There’s only one possible answer: transfer payments to persons who are not employed; “Rueff’s Law” holds up for 70 years.
How can the trend of unemployment rise while the share of income received by employed workers keeps shrinking? Unless policymakers can answer that question when they meet in Detroit March 14-15 for their “jobs summit,” they must admit they don’t know how to cure workers’ biggest problem, which has been worsening for a generation. Indeed, the most striking feature of the jobs summit is its air of pessimism, particularly among those most busily launching new initiatives. President Clinton’s economic advisers, it turns out, don’t even see the problem.
A little thought (confirmed by the facts) reveals that the problem has only one possible cause — the growing wedge of income earned by employed workers but paid to persons who are not employed. Without such “transfer payments,” the combination of rising unemployment and relatively falling take-home pay is flatly impossible. This must also explain much of the inequality of income and slowdown in the growth of real national income since 1973. The good news is that the problem can be cured by throwing the same process into reverse. The bad news is that President Clinton’s plans to expand such transfer payments — particularly through universal health coverage — must worsen the plight of American workers.
European Union President Jacques Delors summarized not only his own massive White Paper on unemployment, but also the attitude of most policymakers heading to the “jobs summit” in Detroit, when he snapped, “If there were a miracle cure, it would not have gone unnoticed.” EU policymakers openly admit that their welfare states have caused unemployment to rise even more sharply than in the U.S. But they fear that trimming lavish benefits would create low-wage economies of their own.
President Clinton, meanwhile, has consistently treated his own initiatives, not as a means of curing workers’ biggest problem, but only of coping with it. “There is no power to protect the people of this country from the changes sweeping the global economy,” he said recently. “I mean, the average 18-year-old is going to change work eight times in a lifetime, anyway, whatever we do. But we do have an obligation to help them.”
The first report of the Clinton Council of Economic Advisers reveals the source of Mr. Clinton’s curious ambivalence. The CEA finds that real growth of national income (not just labor income) has slowed since 1973, and that the distribution of this income has become more unequal — for example, because of differing skills and education. But the CEA does not acknowledge that the income of workers as a group has slipped in relative terms — or that this could explain increasing income inequality. The CEA asserts that “most of the increase in average unemployment over the 1970s and 1980s was due to slack aggregate demand” (page 129). Otherwise, it sees no evidence of declining job security, suggesting that this “perception” is caused by “media accounts” or by “a constant rate of job loss combined with greater income inequality” (page 126). Yet the CEA’s own forecast expects the minimum unemployment rate to be higher once again in this business cycle than in the last.
Commentators who do see the problem have variously speculated that it is due to: the greed of the owners of capital, who are said to be hogging a larger share of national income; the pressures of international trade in equalizing wage rates among countries; the decline of labor unions; investment or total demand which is insufficient to achieve full employment; insufficient national saving; rapid technological change; poor education and training; or an “underclass” promoted and perpetuated by welfare.
Some of these theories could explain a fraction of the slowdown in real growth of total national income, or a fraction of any increased inequality of incomes. (The CEA report cites most of them on this score.) Some are simply false. But none can possibly explain how unemployment can rise while the relative income of workers declines. In fact, classical, Keynesian and neoclassical economics all agree in requiring exactly the opposite.
John Maynard Keynes began his General Theory by saying: “Thus I am not disputing this vital fact which the classical economists have (rightly) asserted as indefeasible. In a given state of organization, equipment and technique, the real wage earned by a unit of labour has a unique (inverse) correlation with the volume of employment” (Harbinger, New York, 1965, p. 17).
The main reason — it is essential to grasp — is the diminishing returns from hiring more labor to produce with any given equipment. If all else is the same, two carpenters with only one hammer may build more houses in a year than one carpenter with one hammer — but not twice as many. But labor and capital receive incomes equal to their contributions to any change in output, or “marginal product.” Therefore as Keynes agreed), “any means of increasing employment must lead at the same time to a diminution of the marginal product and hence of the rate of wages measured in terms of this product” (p. 18).
Moreover, the real wage rate must stop falling once full employment is reached: if no more labor is forthcoming, output can’t increase, so real wage rates can’t fall any more.
So how is it even possible to have an unbroken relative decline in take-home pay while unemployment rises? Must we throw out all labor-market theory, or what? No: to understand what’s going on we must only recognize two important facts.
First, real wage rates can rise, without reducing employment, thanks to improved or increased “organization, equipment and technique” (which were assumed constant in the theory just cited). This means that the unemployment rate varies, not with real wage rates measured in absolute terms, but with the relative price of labor, which is equivalent to labor’s share of the total income resulting from production.
Second, there is a difference between wages and labor income. Standard theory treats the two as interchangeable, and perhaps this simplification was justifiable 60 or 70 years ago. But today we cannot ignore all the taxes and subsidies which are specifically designed to transfer income between and among the owners of labor and capital. To calculate the relative price of labor accurately, we must take taxes and subsidies into account. Yet every labor cost calculation you have ever read assumes, in effect, that workers and owners of capital ignore most taxes, transfer payments, and many costs of earning income. Such calculations assume that labor income is equal to pre-tax pay and fringe benefits plus employers’ payroll taxes; when in fact labor income must equal take-home pay plus transfer payments. Thus neither total labor income nor the income actually received by employed workers has (to my knowledge) been accurately measured — although this would seem to be the crux of the current debate.
To estimate labor income on this basis, I went to the national income and product accounts, and added pretax employment income (including fringe benefits actually received by employed workers, and the government’s estimate of self-employed labor income, which had to be reconstructed before 1947), plus transfer payments to persons, minus labor’s share of personal and payroll taxes. The take-home pay of employed workers will equal total labor income minus after-tax transfer payments to persons who are not currently employed. Because of data limitations, especially going back in time, such a calculation can be only approximate. But it is extremely enlightening.
Graph 1 shows that, exactly as theory requires, there is an extremely close correlation between labor’s share of national income and the unemployment rate. Whenever labor’s share of income increases, unemployment goes up; whenever labor’s share decreases, unemployment goes down. Labor’s share of national income exceeded 90% at the depth of the Great Depression; at the same time, unemployment peaked at 25%.
Moreover, again as theory predicts, there is a limit, set by full employment, below which labor’s share of national income cannot fall. The lowest share of national income ever received by labor since 1929 was about 64% — and it corresponded to the lowest unemployment rate on record: 1% at the peak of the World War II boom. Since then, the upward trend in labor’s share of national income has been mirrored by the rising trend in unemployment. On average, each 1% rise in labor’s share of national income has been accompanied by about a 0.8-percentage-point rise in the unemployment rate.
Yet while labor’s share of national income has risen by 7 percentage points since the war, the share received by employed wage-earners has declined by 9 percentage points. The entire difference is due to transfer payments to persons who are not employed. This is exactly what theory requires: without such transfer payments, any increase in unemployment must be associated with a rise in take-home pay as a share of national income. (That’s what happened during the Great Depression.)
In the graphs I have omitted veterans’ benefits (which are not universally available based simply on work status or need), to show the effect of civilian transfer payments. But either way, the correlation between unemployment and labor’s share of national income is about 90 percent (Graph 2), and the correlation of changes in both is almost as high (Graph 3).
The statistical relationship between the relative price of labor and unemployment is not new; it was known in the 1930s and 1940s as “Rueff’s Law,” because French economist Jacques Rueff first used it in 1925 to explain Great Britain’s unprecedented post-World War I unemployment (“Les variations du chomage en Angleterre,” Revue Politique et Parlementaire, December 1925). Other researchers quickly found the same relationship (despite poor data) in at least a dozen other countries (Jean Denuc, “Les fluctuations comparees du chomage et des salaires dans quelques pays de 1919 a 1929,” Bulletin de la Statistique Generale de la France, April-June 1930). This impressed Keynes, among others, who cited the relationship in his General Theory (p. 276). (How the American economics profession came in the 1960s to believe in a “Phillips Curve” tradeoff between unemployment and inflation — first suggested by Irving Fisher in 1926 [“A Statistical Relation Between Unemployment and Price Changes,” International Labor Review, June 1926] but refuted by Rueff — is one of life’s little mysteries.)
Since all this is exactly what theory would predict, “the astonishing thing is not that this relationship exists,” as Rueff remarked, “but that it should astonish anyone” (De L’Aube au Crepuscule, PLON, Paris, 1977, p. 96). However, conventional labor cost calculations — typically defined as gross labor compensation as a share of business-sector GDP — have been unable to explain most of the variations in unemployment, simply because they ignore most taxes and subsidies (Graph 4). All that I have done here is to calculate Rueff’s Law for the United States, taking taxes and transfer payments into account: a calculation which anyone can verify (or, indeed, refine).
Transfer payments alone explain the relative decline of take-home pay for workers as a group; but they do not entirely explain the relative decline of take-home pay for the average worker. Since the mid-1960s, the workforce has grown more than 20% relative to the total U.S. population — due the Baby Boom’s size, increased labor force participation, and reduced birth rate. Therefore, take-home pay per worker has fallen by about 20% more, relative to national income per capita, than is explained by the decline of take-home pay as a share of national income.
However, such comparisons can be misleading, especially when they involve a change in labor force participation. Suppose a one-earner family earns a net $40,000 in wages, and then the other spouse gets a job netting $20,000. Average income per worker has fallen from $40,000 to $30,000, or by 25%; but average wages per capita have risen from $20,000 to $30,000, or by 50%. (Of course, neither statistic reflects the non-monetary costs and benefits of the change.) There is some demographic effect on wages from the size of the Baby Boom; it should now be at its peak and decline over the next 25 years, because the following generation is much smaller. But to be accurate in assessing the income of workers as a group, rather than some class of workers, it is better to use workers’ share of national income rather than comparing average income per worker with national income per capita.
The increase in labor’s relative share of national income is related to the sharp slowing of real national and labor income in absolute terms. This is because not only wage rates and employment, but also output and income, are all tied in a unique relationship. Wage rates and employment are inversely related; but output and income (including labor income) are positively related to employment. Therefore, whatever caused the rise in labor’s share of national income must have caused not only the rise in unemployment, but also much of the slowing of real output and income growth relative to the long-term trend.
We can gauge this effect by comparing labor’s share of national income with the deviation of real national income from its 1929-93 growth trend (an annual average of about 1.8% per capita). (See Graph 5.) Whenever labor’s share of national income rose by 1 percent, national income fell an average of about 2 percent. This is just as apparent when comparing changes in national income with change in labor’s share of national income (Graph 6). The correlation is about 90 percent, and would be nearly perfect if we used a slightly curved trend line. But this means that whenever labor’s share of national income rises by 1 percent, real labor income falls by 1 percent. That is, real labor income and take-home pay are positively related to national income, and inversely related to labor’s share of national income (Graph 7).
Together with our earlier findings on unemployment, this confirms “Okun’s Law,” which relates employment to output: Okun’s Law says that when real output (and therefore income) rises 2% faster than the trend, unemployment falls 0.7 to 0.8 percentage points (Graph 8). But theory, as well as our calculation, shows that Okun’s Law cannot “work” without Rueff’s Law: both the change in output and the change in employment are necessarily related to an opposite change in labor’s share of national income. (In fact, the statistical evidence for Rueff’s Law is even stronger than for Okun’s Law.)
Thus, whatever has caused labor’s share of national income to rise, by about 7 percentage points above its minimum observed level, has not only raised unemployment about 6 percentage points, but also reduced national income about 14 percent, and labor income about 7 percent, below the long-term trend.
But what is the cause? It is easy to show that the rising postwar trend of unemployment is due to efforts by governments to legislate wage rates in real, not just money terms.
The two main methods of legislating wage rates are minimum wage laws and transfer payments to the unemployed. Both put a floor under market wage rates, which requires employment (and therefore output and national income) to be cut back until labor’s contribution to any change in output equals the legislated wage rate. Transfer payments to employed workers or to persons outside the labor force reduce hours worked voluntarily, but do not increase unemployment, since receiving the benefits requires either having a job or leaving the labor force.
Therefore, when we look at transfer payments, we should find that transfer payments to the unemployed raise labor’s share of national income, while transfer payments to persons outside the labor force reduce take-home pay as a share of national income. (Both should reduce employment and lower real national income.) And this is in fact what the data tell us. This is exactly what Graph 9 shows. Most of the rise in labor’s share of national income since the Second World War is due to the rise in transfer payments to the unemployed (unemployment insurance and welfare to the able-bodied); while the decline in take-home pay as a share of national income is due to the rise of transfer payments to persons outside the labor force (chiefly Social Security, Medicare, other government retirement pensions, and benefits to the disabled).
With the help of Rueff’s Law, we can pin down the causes of the rise in unemployment in the United States still more precisely. The role of specific benefits to the unemployed in causing unemployment, will be revealed by the share of such benefits in national income. Similarly, the role of the minimum wage in causing unemployment is revealed by calculating aggregate wages paid at the minimum wage as a share of national income.
As Graph 10 shows, the entire upward trend of labor’s share of national income since 1960 is accounted for by welfare to the able-bodied. Unemployment insurance has no upward trend as a share of national income, and in fact has added nothing to unemployment since its institution. Except in 1975, when benefits were temporarily expanded to an unusual degree, the entire unemployment insurance system has never added more than about 1 percentage point to the unemployment rate. But welfare to the able-bodied has mushroomed. Based on its share of national income, welfare now adds more than 3 percentage points to unemployment: more than 2 percentage points have been added since the 1960s. And more than 1 percentage point of unemployment was added just since 1988 — thanks in large measure to the Bush administration’s acquiescence to large benefit increases, particularly for Medicaid.
Figures on aggregate wages paid at the minimum wage are available only since 1979. They show that the minimum wage added about 1 percentage point to unemployment in 1980, but that this effect virtually disappeared by 1990 as the result of “benign neglect” — holding the wage constant at $3.35 an hour while national and labor income rose. According to the Bureau of Labor Statistics, by 1990 fewer than 1 percent of all American workers were paid at the minmum wage, representing only a fraction of a percent of national income. (Some workers, exempt from the minimum wage, were paid less, but most were paid more.) The minimum-wage hike from $3.35 to $4.25 in 1990-91, again under the Bush administration, added about half a percentage point to theunemployment rate. But this effect should decline over time, as long as the minimum wage is not raised or indexed.
The remaining variation of labor’s share of national income — and therefore of unemployment — shows a cyclical variation, but no significant upward trend since the Second World War. Without the effect of all transfer payments to the unemployed, the unemployment rate would not have exceeded 4 percent during the last business cycle, and would have fallen as low as 2-1/2 percent.
Now, as we saw, President Clinton’s Council of Economic Advisers argues that “most of the increase in average unemployment over the 1970s and 1980s was due to slack aggregate demand.” If true, this would mean that unemployment could be reduced, and national income substantially increased in real (not just dollar) terms — albeit at the cost of higher inflation — merely by printing and spending more money. And in fact, the Clinton people have spent much effort lately telling our trading partners to put aside labor-market reforms and instead — you guessed it — “stimulate aggregate demand.”
But the rising postwar trend of unemployment — above all during the record peacetime inflation of the 1970s and early 1980s — cannot possibly be explained by “slack aggregate demand.” The last person who would agree to such a notion is John Maynard Keynes. It’s true that Keynes and his adversaries disagreed about what caused fluctuations in national income and output, and that it had something to do with aggregate demand. But that argument doesn’t apply here.
Crudely put, the classical economists argued that unemployment was caused by the failure of money wage rates to adjust to the existing level of money spending; while Keynes argued that it was a failure of policy to adjust total money spending to the existing level of money wage rates. Stripped of secondary complications, the two theories amount to the same thing in relative terms (as we saw, both involve a decline in labor’s share of national income as the necessary complement to rising employment, wages and national income) — but with very different price levels and social consequences. The postwar “neoclassical synthesis” tried to combine both possibilities into a single theory.
During the early 1930s, when the collapse of the international monetary system caused prices to fall sharply, Keynes’s view was more correct in practical terms — in the sense that it would have been better to prevent the 50% deflation of prices than to expect money wages to adjust to it, even temporarily.
But that’s not the case now — and not only because prices have been continuously rising. The big difference is that in Keynes’s day, minimum wage laws and unemployment benefits were fixed in money terms: the British “dole” was so many shillings a week, for example. A fall in the price level (for example, after Great Britain returned to the pre-World War I gold value of the pound) would raise the “real” value of the dole or the minimum wage, increasing unemployment; a rise in prices would lower the minimum wage or benefits in real terms, reducing unemployment. Engineering a deliberate rise in prices, as Keynes in effect proposed, instead of changing wage and benefit laws, may have been messy and perhaps cynical — but it was effective in reducing the unemployment caused by those laws.
But such a strategy was effective only until people (understandably) tried to gain legal protection from the inflation caused by “demand management” policies. Since the early 1970s in most countries, minimum wage laws and especially transfer payments have been indexed to the price level, or else (as with medical benefits) provided “in kind” regardless of the money cost. This effectively fixes wage rates in real terms, no matter what happens to prices.
Keynes himself predicted the result: “If, as in Australia, an attempt were made to fix real wages by legislation, then there would be a certain level of employment corresponding to that level of real wages,” he wrote in his General Theory. This is why the “Phillips Curve” tradeoff between inflation and unemployment seemed to work for awhile, and then disappeared altogether in the inflationary 1970s — that is, once indexing became almost universal.
Keynes stated explicitly that fixing real wage rates would render “Keynesian” policies useless for anything but causing inflation: “Moreover, it would, in this event, be impossible to increase employment by increasing expenditure in terms of money; for money-wages would rise proportionately to the increased money expenditure, so that there would be no increased expenditure in terms of wage-units and consequently no increase in employment.”
Thus, it is fair to call the Clinton Administration’s view of labor markets idiosyncratic; but to call it “unreconstructed Keynesianism” is to miss the whole point. The Clinton labor-market strategy was designed, as it were, by the first generation of Keynesians who never read Keynes.
The Clinton strategy violates three simple rules which, if followed, would restore full employment, raise real national income and increase workers’ share of it. The first is that transfer payments, though sometimes necessary, must always be a last resort, not the normal source of income. The second is that all benefits must be fixed in money terms, not indexed or provided in kind. The third is that policies designed to provide a “living family wage” must avoid fixing wage rates and be tied to current employment.
First, transfer payments must be a last resort simply because every real benefit has a real cost. As we saw, benefits to the unemployed increase unemployment; benefits to persons who are employed or outside the labor force reduce hours worked. In every case this means lower employment, lower national income and lower labor income. To the extent that transfer payments substitute for provision out of private saving, insurance, or mutual family aid, they reduce real income in another way: by reducing investment in physical or human capital (e.g., purchasing equipment or raising families). And transfer payments always squeeze employed workers in relative terms, because they require that provision for human needs come out of labor’s rather than capital’s share of national income.
Second, benefits must be fixed in money terms. As we saw, if benefits to the unemployed are indexed, then real wage rates are fixed — and this fixes employment, output and income at a lower level. If benefits to those outside the labor force are indexed, it means that non-workers have protection from inflation which employed workers — whose take-home pay is reduced dollar-for-dollar — do not. Both fairness and efficiency demand that benefits be fixed in money terms and raised only when it can be afforded. That’s how Social Security was run for almost 40 years.
Third, a “living family wage” is achievable — as long as benefits are conditioned on being employed and do not seek to fix wage rates. In this case, unemployment and the shares of labor income and take-home pay in national income will remain unchanged. In the final analysis, benefits for lower-income workers are paid by upper-income workers. Examples of this principle which have been supported by the Clinton administration are the earned income tax credit and work requirements for welfare to the able-bodied.
Unfortunately, these positive initiatives are outweighed and nullified by other proposals. Labor Secretary Robert Reich’s proposal to raise and index the minimum wage must not only increase unemployment; it must guarantee the failure of welfare reform — because few trying to leave welfare could find employment at such a wage.
Likewise, some positive plans to speed re-employment are overshadowed by the Administration’s attempt to transform the whole principle of unemployment insurance — from a “tiding-over” of temporary unemployment, into an open-ended “dole” for long-term unemployment. That was exactly the fatal mistake that turned the British system after World War I and the European systems in the 1970s into unemployment traps for millions of workers. Precisely because of its limited and temporary benefits, the current U.S. unemployment insurance system is the best in the world, providing a safety net without destroying the habit of work.
Finally, the Clinton plan for universal health care scores a hat trick by violating all three principles of sound policy. First, transfer payments are made the normal means of providing for health care, not a last resort. Second, benefits are guaranteed in kind, which guarantees that there can be no dollar-cost containment. Third, the plan’s benefits would fix real wage rates and are not tied to employment — and so must further raise unemployment and reduce take-home pay as a share of national income.
If health benefits, like the earned income tax credit, were conditioned on being employed, there would be no increase in unemployment or change in the share of national income received by employed workers as a group — though higher-income workers would once again be socked, and total hours worked would decline.
But the Clinton health plan is not a means of achieving a “living family wage”: it offers a minimum income, whether or not the recipient is employed. “Universal” health care means that a package of benefits, currently worth up to $6,000 a year, will be extended not only to workers but also to the unemployed and to those outside the labor force. Yet the entire cost must come out of the earnings of employed workers. This is why the Clinton health plan cannot avoid increasing the unemployment rate and further reducing take-home pay as a share of national income.
Thus the First Lady was poorly advised in claiming that there is “no evidence whatsover” that either raising the minimum wage or enacting universal health care would have such effects. Mrs. Clinton is free to opine that universal health coverage is worth the cost of higher unemployment and lower take-home pay; but it’s absurd to deny that there is such a cost. (If there’s no cost, why palter? Why not a minimum wage of $500 an hour, and health benefits worth $50,000 apiece?)
Even those who may share many of President Clinton’s goals are forced to paraphrase Bosquet’s remark on the Charge of the Light Brigade: It’s magnificent, but it’s not economics.
In fairness, this blindness is not limited to the administration. As we saw, the past damage has been done under Republican and Democratic administrations alike. And many conservatives, like the Clinton CEA, somehow manage to ignore or at least downplay the very real relative slippage of workers’ incomes. Boilerplate language about “full employment” is in both parties’ platforms — but when was the last time you heard a politician in either party actually mention it as a desirable and achievable goal?
It’s time for policymakers to snap out of their funk. As an explanation of the postwar rise in unemployment there is simply no such thing as “natural” or “structural” unemployment — in the sense of unemployment arising out of the nature or structure of the economy, and therefore beyond the reach of policymakers. There is such a thing as “institutional unemployment” — but it comes precisely from misguided policies which legislate real wage rates, and necessarily lower real labor income. To confuse the two (as is now becoming fashionable) is a counsel of despair: abandon hope, ye who enter the labor market.
Nonsense. An unemployment rate of, say, 3% is eminently achievable, both in America and in Europe, merely by following time-tested principles. And doing so must always raise, not lower, workers’ real incomes, in both absolute and relative terms: Europe, take note.
All that is required is for policymakers to stop moping around “jobs summits,” pull up their socks, and change the misguided policies which, alone, have caused the combination of rising unemployment and relative decline in workers’ real incomes.
John Mueller is a principal of Lehrman Bell Mueller Cannon, Inc., an Arlington, Va., financial market forecasting firm.
Calculating Labor’s Share of National Income Net of Taxes and Transfer Payments
The validity of any theory depends on the ability of anyone to test it against reality. Rueff’s Law is no exception. The purpose of this appendix is to describe how to reproduce the calculations of labor income in the foregoing text.
Most of the data come from the National Income and Product Accounts. In the notation, NIPA table and line numbers are used. For example, 1.1.1 means NIPA Table 1.1, line 1.
Total labor income equals wages and salaries, plus other labor income, plus the BLS estimate of self-employed labor income, plus net transfers to persons, minus personal contributions for social insurance, minus labor’s share of personal taxes. (Employers’ contributions for social insurance are ignored, since they are, in effect, already included as one means of financing transfer payments to persons.) In the graphs, we also exclude veteran’s benefits, though this makes a noticeable difference only in the few years immediately after World War II. Thus the calculation of labor’s share of national income is: (1.14.3 + 1.14.8 + 2.1.15 – 2.1.29 – 2.1.23 – [2.1.24 x labor’s share of personal taxes] + BLS self-employed labor income – 3.12.19 – 3.12.16)/1.14.1.
Self-employed labor income is the difference between business-sector compensation as figured by the BLS for labor cost calculations, and business-sector compensation in the NIPA (1.14.5 + 1.14.7 + 1.14.8). Before 1947, self-employed labor income is calculated as the sum of the products of the corresponding lines in NIPA tables 6.6A and 6.7A, times 1.035 (a grossing-up factor, designed to match the BLS estimate of self-employed labor income for 1947).
Labor’s share of personal taxes is assumed to be proportional to labor’s share of labor’s and capital’s combined taxable income. Labor’s taxable income includes total wages and salaries plus self-employed labor income plus taxable transfer payments. Capital’s taxable income includes proprietors’ income other than self-employed labor income, personal dividends, personal interest, personal rental income, and capital income included in adjusted gross income which is not included in personal income. Labor’s share of personal taxes is therefore (1.14.3 + self-employed labor income + 2.1.15 – 8.24.3)/(8.24.10 – 2.1.23 + 1.14.3 + 1.14.9 + 2.1.13 + 2.1.14 + 1.14.7 + 2.1.15 – 8.24.3). Before 1947, all government transfer payments are assumed to be untaxed, and capital income included in AGI but not in personal income is assumed to equal 1.9% of personal income (2.1.1), as in 1947.
Employed workers’ income is equal to total labor income, minus net transfer payments to persons other than veterans’ benefits (2.1.15 – 2.1.29 – 3.12.16 – 3.12.19). This requires adding the tax on transfer payments to persons, which equals personal taxes (2.1.24) times the fraction of taxable transfer payments (2.1.15 – 8.24.3) in total taxable labor and capital income (see the previous paragraph).
Unemployment benefits equal 3.12.6. Welfare benefits assumed to go primarily to the able-bodied unemployed include Food Stamps (3.12.23), state & local public assistance (3.12.34), and “other” Federal transfer payments (3.12.28), but exclude payments primarily to aged, blind or disabled, such as SSI, Black Lung benefits, etc., and exclude payments to the employed, such as the earned income tax credit.
Figures on population and the labor force used in per capita and per worker comparisons are from the Census Bureau, Commerce Department, and the Bureau of Labor Statistics.
Trade Unions and Take-Home Pay
Many different theories have been put forward to explain how the trend of unemployment can rise while the share of national income going to employed workers declines. Because there is only one possible explanation — transfer payments to persons not currently employed — it is not necessary to refute the other arguments one by one. But it is certainly possible to do so.
For example, it is sometimes argued that the relative decline of workers’ incomes is due to the decline of trade unions. For example, John B. Judis recently asserted in the New Republic: “The unfashionable truth is that the decline of American wages has been largely a result of the decline of American labor unions” (“Why Your Wages Keep Falling,” February 14, 1994).
Now, it is one thing to say that a combination of workers, buttressed by legal privileges, creates a wage premium for unionized workers relative to other workers; but something else again to claim that this increases take-home pay for workers as a group.
Keynes contradicted this idea in his General Theory: “In other words, the struggle about money-wages primarily affects the distribution of the aggregate real wage between different labour-groups, and not its average amount per unit of employment, which depends, as we shall see, on a different set of forces. The effect of combination on the part of a group of workers is to protect their relative real wage. The general level of real wages depends on other forces in the economic system” (p. 14). [emphasis in original]
As Graph 11 shows, the relative decline of real take-home pay per worker does bear a striking relationship to trade union membership. The trouble is that, at least since the Wagner Act of 1935, significant changes in trade union membership have always followed major changes in real take-home pay, by one to three years.
I suggest that the decline of trade union membership is not the cause but the result of the relative decline of workers’ real income. History shows that workers join unions when employment and wages are relatively high — that is, when they have the least to lose — and they desert unions when employment and pay are relatively low. This ought to be good news for unions, since it means there is nothing at all inevitable about the decline of union membership. For better or worse, union membership will rise again if we ever solve the problems of American workers.
Unfortunately, efforts to improve the plight of American workers are hampered precisely by the fact that the interests of workers are represented, if at all, by union leaders rather than by labor leaders. The prevalent attitude among union leaders is that “what’s good for the UAW is good for American workers” — which, I suggest, has no more validity than “what’s good for General Motors is good for America.”
Union leaders on the whole support universal health care — despite the fact that it must necessarily raise unemployment some more and further lower take-home pay of employed workers as a share of national income. If unions do sign on to the massive legislation now congealing in Congress, it may turn out to be the longest suicide note ever written.