Thursday, October 25, 2012

A Serious Look At The December Jobs Report

Economists tend not to be very good at economics. We know this because almost none of them were able to see the $8 trillion housing bubble that was driving the economy from 2002 to 2007. This was an oversight of astonishing importance, sort of like a physicist not noticing gravity.

Their failure to understand the economy has led to enormous misreporting of the December jobs data. There are two basic problems. They fail to accurately put the job growth numbers in the context of the economic downturn and they badly misread the December data leading them to overstate the true growth path we are now on.

Taking the two in turn, the reports were full of the good news that the economy had created 200,000 jobs and the unemployment rate had dropped to 8.5 percent. Creating 200,000 jobs is undoubtedly better than creating 100,000 jobs and much better than creating no jobs at all, but is this good?

From December of 1995 to December of 1999 the economy generated more than 250,000 jobs a month, and that was starting from an unemployment rate of under 6.0 percent. We expect more rapid job growth following a steep downturn like the one we saw in 2007-2009.

In the two years following the 1981-82 recession the economy generated over 300,000 jobs a month. Following the 1974-75 recession, the economy generated more than 340,000 jobs a month in the two years from December 1976 to December 1978, and this was with a labor force that was only 60 percent of the size of the current labor force. So we're supposed to be happy about 200,000 jobs in December?

Another way to think about this is that we currently have a shortfall of around 10 million jobs. If we generate 200,000 jobs a month, then we are cutting into this shortfall at the rate of 100,000 a month, since we need 90,000-100,000 jobs a month just to keep pace with the growth of the population. This means that in 100 months we should expect to be back to full employment. So the champagne bottles for that happy occasion will be dated 2020.

Okay, but this puts too bright of a picture on the data. The 200,000 jobs number reported for December was distorted by unusual seasonal factors, the most obvious of which was the 42,200 job growth reported in the courier industry. This is primarily companies like Fed Ex and UPS who hire additional workers to deal with holiday demand.

In principle seasonal adjustments should remove the impact of seasonal fluctuations, however these adjustments are always based on historical experience. When there is a sharp departure from historical patterns, like the explosion of Internet sales, the seasonal adjustments will not pick this up. We have good reason for believing this to be the case here because in 2010 the Labor Department reported an increase of 46,300 jobs in the courier industry, all of which disappeared the next month. In 2009, it was 30,100 jobs reported in December that all disappeared in January.

Here's the picture:

Employment in Couriers and Messengers (seasonally adjusted)

Source: Bureau of Labor Statistics.

What should we infer from this? We should assume that most, is not all of these 42,200 jobs reported in December will disappear in January. That puts our jobs number around 160,000. There were some other unusual factors that may have pumped the numbers in December slightly. Construction employment reportedly rose 17,000 in December after falling 10,000 in October and 12,000 in November. Did we turn the corner in the construction industry? Well the sector added 31,000 jobs in September. Construction employment is very erratic because of the weather. We had a relatively mild December in the Northeast and Midwest, which means that we would expect better than usual construction employment. Don't bet on this one being part of a trend.

There were a few other anomalies of less consequence in both directions, but a clear-eyed look at the December data puts the job growth at around 150,000. If we take the average job growth over the last three months we get roughly 140,000. Maybe we have a slight pick-up, but probably not much more. At 150,000 jobs a month, the full employment champagne bottles will be dated 2028.

What about the drop in the unemployment rate, surely that is good news? Well the unemployment data come from a separate survey of households. This survey is much more erratic than the establishment survey due to the fact that it has a much smaller sample.There are often large movements in this survey that clearly cannot be explained by movements in the economy.

For example, the survey showed the unemployment rate falling from 4.7 to 4.4 percent in the second half of 2006, a period when GDP growth averaged 1.4 percent. It then rose back to 4.7 percent in the first half of 2007, a period when growth averaged 2.1 percent. The monthly employment changes can be even more erratic. In the four months from July to November 1994, the survey showed the economy adding almost 1.8 million jobs or 450,000 a month. This was a period in which the economy was growing at a healthy, but not spectacular, 3.6 percent annual rate.

More recently, the survey showed employment plunging by 423,000 last June. Fortunately no one thought to seize on that change as marking the start of another recession. Over the course of a year, these erratic movements largely even out. If we look at employment from December of 2010 to December of 2011, it increased by 1,570,000 in the household survey. This is telling us pretty much the same story as the rise in payroll employment over this period of 1,640,000 jobs.

The other point to remember is that the unemployment rate is telling us not how many people are out of work, but rather how many people are out of work and looking for jobs. Many people give up looking for work if they feel their job prospects are hopeless. A better measure for most purposes is the employment to population ratio (EPOP). By this measure, we have made little progress since the trough of the recession.

The 58.5 percent number for December is up just 0.3 percentage points from the trough of 58.2 percent hit last summer. By comparison, the EPOP hit a peak of 63.4 percent in 2006. We still have almost 5 percentage points to go before we get back to this pre-recession peak. Or to put it slightly differently we have made up just 6 percent of the lost ground.

Employment to Population Ratio

Source: Bureau of Labor Statistics.

In short, a serious look at the December report does not provide much cause for celebration. The economy is still in very bad shape and the current growth path provides little hope for much relief any time soon. Economists should know this, but unfortunately few seem to pay much attention to the data. Remember the double-dip recession?

No comments:

Post a Comment