Goldman Provides An Answer To Albert Edwards' Household Employment Survey Concerns
If there is one thing Goldman's Andrew Tilton did today, it was to read the earlier points on payrolls by Albert Edwards. Goldman provides a retort, which is in keeping with the party line, yet which does share some useful NFP information.
Why Does the Household Employment Survey Look So Weak?
Andrew Tilton, Goldman Sachs
The biggest mystery in recent employment figures has been the dramatic underperformance of the household survey of employment. (For readers unfamiliar with the distinction between the two surveys, please see www.bls.gov/web/ces_cps_trends.pdf or our Understanding US Economic Statistics booklet.) Over the past six months, the payroll survey has lost an average of 134,000 jobs per month while the household survey has lost an average 374,000 jobs. Yet, because of very large declines in the labor force, the unemployment rate has “only” risen one-half a percentage point over this period – bad, but much less than the employment loss would suggest. So there is really a double mystery—why is household employment declining so much more quickly, and why are so many people giving up looking for work?
The table below sheds some light on the first question. The first three rows show the payroll survey, the household survey, and the average difference of 240,000 over the past six months. Over the past year (rightmost column), the difference has been somewhat smaller, but still more than 100,000 jobs per month.
However, we need to make two adjustments for a fairer comparison. First, payrolls are about to be revised down significantly. The Labor Department’s preliminary estimate is a downward revision of 824,000 to the level of payrolls in March 2009. Since all of this represents underperformance in the time period since the March 2008 benchmark, it implies that the average monthly drop in that interval was about 69,000 jobs steeper than originally estimated. If we assume as a baseline that this correction is extrapolated forward linearly—note that the actual procedure could well result in a bigger or smaller correction—we would see correspondingly weaker numbers for more recent payroll reports as well (line 4). Second, we need to adjust household data to the payroll definition (line 5) for an “apples to apples” comparison. Doing this does relatively little to the gap over the past six months (it’s still 224,000 jobs per month), but the relatively small gap between the two over the past year (39,000 per month) suggests that much of the recent underperformance is a correction of earlier outperformance. The remainder, as we have discussed previously, we suspect is due to the underperformance of small and mid-sized employers, which the household survey may capture better in real time. (The lag for the payroll survey to capture this underperformance also may help explain the large downward revision.) So, the employment discrepancy looks less mysterious with closer analysis.
What about the discrepancy in labor force participation? Over the past year, the share of Americans participating in the labor force (either employed or actively looking for work) has fallen by 1.2 percentage points to 64.6%, the largest one-year drop on record. As a result, the number of people classified as “not in the labor force” has grown by 3½ million. If these people began looking activity for work, the unemployment rate could rise sharply.
We believe the best explanation for this drop in the labor force is a shift by many younger workers to get additional education or training. Three points in evidence:
1. The drop in participation is particularly evident among the young… The table below shows the drop in labor force participation over the past year by age group, both on an absolute basis and relative to the trend for that group in the years before the recession.
2. …and the less educated. Relative to pre-recession trends, the biggest drop in labor force participation has come among people who do not have a high school diploma. The smallest drop was among those with a college degree.
3. Most of those leaving the labor force say they do not currently want a job. While the category “not in the labor force” has grown by 3.5 million over the past year, only about one-sixth of this increase has come in the “marginally attached” category of people who want a job and have looked for one in the last year. The vast majority are not looking for a job at present.
While not definitive, these facts are consistent with an explanation that lots of younger people see limited employment prospects in the near term, have decided to get additional schooling or training, and are therefore not currently looking for work. Anecdotally, enrollment in postsecondary education has increased, although at the time of this publication we do not have comprehensive figures for 2009-2010 academic year enrollment. Insofar as this explanation is correct, many recent “labor force dropouts” are quite likely to look for work again at some point in the next few years, as opposed to retiring or dropping out permanently.
Even accounting for demographic shifts in the labor force, the recent decline in the labor force participation rate is at the extreme end of prior recessions. We therefore suspect that the participation rate will stabilize over the coming year, implying that outright job growth of more than 100,000 per month will soon be needed to keep the unemployment rate from rising further.