Labor Markets: What We Know and What We Don't
Good evening. It's a pleasure to address a meeting of the Money Marketeers once again. The last time I met with you as a speaker was nearly ten years ago. I was a freshman member of the FOMC then, trying to sort out how the Committee operates, and how I could contribute to its vital function of setting monetary policy. I am struck by what a remarkable period these ten years have been, incorporating as they have the good, the bad, and a little bit of the ugly. We went from a "Goldilocks" economy in a growing world to Asian and Russian debt crises; from stock market euphoria to a sizable and painful stock market correction; from an economy thought in 1994 to have an ever slower rate of productivity growth to one that has passed all expectations even in the face of cyclical swings; and from innocence borne in part out of a sense of geographical security to full immersion in a world of geopolitical uncertainty. This remarkable period has raised challenging questions for monetary policy makers as you can imagine, questions that have changed over time, but that have always revolved around how policy could best contribute to both price stability and to maximum sustainable growth.
Tonight, I want to talk with you about my perspectives on one of the major questions facing policy makers right now — the reasons for the labor market’s unusual performance. One of the key puzzles about U.S. economic growth is the anemic pace of job creation in an economy with solid demand. Indeed, this has been called not the "jobless" recovery -- that term is reserved for the recovery of the early ‘90s — but the "job-loss" recovery. Compared with our current job decline, by this time in the previous recovery — 27 months after the end of the recession — the economy had already created about 2 million jobs.
Tonight, I want to explore with you three questions about this “job-loss” recovery. First, what do we know about the nature of the problem, and how can it best be measured? Second, what are the reasons for such employment patterns — though I will tell you right now, I have found no smoking guns. And, finally, when will growth pick up? Again, no crystal ball here, but my take is that for any number of reasons current labor market weakness is largely a transitory or cyclical phenomenon that will improve.
Let me review how we have arrived at the current puzzling phase of labor market activity. As you will recall, despite the buoyancy of the late 1990s and the subsequent sharp fall in wealth caused by the demise of the dot-com bubble, the recession of 2001 was unusually mild. Output declined just modestly, the unemployment rate peaked at 6.3 percent, and did so well after the recession officially ended. The current recovery began in December 2001, a dark time in our nation’s history, as the Wall Street community knows far too well. No doubt partly as a result of the uncertainties that characterized that time, the early stages of the recovery had a start-stop character. Recently, however, the upturn has gained momentum. In the second half of last year, GDP growth averaged a robust 6 percent, a pace generally thought fast enough to absorb unused resources at a decent clip. Spurred by two rounds of Federal tax cuts and more recently by gains in household wealth, consumer spending strengthened during this period. With a striking rebound in corporate profits, monetary policy stimulus, and modest business tax incentives, the long-anticipated double-digit pick-up in capital spending finally emerged in the third quarter of 2003. Moreover, core inflation has slowed to a rate of about 1 percent, reflecting the spare capacity still available. All in all, these developments bode well for the economy, certainly in the near term.
So in terms of spending and profits, the recovery is proceeding quite nicely. Nevertheless, even with strong demand growth and a double-digit surge in capital spending, total non-farm employment reported by business establishments remains 700 thousand below where it was at the start of the recovery and 2.4 million below its pre-recession high. Although employment has been growing since last August, over the past 5 months job growth has averaged under 60,000 a month. This rate is roughly half the pace needed just to absorb new entrants to the work force, let alone make inroads into the growing supply of long-term unemployed.
This unusually slow pace of job growth raises questions about our measures of employment and labor market slack. Are we really sure that employment is measured correctly? Analysts have at their disposal several measures of employment that do not always agree, at least in the short run, on how well the labor market is doing. Indeed, an unusual divergence between the so-called household and the payroll series has drawn much attention of late. Some observers have found encouragement in the recent pickup in employment recorded in the household survey, which — as the name implies — is based on interviews with households. The optimists suggest — as is certainly possible — that our increasingly “virtual economy” has resulted in growing numbers of contract and self-employed workers. Such workers get counted in the household survey but do not appear directly in the head counts at factories and corporate offices that make up payroll employment, at least on a current month basis.
While I would like to take comfort in this divergence as a possible explanation, I cannot. There are at least two reasons for my skepticism. First, most economists — as well as the Bureau of Labor Statistics, which produces both surveys — believe the payroll measure to be the more accurate survey. It is based on a larger statistical sample, and it queries business establishments rather than relying on less reliable self-reporting by individuals and families. It also includes an estimate of newly-formed business establishments, the mechanics of which have recently been improved. Second, and more importantly, despite their divergence, both the payroll and the household figures tell the same basic story: compared with what one would expect, both surveys are at a low point. That is, whether you use the household data or the payroll data, job growth has been slow as compared with what we would estimate given the growth in GDP.
Some have pointed to the recent decline in the unemployment rate — from 6.3 percent last June to 5.6 percent currently — as evidence of an improving labor market. Of course, 5.6 percent is a relatively low number by historical standards. But unfortunately, unemployment has fallen in large part because many people have stopped looking for work. They have left the labor market, thereby reducing the labor force participation rate to a relatively low level. If those unemployed workers had remained in the labor force or returned and failed to obtain jobs, the jobless rate would currently be near 7 percent.
Now it is true that the largest decline in labor force participation has occurred among younger workers aged 20-24. Perhaps young workers faced with poor job prospects are returning to school. Indeed, we are told that the nation’s community colleges are filled across the country. It’s likely these workers won’t resume work until their schooling is finished, so this 7 percent extrapolation of the unemployment rate may overstate the amount of labor market slack. Still, I don’t expect any material unemployment rate relief to occur when job growth resumes — at least initially, the labor force participation rate will likely also rise, and it could take some time for those reentering the labor force to be rehired.
Others who had earlier pointed to layoffs as a source of concern have taken heart from the fact that the pace of layoffs has appeared to subside. But new BLS data reveal that our current bout of labor market weakness stems not so much from unusually high layoffs or other forms of “job destruction,” but from unusually subdued hiring or “job creation.” Job destruction rose in the recession and fell in the recovery, as expected and as reflected in declining layoffs. But job creation, which had been falling since before the start of the recession, remains notably weak. The decline in layoffs implies that individuals who now have jobs can be more confident of holding onto them. But because firms have yet to resume hiring at a pace that is in line with recent economic growth, people looking for jobs have a hard time finding them. Indeed the duration of unemployment is near its all-time high.
So there you have my summary: overall, just about everything seems to be coming up roses — except for jobs. No matter how you measure it, job creation has been anemic. Why is this? The obvious answer is “productivity.” In the past two years, the growth in labor productivity has been truly startling. Having risen to a 2.5 percent annual rate of growth in the late 1990s, the growth in labor productivity at non-farm businesses has averaged over 4 percent in the past two years and about 5 percent in the last four quarters. These numbers are literally staggering in the context of a mature post-industrial economy.
But saying the answer is productivity is simply offering a tautology. Strong output growth achieved with few added workers means that productivity gains must have been remarkably high. Conversely, unusually rapid productivity growth means that fewer workers are needed to satisfy a given level of demand. Thus, the interesting question becomes why are U.S. firms still striving so hard to achieve ongoing gains in productivity growth? Why have firms remained so reluctant to hire even as profits have rebounded and demand growth has accelerated? The answers to these questions raise some tantalizing issues.
Is it possible that unusually widespread “job restructuring” is creating more than the normal mismatches between the skills workers have and the skills required in available jobs — slowing the pace of job growth? Is it possible that we are simply outsourcing all new job creation to foreign shores? Is it possible that labor is more expensive now, spurring more intensive use of all the equipment and software purchased from the late '90s through the present? And finally, is it probable that firms continue to be unusually uncertain about the future, even as profits have strengthened and capital spending has risen? Let me answer these questions in turn.
Start with the possibility that the economy is undergoing an intense bout of lasting structural change. An historic example drawn from my own District might be New England’s shift from textiles to electronics in the second half of the last century. Such a change usually requires a significant reallocation of workers across firms and industries. But because the skills possessed by the existing pool of workers may be less well matched with the available jobs than previously, such reallocation can take time. During this re-matching period, unemployment may remain elevated, and hiring may be depressed.
Of course, structural change is a fact of life given the dynamic genius of U.S. labor markets, where innovation and productivity gains have allowed the predominant work activity to shift from agriculture to manufacturing, to services, to knowledge-based endeavors. But some analysts suggest that the economy is currently undergoing an unusual amount of restructuring. They note that the rate of permanent job loss has been high relative to temporary layoffs, and that employment has continued to fall in industries that shrank during the recession. In addition, even in the midst of a soft labor market, some skills are reportedly in very short supply. Normally, however, a period of rapid structural change would coincide with significant “job churning” — that is, we would see high rates of both job destruction and job creation as large numbers of workers shift across firms and industries. But as I have already noted, the data on such transitions are quite subdued of late. Fewer than normal jobs are being created, and fewer are being destroyed. While the amount of industrial job loss we've seen suggests some role for structural change, the churning data suggest that restructuring may not be a large part of the problem.
Are U.S. firms "exporting" or "offshoring" vast numbers of jobs by establishing foreign affiliates or contracting with foreign firms to produce goods and services for use in this country? This phenomenon is not new, of course. Manufacturers and, to a lesser extent, service producers have pursued this strategy for years. Indeed, in the late 1990s when labor markets were at their tightest, this strategy was welcomed as a safety valve that allowed the economy to keep expanding rapidly without igniting inflation.
What’s different and has caught the public’s attention is that new communications technology is facilitating the “export” of some relatively high-skilled service jobs — in software design, medical technology, and data analysis, for example — jobs that we once thought were immune to foreign competition. This foreign competition will damp U.S. cost pressures in those industries, of course, but it will also reduce prices for U.S. producers and consumers. And if, for example, more cost effective foreign medical analysis can help keep our rising health care costs in check, I suspect most U.S. health care consumers — both employers and households — would say that might not be all bad.
I do not mean to treat this matter lightly. “Offshoring” is real. It clearly hurts those whose jobs move to foreign countries. Technological change and highly efficient international money and capital markets make it possible, just as these factors have been central to the 30-year decline in U.S. manufacturing jobs. And offshoring is likely to continue. But does it bear a lot of the blame for our current weak job growth? The available data suggest that the answer is no.
To be sure, foreign investment by U.S. firms and imported goods and services have become a bigger fact of U.S. economic life over a long period, and certainly this was true in the 90’s. But it is hard to find a recent increase in either foreign investment or imports relative to domestic activity that suggests that domestic jobs are losing new ground to import substitution. Moreover, this country has a large ongoing trade surplus in services, indicating that the U.S. services sector remains highly competitive internationally. These relationships are largely ignored in the highly charged political atmosphere in which the topic of offshoring is usually discussed. Trade in services is two-way; we buy services abroad, but foreign countries buy more from us than we do from them.
As for the call center and programming jobs that have provoked particular concern in recent months, U.S. imports of IT and other business and professional services from all of developing Asia amounted to about one-tenth of 1 percent of U.S. GDP in 2002. Clearly, this is not immaterial, but it simply isn’t large enough to have had a major impact on U.S. employment levels in the aggregate, despite the rhetoric that suggests otherwise.
Finally, we have looked for ways to “guesstimate” the impact of offshoring on all employment, not just in services. One source of data is a BLS survey that asks establishments with large, extended layoffs the reasons for these layoffs. These data suggest that over the past three years only a small share — just about 1 percent — of extended layoffs come from job relocations offshore, though the way layoffs are categorized suggests this survey should be taken as just a general indicator. The U.S. economy had about 55 million layoffs in the past three years, so if you do the math, even this small share seems like a lot of jobs. But this job loss needs to be taken in the context of the huge numbers of jobs both destroyed and created in our economy each month. Trade creates jobs as well as destroys them. Service jobs have been created, as I noted earlier, and I suspect even manufacturing jobs as well. This layoff survey gives a rough take on gross offshoring flows, but it tells us little about the net impact of offshoring.
In sum, there is no doubt offshoring is having an effect, and it is certainly clear that the impact has been significant in some industries. But offshoring has to be a small share of all job flows in our remarkably vibrant economy. Thus, we are left with the conclusion that our weak job market has its roots primarily in domestic developments.
Is the domestic cost of labor restraining hiring? Ask any business leader about labor costs, and you will hear a chorus of complaints about rising wages, rising health care costs, rising contributions to defined-benefit pensions, and the increasing cost of unemployment insurance. Just yesterday, for example, we held a meeting of 60 or so business leaders in Providence which featured an almost continual expression of concern about rising labor costs.
Unlike many of the candidate explanations, this one contains more than its share of truth. Overall labor costs are rising — at a pace of 3.5 percent at the end of last year. Productivity is rising even faster, however, as I suggested previously. In fact, unit labor costs — the difference between total labor compensation and labor productivity — have been falling for the past two years. But this may not be the whole story. The dynamics of the various components of overall compensation could play a role here, particularly in the decision of firms to hire new workers.
Take pension costs — they are rising at a 12.5 percent pace. In the good old days of the 1990s, an employer could use the capital gains that go with a rising stock market to fund current pension liabilities and to help with what for many had been chronically underfunded pensions. Now the employer with a defined benefit plan must again kick in some of the firm’s own resources. Health care costs have been rising quite rapidly of late as well — 11 percent or more for the average employer. Together with pensions and other forms of non-salary compensation, these costs make up nearly 30 percent of the overall wage bill — a share that grew last year in the aggregate at a nearly 7 percent pace. Moreover, these are all costs over which the employer arguably has little or no control. It seems clear to me that there is some logic to concerns about the growing cost of labor, even in a world of rapid productivity growth.
Indeed, one explanation for slow job creation I find particularly appealing is that, in the face of growing labor costs or the real difficulty of finding the right skills at the right price, employers have a substantial incentive to take advantage of all the technology they acquired in the late '90s and more recently. As I’m sure you know, it takes a considerable period for widespread and effective adoption of a new technology to occur. Some have suggested that the advent of the Internet, combined with rapidly expanding telecommunications capabilities and falling costs, represents a level of technological change not seen since the arrival of the mainframe in the late '60s. Given the many anecdotes we've all heard about businesses working faster and harder, streamlining their supply chains and reorganizing manufacturing and service processes, it would appear that finding new ways to use these new technologies remains a driving force for many companies. Still, at some point, demand will overtake the ability of firms to continue to use labor ever more efficiently. Sooner or later if demand persists, job growth will have to occur.
That takes me to the last reason many have given for the lack of job creation — continuing uncertainty. Unfortunately, the recent bombings in Madrid remind us that the world remains a dangerous place. Moreover, while monetary and fiscal stimulus have been important to this country's growth in the last two years, policy cannot sustain robust growth by itself. Businesses may be unsure about the sustainability of the recovery in the future as fiscal stimulus fades away. If so, they might be holding off on hiring new workers until the shape of the recovery becomes clearer, despite the reassurance of growing profits.
Some argue that the recent resurgence of capital spending suggests uncertainty is not the whole story. Why would businesses have increased capital spending at double-digit rates, as they did in the second half of 2003, if they were so uncertain about future demand prospects? This may not be such a puzzle. To the extent that such capital spending is a critical element in businesses’ drive to improve efficiency and contain labor costs, as I have just suggested, the investment resurgence makes some sense, even in the face of uncertainty. This observation may not completely resolve the tension between growing capital investment and residual uncertainty, but it likely resolves some of it.
Perhaps the best we can do, then, in trying to explain our lackluster labor market is to accept a mixed bag of explanations. The recent weakness in hiring could stem in part from structural change. Outsourcing surely contributes some, but not much, to that picture. And rising labor costs likely play some role as well. But I find the most compelling explanations for weak hiring to be firms’ continued ability to eke out productivity gains from their ongoing capital acquisitions, combined with a hesitance to hire that is the result of lingering uncertainty about both external risks and the sustainability of demand over the near term.
These explanations may help us to understand recent developments, but what do they portend for the future? Here, I think the news is promising. It seems unlikely that we are entering a period of indefinite super-human productivity gains in which we will never again see any material job growth. While some explanations for slow job formation have structural components, it is self-evident that none of these hypotheses are durable over time.
If demand continues to be solid as is widely expected — and I share the consensus expectation of 4-percent GDP growth or better this year — then employment will respond. Employers will need to expand their work force and at some point the competitive dynamic to secure the best worker for the job will feed on itself.
What does this mean to me as a policy-maker? For some time now, monetary policy no matter how measured has been accommodative, as the economy has weathered a recovery that until recently has been long and slow. Indeed, its recent signs of strength are due in part to monetary stimulus. However, with a continuing solid pace of growth, this level of monetary accommodation could at some point become incompatible with the core objectives of the central bank. Thus, I would argue that while patience is a virtue, so too is vigilance.
Looking back to the 1930s when your club was new, the economic and geopolitical outlook must have seemed far more uncertain than we can even imagine today. At that time, Marcus Nadler, your group’s founder, advised those who were forecasting U.S. economic activity in the depths of the Depression in the following ways: “You’re right if you bet that the United States Economy will continue to expand. You’re wrong if you bet that it is going to stand still ...” Marcus Nadler got it right, of course. And his advice remains good — for now and for the long haul — as our extraordinarily flexible, productive economy continues to prove decade by decade.