574 April 13, 1987 THE CHICKEN Lll TLE THEORY OF THE VANISHING MIDDLE CLASS Bruce Bartlett E.L. Wiegand Fellow INTRODUCTION Is America's middle class an endangered species? Yes say a number of vocal economists, politicians, and journalists who warn that growth of the U.S. service sector is destroying the middle class. They point to what the see as a decline in America's replaced, they say, by lower-paid service 'obs. he result they predict is the polarization of the U.S. into a two-tier society of rich an dT poor. Although not linked direct l y to trade such arguments fuel the pressure for protectionism to maintain traditional manufacturing jobs in steel, textiles, autos, and many other industries manufacturing base, with its concomitant loss of we Y I-paid jobs. These jobs are being Sky Not F a lling. Such warnings are as valid as Chicken Little's. The sky is not fallin on America's middle class. For one thing, the movement toward services is a long-term tren 8 that the U.S. shares with other nations, including Japan. For another, the service se c tor is not made up solely of low-paid jobs, nor does its growth come atthe expense of manufacturing, which actually is doing quite well. And finally, there is no evidence at all that the middle class is eroding. The fact is that the middle three-fifths of the population ranked by income, receive about 52 percent of total national income, a proportion that has been virtually unchanged since the Census Bureau began keeping such statistics in 19

47. A 1. See, for example, Jane Seaberry Middle-Class Dream Fade s for Some The Washlnaton Post January 4,1987 The Polarization of America: The Loss of Good Jobs. Fallina Incomes and Risina lneauality (Washington, D.C Industrial Union Department, AFL-CIO, 1986 Is the Middle Class Shrinking?'.Time November 3,1986, pp 54 - 56; Katharine L. Bradbury, The Shrinking Middle Class New Enaland Economic Review September/October 1986, pp. 41 -55; David Wessel U.S. Rich and Poor Increase in Numbers; Middle Class Loses Ground The Wall Street Journal September 22,1986; Robert Kuttner A Shrinking Middle Class Is a Call for Action Business Week.

September 16,1986, p. 16; and Barbara Ehrenreich Is the Middle Class Doomed The New York Times Maaazine September 7,1986. similar anal sis of annual earnings by the Bureau of Labor Statistics ind icates that the of total employment that they did ten years ago. In short, despite anecdotal evidence to the contrary, there is nothin in the aggregate data to indicate that recent changes in the Congress therefore should not fall for the false image of A m ericans being forced to exchange their well-paid jobs for positions as check-out clerks or short order cooks. While the U.S. economy faces some problems, a vanishing middle class is not one of them THE TREND TOWARD SERVICES middle thir J of workers, ranke d by earnings, make up almost exactly the same percentage economy, such as the s a ift from manufacturing to services, are eroding the middle class.

Services have been the main source of job growth in recent years. Employment in manufacturing fell from 21 million in 1979 to 19.2 million in 1986, although this is an increase from the 1982 low of 18.4 million jobs. Total employment in goods-producing industries, including mining and construction but excluding agriculture, peaked in 4 979 at 26.5 million jobs , falling to 24.9 million last year.

Although manufacturing employment continues to strengthen, virtually all of the net employment qrowth in the U.S. econom in the last few years has been in services, a broad category that includes transportation an J pub lic utilities, wholesale and retail trade finance, insurance, and real estate, government, and wide variety of other occupations.

Employment in this category has risen by more than 10 million jobs just since 1980, up from 64.7 million to 75.2 million in 19

86. Thus 75 percent of all nonagricultural workers in America are employed in jobs classified as service producing. As Table 1 indicates, this is part of a long-term trend in the U.S. economy starting at least in the 1860s, when agricultural employment began its steep decline Why have services grown ,so rapidly? The simple answer is that, as an economy grows and matures, there is greater demand for services. As Table 2 illustrates, the consumption of services has increased dramatically over time, from 3 3 percent of total personal expenditure in 1950 to over 52 percent in 1986 The reasons for this growth in service expenditures 1) The desire of people to consume more goods seems limited as their income rises.

Noted Adam Smith: I'pe desire of food is limit ed in every man by the narrow capacity of the human stomach Households do not typically buy more and more food as their income rises, for instance, but tend instead to consume food in restaurants rather than at home.

Similarly, the demand for other basic goods tends to taper off once a certain level'of wealth is achieved 2) Sophisticated goods create a demand for services. Buying a car, for example, leads to 3) A growin and more complex economy increases the degree of specialization, with years of mainten a nce and repair services manufacturing a rms contracting out such services as accounting, personnel, advegsing and data processing that formerly were taken care of in-house or simply neglected 2. Adam Smith, Tm (17761 (New Yo& Random House, Modem Ubrary, 1 933, p. 164 3. Victor Fuchs, The Senrice Economv (New York: National Bureau of Economic Research, 1968 p. 4; Richard 6.

McKenzie and Stephen D. Smith, The Good News About U .S. Production Jobs (St. Louis, Missouri: Center for the Study of American Business, 1986 2Figure I 50 45 40 35 30 25 20 15 10 5 0 Distribution of Income In the US. m5 8 1-11 1.

11. I 60 62 64 66 68 70 72 74 76 70 SO 82 84 61 63 65 67 69 71 73 75 77 79 81 83 Year LEGEND Lowest 5th Second 5th Middle 5th Fourth 5th 11.11 1.11 I Highest 5th Source: Census Bureau Period 1860-69 1870-89 1890-99 190049 191 0-19 1920-29 1930-39 1940-49 1950-59 1960-69 Table 1 P ercentages of U.S. Labor Force Employed by Industry Aariculture Manufacturina Services 60 20 20 50 25 25 42 28 20 37 30 33 31 31 38 27 34 39 22 31 47 17 31 52 09 34 57 06 32 62 I Includes Forestry and Fisheries Includes Mining and Construction Source: US. Department of Commerce, Bureau of Economic Analysis, Lona Term Economic Growth. 1860-1970 (Washington, D.C US. Government Printing Office 1973), p. 101.

Table 2 Services as a Share of Personal Consumption Expenditures Year Percent 1950 32.8 1960 40.5 1970 44.4 1980 48.0 1981 1982 48.8 50.1 1983 50.5 1984 50.5 1985 51.4 1 986 52.2 Source: Commerce Department, Bureau of Economic Analysis 4Indeed, business services have been one of the fastest growing areas of employment and are projected to be the fastest g rowing area of employment growth over the next ten years.

The growth of services thus is a trend associated enerally with economic development rather than one unique to the United States. Predicta B ly, therefore, the same pattern is found in other countri es. As Table 3 illustrates, employment in services has grown sharply Countrv Spain Table 3 Employment in Services as a Share of the Labor Force 1965 1980 32 46 Ireland 41 48 Italy 34 48 New Zealand 51 56 United Kingdom Belgium Austria Netherlands France J a pan Finland West Germany Denmark Australia Sweden 50 48 36 50 43 42 41 42 49 52 46 59 61 50 63 56 55 53 50 61 61 62 Canada 57 65 Norway 48 62 Switzerland 41 55 United States 60 66 Weighted Average 48 58 Source: World Bank and International Labour Office 4 . Valerie A. Personick A Second Look at Industry Output and Employment Trends Through 1995 Monthlv Labor Review, November 1985, pp. 27-28; Wayne J. Howe The Business Services Industry Sets Pace in Employment Growth," Monthlv Labor Review, April 1986, pp. 2 9 -36 5in every Western industrialized nation. In fact, between 1965 and 1980, the Service jobs increase in Japan was over three times greater than the increase in the U.S THE QUALITY OF SERVICE JOBS While there is no dispute that there has been a dramatic e xpansion in employment in the U.S., thanks to the increase in service jobs, there also has been concern about the nature and wage rates of service jobs. The popular conception is that millions of Americans have been forced to swap well-paid manufacturing j obs for cash register positions at fast food restaurants At first glance, this seems plausible, but the data hide more than they reveal. In 1986 earnings in manufacturing averaged $396.01 per week compared with $265.20 per week in services. Thus, to the e x tent that lower-paid service jobs "replace" higher-paid manufacturing jobs, many Americans would seem to have difficulty maintaining a middle-class standard of living. A recent report commissioned by the Democratic members of Congress's Joint Economic Com mittee has generated wide publicity for this argument. According to the JEC study, six out of ten new jobs created during the current expansion pay less than $7,000 per year. The problem is that this study is seriously flawed.

Simplistic generalizations ab out manufacturing versus services mask important distinctions about the quality of such jobs. Moreover, the relationship between wage rates and family incomes is more complex than mi ht be imagined. Examining these issues in detail provides a very differe n t picture oft a e economic impact of services High Paid Services. In the first place, services include not only traditionally low-paid jobs in the retail trade but also many of the highest-paid jobs available, such as those in law, computers, advertising, and medicine. In addition, it is the higher-paying service jobs rather than lower-paying unskilled service jobs that are expanding most rapidly.

It turns out that much of the decline in manufacturing has been in the lower-paying manufacturin positions suc h as those in textiles and leather products. Moreover, the relative jobs, leaving the relative position of the middle class unchanged. For example, the proportion of total employment by workers classified as professional and technical will rise from 16.3 percent in 1982 to 17.1 percent by 19

95. By contrast, She proportion of workers classified as laborers will decline from 5.8 percent to 5.5 percent.

The pattern of service earnings, meanwhile, has been influenced strongly by the high proportion of part-t ime jobs in this sector. Example: Some 20 percent of service jobs are part-time, compared with less than 5 percent in manufacturing. Part-time jobs, in turn generally pay less than equivalent full-time jobs in the same business. Obviously, a large number o f part-timers pulls down the average level of wages, giving the false impression that typical full-time workers are suffering an erosion of income contraction o B higher-paid jobs has been matched by a more than equal expansion in well-paid 5. Japan went f rom 42 to 55, an increase of 31 percent. The U.S. went from 60 to 66, an Increase of 10 percent 6. Barry Bluestone and Bennett Harrison, The Grea t American Job Machine: The Proliferation of Low Waae Emdovment in the U.S. Economv (Washington, D.C.: Joint Economic Committee, December 1986 7. Neal Rosenthal The Shrinking Middle Class: Myth or Realw Monthlv Labor Review, March 1985, pp. 3-

10. P 6Erroneous Study. In a serious error of scholarship, the Joint Economic Committee study fails to distin uish betwee n part-time and full-time emplo ment in its analysis of service wa 8s it turns out that only 22.5 percent of jobs created in the 1979-1984 period were iathe low-paid category, rather than the 58 percent share indicated earlier in the study The study made s everal other critical errors. Example: by comparing job growth between 1979 and 1984, Harrison and Bluestone chose the two most extreme years possible, thus obscuring the long-term trend, which indicates no change in the distribution of jobs. Using the Ha rrison-Bluestone data, Table 4 puts the numbers into a continuous series, instead of merely comparing two base periods, 1973-1979 to 1979-1984, as they did.

From the underlying trend, it becomes clear that Harrison and Bluestone carefully selected only the periods that ave apparent support to their argument. An examination of the Only at t a e very end of the study are part-timers bro z en out separately. When this is 8 one whole period decisive B y refutes their contention.

Table 4 Percentage Distribution of New Jobs by Earnings Low Middle Hiah I Year 1973 1974 31.8 51.6 16.6 32.0 52.6 15.4 1975 32.2 52.6 15.1 1976 31.7 52.8 15.5 1977 32.2 51.6 16.2 1978 31.2 52.0 16.9 1979 30.6 53.0 16.3 1980 31.9 54.1 14.0 1981 33.0 53.0 14.0 1982 33.0 54.0 13.0 1983 32 . 1 53.3 14.6 1984 32.4 52.6 14.9 1985 31.4 52.6 15.9 Source: Congressman Daniel E. Lungren and Christopher Frenze, The Chairman's Commentary,' House Republican Study Committee, March 2,1987, based on Harrison and Bluestone's data and categories of high, mi ddle and low-paying jobs 8. Bluestone and Harrison, OD. cit p. 41 7As can be seen, the proportion of jobs in the low-paid category actually has declined during the Reagan years from 33 percent of new jobs created in 1981 to 31.4 percent in 19

85. Over a longer period, it is uite clear that there has been no basic change in the distribution of jobs--a fact ignore a by Harrison and Bluestone. It appears that the P ight on an important economic issue.

Table 4 also indicates the importance of cyclical factors in this form of analysis.

Comparing a year during a cyclical peak in the economy with a year during a cyclical low point might well show an increase in the proportion of low-paid jobs in the econom number of low-paid jobs, however, likely will decline during the current upswing in t e business attacking the conclusions o f the Harrison-Bluestone study rincipal goal of the JEC report was to embarrass the Reagan Administration, not to shed Y; The cle as more well-paid jobs are created. This was emphasized strongly by Janet Notwood, x ead of the U.S. Bureau of Labor Statisti c s, in a pew York Times article Ignoring Famil Income. Finall Harrison and Bluestone chose to adjust their d income data for in k ation using the 8 onsumer Price Index CPI rather than the more appropriate Personal Consumption Expenditure Index (P E). It is widely recognized by economists that the CPI tended to overstate inflation during the 1970s. Thus using the CPI rather than the more accurate PCE index, tended to understate the real incomes of workers in the study. V$en the PCE deflator is used instead, the low-pay trend mainly disappears.

The data do not indicate, moreover, that families general1 are drifting downward out of the middle class, even when an individual takes a lower-pai cy job. Studies that purport to show such a drift othewise rely almost exclusively on earnings, rather than family income data. Yet the relationship between wage rates and family income is tenuous, largely because of changing famil size and the proliferation of two-earner families. Thus,even if high-wage jobs, it would not n e cessarily imply that families with middle-class incomes were declining in number for the sake of argument, t K e idea were accepted that low-wage jobs were replacing High Productivity Services. Another widel -held misperception is that service low-quality work environment. The service sector in fact is highly capital intensive, and the rod ivity growth of service workers compares well to manufacturing industries have low productivity and low capita Y intensity, contributing to an allegedly wor R ers. yF Ad m ittedly, the overall level of productivity ap ears lower in the service sector than in the manufacturing sector, but much of this may E e due simply to the difficulty in measuring service sector productivity. It is theoretically easy to measure output in manufacturing 9. Janet L No~wood The Jobs Machine Has Not Broken Down The New York Times, February 22,1987 10. See Warren Brookes, "Low-Pay Jobs: The Big Lie," The Wall Street Journal March 25, 1987 11. Richard

1. Kirkland, Jr Are Service Jobs Good Jobs F ortune, June 10,1985, pp. 38-43; Ronald E. Kutscher and Jerome A. Mark The Service-Producing Sector: Some Common Perceptions Reviewed,' Monthlv Lab0 r Review, April 1983, pp. 21 -24 A Productivity Revolution in the Service Sector Business Week September 5 1983, pp. 106,108 8since the number of units produced per worker can be counted. This task is much harder in services, where the product is less tangible. And there is no known way to measure productivity in government, a major area of service employment.

Finally, an employment shift away from manufacturing toward services improves the quality of life for most working Americans. Working in clean, air-conditioned offices general1 would be considered an improvement over physical labor on a noisy assembly the mselves as a point in favor of services line. Alt ough this is seldom mentioned by scholars, it is often stressed by workers THE HEALTHY STATUS OF MANUFACTURING Virtually all discussion of the "decline" of the U.S. manufacturing sector concentrates on emp loyment. But the real measure of manufacturin of course, is output. This shows no decline. Manufacturin as a share of Gross Nationa 7 Product has held remarkably steady for decades, as Ta 73 le 5 demonstrates.

Table 5 Year Real Manufacturing Output as a Pr oportion of Real GNP Percent 1950 21.4 1960 20.4 1970 21 .o 1980 20.9 1981 20.8 1982 20.1 1983 20.6 1984 21.4 1985 21.7 Source: Commerce Department, Bureau of Economic Analysis The reason for this is that rising productivity in the manufacturin sector all o ws more goods to be produced by fewer workers, just as rising productivity a as allowed agricultural 9employment to fall from 60 percent of the labor force in 1860 to less that one-tenth that number today. In fact, contrary again to what may be popular pe r ception, U.S manufacturing productivii is the highest in the world U.S. Productivity Lead. To be sure, the rate of U.S. productivity gains has lagged behind other countries in recent years mainly is a function of measuring techniques manufacturing output p er worker were 100 and each count increased productivity were 50 productivity wou 7 d have grown 2 percent risen 4 percent. This explains why America may seem to be lag ing behind Japan, while in fact the U.S. is more productive. Japan suffered considerab l e 8 estruction in World War II while the U.S. was essentially undamaged, so Japan started the peacetime era from a much lower level of productivity. Thus Japan's amazing productivity growth rates can be explained mainly as "catching-up Yet despite Japan's double-digit growth rates, its overall level of manufacturing productivity remains below the U.S., as Table 6 indicates.

Table 6 Growth in Manufacturing Productivity, Selected Countries Country 1960-73 1973-84 1980-84 1982-84 Productivitv Level Increase i n output per hour at an annual rate U.S. 03.2 1.2 4.0 5.8 100.0 Canada 04.7 1.6 2.4 5.2 085.7 France 06.5 4.6 4.7 4.6 081.3 Germany 05.9 3.8 3.1 4.7 090.4 Italy 07.3 3.7 3.5 4.4 084.1 I Japan 10.5 7.0 6.8 7.3 .093.3 U.K. 04.3 1 .o 5.3 5.3 059.3 I984 Sourc e: Mdly McUsic U.S. Manufacturing: Any Cause for Alarm New Enaland Economic Review Federal Reserve Bank of Boston, January/February 1937, p.

10. G The table indicates that in 1984, a Japanese manufacturing worker produced only 93.3 percent as much per hou r as an American manufacturing worker, even though the rate of manufacturing roductivity in Japan has grown more rapidly over the ostwar period University economist William Baumol sug ests that productivity levels in all countr;ps raising the leve P of Ja p anese productivity closer to that of the U.S. In 8 eed, Princeton eventually will converge as capital and tec a nology spread throughout the world 12. William J. Baumd A Modest Decline isn't All That Bad The New York Time February 15,1987 10The table also shows that, although America's productivity growth lagged behind its major trading partners for most of the 1960s and 1970s, since 1980 the U.S. productivity growth rate in manufacturin compares well with competing Fpuntries. Indeed, several CONCLUSION: T H E AMERICAN MIDDLE CLASS IS THRIVING reports point to a new era o B growth in U.S. manufacturing A number of recent studies have examined the question of whether the middle class is declining,fnd concluded, in contrast to the popular mythology, that the mi d dle class is alive and well popular notions about job polarization dissolve on close examination here's no Brookings Institution 6 enior Fellow Robert Z. Lawrence concludes Contrary to the common perception, the pro ortion of full-time workers with middle - class earnings in the earnings in the rest of the economy And an examination of the data b U.S. Bureau Labor Statistics economist Neal Rosenthal also found no evidence of midd r e-class decline Newsweek economics columnist Robert Samuelson, for example, o b serves Most pronounced trend for 'obs to cluster at the top and bottom of the pay sca T e.1115 Similarly production of goods is exact P y the samg as the proportion of workers with middle-class The growth of the service sector is, in fact, a natural devel o pment in the U.S. economy, and it is associated with rising, not declining wealth. America IS experiencing erosion neither in the manufacturing sector nor in the middle class. Those who make such arguments seem determined to ignore the evidence to further their own political agenda 13. U.S. Department of Commerce, International Trade Administration, 1987 U.S. Industrial Outlook (Washington D.C U.S. Government Printing Office, 1937 Barnaby J. Feder Production Returning to U.S The New York Times February 18, 1 987 Manufacturlng--Dead or Alive The Global SDectator, Bear, Strauss 81 Co January 29,1987 why Manufacturing Will Revive Business Week, January 12,1987, pp. 6648; and Robert J. Samuelson, "Glimmers of Manufacturing Revhl November 12,1986 14 The Shrtnklng M iddle Class and Other Myths Research ReDortg, American Institute for Economic Research, September 15,1986, pp.77-79; Patrick J. McMahon and John H. Tschetter The Declining Middle Class: A Further Analysis Monthlv Lab0 r Review, September 1986, pp. 22-27; William Baidwin Chicken Me's Income Statistics Forbes, March 24,1986, pp. 68-69; Sar A. Levitan and Peter E.

Carlson Middle Class Shrinkage Across the Board, October 1984, pp. 55-59; Robert

2. Lawrence Sectoral Shifts and the Size of the Middle Class The Brookinas Review, Fall 1984, pp. 3-1 1 The Myth of the Vanishing Middle Business Week, July 9, 1984, pp. 83,86; and Robert J. Samuelson Middle-class Media Myth National Journal, December 31,1983, pp. 2673-2678 15. Robert J. Samuelson The Myth of the Miss ing Middle Newswee k, July 1, 1985, p. 50 16. Robert

2. Lawrence The Middle Class is Alive and Well The New York Timeg, June 23,1985 11