Former Vice President of the Federal Reserve Bank of Atlanta, Bob Eisenbeis, argues the Fed's best course of action this September is for an itty-bitty tiny taper, lest the market suffer shock...
Right now, the FOMC states it will hang the initiation and pace of the tapering of its $85 billion monthly asset purchase program on “incoming data,” and more specifically, on data reflecting the condition of the labor market. Markets and the press have placed undue focus on the most current release of the jobs numbers, as if this report will determine policy. But the market’s emphasis on short-run jobs data as the key trigger for policy is too simplistic for several reasons.
The first reason is that the headline jobs number lacks precision. The employment numbers are taken from two sources. The more frequently cited number is the BLS’s CES (Current Employment Statistics), or establishment series. The other source is the BLS’s CPS (Current Population Survey), or household survey, which is used in computing the unemployment rate. Both of these monthly statistics are subject to significant measurement errors. The error rate in measuring the monthly change in employment is smaller for the establishment survey than for the household survey, in large part because the sample size is larger. BLS states that the household survey is broader in coverage because it includes categories of workers, such as self-employed people, agricultural workers, and others who are not included in the establishment survey. In either case, the potential estimation errors are large for two main reasons: measurement errors and problems with the seasonal adjustment factors. For the CPS the 90% confidence interval for the headline unemployment rate is + or - 0.2, and for the more widely publicized CES it is + or - 90K. In other words, any change in the unemployment rate smaller than 0.2 percentage points is not statistically different from zero, and for the jobs number any change less than 90K is not statistically different from zero. Thus, for the 169K change in the jobs number for August, the actual increase could have been as small as 79K or as large as 259K. Similarly, for the most recently reported unemployment rate, which reportedly declined from 7.4 to 7.3, the actual unemployment rate is expected to lie somewhere between 7.5% and 7.1%. The reality is that a more precise estimate might find that unemployment, in fact, had either declined, increased, or not changed at all. Given the size of the margin of error, for the press or the markets to get excited about the most recent job numbers is to make much ado about nothing.
But the situation is even more complex when it comes to the jobs numbers, because they can also be subsequently revised. So what you thought may have happened may not be reality at all. The current CES release is revised twice in the succeeding two months, with the first revision coming one month later and the second two months later. Why is this important? Consider that the first reported jobs number for July 2013 was 162K. That, combined with the August number of 169K, would suggest a stable and perhaps improving job market. But the July number was revised down from 169K to 104K after one month. So July was not necessarily what we thought it was, and perhaps August may not be either, especially since other information from the NFIB survey of small businesses, for example, suggests that job creation was, at best, tepid. Furthermore, we also learned that the numbers for June had been revised down, not once but twice. The initial reported jobs creation number for June was 195K. That number was revised to 188K when the July data came out; and the final number, which came out with the August report, ended up at 172K.
So what did the FOMC actually know at the June meeting, which concluded with Chairman Bernanke’s press conference and all the initial talk about the Fed’s beginning the tapering process? The following table shows the path for job creation from the CES survey that was available at the June FOMC meeting.
At the June meeting the FOMC had only the following numbers:
April, 149K (down from first release of 165K)
March, 142K (up from first release of 88K and from second release of 138k)
February, 332K (up from first release of 236 and from second release of 236)
According to that data, the economy had created on average 199K jobs the previous four months – a reasonably positive picture and perhaps enough to justify the tapering discussion at the June press conference. Fast-forward to what we knew going into the September FOMC meeting. We have the following data, including the revisions:
July, 104K (down from 162K)
June, 172K (down from first release of 195K and from second release of 188K)
May, 176K (flat from first release of 175K but down from second release of 195K)
April, 199K (up from first release of 165K and up from second release of 149K)
The picture now suggests an economy that created on average 165K jobs for the previous four months, not the more robust 199K jobs that was the input for the June FOMC meeting. Would we be having the same discussion regarding the need for tapering if the FOMC had had access in June to the revised, more accurate employment picture?
There are obviously other dimensions to the employment situation besides the headline numbers and the employment path they imply. We know that there remains a huge gap between headline numbers and the situation on the ground when we consider the large number of people who are underemployed – who are working part-time but want full-time jobs (as reflected in the U-6 statistics) – and the continual decline in the participation rate, as more people give up looking for work. Similarly, and on a more positive note, today’s data for new claims for unemployment insurance dipped below 300K to 292K. The last time this happened was in April 2006 before the start of the financial crisis. While this release may be viewed positively by markets, the weekly claims numbers are also quite volatile, with a mean absolute deviation of 15K from the end of the most recent recession until today. As well, the data are subject to subsequent revisions. All of these factors enter into a full assessment of the employment picture, as FOMC participants have struggled to make clear.
So what does the readjusted, still-emerging employment picture mean for Fed policy at next week’s meeting? The economy is not growing at a rate that will create jobs at a sufficient pace to reduce the unemployment rate and improve the employment situation until well into 2014 or perhaps 2015, which is a picture consistent with the FOMC’s previous projections. But the Committee has also convinced markets that tapering is imminent. Under the circumstances, backing off from that message would be disruptive to markets. On the other hand, a very modest tapering, on the order of a reduction in purchases of $5 billion a month, would enable the Committee to act without shocking markets that have already priced in a tapering. It would also preserve flexibility for the reconstituted 2014 FOMC to formulate and implement its own policy, largely independent of what the current FOMC has put in place.