratio rises, recovering to that high level becomes harder and harder. Of course, the latest recession belies that, though in fairness, it was a much more severe downturn.
2. it has, over time, become easier to fire employees and move the jobs to jurisdictions where the new employees are less likely (i.e., able) to complain. In the 50s and 60s, jobs started migrating from the Midwest to the South (more union to less union), and now they’re migrating out of the country altogether. That makes it easier for a company to operate with fewer employers for extended periods of time, and thus expand for a while after the end of a recession without bringing on new workers.

Your thoughts?

Comments (10) | |