16 June 2014
All examples in this report are hypothetical interpretations of situations and are used for explanation purposes only. The views in this report reflect solely those of the authors and not necessarily those of CME Group or its affiliated institutions. This report and the information herein should not be considered investment advice or the results of actual market experience.
To interpret or anticipate Federal Reserve policy actions, one must inhabit the mind of the Fed chairman, and see what he (or she) sees. Under Paul Volcker, money supply data ruled the Fed from the late 1970s through the mid-1980s. Labor market data moved to the fore under Alan Greenspan, then rose to target status under Ben Bernanke. Now, Janet Yellen is elevating discernment of US labor market data to new heights.
In this note, we discuss how to analyze labor markets from the perspective of monetary policy-making, and we highlight crucial differences in interpretation that could fuel debate, both inside and outside the Fed. Our central theme is that interactions among labor market conditions, economic growth, inflation, and inflation expectations are inherently dynamic. In the words of the venerable Joan Robinson, "There is no such thing as a normal period of history. Normality is a fiction of economic textbooks."1 Accordingly, any interpretation of labor market data built on assumed reversion to some "average" or "normal" economic state is likely to be wrong.
To suggest how wrong it might be, we contrast the "standard view," premised on a post-recession return to normality, against a "dynamic view," which highlights the challenges of interpreting labor market data in light of the aging of the work force, secular deceleration of labor force growth, and the impacts of technological innovation upon business structure and hiring practices in the retail and service sectors. (Throughout, interested readers may consult footnotes for technical details of data collection and compilation.)
In terms of debate over US monetary policy, our interpretation suggests that US labor markets are more robust than the standard view, which assumes reversion to a fictive normal state, might lead us to conclude. It also suggests, more importantly, that the implications for inflation are decreasingly clear, because macroeconomic linkages between labor market conditions and inflation have loosened up in recent decades. Indeed, our analysis encompasses two very different scenarios for how interest rates and bond markets might react, if and when the Fed begins raising short-term interest rates.
If the Fed moves to tighten credit market conditions in response to improving labor market conditions, but without core inflation rising above 2%, the result is apt to be a flattening of the yield curve. That is, bonds would not necessarily sell off just because short-term interest rates move from near-zero to 0.5% or 1.0%.
Alternatively, if improving labor market conditions prompt the Fed to raise short-term interest rates, and with inflation rising through 2% toward 3% or higher, market participants may conclude that the Fed is "behind the inflation curve," resulting in an up-shift in yields across the entire term structure. One variant of this scenario is that the yield curve might even steepen temporarily in anticipation of a belated credit-tightening move by the Fed.
Either way, what will matter for bond values is whether improving labor market conditions coincide with mounting inflation pressures. The key challenge in this is that strengthening labor markets are likely to prove an unreliable indicator of future core inflation.
In what follows, we interpret various familiar labor market performance gauges, starting with long-established favorites from the Greenspan and Bernanke eras. Then we highlight two more -- labor force growth, and aging of the labor force -- that we believe are under-appreciated. Lastly, we turn to the macroeconomic linkage between inflation and labor market conditions, and how it could impact Fed policy decisions and market reactions.
II. Greenspan and Non-Farm Payroll Data
Fed Chairman Greenspan came to the job from the forecasting and consulting business, and he was a big believer that the payroll employment data was a critical statistic for understanding the cycles in the US economy. During his leadership at the Fed, the monthly payroll employment data release, on the first Friday of each month, eclipsed the previously favored data -- US money supply -- and all other economic data as the most watched statistic about the US economy.
The 2007-2009 period of financial distress led to an especially deep reduction in employment. Compared to 60-year historical averages, non-farm payroll growth since 2010 has not been as robust as many had hoped, indicating continued slack in the US labor markets.
Figure 1: Non-Farm Payroll Employment2 from 1954
Figure 2: Recent Payroll Employment
Figure 3: Private Sector Employment
Figure 4: Government Sector Employment
The recovery that started in late 2009 has been strikingly different than previous recoveries, and to appreciate one of the major differences the composite non-farm payroll data needs to distinguish between private and government sector job creation. Job growth has been quite strong in the private sector. The challenge with the total payroll data has been the significant contraction of workers in the federal, state and local government sector (including the postal service). That is, this economic expansion differs from all other post-WWII growth periods by having to cope with a loss of 850,000 government workers, while other expansions benefited from positive job growth in the government sector.
III. Bernanke and Unemployment Rate Data
Fed Chairman Bernanke sought to make the Fed's decision process more transparent. As part of this effort he introduced the concept of forward guidance based on economic data related to the Fed's dual mandate of maintaining price stability and encouraging full employment3. Bernanke chose to emphasize the unemployment rate as the single measure representing the labor market objective of the Fed, and in December 2012 he set achievement of a 6.5% unemployment rate as a potential milestone related to the economic health of the labor markets.
Compared to post-WWII historical averages, an unemployment rate above 5.5%, give or take, is considered a sign of a slack labor market. Since the current rate (April 2014) is above 6%, there is room for improvement in labor market conditions.
Figure 5: Civilian Unemployment Rate4
Figure 6: Pace of Unemployment Rate Decline
As with non-farm payroll data, the pace of the decline in the unemployment rate from its recession peak of about 10% to the current level just above 6% has been limited by substantial job losses in the government sector. Nevertheless, a 3.5% decline in the unemployment rate over the first four years of the economic recovery would generally have been considered a solid improvement if it had not been for the depth of the recession and height of the unemployment rate at its peak. The standard view focuses on the absolute level of the unemployment rate compared to set threshold, while the dynamic view emphasizes the pace of the decline in the rate given some powerful headwinds from the government sector. We have a typical two-handed economist perspective on whether the glass is half-empty or half-full.
IV. Yellen and a Plethora of Nuances
As noted previously, at the first Federal Open Market Committee (FOMC) meeting at which she held the gavel (March 2014), new Fed Chairwoman Yellen eliminated forward guidance based on one or two economic indicators, and she adopted a much more nuanced approach to interpreting labor markets. Moreover, given that core inflation is subdued by any measure, the current focus of the Yellen Fed is on encouraging full employment and gauging the degree of slack in the labor markets by looking at a wide variety of information and weighing it all subjectively. In this section, we provide a compendium of charts on the data that Yellen has from time to time discussed in her speeches or in the question and answer periods following her remarks at press conferences.
Figure 7: US Labor Force Participation Rate5
The US labor Force participation rate has been trending downwards since 2008. Many labor economists link the labor force participation rate with the idea that there is a large pool of discouraged workers. As citizens are unable to find employment over a significant period of time, they may become discouraged and then drop out of the labor force. A declining labor force is a sign of a weakened labor market.
All may not be what it seems with regard to the participation rate this time around. The declining participation rate could be a function of less-educated or manual workers from the baby-boomer generation dropping out early. It may also be indicative of a longer-term correction occurring. In the boom-boom years preceding the financial crisis of 2008, there is the possibility that many people entered the workforce who might not have otherwise been working if the times had not been so exceptionally good due to excessive mortgage lending induced by the Greenspan's Fed extended period of very low rates along with the relatively lax supervision of the expansion of risk-taking in the banking sector. The financial crisis may have been the catalyst for a secular correction as older part-time workers, stay-at-home parents, etc. and other marginal, less skilled, or manual workers began to leave the labor force as finding and keeping a job became more difficult. Put another way, the labor force data at the end of 2006 of 152,732,000 may well have been a significant overstatement of the trend size of the labor force, because it contains so many marginal workers drawn to jobs due to the housing boom. If this interpretation is correct, then we may be over-estimating the long-term full-employment labor force and the size of the output gap.
Figure 8: Part-time Employment
Part-time employment sky-rocketed after the financial crisis. The standard thought is that as employers suffered financial losses they downgraded some workers from full-time to part-time status while laying-off others. As this theory goes, in a recovery period, the part-time workers would be returned to full-time status as profits recovered, and then later the laid-off workers would be rehired. The consensus, this time around, would argue that the failure of the part-time workers to convert back to full-time jobs as the expansion has progressed is a sign of companies lacking confidence in the economy and therefore not willing to take on full-time employees.
Higher part-time employment levels could be indicative of a labor market that is in the process of experiencing fundamental and structural changes. The increasing importance of Internet online sales and e-commerce in consumption and spending patterns has probably reduced sales at traditional brick and mortar stores. As a response, and while companies struggle to find a new business model that better balances e-commerce relative to storefront presence, a more flexible part-time workforce is better suited for the way many businesses, especially retail business, are now functioning.
There may also be a labor supply dynamic at work at well. An older work force may have a larger percentage of workers, compared to a younger work force that would actually prefer part-time work in their later working years. That is, they want to keep working, but not necessarily full-time, especially in those special cases where the part-time jobs offer health insurance or other desirable benefits.
Figure 9: Length of Unemployment
Figure 10: Duration of Unemployment
The length of time displaced workers remain unemployed rises in a recession and then declines. The average and median length of unemployment are still elevated, which would be considered a sign of slack in the labor markets.
There is a real division occurring in the labor markets between what Salman Khan6 refers to as "mind" workers and we call problem-solvers, relative to workers, even skilled workers, that largely perform repetitive tasks. The labor market appears quite tight for problem-solvers, admittedly a much smaller proportion of the labor force relative to other workers. This kind of structural change vastly complicates overly simplified interpretations of labor force slack. We also should note that short-duration unemployment, that is of 5 weeks or less, suggests a very tight labor market, while long-duration unemployment, 27 weeks or more, is indicative of the kind of structural change that might be occurring with less older and less skilled workers not being able to find suitable work at all.
Figure 11: Education Level: College
Figure 12: Education Level: Below High School
Figure 13: Turnover Rates7
Figure 14: Unemployed and Job Openings
Turnover rates - both voluntary (Quit rate) and involuntary (Layoff rate) -- indicate very different aspects of the labor market. With the voluntary quit rate being a strong indicator of confidence of workers within the labor force. Since the quit rate is rising and the layoff rate is falling, this indicator is sending positive signals about improving labor market health and economic confidence.
Each generation of workers seems to bring a different set of expectations about employment and corporate loyalty. The parents of the baby boomers were quite loyal to their employers, while baby boomers have shown an increased willingness to move jobs and have become much more cynical about corporate motives. The new generations X and Y and millenials may define their career and work expectations differently yet again. These newer generations, on average, may not have the prospects to achieve the income and wealth their parents achieved. We will have to watch for new patterns to develop, but reduced income expectations appear to be increasing the incentives for entrepreneurial efforts at a younger age than in previous generations.
V. Indicators Needing More Attention
Virtually everyone that studies the US labor market has some awareness that the slowing of the nation's population growth and current immigration environment has led to a much slower growth in the labor force as well as an aging of the labor force. Our own observations, however, have led us to the conclusion that while many analysts are aware of these trends, they often do not factor them into their analysis, especially when deciding how to interpret labor market data patterns that cross the divide in the 1970s from when labor force growth was accelerating to the current period of deceleration.
Figure 15: Growth of Labor Force
Figure 16: Age of Labor Force
Most standard perspectives on labor market do not appropriately take account of the declining growth rate of the labor force or the aging of the labor force. This is a notable omission.
A slowing growth rate for the US labor force has some important implications for the potential real GDP growth of the country. One way to think about real GDP potential is to look at the arithmetic, which says that long-term trend potential real GDP growth either comes from labor force growth or from growth in labor productivity. In turn, labor productivity can be influenced by technological innovations and capital investment. For a modern and mature industrial country such as the US, the returns from additional capital investment are less than in emerging market countries. And the application of new technologies in the US is very rapid and very advanced already. Thus, achieving a long-term average trend growth rate of labor productivity above the 1.5% to 2.0% range would seem to be a stretch. That is, with labor force growth slowing toward 0.5% per year, trend potential real GDP growth may be in the 2.0% to 2.5% range. Of course, this is not set in stone. Immigration can increase the labor force growth rate and raise potential real GDP. Resource discoveries, such as the oil and natural gas boom in the US can raise potential real GDP for as long as a decade or more. And the pace of technological innovation is also dynamic, and could well accelerate.
And then there are the challenges of an aging demographic pattern. US citizens appear to be vastly underprepared for retirement. Defined-benefit pensions (set amount of guaranteed income per year on retirement) have been "out" for a while, and "in" have been 401ks and other types of defined contribution plans (save as much (or as little) as your ability or willingness dictates). With the average lifespan creeping upwards, and the cost of living shifting toward health care as one ages, the rising median age may be indicative of the reality that many will have to work longer than the generations before them. This factor has the potential to influence labor force dynamics through the channels and incentives related to retirement savings and cost of health care.
VI. Fed's Dual Mandate and Changing Relationship of Inflation Dynamics and Labor Markets
So far, our analysis has focused on reinterpreting labor market conditions in a more dynamic manner taking into consideration the decline government jobs since 2009, the decelerating growth of the labor force since the 1970s, the corresponding aging of the labor force, and the critical importance of impact of the information age on business models and hiring practices. Our conclusions can be summarized as follows:
- US labor markets may have some slack still left over from the deep recession, but the degree of underutilized resources may be much less than the standard interpretation suggests.
- The private sector has been creating jobs more or less in line with the pace of the previous economic expansion.
- Labor market weakness has been dominated by the government sector, which has shed about 850,000 jobs during the expansion, rather than adding jobs as would have been a typical case. Our perspective is that the Fed's QE programs from 2010 onward were never going to stem the job losses from over-extended state and local governments and the US Postal Service.
- The last time the unemployment rate went above 10% (1982), it took about five years for the unemployment rate to decline below 6%, and the economy is on that same pace this time around following the peak unemployment rate at 10.0% in October 2009.
- Additional labor market indicators, including part-time employment, participation rates, and many other indicators are often cited, in our view incorrectly, as suggesting a slacker labor market than indicated by the unemployment rate or the pace of job creation. We strongly disagree, arguing that once one takes into account the aging of the labor force and the very slow growth of the labor force, then an interpretation of a wide variety of labor market indicators supports the view of a relatively robust and healthy market adjusting appropriately to some very new and difference dynamics.
There is, however, one critical dynamic pattern that we have so far not discussed and that many labor market analysts often fail to appreciate. Namely, the association between labor market developments and inflation. Labor market slack may work to reduce inflation just a little, but that relationship is very weak, complex, and not particularly dependable. This means that the standard view of a powerful trade-off between labor market health and price stability may need to be reconsidered, especially in light of the dual mandate of the Fed to encourage full employment and low inflation.
Since 1995, the core inflation rate, excluding food and energy, has remained quite steady, bounded by a 1% to 2% range. Essentially, the US economy has experienced two decades of low and stable inflation. What is notable about this 20-year period is that it encompassed the tech wreck in the stock market in the late 1990s, the housing boom of 2003-2007, and the financial disaster of 2008. Unemployment rates moved up and down during the period with some very wide swings. Through it all, the core inflation rate was remarkably stable.
One can see this is some very basic, yet revealing, statistical relationships. Take the ordinary least squares (OLS) relationship between core inflation and the unemployment rate before and after the mid-1990s. T-Statistics are in parenthesis underneath the estimated coefficients.
EQ #1: January 1961 through December 1994
Core Inflation = 0.008 + 0.58 x Unemployment Rate
EQ #1: January 1995 through April 2014
Core Inflation = 0.023 - 0.10 x Unemployment Rate
There are a few things to note from these basic statistical associations. First, prior to 1995, a 1% higher unemployment rate was associated with a 0.58% higher core inflation rate. This non-intuitive relationship is dominated historically by the stagflation of the late 1960s and 1970s, where higher levels of unemployment and higher inflation developed at the same time. That stagflation relationship, at least statistically, has disappeared. From 1995, there is actually a small negative relationship, which is more intuitive, but it is not powerful. A 1% higher unemployment rate would be associated with only a 0.1% lower core inflation rate, and vice versa for a lower unemployment rate. The statistical relationship is dominated by the stability of the last two decades in the core inflation rate. That is, the estimated constant term is 2.3% or just a little above the trend inflation rate. From a statistical perspective, when most of the "fit" is captured by the constant term in an OLS regression, it means that explicit fundamental factor(s) are not providing much additional information or useful associations. The tighter the "fit" for the estimated constant term, the greater our ignorance about what factors actually drive core inflation.
Figure 17: Core Inflation and Unemployment
The relationship between labor markets and core inflation needs much more study and examination than is appropriate for this research report. Some very basic statistical associations suggest that the link from a slack labor market to lower inflation or the link from a tight labor market to higher inflation has been exceedingly weak over the last 20 years of price stability. This observation, if supported by further research, has implications for Fed policy and how the dual mandates of full employment and price stability are managed. Specifically, the typical view that there is powerful trade-off between fighting inflation or possibly causing more unemployment may not be accurate. If this is the case, the Fed may need to examine the links between policy actions in a much different manner from the past. Quantitative easing may not have been of much use at all for labor markets; yet prolonged periods of extremely low short-term rates may risk asset price booms instead of rising core inflation.
What all this means for Fed policy is rather intriguing depending on what the actual catalyst is if/when the Fed decides to abandon its near-zero target federal funds rate and initiate a step by step process of pushing short-term money market rates a touch higher. We suggest two scenarios.
First, labor market conditions may improve, but core inflation stays around 2% and inflation expectations remain well grounded. In this case, we would consider the labor market conditions as the main catalyst for a rate decision. And without rising inflation expectations, the decision to raise the target federal funds rate might not be accompanied by higher bond yields, so the yield curve flattens.
Second, labor market conditions may improve, core inflation starts to rise a little and inflation expectations rise even more toward 2.5% or 3.0%. In this case, we would consider the rising inflation expectations as the main catalyst for a Fed tightening decision. And we think there is a good possibility that market participants would view the Fed as being behind the inflation curve, suggesting an upward shift in both short rates and longer-term yields.
In conclusion, we return to our opening theme. Analysis based on some concept of what was once considered "normal" is a very dangerous approach given the dynamic nature of markets. This would especially apply to labor markets, which are clearly undergoing huge structural change associated with demographic patterns and the changing business models in the Information Age. Moreover, we would caution against taking the perceived trade-off between unemployment and inflation to heart, as that relationship appears weak and undependable based on our analysis.
Samantha Azzarello, Economist, CME Group, also contributed to this report.
1 See Contributions to Modern Economics (1978), Chapter 1, page 3, en.wikiquote.org/wiki/Joan_Robinson. One of the leading economic scholars of the last century, and probably the most influential economist not to receive a Nobel Prize, Joan Violet Robinson (1903-83) was famous for her path-breaking work on the theory of imperfect competition. Cambridge University denied her a full professorship until late in her career, notably after her husband, economist E. A. G. Robinson, had retired from the faculty. The Nobel Prize committee is rumored to have passed her over repeatedly because they felt her political views lay too far to the left. Despite the glass ceiling that confined her, her intellect and theoretical work earned her the admiration of her colleagues and a prominent place in the history of economic thought.
2 The non-farm payroll data comes from the establishment survey which asks for the number of employees on company payrolls on the 12th of each month. Typically, big companies are assumed to do a much better job of timely reporting than smaller companies, so estimates are made about the rate at which new companies are started and the rate at which small companies fail. This birth/death assumption for small companies embedded in the data calculation process can be subject to significant errors. Payroll data can be substantially revised, especially during years in which the economy goes into or comes out of a recession - namely the times during which the birth/death assumption is most likely to be incorrect.
3 The concept of economic targets being set by central banks for policy guidance has been highly debated. Our favorite contribution to this debate came from Charles A.E. Goodhart, who argued that designating a single economic indicator as a target will have the effect of changing the meaning of the indicator, since markets will react differently once the target has been identified. This concept became known as "Goodhart's Law", and it has its parallel in physics with the Heisenberg Principle - that when one measures something, one changes it. In any case, Bernanke was willing to break Goodhart's Law, while the first thing Yellen did was abolish single indicator targets and move back into compliance with the law.
4 The civilian unemployment rate is a percentage calculated from two large numbers estimated from the household survey - namely the number of unemployed divided by the total size of the active labor force. The household survey is based on calls and contacts with tens of thousands of people across the United States. Responders to the household survey are asked if they are employed. If "yes", they are counted in total employment. If the responder answer "no", then a follow-up question is asked: "Are you actively looking for work?" A "yes" answer means the responder is included in the labor force, and a "no" answer excludes the responder from the labor force count. This survey has considerable volatility from month to month for the two big numbers - total employment and total labor force - so the ratio of unemployed to the total labor force is better viewed in terms of its trend rather than one month's piece of data.
5 The labor force participation rate is the ratio of the estimated size of the labor force relative to the working age population. The most significant change in the labor force participation rate came in the 1970s and 1980s as women entered the work force in unprecedented numbers.
6 See Salman Khan's "The One World Schoolhouse", Hachette Book Group. 2013.
7 The quit rate is calculated as the number of quits as a percent of total non-farm employment, while the layoff rate is calculated as the number of firings as a percent of total non-farm employment.
Disclosure: The author has no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it. The author has no business relationship with any company whose stock is mentioned in this article.