There was a time when Bank of America's archoptimist David Bianco would take any economic data point, no matter how fecal mattery, and convert it into 24-carat gold. Then, in late 2011 Bianco was fired because the bank realized that its only chance to persevere was if the Fed proceeded with another round of QE, (and another, and another, ad inf) and as such economic reporting would have to lose its upward bias and be reporting in its natural ugly habitat. And while many other banks have in recent days become content with every other central bank in the world easing but not the Fed in an election year due to the risks of record gas prices, BAC's push for QE has not abated and in fact has gotten louder and louder. So exposes us to some oddities. Such as the firm's 29 year old senior economist Michelle Meyer literally demolishing any myth that yesterday's job number was "good." Needless to say, this will not come as a surprise to Zero Hedge readers. Nor to TrimTabs, whose opinion on the BLS BS we have attached as exhibit B as to the sheer economic data propaganda happening in an election year. Yet it is quite shocking that such former stalwarts of the bullish doctrine are now finally exposing the truth for what it is. Presenting Bank of America as we have never seen it before - throwing up all over the Bureau of Labor Statistics.
Booming or treading water?
The funny thing about economists is that they can look at the same set of data and draw completely different conclusions. This is partly because of different models and assumptions but also because of the wide range of data to analyze. Nowhere is this more true than in the labor market. There are a number of different indicators, including the unemployment rate, non-farm payrolls, initial jobless claims, the employment-to-population ratio, the diffusion index, and the list goes on.
In this piece, we examine the most popular indicators and summarize their message about the labor market and the overall economy. We focus on the February employment report, which showed a solid gain of 227,000 payrolls but an unchanged unemployment rate of 8.3%. Looking across the range of recent data, we argue that there are signs of true healing in the labor market. However, challenges persist which will make a more robust recovery difficult.
The magic rate
The most popularly cited labor market indicator is the unemployment rate. Normally, the level of the unemployment rate is a good measure of slack in the economy and the change in the unemployment rate is a good gauge of the speed of growth. Today, we see three reasons for caution in interpreting the unemployment statistics.
First, and most commonly observed, the unemployment rate is not only capturing changes in employment, but it is also measuring changes in labor force participation (Chart 5). The labor force participation rate measures the percent of the working-age population that is either working or searching for employment. There are two factors which determine the labor force participation rate:
demographics (such as baby boomers entering the early retirement years) and the health of the economy (perceptions of whether it is possible to find a job).
Since the recession began in December 2007, the labor force participation rate has declined over 2.0pp to 63.9%, of which 1.2pp was due to population shifts into age groups that tend to work less (Table 3). Since we cannot fight the aging process, the downward pressure from the aging population will continue.
But, we should expect at least part of the cyclical drop in the participation rate to reverse as the labor market heals. If the current pace of job creation persists and people start to perceive that there are greater job opportunities, discouraged workers will return to the labor force. This includes many young adults who went back to school rather than face a very tough job market. This means that as job growth picks up, so will the participation rate. That in turn means the drop in the unemployment rate will slow even when job growth accelerates.
Second, not all unemployment is the same. Most of the gain in unemployment comes from record high “long-term” unemployment. As Chart 6 shows, there are a record number of people who are unemployed for greater than six months. In contrast, the short-term unemployment rate has fallen sharply and returned to normal levels. This suggests labor misallocation – those with certain skill sets can find employment in short order while others can struggle with unemployment for some time.
Finally, the unemployment rate does not capture the quality of jobs. For example, it does not distinguish between full-time and part-time jobs. During the recession there was a big increase in part-time workers, which has remained high. These workers are counted as “employed” even though in reality they are half unemployed. In addition, it does not provide information about the quality of jobs in regards to compensation or skill set. This all matters because the standard inflation models use the unemployment rate as a way to determine the amount of slack in the economy, and hence wage pressure. It is therefore useful to control for these other factors when using slack models for inflation.
Bottom line: Don’t be surprised if the unemployment rate remains sticky even once job growth accelerates.
Also receiving the top billing among labor market statistics are non-farm payrolls. This measures net job creation: the number of jobs created less destroyed. It tends to be a volatile number and difficult to forecast. Nonetheless, it is one of the most influential indicators for the markets. There are two main problems with nonfarm payrolls.
First, there are significant revisions, both in the subsequent two months and also with annual benchmarks. To put this into context, the initial estimates showed job growth of 1.38 million least year; after revisions this was up to 1.82 million. This means that simply from monthly revisions, there were another 440,000 jobs created. Remember the September report which showed a payroll print of zero? That has since been revised to 85,000 job growth.
Second, there are often seasonal distortions. Poor weather conditions during the survey week, such as a snow storm, will make it difficult for people to report to work. If a non-salary worker does not receive compensation during the survey week, it will count as a job loss. The BLS attempts to control for normal seasonal swings, but sometimes that creates distortions by over-adjusting the data. This may be playing a role in the recent data given the abnormally mild winter. As Chart 7 shows, there were considerably fewer reports of people not reporting to work as a result of the weather this winter. The average from December through February this year was 170,000 workers compared to the historical average of 290,000 and clearly well below the last two years.
There are also quirks in the data unrelated to seasonal factors. For example, December payrolls were boosted by a spike in hiring of couriers and messengers as a result of online shopping during the holiday season. However, the BLS retroactively adjusted the seasonal factors and that spike has since been revised away. But, on the day of the release, this made it more challenging to interpret the report. Also, changes in hiring schedules in the public schools can create volatility in the data.
And for a secondary perspective, here is TrimTabs:
TrimTabs’ Says Payrolls Grew So-So 149,000 in February, while BLS Reports Robust 227,000 New Jobs
Real-Time Withholding Tax Data Does Not Support Big Growth Estimates
BLS’ Huge February Seasonal Adjustments (+1.53 Million Jobs) Likely Inflate Results
TrimTabs estimated that U.S. payrolls grew 149,000 in February, down from its revised January estimate of 181,000. Meanwhile, the Bureau of Labor Statistics (BLS) reported the U.S. economy added a better than expected 227,000 jobs. In addition, the BLS revised its January total up from 243,000 to a whopping 284,000 While TrimTabs’ results point to an economy stuck in slow growth mode, the BLS results point towards healthy acceleration of economic growth.
Obviously, the huge differences between the two estimates have us re-evaluating our model inputs. We did discover a problem with our January 2012 results due to the delay in implementation of the 2% payroll tax reduction in January 2011. The delay resulted in an overwithholding of payroll taxes in January 2011 and a subsequent correction in withholding tax collections in February 2011. In addition, we believe our November and December results were challenged by a larger than expected decline in seasonal bonuses. We are in the process of revising our employment estimates from November 2011 through January 2012 and will report our results this coming Tuesday, March 13, 2011.
Despite the many challenges confronting the TrimTabs’ estimate, the BLS estimate also faces numerous difficulties this time of year. Given these difficulties, we believe the BLS is likely overstating February job growth due faulty seasonal adjustments.
The BLS’ February seasonal adjustment is the second largest of the year totaling +1.53 million jobs while the January seasonal adjustment is the largest of the year totaling +2.16 million jobs. To put these huge seasonal adjustments into perspective, labor turnover during any given month is about 4 million people hired or fired. The size of the seasonal adjustments in January and February relative to the number of people hired or fired is a whopping 53% and 39%, respectively.
Looking forward, withholding tax data during the month of March will be much cleaner providing better view of growth in wages and salaries and employment because the volatility from the 2% payroll tax change in January and February 2011 will disappear from the data. Similarly, the BLS seasonal adjustments decline significantly in March and April reducing this source of error to their survey based results.
TrimTabs’ believes the anemic wage and salary and job growth results in February support the view that the economy remains stuck in slow growth mode due to numerous economic headwinds confronting the economy. Those headwinds are sluggish real wage and salary growth barely higher than inflation, elevated unemployment, waning government support, reduced and expiring tax incentives, contracting state and local governments, elevated fuel prices, and a sluggish housing market.