Goldman On The Reality Of The Jobs Market
Some prefer to see the 'employment' glass half-full, some half-empty, and others see the glass smashed into a million shards on the keynesian kitchen floor. The zealousness with which the 'number' has been dismissed and praised has generated more questions than answers. Goldman's Jan Hatzius addresses the question of the pace of progress in the labor market, the reasons for the contrast between GDP and employment, the amount of slack left, and the implications for Fed policy.
Via Jan Hatzius, Goldman Sachs:
Q: How quickly is US employment really growing?
A: This is not an easy question to answer because there are a number of different measures of employment growth. However, a composite estimate is shown in Exhibit 1. It is based on a principal components analysis that summarizes the information in the establishment survey of employment, the household survey of employment, and the employment components of the ISM surveys for the manufacturing and nonmanufacturing sector. We adjust the nonfarm payroll data linearly for the preliminary benchmark revision published on September 28, which indicated that cumulative payroll growth from March 2011 to March 2012 was 386,000 higher than initially estimated. (We do not include any adjustments for subsequent months.) We include household employment growth both on a headline basis and adjusted for the definitions of the nonfarm payroll data. (In both cases, we use data smoothed for the annual changes in the so-called population controls at year-end.)
Exhibit 1 shows that since early 2010, monthly US employment growth has mostly fluctuated around a 175,000-per-month average, which corresponds to about 1.5% at an annual rate. Job growth was above this rate in late 2011 and early 2012 but then slowed notably in the spring, probably at least in part due to distortions in the seasonal factors and payback for the warm winter. However, September looks more encouraging again, mostly because of the much stronger-than-expected household survey.
Exhibit 1: Underlying Jobs Trend Has Fluctuated around 175,000
Q: Doesn't a 1.5% job growth rate seem high relative to real GDP growth, which has only averaged 2.1% since early 2010?
A: Yes, it does seem high. From an accounting perspective, it is always possible to "explain" a contrast between strength in job growth (or the closely related measure of total hours worked) and weakness in GDP growth via a slowdown in productivity growth. Indeed, since early 2010, nonfarm labor productivity growth has averaged only 1.1% (annualized), a number that is actually slightly below the disappointing trend of the period from the mid-1970s to the mid-1990s. But that simply raises the follow-up question why productivity has been so weak. From a more fundamental perspective, we see four potential explanations:
- Mismeasurement of output. One might worry that the preliminary GDP data could be understating the true pace of activity growth. However, it is difficult to find much evidence for this. Real gross domestic income (GDI)--an alternative measure of output that has been shown to perform at least as well as GDP--has only grown marginally faster than real GDP (2.3% vs. 2.1%) since early 2010. There is also no obvious sign that survey measures of activity such as the ISM manufacturing and nonmanufacturing surveys have been running ahead of where one might expect them to be in a 2%-2.5% growth environment.
- Catch-up hiring. In March 2012, Fed Chairman Bernanke discussed a closely related puzzle, namely the rapid fall in the unemployment rate despite sluggish GDP growth. His preferred explanation seemed to be that employers, having reduced their headcount excessively during the most frightening phase of the crisis, were now forced to re-hire workers as the environment was becoming more predictable. This remains a possible explanation, although it is becoming less attractive the more time passes.
- Capital scarcity. Another explanation is a temporary slowdown in capital accumulation resulting from the crisis. Even in 2011, when the crisis was already clearly in the rear-view mirror, the real stock of private nonresidential fixed assets grew just 0.9%, the slowest pace since the end of World War II except for the years 2009-2010. If this slowdown results from the underutilization of capacity over the past few years and/or the reduction in credit availability, it should gradually unwind. After all, utilization rates have been rising gradually, and credit availability for businesses has been improving. For example, today's NFIB survey showed that only 6% of small firms viewed credit as "harder to get", the smallest proportion since February 2008. (Despite the wording, we would interpret this survey as a measure of the level of credit availability, not the change.)
- More lasting productivity weakness. Finally, it is possible that productivity growth has slowed for more lasting reasons. These might include exogenous variations in the pace of technical progress or a lasting increase in uncertainty (and therefore a lasting reduction in investment rates) on the part of businesses. If so, the gap between GDP growth and job growth might stay slim on a more permanent basis.
Q: How quickly is the rebound in employment reducing the slack in the labor market?
A: The answer depends on the measure of labor market slack one employs. The unemployment rate has come down reasonably quickly since late 2009, from a peak of 10.0% to "only" 7.8% now. If we use the midpoint of the FOMC's central tendency for the long-term unemployment rate of 5.6%, this means that according to the unemployment rate, half of the slack in the labor market has now been eliminated.
However, other measures imply less progress. For example, the employment/population ratio has barely risen following a huge decline in 2007-2009. Some of this weakness undoubtedly reflects the continued aging of the population, which is lowering the labor force participation rate on a structural basis. But even on a demographically adjusted basis, participation has only unwound about one-fourth of its prior four-point decline.
Q: How is the pace of labor market improvement likely to change?
A: At least in the near term, we expect it to slow because of weaker GDP growth. Fiscal policy is likely to tighten significantly in early 2013, and the "multiplier" effects from this tightening are likely to be quite sizable. We therefore expect a renewed slowdown in real GDP growth to 1.5% (annualized) in early 2013, with risks to the downside.
However, we would note that the signals from the labor market data themselves are looking more encouraging. As we have noted recently, the flows between employment and unemployment provide a reasonable amount of information about short-term changes in the unemployment rate. When inflows into unemployment are low and outflows from unemployment are high, the headline unemployment rate has a tendency to decline, and vice versa. In particular, the "flow-consistent" unemployment rate--calculated as the rate at which the actual rate would settle over time if the inflow and outflow rate stayed unchanged--is a decent leading indicator for the actual rate. As recently as three months ago, this simple model was pointing to modest increases in unemployment. Now, however, it is again pointing to declines, as shown in Exhibit 2.
Exhibit 2: Labor Market Flows Send an Encouraging Message
Q: What are the implications of all this for the Fed?
A: The various uncertainties--about the pace of employment growth, the link between employment and output growth, and the amount of slack remaining in the labor market--provide one reason why Fed officials might not want to adopt "hard" thresholds that link the unemployment to specific monetary policy measures, or at least why it may be quite difficult to find agreement on any particular numbers. At this point, we therefore do not expect such thresholds to be adopted in the short-term, e.g. at the October or December 2012 FOMC meetings, despite the fact that the September FOMC minutes showed "many" meeting participants in favor of such a move.
Source: Goldman Sachs