Getting Back To Full Employment

Getting Back to Full Employment: A Better Bargain for Working People 5Misunderstanding the impact of the American Recovery andReinvestment Act (ARRA) – the 2009 stimulus was an effective jobcreator, though not all of its programs were as effective as others.Research has given us a good idea as to which measures had the biggestbang-for-the-buck on job creation, and instead of caterwauling aboutthe “failed stimulus” we should take advantage of this information tobring down the unemployment rate.3Yet, it will likely take more than short-term stimulus measures tomaintain full employment in the future. While there is no evidence that thestructural problem of a pervasive skills mismatch is holding us back on thesupply side of the labor market, we are facing structural deficiencies on thedemand side. That is, the market is not providing enough gainful employmentopportunities for all comers into the workforce.It just so happens that at the same time we have a crisis of deficientstructural demand, we also have a crisis of crumbling infrastructure. America’sonce-world-class systems of transportation (roads, bridges, railroads,airports), the power grid, the water supply, public buildings and spaces(schools, parks, offices, and sidewalks), and research and development aredeteriorating, and it is hurting our productivity and our living standards.Addressing the poor state of our infrastructure in a period of highunemployment would be a perfect marriage of demand and supply. But we canalso be more forward looking. Our per capita energy use is about twice as highas energy use in European countries that have comparable living standards.Much of our excess use is simply due to waste, and we can create well-payingjobs by resolving to make our homes, cars, offices, and other buildings moreefficient (Pollin 2012).We can also look at changing the structure of work as a way togenerate jobs. A full-time job in the United States typically means workingmany more hours a year than it does in Germany, the United Kingdom, orother wealthy countries. In those countries, four to six weeks a year of paid3 Michael Grunwald’s book, The New New Deal, provides an extensive analysis of theeffectiveness of the Recovery Act.

6Dean Baker and Jared Bernsteinvacation is the norm, and nearly everyone can count on paid sick days and paidparental leaves. As a simple arithmetic proposition, if everyone worked 20percent fewer hours and cut back their work accordingly, then employerswould have to create roughly 20 percent more jobs. The real world is morecomplicated, but the basic logic holds, other things being equal: If the typicalworker puts in less time on the job, more people will have jobs.We should also recognize that some people will find it almostimpossible to find jobs given the current state of the economy. In particular, inmany areas the teen unemployment rate exceeds 50 percent. In effect, thereare no jobs for teens. It is not fair to ask them to wait until the policymakerscan figure out how to fix the economy. We should be looking to give youngpeople jobs now, even if that means the direct creation of jobs by thegovernment. Conservatives can disparage these as “make-work jobs,” but thereis solid, conventional economics behind this: If the market fails to providethose willing to work a chance to contribute to national output, then policymust intervene to fix that market failure, and in this case there is the addedbenefit of giving people an opportunity in life. It’s not the fault of the typicalunemployed teenager that virtually all of the economic authorities failed torecognize an 8 trillion housing bubble.We believe there are few if any economic policy issues as important asfull employment. It is essential for reducing the income stagnation that hasbeset the middle class, reducing poverty rates among working-age families,pushing back against economic inequality, and improving our fiscal outlook.Today’s dysfunctional politics are doing nothing helpful to get us there. To thecontrary, policymakers in advanced economies are embracing “austerity”measures that push the other way.That is why it is important to lay out the case for full employment andthe path for getting there. The logic and evidence for full employment arestrong, and someday, hopefully soon, logic and evidence will matter again.

Getting Back to Full Employment: A Better Bargain for Working People7Chapter 2Evidence of the Benefits of Full EmploymentThe historical record in the United States supports the notion that,when labor markets are tight, the benefits of growth are more likely to flow tothe majority of working people. Conversely, when there’s slack in the jobmarket, as has been the case more often than not in recent years, workingfamilies fall behind.Job markets operating below full employment are not confined torecessions. Business cycle expansions over the past 30 years have featuredlabor markets with too much slack to provide workers with the bargainingclout they need to claim their share of the growth they’re helping to produce(the later 1990s were an important exception). Moreover, the last threerecessions have been followed by initially weak “jobless” and “wageless”recoveries, implying that, in recent years, incomes and wages have failed toget much of a lift in bad times or good ones.Indeed, the post-1970s period of slack job markets has also been aperiod of low- and middle-wage stagnation and rising wage and income (andwealth) inequality. The absence of full employment in most years since the1970s was not the only factor in play; the reasons for the rise in inequalityinclude globalization, technological change, a bubble-driven finance sectorclaiming disproportionate profit shares, declining unions, a falling value of theminimum wage, and more.

8Dean Baker and Jared BernsteinBut slack employment and its corollary – diminished bargaining power– get overlooked, in no small part because policymakers assume fullemployment is out of their control, though it is decidedly not. To give up onfull employment is a mistake, because in an economy in which collectivebargaining is minimal in the private sector and under siege in the public sector,full employment is the only route for working Americans can get ahead. Risingliving standards for the majority require a labor market that is tight enough toforce employers to raise compensation to the level where they can attract andkeep the workers they need. Whenever that force has been in place, workingpeople have done much better than when it’s been absent.Growing together or growing apart,and the role of full employmentAs discussed in Chapter 1, economists don’t have a good track recordin terms of quantifying a reliable definition of full employment or the costs ofsetting the NAIRU – the unemployment rate generally associated with noninflationary full employment – so high that it sacrifices growth and jobs,particularly for less-advantaged persons whose incomes are closely tied to theunemployment rate.We employ two methods to estimate where “inflationary” fullemployment might kick in. In this section we compare the CongressionalBudget Office’s NAIRU measure, plotted in Figure 2-1, to the actualunemployment rate. We do not claim that CBO’s (or anyone else’s) NAIRU isthe correct measure of full employment, but we’re letting it stand in for thisconcept for comparative purposes. In the next section we use changes in theactual unemployment rate to explore the relationship between movements inthe unemployment rate and wage trends for different groups of workers.

Getting Back to Full Employment: A Better Bargain for Working People9FIGURE 2-1Unemployment and the 96Q31999Q12001Q32004Q12006Q32009Q12011Q30Source: Congressional Budget Office and Bureau of Labor Statistics.Figure 2-1 shows that unemployment was generally lower before1980 than it has been since. Why is that? Demographics don’t explain thistrend, because the workforce has in general gotten older and better educatedover these years, and older people and those with higher education levels havelower-than-average unemployment rates. Globalization, which leads to theloss of factory jobs, and immigration of less-skilled workers may have played arole, but the larger story has to do with booms, busts, and macroeconomicpolicy mistakes.The post-1979 period includes the two worst recessions that occurredover the time depicted in this figure (the Great Recession and the early 1980s"double dip" recession). It also captures the so-called “jobless recoveries”coming out of the early 1990s recession, the early 2000s recession, and themost recent one. The current, large gap between actual unemployment andthe rate associated with full employment is clear at the right-hand side of thefigure.One way to quantify the differences between the pre- and post-1979periods is to count the number of percentage points that actual unemploymentwas above or below the estimate of full employment in each period. As

10Dean Baker and Jared BernsteinFigure 2-2 shows, between 1949 and 1979 the unemployment rate wasbelow the NAIRU more than it was above it, to the tune of 15 percentagepoints. This implies tight labor markets. A different pattern prevailed post1979: Unemployment was 31 percentage points above the NAIRU from 1980to 2012, though about half of those points are due to the Great Recession. Inthe pre-1980 period, the unemployment rate was below the full-employmentbenchmark in 84 out of 124 quarters; in the latter period, the unemploymentrate was lower in just 39 out of 132 quarters.FIGURE 2-2Cumulative Percentage Points Above or Below 07Source: Congressional Budget Office and Bureau of Labor Statistics.What was the impact of those very different labor market regimes on wagesand incomes of working families? Figure 2-3 plots low, middle, and highincomes, with its 1947 value set to 100. The trends reveal differences in thenature of income growth over the two periods. When labor markets weretighter, incomes for these different income classes grew together; when jobmarkets were slack, incomes grew apart.

Getting Back to Full Employment: A Better Bargain for Working People11FIGURE 2-3Low, Middle, and High Incomes, 1947-2011350300High250MiddleLow1947 470Source: Congressional Budget Office and Bureau of Labor Statistics.Again, other factors are at play besides the unemployment rate. Butthe correlations clearly show income growth, especially middle- and lowerincome growth, is associated with tight labor markets.Trends in the later 1990s illustrate this point. Figure 2-1 shows thatunemployment stayed below the CBO NAIRU for a number of years duringthis period, meaning job markets were tight. This dynamic led to real familyincome growth for all families at all income levels (though the top grewfastest, meaning inequality continued to increase in these years). Certainly thefactors that economists argue are holding back the income growth of middleand low-wage workers, such as globalization and technology, were in play inthose years. Yet strong labor demand created enough pressure to ensure thatlow- and middle-wage workers were able to get ahead.Figure 2-4 shows the results of a statistical exercise to test thecorrelation between full employment and trends in real income for differentgroups of families. Specifically, the exercise examines the relationship betweenchanges in real income by income group and the “deviation from fullemployment” – movements of the unemployment rate above or below thefull-employment benchmark (see data appendix for more details).44 Since the CBOs NAIRU, as shown in Figure 2-1, is fairly constant, this exercise is similarto using the actual unemployment rate rescaled by a constant.

12Dean Baker and Jared BernsteinFIGURE 2-4Impact of Higher Unemployment on Family Income and rce: Authors’ analysis of Census, BLS, and CBO data. See appendix for more info.The pattern of the bars shows that the lower your family income, themore you lose in slack labor markets. For families in the 20th percentile, foreach percentage point that the unemployment rate was closer to fullemployment, incomes grew 2.2 percent.The correlation was strongest for these families; at the median incomegrowth was about a third less; and for high-income families (the 95thpercentile) growth was two-thirds less. For African American families theimpact was similar to that for low-income families, and white families sawgains equivalent to gains at the median.The last bar is particularly important as it explicitly measurescorrelation between slack labor markets and the growth of income inequality,measured here as the ratio of high to low incomes. One extra point of labormarket slack is associated with a 1.6 percent increase in the ratio of high tolow incomes. Below, we see this same type of relationship in data on earnings,suggesting an important linkage between slack job markets, uneven wagegains, and income inequality.At least for working families, the mechanism upon which thesecorrelations rest is the paycheck, and that in turn is a result of two important

Getting Back to Full Employment: A Better Bargain for Working People13factors associated with tighter job markets: more hours of work, and higherhourly wages. Both are especially important for less-well-off households.FIGURE 2-5Annual Hours Worked: Response to 10% Lower 0%Lowest FifthMiddle FifthTop FifthSource: Authors’ analysis of U.S. Census Bureau microdata (Annual Social and EconomicSupplement) provided by Economic Policy Institute.Figure 2-5 focuses on annual hours of work among low-, middle-,and high-income households (summing hours of work across households).Each bar represents the percent change in annual hours given a 10 percentchange in the unemployment rate.5 The average jobless rate over the period(1975-2011) was 6.5 percent, so a drop in the unemployment rate to slightlybelow 6 percent would raise annual hours for low-income workers by around2.5 percent. The impact for middle-income workers is about half that, and forhigh-income workers about half that again. In other words, the benefits of adrop in unemployment accrue disproportionately to lower-income families.This result is an average over the full period. Further analysis lookingspecifically at the full-employment years of the late 1990s reveals a particularlylarge impact on hours worked by families below the 20th percentile (Figure2-6). Hours worked were up 17 percent, representing over 100 more hours ofwork in 2000 compared to 1996. At high-income levels, hours were virtuallyunchanged.5 The bars in the figures are regression coefficients from regressing the log change in annualhours for each fifth and the log change in the unemployment rate. Each coefficient shownwas statistically significant beyond the 0.01 level.

14Dean Baker and Jared BernsteinFIGURE 2-6Change in Annual Hours Worked, 1996-2000, by Income Fifth20%17%15%10%5%5%2%1%1%Third QuintileFourth QuintileFifth Quintile0%First QuintileSecond QuintileSource: Authors’ analysis of U.S. Census Bureau microdata (Annual Social and EconomicSupplement) provided by Economic Policy Institute.Over the comparable period in the 1980s (1985-89), hours for thebottom group were up only 8 percent, and in the 2000s they were essentiallyflat, even as the economy expanded. In other words, full employmentprovides the opportunity for the lowest-income workers to expand their laborsupply. Contrary to conservative stories about how low-income people don’twant to work, these dynamics suggest that given the opportunity, they are themost eager.As we undertake this analysis, about 12 million are unemployed and 8million are working part time but would prefer to be full time. The trends inthese figures underscore how much pent-up labor supply there is in thecountry when the economy is poorly managed and the unemployment rate ishigh.We now turn to the impact of unemployment on hourly wages. Thenational data used so far are limited both by few observations and the inabilityto capture geographical variation. The national unemployment rate is, ofcourse, an average of jobless rates from across the country, weighted by therelative size of the local workforce. It is useful to tap that variation across placeand time, as we do here to examine the correlation between unemploymentand real wage growth.The next set of figures show the results of analyses of therelationship between the real wages of workers at the 20th, 50th, and 90th

Getting Back to Full Employment: A Better Bargain for Working People15percentiles and the unemployment rates in the state where they live, over theperiod from 1979 to 2011. We provide more detail in the technical appendixto this chapter but for now, the numbers on top of the bars represent thepercent change in real wages with the unemployment rate falls one point.Across many countries and many time periods, these analyses have shown thesame consistent relationship: Higher unemployment rates mean lower realwages.6The patterns here are consistently similar: the less you earn, the moreyou need a tight labor market to get ahead. Figure 2-7 shows that a 10percent decline in unemployment, say from 5 percent to 4.5 percent, isassociated with a 10 percent increase in the real 20th percentile wage, sayfrom 10 to 11.FIGURE 2-7Coefficients on Hourly Wage Variables, Low, Middle and HighWage 000.00020th50th90thSource: Authors’ analysis.Similar to the relationship we saw in Table 2-1, the wage level’sresponsiveness to unemployment falls as wages get higher, with the effect formid-wage workers half that for 10th percentile workers. For wage earners atthe high end of the pay scale, there’s virtually no impact of unemployment onwage levels.6 Our regressions follow the recent work of Manchin and Gregg (2012). For a thoroughdiscussion of the t

policy must avoid full employment because it will cause inflation, or that it must tolerate a cruelly high level of unemployment simply to avoid a slight risk of inflation. In the conventional view, the unemployment rate associated with full employment and stable inflation is called the non-accelerating inflation rate of unemployment, or NAIRU.