An off-kilter expansion

Slack job market continues to hurt wage growth

by  Jared Bernstein and  Lee Price

A few weeks back, on a late July morning, the U.S. government released two statistical reports that, taken together, provide an excellent microcosm of the current U.S. job market. One report showed that the overall economy was expanding smartly, suggesting the four-year-old economic expansion was solidly on track.

The other report, however, painted a very different picture. This report—on wages and compensation—revealed that average wage growth had yet again failed to beat inflation. In fact, the wage growth cited in the report was tied with the previous three quarters’ results for the worst on record.

This disconnection between production trends and wage trends has persisted for some time. For more than a year, growth has been solid and productivity has been stellar, yet the wages of many workers have lagged behind. This persistent and growing gap between earnings and growth often gets overlooked by central bankers and financial economists, whose main concerns tend to be the rate at which the economy is expanding and whether the economic climate is too cool, too hot, or just right. And these analysts have been uniformly very satisfied with our recent economic progress. In fact, in response to the July GDP report, the research director at the Federal Reserve Bank of Minneapolis quipped, “It’s kind of boring around here because the economy looks so good” (Minneapolis Star Tribune 2005).

If it weren’t for the wage and employment results that this report documents, it would not be a challenge to build support for this view. Unemployment, at 5.0% in July, is low in historical terms. Core inflation—price growth omitting the volatile food and energy components—is well contained. Productivity growth has slowed somewhat in recent quarters, but it’s merely coming off its recent Elysian heights that few analysts believed to be sustainable and is still growing at a very healthy pace of 2.3% over the past year.

But once we incorporate the wage and employment trends, the case for a “boringly” great economy dissolves. Most non-retired families depend on their earnings to support their living standards, and when hourly wages are lagging inflation, the only way for them to get ahead is to work more hours. Moreover, the gap between productivity and wages implies a fundamental imbalance in the expansion. When those partially responsible for the growth in the economic pie—the American workforce—consistently end up with slimmer slices, a sense of unfairness dominates people’s sentiments about the state of the economy.

This can be seen in recent poll results, showing greater dissatisfaction with the state of the economy than might be expected by just looking at the “top-line” statistics on unemployment, GDP, or productivity. For example, a Wall Street Journal/NBC News poll reported that the “share of the public approving of President Bush’s handling of the economy has gone down, to 39% in July from 47% in January” (Wall Street Journal 2005). As reported in the Washington Post, a CBS News poll from the first week of August found that “52% of respondents disapproved of Bush’s handling of the economy, while 42% approved…Just 20% of those polled said the economy is improving; 32% said it is getting worse” (Washington Post 2005). A Washington Post /ABC News poll showed that when the recession ended in November 2001, 13% said the state of the economy was poor. In mid-August, that was up to 21%.

This report presents a set of wage and compensation trends in detail and then explains the underlying factors driving these trends. The conclusion drawn from this analysis is that, despite what appears to be a low unemployment rate, considerable slack remains in the job market. Earlier research shows that a truly tight job market plays the critical role of broadly distributing the benefits of growth (Bernstein and Baker 2003). Specifically, in a full-employment job market, with the number of jobs available matched fairly tightly to the number of workers, employers tend to bid compensation up to recruit and keep the workers they need. That mechanism, inoperative thus far in this expansion, is one way of ensuring that the benefits of growth are fairly shared.

The main findings of this analysis are:

• Recent wage growth is compared to three benchmarks: trends since mid-1995, inflation, and productivity. In every case, wages are performing worse now than a few years ago.
• Wages for most workers are falling slightly behind inflation and well behind productivity growth, implying lower living standards and greater inequality.
• The main factor behind the weak wage performance is the fact that considerable slack remains in the U.S. job market, with an estimated employment deficit of 3.2 million.
• The earnings and employment of African Americans are particularly sensitive to job market slack, and African American employment rates and real earnings have recently fallen faster than the average.
• Employment and labor force participation rates remain depressed, even for groups with very high levels of labor market attachment, such as prime-aged married men with children and the college educated.
• Because the overall labor market was not in a bubble in the 1998-2000 period, it should be able to return to the conditions of the late 1990s without problems, particularly in light of the intervening five years of strong productivity gains.

Wage trends: slower nominal growth as wages lag inflation and productivity

There are at least three ways to evaluate the progress of wage growth thus far in the expansion. The first is historical: how fast are wages growing now relative to an earlier period? This question helps shine light on the relationship between evolving job market conditions and wages, with tighter labor markets usually leading to faster wage growth. Second, wages are examined relative to inflation, which addresses the question of whether workers’ buying power is increasing. Finally, real wage and compensation growth is compared to productivity. This comparison goes beyond the evaluation of buying power and looks at the extent to which workers are broadly benefiting from the increase in the economy’s productive capacity.

In the current economy, workers are generally falling behind on all three of these measures. Nominal wages and compensation (before accounting for inflation) are growing more slowly in recent quarters, and for some series, are lagging inflation. There is also a large and growing gap between productivity and real pay growth in all but one data series.

Slowing nominal wage growth reveals ongoing labor market slack  
As has been shown in much economics literature, wage trends respond to the tautness of the job market (Katz and Krueger 1999; Blanchflower and Oswald 1994). Intuitively, when labor supply exceeds demand—when there are too many jobseekers and too few jobs—workers’ bargaining power is diminished, and they are generally less able to push for faster wage growth. Conversely, at full employment, employers often have to bid wages up to recruit and keep the workers they need. Thus, wages are expected to accelerate in periods of tight job markets and decelerate in loose job markets. We are allegedly in such a period now, with low unemployment and solid GDP growth. But recent wage trends do not corroborate this scenario.

<
p> Since inflation is affected by wage growth itself along with other factors like commodity prices (e.g., oil) and profit margins, this analysis begins with a focus on nominal wage growth.

Figure A and Table 1  show the annual nominal growth rate of various wage series: the Employment Cost Index (ECI), including both the compensation and wage and salary components; the average hourly wage of production, non-supervisory workers (called average hourly earnings in Figure A and Table 1); and (Table 1 only) the median weekly earnings of full-time workers, by gender, for all wage and salary workers, with African Americans shown separately. (Since the quarterly median series is fairly “noisy,” the more-stable trend component is used.) The ECI series cover the full civilian workforce (excepting those who work for the federal government), and the production worker series covers the 80% of the workforce in blue collar and non-managerial occupations. The three periods represent the strong labor market of the latter 1990s, the recession and jobless recovery, and the most recent period of sub-par job gains. 

Most of the wage series show some degree of deceleration in the recession/jobless recovery, and, importantly for the view of current conditions, all decelerate further in the more recent period. That is, instead of speeding up, as might be expected given low unemployment and solid GDP growth, wage growth has slowed in recent quarters. In fact, wages are growing more slowly now than at any point since the mid-1990s, especially for minorities. This report discusses minority labor market conditions throughout because, as discussed in more detail below, the absence of a minority full-employment response in wages provides compelling evidence that slack remains in the job market.

While earnings growth has decelerated, the rate of compensation growth has stayed relatively strong (though it has come down slightly in the most recent period). This however, is likely due less to labor market pressures than to higher employer costs for health care and defined-benefit pension plans. In theory, a tight labor market could cause total labor compensation to increase faster even as wage gains decline because workers opt to take more of their compensation in the form of fringe benefits. In practice, however, it seems implausible that employers are responding to a tight labor market by offering higher payments for fringe benefits and scaling back wage gains. Note that (a) about half the workforce does not have employer-provided health or pension coverage, so this reason doesn’t explain their decelerating wages; (b) at the same time, employers are shifting health costs onto workers through higher premiums and co-pays; and (c) the increase in defined-benefit contributions are legally required to make up for stock market losses and do not fund increases in future defined-benefit pension payouts. Employers are paying more for health and pension plans not because of a tight labor market but because of cost pressures from outside the labor market. Thus, wage growth is the best indicator today of tightness in the labor market.

Real wages decline  
One of the most important benefits of the full-employment period in the latter 1990s was the accompanying increase in real wages for groups of workers whose wages had stagnated, at best, in earlier periods (Bernstein and Baker 2003). As shown in Table 2, real median weekly earnings grew 1.3% and 1.4% per year for men and women, respectively, over the latter 1990s. These gains stand in stark contrast to the 1979-95 trend of -0.7% for men and 0.4% for women.

Other than for the median male, the momentum of the latter 1990s, the deceleration of inflation, and very fast productivity growth kept real wages and compensation growing over the recession/jobless recovery. However, nominal wage growth slowed most recently (Table 1), and inflation has accelerated. The result is falling real wages over the past two years, even while productivity continued to post historically high growth rates.

Note that wage trends lag labor market conditions—as just pointed out, this dynamic property helped to keep wages growing through the recession. But the recent deceleration in nominal wages in tandem with the historically large gap between productivity and real wage growth suggests that it takes more than lags in wage trends to explain the current wage situation. Ongoing slack in the job market is surely at play here as well.

Figure B  plots a number of real wage and compensation series, along with productivity growth, with all series indexed to 2000q1. Productivity grew at a particularly fast clip over this period, and none of the series match its pace. This is not particularly surprising, given the rocky job market over these years, which encompass both the recession and the jobless recovery. And the two compensation series—which combine wages and benefits—do show some real growth in recent quarters (though this is less the case for the ECI compensation series). But what is notable about the figure is the extent of the gap between productivity growth and real wages.

The labor market context for the disappointing wage trends

As noted in the introduction, polling data reveal that despite lots of “happy talk” from those observing the economy from 30,000 feet, there is considerable dissatisfaction with the current economy. For example, the Washington Post/ABC News Consumer Comfort Survey shows that the share who describe the economy as “not-so-good” or “poor” is one percentage point higher today than at the end of the last recession.  Over the same period in the last expansion, it was down 11 percentage points. The widespread dissatisfaction is likely tied to the fact that, for many working families, pay is not keeping up with inflation.

But why, in a recovery widely touted as very much on track, are wages lagging so far behind? The evidence presented in this section of the report suggests that the problem lies with the job market, which continues to be too slack to generate the pressure needed to raise the wages of many to at least the level of inflation, if not inflation plus productivity.

The debate about slack in the labor market  
This discussion brings up a behind-the-scenes debate taking place among economists as to the extent of slack in the job market and whether it threatens to become too tight, thus leading to inflationary pressures. Though this argument is somewhat arcane, the stakes are high. It helps to understand the two sides in the debate.

Since the current economic expansion began in November of 2001, job growth has been disappointing. For 18 months, employment continued to slide even as the rest of the economy recovered.

Since mid-2003, payrolls have been expanding on a monthly basis, but the rate of growth has been well below that of prior recoveries (Price and Vasavada 2005). This has led many analysts to note that a significant gap has developed between the actual number of jobs created and the number the labor market would have if it were operating closer to its full potential (Bradbury 2005).

At the same time, the unemployment rate never reached the heights of past rece
ssions or weak recoveries, and since July 2004, it has been at or below 5.5% (unemployment was 5.0% in July 2005). These are historically low unemployment rates, and they raise the question: how is it that unemployment rates remain low yet job growth remains weak?

One technical answer is that the two measures come from different surveys, but this is unconvincing because, while the two surveys vary greatly month-to-month, over the longer term they are both telling a similar story.¹ In fact, there is little disagreement among economists regarding the facts of historically slow job growth accompanied by low unemployment. One explanation for this phenomenon is that millions have dropped out of the labor market, and that the absence of these potential jobseekers has helped dampen the growth of unemployment (only those actively seeking work are counted among the unemployed).

It is here that the disagreement begins. One camp of labor market analysts believes that the current job market is relatively tight, meaning that the number of jobseekers is roughly aligned with the number of available job openings, and that the diminished rate of job growth is about what’s needed to keep up with the growth of the labor force. Regarding the fall-off in labor force participation, those in this camp argue that participation was artificially inflated by an economic bubble at the end of the last business cycle. The reasoning is that excessive demand, driven by speculation in financial markets, drew too many people into an overheated job market. Under this scenario, once the bubble burst, the labor market should expect a drop in participation to levels more consistent with its historical experience.

The other camp argues that the job market is significantly underperforming. Labor force participation rates are depressed by weak employment conditions, and the rate of monthly job growth, on average, is well below potential. This wasted potential is quite costly. From a macroeconomic perspective, it creates a drag on growth as current and potential work skills are underutilized. From the perspective of working families, particularly minorities, it is leading to declining real wages, greater inequalities, and diminished living standards, especially relative to what would occur with full employment.

This study argues that the current U.S. labor market problems are the result of too little demand to absorb the available resources, i.e., employers are demanding fewer hours of work than the workforce is willing to contribute.² Unemployment is artificially depressed by low labor force participation rates that are themselves a function of the weak job market. We find no structural reason—no fundamental change in key economic relationships—for the diminished path of job growth, and thus conclude that a significant jobs deficit exists.

Evidence of labor slack  
Starting with a set of measures inconsistent with a tight job market and indicating a market far from full employment, this report derives an estimate of the number of persons “missing” from the labor market, i.e., people normally expected to be working in a more robust expansion.

The first notable fact is that despite a slight recent uptick, gross job creation is close to a historical low point (Figure C ). As this is solely a measure of labor demand (and says nothing about supply), low job-creation rates alone do not imply slack in the job market. It does, however, suggest weaker-than-expected labor demand. Note that while this series only begins in the 1990s, the current rate of job creation is much lower than that of the last cycle. However, if there are fewer jobseekers, then the market presumably can achieve full employment with less job growth. Thus, a critical question to ask is whether there has been a secular downshift in the share of persons who want to work.

Other evidence that employment weakness is cyclical rather than secular comes from the analysis of various groups for whom there appears to be no other plausible explanation for their absence. For example, the employment rate of prime-age married men with children, a group that consistently posts the highest employment rates, fell 1.4 percentage points between 2000q1 and 2005q1, from 93.7% to 92.3%. What else other than lack of job opportunities could be keeping these men out of jobs? Note that the decline is comparable to the 1.5 percentage-point drop for all prime-age workers, including men and women, married and unmarried, with and without children. It seems implausible that so many people would voluntarily withdraw from employment so quickly.

Figure D  shows evidence of depressed employment activity among another group whose employment we would expect to be robust if the market were truly at full employment: college graduates.

These workers are usually in high demand relative to those with lower education levels, and, in fact, their employment rates remain the highest (and their unemployment rates the lowest) of any educational groups. But note that their employment rates fell in the recession and have yet to recover. Young college graduates are presented separately because this group—with newly minted skills paying, presumably, a considerable premium—is particularly unlikely to drop out of the job market for secular reasons.

In a similar vein, employment rates for African Americans are further below their peak than other groups, and this is a fairly clear sign of lack of pressure in the job market. Minority employment rates, especially those of African Americans, tend to be particularly responsive to cyclical conditions in the job market. In the run-up to the strong labor market of the latter 1990s, African Americans made particularly large gains, and the reverse is occurring now. In the second quarter of this year, African American employment rates were down 2.7 percentage points compared to 1.6 percentage points overall.³ This recent trend is inconsistent with the view that the market is at full employment. In fact, these minority trends provide an important glimpse at the cost of accepting current conditions as the norm: to do so consigns over 700,000 African Americans to non-employment status.

Another revealing piece of evidence that militates against full employment is the share of the unemployed who are “long-termers,” i.e., those who have been looking for work for at least six months. Historically, unemployment rates in the current range—5.0%-5.5%—have been associated with long-term shares of 12% on average. Over the past year, that average has been 20.6%.

The weight of this evidence leads to the conclusion that the market is underutilizing its full labor potential. It is a challenge, however, to quantify the degree to which this is occurring. It is demonstrably the case that a substantially lower share of the population is employed today than was employed in the late 1990s. At the same time, however, more young people are enrolled in school today than at that time. In addition, there has been a long-term downward drift in the share of prime-age men employed.

Figure E  demonstrates a very substantial labor slack for three distinct groups—young people ages 20 to 24, women ages 25 to 54, and men ages 25 to 54—based on their July 2005 employment rates compared to those toward the end of the last expansion. Increased school attendance of young people and the long-term trends of prime-age men and women are taken into account. Finally, to avoid basing estimates on a single year, an average is taken of the last three years preceding the 2001 peak. The d
ata appendix explains the specific conservative assumptions used to make these estimates. As a result of this exercise, it is estimated that the labor market is currently providing 3.2 million fewer people with employment than it should0.9 million for people ages 20-24, 1.3 million women between the ages of 25 and 54, and another 1.0 million men ages 25-54.

Conclusion

The data presented on wages, employment, and the current jobs deficit portray a labor market that still has a long way to go before it is performing at potential. Despite the fact that measured unemployment is historically low—5.0% in July—slack remains in the job market. The United States has an employment deficit in the neighborhood of 3.2 million. Lack of strong creation of job openings has been depressing labor force participation rates and, since these missing jobseekers are excluded from the unemployment count, artificially lowering the unemployment rate. Nominal wage growth has slowed for everyone, but especially for minorities who also have lost the most in terms of employment rates. The fact that the job market is especially failing to provide ample opportunities for minority workers—a group that disproportionately benefits from truly tight job markets—is a particularly strong sign that we have yet to reach full employment.

Though the jobless recovery has ended, job growth in recent quarters has not been strong enough to absorb enough of this slack to stimulate much-needed wage growth. The policy implication of this point is that the Federal Reserve should not be satisfied with the 3.6% GDP growth rate over the last year because it has done little to reduce labor slack. To tighten the labor market, the economy needs a number of quarters with growth above 4.0%, something the Fed should accommodate.

The conclusion regarding labor slack results from a drop in employment rates, as explained in the data appendix. Some analysts argue, however, that such comparisons set the benchmark too high, since the labor market was in a bubble prior to the last recession. When it is pointed out, for example, that African Americans had a 60.4% employment rate for the three years prior to the last recession, 1.7 percentage points higher than today, this argument holds that those three years are an unfair comparison because they were artificially inflated by the bubble. If analysts were to accept this reasoning, and compare African American employment rates to their level in 1995, for example, the data would find that the employment rates are actually slightly ahead of where they were back then.

We do not believe that the evidence can support this view. We, of course, acknowledge the bubble in the financial markets at the end of the last expansion. Before every business cycle peak, serious economic imbalances have developed and tipped the economy into recession and the last expansion was no different. But it is a serious mistake to conflate the sources of demand tied to asset price bubbles with its impact: the lowest unemployment rate in decades, robust job growth, broadly shared wage and income growth, and all of the above without rising inflation associated with an overheated job market. Despite the strength of the labor market for working families, the labor market was not unbalanced prior to the last recession. Strong productivity growth had accompanied wage gains. Moreover, because the last four years of productivity gains have not been accompanied by comparable wage gains, there is ample room to draw people back into the labor market with even higher pay.

A decade ago, many macroeconomists warned of dire consequences if unemployment fell much below 6%. Fortunately, the Federal Reserve did not heed those warnings. The argument that the Fed should prevent further tightening of the labor market due to inflationary concerns is as wrong today as it was a decade ago. Slack exists today comparable to that found with unemployment rates above 6% in the past. By mistaking an underperforming job market as the best of all possible worlds, we risk permanently sacrificing millions of hours of productive human capital and the potential output those workers could contribute. Millions would be prevented from reaping the benefits of the ongoing economic expansion. These millions are not exclusively the unemployed or underemployed. They are also the far larger number of incumbent workers who, in a weak labor market, lack the bargaining power to claim their fair share of the growth.

These are the dynamics behind the unbalanced recovery. Until the engine of economic growth lifts the fortunes of all working families, not just the chosen few, the economy will continue to disappoint a majority of working families.

— Labor Day 2005

 

The authors thank the Rockefeller Foundation, the Charles Stewart Mott Foundation, the Joyce Foundation, the John S. and Catherine T. MacArthur Foundation, and the Foundation for Child Development for their support. We thank Yulia Fungard and David Ratner for research assistance. Danielle Gao provided programming assistance, and Jason Faberman and Claudia Deane provided helpful data.

Data appendix: estimating labor slack

The employment rate4 can be used to provide a better gauge of additional labor capacity than does either the unemployment rate or the labor force participation rate (LFPR). The employment rate reflects how many people were actually willing to work in the recent past. Both the unemployment rate and the LFPR reflect the willingness of people to persist in looking for work under varying conditions of job availability. Labor economists have become accustomed to using the unemployment rate as a proxy for the tightness of the labor market because they had virtually no information on the supply of new jobs. There is now a new data series on job openings that suggest a remarkably poor supply of new job openings in recent years. Why—in a period in which jobs are being created so slowly—should the number looking for work be taken as the best measure of capacity for more work? Surely the number that has worked in the recent past provides a better gauge.

Analysis of the available slack in the labor market requires separate attention to five distinct groups: young people whose school enrollment has risen but not as much as their employment has fallen; prime-age men and women who have had differing trends in recent decades; and older men and women whose attachment to the labor market took a U-turn in the 1990s.

Over the last several years, young people have been attending school at higher rates and those in school have lower employment rates than those not in school. The school enrollment rate among people ages 20 to 24 has risen from 33.3% in the last three years of the last expansion to 34.7% in the first quarter of 2005 (Figure A-1.) At the same time, the employment rate of this age group has decline from 70.6% to 66.0%. Thus, while the schooling rate rose 1.4 percentage points, the employment rate fell 4.6 percentage points. The increase in school enrollment has contributed very little to that decline. The employment rate would have been 70.3%—0.3 percentage points lower—in the last three years of the last expansion if the share of the population in school had been as high as today.⁵ Put another way, there would be an additional 0.9 million people ages 20 to 24 who would have employment today if their employment rate had recovered to the rate for the last three years of the most recent expansion and the school enrollment rate had always been as high for those three years as today (Figure A-2 ).

Next we turn to the situation for the 63 million women between the ages of 25 and 54 (often termed “prime age” in terms of the labor market). Their employment rate rose steadily from 33.0% when the data began in 1948, to 50% by 1973, to 70% by 1988. Their gains slowed in the 1990s, but by the last three years of the last expansion, their employment rate averaged 74.1% (Figure A-3 ). (That employment rate was notably higher than their average 69.9% employment rate at the end of the 1980s.) In the most recent month, their employment rate of 72.0% remained 2.1 percentage points lower than at the sustained peak. Another 1.3 million women would have employment today if they were employed at the same rate as the last three years of the last expansion.

The employment rate of prime-age men initially declined more sharply than for women during the last recession and following year and a half (Figure A-4 ). For the last two years, however, the employment rate for men has recovered somewhat, from a trough about 3.0 percentage points below the sustained peak to a shortfall of 1.9 percentage points today. If the downward drift in their employment rate is extended from the end of the 1980s expansion to the end of the latest expansion, the current shortfall is still 1.7 percentage points. For those 61.2 million men, the difference between the employment rate at the end of the last expansion and that today represents 1.2 million fewer men employed. We have chosen a more conservative methodology that takes account of the slight decline in their employment rate from the end of the 1980s to the end of the 1990s. This method yields a shortfall of 1.0 million people employed.

There does not appear to be measurable labor slack among those 55 and older. For both men and women, employment rates began to rise in the early 1990s and have continued right through the recession and beyond (Figure A-5 ). The employment rates for both women and men have risen about 4percentage points since the start of the last recession in March 2001. There is no reason to believe that people in this age group will withdraw from the labor market as the economy improves. As the labor market tightened in the late 1990s, they did not withdraw from the labor market, but instead flocked into it. Because there appears to be a structural shift occurring, it would be a mistake to include them in calculations of how much the current employment rate falls short of the level reached prior to the last recession.

In summary, the labor market is currently providing an estimated 3.2 million fewer people with employment than it should—0.9 million for people ages 20-24, 1.3 million women between the ages of 25 and 54, and another 1.0 million men ages 25-54.

Endnotes

1. Over the past year-and-a-half (January 2004-July 2005), job growth in the Bureau of Labor Statistics household survey has averaged 193,000 per month, while that of the payroll survey has averaged 186,000; both values reflect growth rates well below historical levels for this stage of recovery.

2. In this sense, we argue that key job market variables are cyclically, not structurally depressed. However, the distinction between the terms cyclical and structural is easily blurred. For example, the job market can consistently operate below full employment for structural reasons, such as long-term trade imbalances or monetary authorities who prevent growth strong enough to reduce the unemployment rate to lower rates consistent with stable inflation.

3. Quarterly data are useful in such a comparison because monthly data for minorities tend to be too erratic.

4. A group’s employment rate is the number employed divided by their population.

5. The actual employment rates are taken for those in school and not in school in the most recent quarter (the first quarter of 2005), and those rates are multiplied times the number of people ages 20 to 24 who would have been in school or not if the school enrollment rate had always been the rate of the most recent quarter.

References

Bernstein, Jared and Dean Baker. 2003. The Benefits of Full Employment. Washington, D.C.: EPI.

Blanchflower, D. and A. Oswald. 1994. The Wage Curve. Cambridge, Mass: MIT Press.

Bradbury, Katherine. 2005. “Additional Slack in the Economy: The Poor Recovery in Labor Force Participation During the Business Cycle.” Boston, Mass: Federal Reserve Bank of Boston. http://www.bos.frb.org/economic/ppb/index.htm

Katz, Lawrence and Alan B. Krueger. 1999. “The High-Pressure U.S. Labor Market of the 1990s.” Brookings Papers on Economic Activity. Washington, D.C.: Brookings Institution.

Minneapolis Star Tribune. 2005. Economy slows, but experts expect revival. July 30.

Price, Lee and Sujan Vasavada. 2005. Last twelve months of job growth trail similar periods of previous expansions . Washington, D.C.: EPI. July 20.

Wall Street Journal. 2005. Job data might be a salve for Bush. August 8.

Washington Post. 2005. Economic news isn’t helping Bush. August 6.