Comment 45: What the employment statistics tells us and what it does not
The equity and financial markets are possessed by the monthly release of the employment data. If the number of jobs added in the non-farm payroll survey is large, the markets anticipate a robust economic growth and a heightened inflation pressure ahead. If the job creation is weak, the markets fear that an economic slowdown is imminent but hail a benign inflation picture. In one word the markets are viewing the employment report as an important leading economic indicator. The financial media are also jumping on the bandwagon by providing economic commentators that laugh and cry according to the monthly release of the employment data. This kind of fanfare about the employment report, on the other hand, has created substantial confusion and cynicism, since recent employment reports show a sustained healthy job creations whereas the GDP report is displaying an unmistakable picture of an economic slowdown. In this comment we will discuss the true meaning of the employment data and what they are saying about the current and future economic condition.
Monthly changes of the number of jobs in the non-farm payroll survey reflect the collective wisdom of business managements about the future course of the economic conditions in general and the prosperity of their own businesses in particular. Business managements are not prophets. They look at the orders and sales streams of their own businesses and listen to the fortune-tellers on the financial media to form their views. They are trend followers rather than trend setters. Thus the number of jobs created monthly is usually a lagging economic indicator. From the following study of the data since 1992, we only see a small number of cases that the job creation data has served as the coincident indicator, but never as the leading indicator. Apparently it is dangerous to consider the job data as a leading indicator and make investment decisions according to the ebbs and flows of the monthly job report.
Monthly job creation data of the non-farm payroll survey are volatile and subject to frequent and substantial revisions. In order to extract the underlying trend of the data we take 12 month moving averages of the monthly job creations of the non-farm payroll survey and plotted them as the purple curve in the graph at the right. The graph covers the period from the beginning of 1992 to June of 2007 (Note: the graph may be updated later on, but not the content of this comment). The purple curve includes both the job creations in the private sectors and the public sectors. The public sector jobs are supported by tax revenues. If the public sector jobs increase, more taxes need to be raised, and thus less money for the private sector employees to spend. As a whole the increase of public sector jobs will be a wash and does not imply a more robust economy. Thus we must look at the private sector jobs only if we want to relate the job creations with the real economy. 12 month moving averages of monthly job creations in the private sector alone are plotted as the green curve in the graph. For comparison, 4 quarter moving averages of quarterly real GDP growth rates are plotted as the red curve. A close inspection reveals that the purple curve and the green curve are synchronized from 1992 up to 2005, but diverged since 2006. It does not matter whether we use the purple curve or the green curve before 2006, but since 2006 it is the green curve that is synchronized with the ups and downs of the 4 quarter moving averages of real GDP, the red curve. Thus we must use the private sector job data only in order to correlate the job data with the real economy. That is the reason why some are frustrated when they use the purple curve, contaminated by the public sector jobs, to compare with the real economy and failed to find correlation between the two after the end of 2005.
Once pinpointed that the private sector only curve, the green one, is the curve to be compared with the real economic growth, let us go through the ups and downs of the two curves (green and red) to see how they are correlated. The growth rate of real GDP had notably dipped during 1993, but the green job creation curve of the private sectors had not shown any dip during the period. A detailed inspection of the real GDP components at that time revealed that the sharp runup and then the dip of 1992 to 1993 was due to the swing in the government spending but not in the private sector. Thus domestic US businesses naturally ignored such swings in the government spending and continued to add workers through that period, that is, the early stage of an economic recovery. The weakness of real GDP growth in 1995 was due to the weakness of consumer spending that directly affected domestic businesses. The business managements were slow to realize the economic down turn, and thus the green curve lagged behind the red real GDP curve by a few months in the slump of 1995. At the bottoming out and then the rebound processes of the real economy, business managements were also a little late in their response, causing the green curve to lag behind the red one. After the long stretch of steady economic growth from 1996 to 2000, the stock market crashed first and served as a warning sign, so businesses were able to respond at the earliest sign of a economic slow down, and immediately cut the job creation, causing the green curve to slump simultaneously with the red one. However, as the real economy turned up in 2002, businesses were again lagging behind. By 2006 FED had raised interest rates many times, and businesses have naturally become jittery again. Under those uneasy conditions, they reacted quickly to the economic slow down and cut back on new hirings. Thus the green curve has come down with the red one in 2006. In recent months the green curve has flattened out; the real GDP data for the second quarter of 2007 is not yet available. What can we say about the upcoming real GDP? From the close synchronization of the red curve and the green one since the beginning of 2006, we may expect that the red curve will also go flat in the second quarter of 2007. Since the red curve is the 4 quarter moving average of real GDP growth rates, the second quarter 2006 data (+2.6%) needs to be purged and the second quarter 2007 data added to calculate the red curve for the second quarter of 2007. Thus we may expect that the real GDP in the second quarter of 2007 will grow about 2.5% from this consideration. The real personal consumptions in April and May were quite weak. Unless those data are revised upward substantially or the June data shows an unexpected surge, the second quarter personal consumption expenditure will grow less than 2.5% (annualized rate) from the first quarter data. If the decline in the fixed asset investment sector moderates substantially, we may expect a real GDP growth rate between 2.0% to 2.5%, not very far from the estimate based on the employment data. If the growth rate of real GDP persists at a low level, like below 2.0% that is the lower end of the band as anticipated in Comment 38, through the second half of 2007, we should expect that the average job creation in the private sectors will also come down to near 100,000 or below in the latter half of 2007.
Along side the non-farm payroll data, the job data based on the household survey are also released monthly. The household survey is based on a survey of 100,000 households, thus roughly a survey of 200,000 workers. This should be compared with the non-farm payroll survey that surveys 160,000 businesses; the number of workers surveyed is many times that of the household survey. The small number of workers surveyed in the household survey makes the monthly changes of the number of workers reported gyrate much more widely than the non-farm payroll survey, and render the household survey not useful in extracting short to medium term trends. However, there are some ratios calculated in the survey that are not subject to the wide monthly gyrations. The best known is the unemployment rate. However, the unemployment rate is always controversial due to the question of how to define a worker as unemployed. In the household survey only jobless workers who are diligently looking for jobs are classified as unemployed. In the US unemployment benefits only last for a short period. After the exhaustion of the unemployment benefit, many jobless workers tend to look for job less diligently, and thus are classified as not-unemployed by the household survey. This should be compared to the cases of Western Europe where the unemployment benefits are much more generous and last for many years, so unemployed workers are forced to look for jobs more diligently in order to qualify for the benefit for a long time, so are properly counted as unemployed. Thus the unemployment rates in the US are always much lower than in Western Europe. The monthly values of the unemployment rate without any smoothing process are plotted as the black curve in the graph; the scale of the black curve is upside down as the black % numbers at the left margin of the graph show. We can see that the unemployment rate falls steadily during good economic times and rises in a recession with some time delays, but does not trace out detailed ups and downs of the real economy. The unemployment rate was above 7% in 1992, at the time that the US economy had just come out of the recession of the early 1990's. The rate dropped as the economy recovered. The 1995 economic slowdown only caused the rate to stagnate, not to rise. The unemployment rate then marched down steadily through the long lasting economic expansion of the late 1990's. The rate rose as the recession of 2000-2002 set in, and started to drop again in 2003 as the current recovery has gathered steam. The economic slowdown of 2006 seems to cause the unemployment rate to stagnate at 4.5% level.
Another important rate contained in the household survey is the labor force participation rate. This rate gives some insight of the labor structure of the society that neither the detailed non-farm payroll survey nor the gross unemployment rate can provide. This rate is simply the ratio of the number of people in the labor force over the working age population. The labor force is the sum of all employed plus the unemployed as defined by the household survey. Monthly values of the labor force participation rate are plotted as the blue curve in the first graph. From 1992 to 2001, the labor force participation rate more or less synchronized with the unemployment rate, and has dipped together as the 2000-2001 recession set in. However, the labor force participation rate has failed to bounce back as the economic recovery takes hold, and started to go down again recently as the growth rate of real GDP drops. To uncover the reason of this divergent behavior of the labor force participation rate, semi-annual average of this rate is plotted as the red curve in the second graph at the right. The semi-annual averages of the real GDP in the conventional logarithmic scale are plotted as the black curve in the graph. There is some tendency for the peaks and the valleys of the red curve to align with those of the black curve. However, besides the notable divergence after 2000 as mentioned above, there is also a notable dip from 1956 to 1965 in the red curve that is comparable in the severity and the length with the current slump of the rate. To uncover the underlying reason of that earlier dip, the annual ratios of the number of students of ages 18 or above over the non-institutionalized (means not in prisons) civilian population are plotted as the blue curve in the graph; this student related rate is called "higher education student ratio" here. From 1956 to 1965 the higher education student ratio shot up nearly 1.5%, matching the decrease of the labor force participation rate of similar amount during the period. Thus we can attribute the dip of the labor force participation rate in the period of 1956 to 1965 to the increased rate of high school students attending colleges and universities. The reason of this rush toward higher education at that time was two fold. At the beginning of that college boom the whole society of USA was encouraging young people to attend colleges from the fear that USSR had leapfrogged ahead in science and technology as manifested in the Sputnik shock. At the tail end of this period, the baby-boomer generation came into age and started to attend colleges, thus boosting the higher-education student ratio. The sudden jump of the blue curve from 1964 to 1965 was due to the rush to attend colleges to avoid Vietnam War drafts on top of this natural phenomena of the push from the baby-boom generation. It is interesting to observe that the labor force participation rate started to turn up strongly since 1965 and peaked around 1990 with short interruptions matching the onset of several recessions. This strong rise of the labor force participation rate was achieved in spite of the continuous rise of the higher-education student ratio. The first factor of this strong upward movement was due to the influx of the baby-boom generation into the labor market. This baby-boom effect should have peaked around 1976 as the higher education student rate (the blue curve) indicates. Another factor behind the great boost of the labor force participation rate was the liberalization of women that induced many women to leave home and joined the labor market. Those two factors probably overwhelmed the not-so-strong economy of the 1970's and boosted the labor force participation rate up to 64% by 1980. After the high-inflation recession of early 1980's, the first phase of the run-away trade deficit set in under the inducement of the globalization. This borrow-from-foreigners-and-spend routine promoted the creation of service jobs that had accomodated more women and many youngsters from the mini-baby boom into the labor market, so the labor force participation rate soared further. The labor force participation rate only surged moderately during the economic boom of the late 1990's. This was the manifestation that the late 1990 economic boom was the result of the second phase of the runaway trade deficit, and many manufacturing jobs were vanishing fast. After the 2000-2002 recession, the higher education student ratio surged by about 0.5%, but the labor force participation rate dropped more than 1%. This means that there is a drop of 0.5% in the rate that cannot be accounted for by the increased enrollment into colleges. We can easily see that many people are forced to retire early from the restructuring and the off-shore outsourcing. Even those increased enrollment in higher education probably come from the young people that cannot find adequately paid jobs and opted to attend higher level educational institutions not necessary from their own desire; this effect is vividly visible in the high tech industry that is reluctant to hire newly graduated scientists and engineers in favor of lower paid talents in Asian countries. Those forced early retirees and students parked in the higher level educational institutions are, of course, not counted as the labor force in the household survey so the unemployment rate is kept low, but the effect shows up in the labor force participation rate.
This analysis does not paint a rosy prognosis for the labor force participation rate. Contrary to the wish of many financial observers, economic cycles are not repealed by the globalization, though the length of an economic cycle has definitely been lengthened. As the next recession hits, the stagnant labor force participation rate will slump one step further and will not be able to bounce back under the increasing pressure of the runaway-trade-deficits. Unless the current globalization scheme is drastically revised and the runaway US trade deficit stopped at its track, the labor force participation rate will fall substantially from the current level by 2020 from the pressure of the trade deficit alone. We should note that from 2010 and on the baby boom generation that are fortunate enough not to be forced to retire early will start to retire. This will put more pressure on the labor force participation rate. We should not be surprised if this rate is heading toward 50% level within two decades. In an extreme trade deficit society as depicted in article 7 the labor force participation rate would have dropped to the single digit level, but the unemployment rate can still be held around 5% level by simply classifying the majority of the idled citizens as not-unemployed according to the definition of the household survey. On the other hand, in the extreme trade surplus countries of article 7, the labor force participation rate must be extremely high and citizens in such societies must work very hard in order to support the parasite-like life style of the citizens of the extreme trade deficit society.