Tuesday, October 16, 2012

US Unemployment, Garter Belts, and Lemonade Stands

When the US unemployment number came out for September it was followed by a firestorm as hot as a kimchi pot in August. The Democrats were delighted that the unemployment rate had fallen below 8%. They were even more delighted that the number of people finding employment had increased by a whopping amount. Who could argue with that? The Republicans, of course! Digging deeper into the September number they found that much of that spectacular increase could be explained by part-time employment and by people moving into self-employment. One interpretation was that the reduction in the unemployment rate was a negative signal – revealing that desperate people had moved from unemployment to part-time work or had at last gasp started their own lemonade stands.

So which one is right about the implication of a falling unemployment rate in September? NEITHER IS RIGHT.  Both are wrong. Both are wrong for the simple reason that you cannot judge what is happening in the US economy from one month’s sample. The household employment survey is notoriously small and therefore it has a lot of variability from month to month. The huge increase in September, for example, followed weakness in the month before. It is possible that some of the employment that was reported for September actually came at the end of August and that the former was underestimated and the later overestimated. This happens all the time with monthly data. It is also possible that the sample was biased by unusual outcomes. For example, weather, holidays, timing of back-to-school sales, and other unusual factors may have made September 2012 different from past Septembers.  
So what we usually do is try to interpret monthly changes as best as we can but realize we need several months of outcomes before we can assign a new trend to the economy. September numbers say almost nothing about the direction of the economy.
We need to look at longer trends in this data. So I took a little digital trip to the Bureau of Labor Statistics (www.bls.gov). They are nice enough to have relevant data back to 1612 when Columbus discovered Plymouth Rock.  Because of my great respect for Great Britain and statistics, I decided to confine my hunt for data to years since 1990. And I found some interesting things.

First, let’s back-up and define what we are talking about. Each month the Labor Department surveys companies about payroll jobs. That number is reported to the public despite the fact that it does not enter into the calculation of the unemployment rate. The payroll jobs number was not in dispute in September.  It showed some modest employment growth at non-agricultural companies as it has been doing of late. This post is not about the payroll jobs number.
The Labor Department conducts a second survey each month. In this case they talk to a sample of households and ask the persons in that household about their status in the labor market. Do they have a job? Is it part-time or fulltime? Is it for a company or is it self-employment.  Does it involve Jack Daniels? Are they now looking for a job? Does the dress requirement of the job invovled high heels and garter belts? And so on. Based on the answers the government estimates the labor force, employment, unemployment, and the unemployment rate.

Second, rather than focus on month-to-month changes – let’s see how some of these key labor market indicators have behaved over more extended time periods. The largest employment category published is what they called Wage and Salary Employment in non-agricultural companies. In September of 2012 it was estimated to be 111,024. Yes, it leapt by a considerable amount compared to August of 2012. But what has it been doing over a longer time period? One benchmark is before the recession began when it was 113,298 in September of 2007. We can conclude, therefore, that household employment at firms remains lower in 2012 by 2.3 million employees.  So we are not back to where we were in 2007.  Employment in this category decreased to 106.5 million in 2009 and has been improving since the worst of the recession.  It increased in 2010, 2011, and 2012. This gradual growth is very compatible with what we have seen from the payroll survey employment figure – modest growth but so far we have not caught up to employment levels of the 2007.
Of considerable interest is the measure of part-time employment. Plotting it over time reveals that it is very counter-cyclical – meaning that it increases in recessions and then dramatically reverses in strong economic years. In the strong economic growth between 2004 to 2006 the number of part-time jobs held by household in the USA decreased by 18%. From 2006 until now part-time employment has increased by more than 100%! Now that is volatility.

But more than the volatility is the number. Between 2007 and 2009 the number of wage and salary workers decreased by 6.8 million. Part-time jobs rose during those two years by 4.5 million. Bringing us up to present, between 2006 and 2012, the number of wage and salary jobs decreased by 2.3 million persons. The number of part-time jobs increased by 4.1 million. Clearly from 2006 to 2012 there was a conversion away from full-time work to part-time.
But what about the changes occurring more recently? From 2009 to 2012 the number of W&S jobs increased by an impressive 4.4 million jobs. But we don’t know how many of those were fulltime positions. We do know that during that time the number of part-time jobs remained stubbornly high.

What we take from this analysis is a very different story than what we get from one-month changes. While things have improved for wage and salary workers in the last few years we still have about double the number of part-time workers we usually have. Progress is being made. But we are still not back to past benchmarks and that is why the unemployment rate is still much too high to satisfy anyone. Obama can gloat that employment does seem to have turned but clearly we are not back to levels of the past and we have many too many people working at part-time employment who don’t want to be there.
That leaves the issue of the self-employed. While this category of employment did increase in September 2012 there is no clear pattern in recent years. It did rise gradually between 1990 and 2007 and then has generally fallen. Since 2007 it is down by about 7%. It fell by about half a million jobs from September 2011 to September 2012. This deviation from the longer term trend suggests less confidence in self-employment.

In short, employment is growing but too slowly. Households are finding more opportunity at non-agricultural firms but many of these are part-time positions. Households have slowed their attempts to start their own businesses. The economy needs more growth to make a difference. The parties should argue about the best way to create sustainable growth – not about September or October’s employment number.
Note: this post is long enough without getting into the related issues of the ideal way to measure unemployment. The unemployment rate that gets the most attention each month considers a part-time job the same as a fulltime job. Alternative measures do not. Furthermore, other measures estimate how many people belong in the unemployed category who dropped out of the labor force. As a result there are measures of unemployment that range from 17% to more than 20%. That is the topic for a future blog.


  1. Fantastic Larry. Thanks for clear explanation. I always knew that both sides would spin a story (not surprising).

  2. You nailed it. Our would be leader's prep people are advising them to focus on the nitshit because they do not have solutions for the real isssues. Meanwhile most businesses in the US have semi frozen buying policies whether or not the consumers buy more cars or new i-phones. The nitshit can be understood by the public and is easy to spin in different ways.

  3. No not surprising Venkat. Thanks for the comment!