Good Times Bad Times: Postwar Labor’s Share of National Income in 16 Capitalist Democracies
Tali Kristal
StanfordUniversity
Working Paper 09-3
September, 2009
The Center for the Study of Poverty and Inequality is a program of the Institute for Research in the Social Sciences (IRiSS). Support from the Elfenworks Foundation gratefully acknowledged.
1
Good Times Bad Times:Postwar Labor’s Share of National Income in 16 Capitalist Democracies
Abstract
This paper returns to a classic question of political economy – the zero-sum conflict between capital and labor over the division of the national income pie. A detailed description of labor's share of national income in sixteen industrialized democracies uncovers two long-term trends: an increase in labor’s share in the aftermath of World War II, followed by a decrease since the early 1980s. In this paper Ipropose a model of the relative bargaining power of capital versus labor towards an understanding of the dynamics of labor’s share. Time-series cross-section equations predicting the short- and long-term determinants of labor’s share support the vast majority of the theoretical arguments. The results, obtained for sixteen developed countries over the period 1960-2000, suggest that the common trend of labor's share dynamics is largely explained by indicators for inter-class struggle in the economic (i.e., unionization, strike activity), political (labor-affiliated government, civilian spending), and global spheres (southern import, foreign direct investments), as well as indirect indicators for the intra-class struggle among workers (bargaining decentralization).
Good Times Bad Times: Postwar Labor’s Share of National Income in 16 Capitalist Democracies
Introduction
This paper fills a lacuna in inequality research by systematically analyzing the division of national income amongst the two broadly dominant social classes that cooperate in its formation– capitalists and workers. Although it is a classic question of political economy, the zero-sum distribution of national income between labor’s compensation and capitalists’ profits (typically called "labor's share of national income")was a relatively neglectedtopic until recently. One possible explanation for its “disappearance” from public and scientific discourse is that labor's share of national income has widely been regarded by economists as one of the great constants of economic development. That is rather misleading, as this paper shows, for labor’s sharehas functioned as an indicator of the fluctuating position of workers in the distributional struggle with capitalists over the course of the postwar period.
Recently, however, labor’s share has been making a comeback in economic research (Atkinson, 2009). Much of this revitalization has been driven by evidence from the last two decades which controverts the usual argument that economic expansion will make most workers better off. The growth of productivity has expanded total income, but in many countries average real wages and employment are flat or even falling. Meanwhile, income growth has occurred mainly in capitalists' profits and at the very top of the wage distribution, sharply increasing income inequality (Piketty and Saez, 2006; Wolff, 2003). As a result, there has been a large and persistent reduction in labor’s share of national income in most developed countries (Economist, 2006; Blanchard, 1997). Nevertheless, except for a few studies, little is known about the scope of the decline in labor's share and its timingacross countries. Furthermore, we know relatively little about labor's share during the ascendancy of social-democratic projects in the aftermath of World War II. The aim of this paper, therefore,is to filla lacuna in inequality research by systematically analyzing labor's share in sixteen industrialized democracies over the period 1960-2000.
There is a lack not only of studies documenting the distribution of national income between workers' compensation and capitalists' profits, but also, with few important exceptions, of rigorous studies that explore the dynamics of labor’s share. Recent studies suggest two alternative hypotheses for the mechanism behinds the decline in labor's share. Computer and related information technologies are assumed to have benefited capital productivity more than labor and, by implication, to have reduced labor's share (Acemoglu, 2002; 2003; Blanchard, 1997). Alternatively, it has been hypothesized that the decrease in the bargaining power of workers in the labor market is the main potential explanation for the current decrease in labor’s share (Blanchard and Giavazzi, 2003). Although studies have shown that workers' collective action in the labor market through unionization and strike activity did in fact increase labor's share in the U.S. (Kalleberg, Wallace, and Raffalovich, 1984; Rubin, 1986; Wallace, Leicht and Raffalovich, 1999) and in Israel (Kristal, 2008), this line of argument has not yet been tested in a longer historical and wider geographical frame.
This paper suggests a broader conceptualizationof the relative bargaining power of capital versus labortowards explaining the long-term trends in labor’s sharethatinvolves a study of the inter-class and intra-class distributional struggles in the economic, political and global spheres. In the economic and political spheres,the relative bargaining power of capital versus labor is directly affected by labor movements that mobilize workers’ collective strength and exercise that strength via policies they seek to advance. Unionization, strike activity, strong labor-affiliated political parties that promote social legislation, government social spending – these are all the feasible means at labor's disposal to increase its share of income. Yet workers' success in extracting economic rewards from employers through these means depends in part on their collective ability to reduce intra-class competition. An additional front in the distributional struggle that is relevant to understanding labor’s share is more global. Economic integrationinto internationalmarkets for goods, capital, and labor over the last two decades has probably boosted capital's bargaining power relative to labor. Together, I refer to these inter-class and intra-class distributional struggles as a class struggle. I apply this conceptualization of class struggle towards an understanding of labor's share in sixteen capitalist democracies over the period 1960-2000.
The rest of the paper is organized as follows. In the next section, I describe the long-term patterns of labor's share that emerge over the period 1960–2000. In Section Two,I discuss the longstanding assumption of neoclassical economic theory that labor's share is constant and its recent reevaluation. In Section Three, I introduce the conceptualization of the relative bargaining power of capital versus labor towards explaining the dynamics of labor’s share. In Section Four,I describe the sources of data and the method applied. In Section Five,I apply the framework developed in Section Three to analyze changes in labor’s share in sixteen developed countries between 1960 and 2000. Section Sixcontains the conclusions.
LABOR'S SHARE OF NATIONAL INCOME IN CAPITALIST DEMOCRACIES, 1960-2000
By way of introducing the dataset for the dependent variable, in Figure 1 I present plots of the trends in labor's share for the sixteen capitalist democracies for which data are available. Labor’s share is measured by labor's compensation (wages, salaries and fringe benefits) as a percentage of Gross Domestic Product. According to national accounts definitions, GDP is the sum of labor's compensation and capitalists’ profits (profits of firms and public-sector business enterprises and income from dividends, interest, and rent).[1] Thus, if labor’s share rises (declines) by a certain amount, capital's share must decline (rise) by the same amount. Realized capital gains through the exercise of stock options and other assets, which became key components of top executive remuneration in the 1990s (Piketty and Saez, 2006), are generally not counted in the national income accounts and therefore are not part of either labor's compensation or capitalists' profits.
<Figure 1>
It is prominently apparent that since the early 1980s labor’s share has decreased in the Anglo-Saxon countries, in Continental Europe and even in Scandinaviancountries. On average, labor’s share declined by almost 8 percentage points since the early 1980s, from 73 percent in 1980 to 65 percent in 2000. In fact, labor's share in 2000 was the smallest it had been for at least four decades. The scope and timing of the decline in labor's share differ across countries, but the downward trend is clear. In Austria, Belgium, Finland, Ireland, and Sweden – where labor's share was relatively high during the 1970s – its decline during the 1980s and 1990s was rapid and substantial. Labor's share declined in these countries from9 (Sweden) to 22 (Ireland) percentage points between 1980 and 2000. Countries such as France, Germany, Italy, and Japanshow a more moderate decrease in labor’s share, between 4 and 9 percentage points since the early 1980s. In the U.S. and the UKthe decline in labor’s share was more gradual and amounted to only 2 percentage points. The increasing salaries of top managers and some professional workers in these countriesmay explain this moderate decline in labor’s share.[2]
While labor's share in the UK has not changed much since the 1970s, during the first half of the twentieth century labor's share in the UK showed a distinct upward trend, from 56 percent in 1913 to 72 percent in 1964 (Matthews, Feinstein, and Odling-Smee, 1982). In the U.S.too,labor's share of national income increased by 5 percentage points from the end of World War II until the late 1960s (Krueger, 1999). Systematic data on labor's share for the period before 1960 inother developed countries are not readily available, yet Figure 1 shows that the increase in labor’s share in the postwar period is not unique to the UK and the U.S. During the 1960s and 1970s labor’s share increased in most countries. In Denmark, Finland, the Netherlands, Norway,[3] and Sweden, labor’s share increased by from4to 6 percentage pointsduring the 1960s. During the 1970s, labor’s share increased also in Australia, France, Germany, and Japan.
To explain what is behind the upward and downward trends in labor’s share,Table 1presents countries' annual growth in real earnings and fringe benefits per employee and labor productivity. The rationale here is that growth in productivity increases national income; the question, though, is how much of that growth translates into wages and fringe benefits. Contrary to the common economic assumption that compensation tracks productivity, in all countries real earnings and fringe benefits have increased slower than labor productivity during the last two decades. InFrance, for example, while productivity increased by 2 percentage points on average during the 1980s, labor's compensation stagnated. Since productivity growth expands total income, slow income growth for workers implies faster income growth for capital. Thus, labor’s share has declined over the last two decades, since real earnings and fringe benefits did not track average productivity growth. The 1960s and the 1970s reveal some variance between countries. In Sweden, to take one example, labor’ share increased due to a more rapid rise in labor’s compensation relative to productivity. Since real earnings and fringe benefits increased in close conjunction with productivity in the U.S., labor’s share remainedrelatively constant during the 1960s and 1970s.
<Table 1
All in all, the general conclusion from the available evidence is that there is no real long-term stability in labor's share of national income during the postwar period. Two common long-term trends have been documented since the end of World War II in rich countries: an increase in labor’s share during the 1960s and 1970s (or earlier), followed by a decrease since the early 1980s. Figure 2 presents boxplots of the yearly spread of labor's share in 16 developed countries from 1960 to 2000. The boxes for each year mark the 25th and 75th percentiles of the distribution of labor’s share in the 16 countries. From 1960 to 1976,the median labor’s share increases by about 3 percentage points. A general decline in labor’s share begins in the late 1970s, with the median share decreasingby about 6 percentage points. Below I advance the argument that changes in capital's relative bargaining power versus labor's explain thelong-term patterns of labor's share that emerge over the period 1960–2000. But first, in the next section I elaborate on the economic mainstream explanations for the dynamics of labor’s share.
<Figure 2>
FACTORS' SHARES: FACTORS OF PRODUCTION AND THEIR SHARES IN PRODUCT
The extensive public and scientific discourse on the causes of rising inequality of earnings in developed countries may conceal the fact that interest in income distribution was marginalized in economic research until the early 1980s (Atkinson, 1997; Gottschalk and Smeeding, 1997). With a few important exceptions, sociology until recently has also been “strangely and remarkably silent” (Morris and Western, 1999, p. 624) regarding earnings inequality. This lack of interest reflected the view that earnings inequality in the U.S. showed little change between the end of the 1940s and the mid-1970s. Like earnings inequality, the peripheral nature ofnational income distribution between capital and labor may be an empirical outcome of what has long been regarded as a "stylized fact" of economic growth – the constancy of labor's share of national income (Kaldor, 1961).[4] Keynes, for example, wrote that “the stability [italics mine] of the proportion of the national dividend [income] accruing to labor…is one of the most surprising, yet best-established, facts in the whole range of economic statistics” ([1936]1973: 408-409). The constancy of labor’s share is derived by adopting a standard neoclassical Cobb-Douglas aggregate production function which assumes that labor and capital are paid their marginal productivity,that labor and capital are neither substitutes nor complementsfactors in the production process,and that the factors' shares are determined entirely by the production technology which increases the marginal product of capital and labor by the same amount.
Yet recently the longtime“stylized fact”regarding the constancy of labor’s share turned out to be not true for many developed countries. Two alternative explanations for the decline in labor’s share since the early 1980s have been advanced; both of them relax the Cobb-Douglas production function assumptions. The first line of explanation suggests that the increase in wages in the late 1960s and early 1970s led to capital-biased technological change (hereafter: CBTC), which has benefited capital productivity more than labor (Acemoglu, 2002; 2003; Blanchard, 1997). That is, the increase in wages not only led firms, over time, to move to technologies using less labor and more capital, but may have prompted them to develop new labor-saving technologies to avoid high labor costs. Once this technical change takes place, firms gradually reduce their labor demand.[5] According to the CBTC hypothesis, we should expect that the rise in productivity since the late 1970s is negatively associated with labor's share, while during earlier years productivity growth was positively associated with labor's share. Supporting only the first part of this hypothesis, Bentolila and Saint-Paul (2003) find a negative effect for productivity on labor's share in industry data for 12 OECD countries over the period 1972-93. While it seems likely that technological changes had someeffect on labor’s share, the CBTChypothesis by itself cannot explain the current decline in labor’s share. The main problem for the CBTC hypothesis is that computer and related information technologieshave advanced continually since the early 1980s,despite wage moderation during the 1980s and 1990s. CBTC also fails to explain why countries that are relatively similar from a technological point of view differ in the scope of the decline.
As an alternative to the CBTC hypothesis, Blanchard and Giavazzi (2003) argue that the explanation for the decline in labor’s share lies in workers’ bargaining power. Based on the assumption that wages not only reflect marginal productivity, but are also determined by bargaining in the labor market, their model implies that,starting in the mid-1980s, a decrease in the bargaining power of labor led to a reduction in the real wage at a given level of employment, and, by implication, a reduction in labor's share. While the bargaining hypothesis seems to fit well with the trend of decline in both unionization and labor's share, this relation has not yet been tested empirically. Moreover, by focusing solely on workers’ bargaining in the labor market, Blanchard and Giavazzi (2003) overlookmuch of the historical sequence of labor movements in developed countries such as labor-affiliated political parties and the expansion of the welfare state. This paper, therefore, suggests a broader conceptualization of the relative bargaining power of capital versus labor towards explaining labor’s shareand empirically tests it for 16 capitalist democracies over the period 1960-2000.
DISTRIBUTIONAL STRUGGLES AND LABOR'S SHARE OF NATIONAL INCOME
In the following pages,I develop the conceptualization towards an understanding of the dynamics of labor’s share. Figure 3presents the study’s hypotheses, which will be tested on sixteen countries over the period 1960-2000. Based on the theoretical discussion below, two basic arguments underlie my analysis. First, labor’s share is a function of shifts in macroeconomic variablesand class struggle indicators on the political, economic and global fronts. Second, the effects of labor organizations on labor’s share depend on the intensity of the intra-class competition, and in particular on the level of fragmentation within organized labor. While indicators for inter-class and intra-class struggles are expected to explain the upward and downward long-term trends in labor's share over the postwar period, the business cycle is likely to explain its short-term fluctuations.
<Figure 3
Macroeconomic Business Cycle
That the business cycle shapes the relative bargaining power of capital versus labor and therefore affects labor’s share is an argument put forward mostly by Marxist economists. Specifically, labor’s share is expected to be inversely related to rapid economic growth, high rates of unemployment, and rising prices. Labor's share should decline throughout economic expansions because rises in wage rates are limited (by institutional constraints such as fixed-term wage agreements), while innovation and, therefore, productivity grow more rapidly than wages (Hahnel and Sherman, 1982; Sherman, 1979; Weisskopf, 1979). According to this argument, the negative relation between productivity growth and labor's shareis due to institutional constraints in the labor market and not necessarily an outcome ofhigher capital productivity, as argued by the CBTC hypothesis. Unemployment should reduce labor's share basically by decreasing labor's bargaining power (Broddy and Crotty, 1975; Glyn and Sutcliffe, 1972; see also Korpi, 2002), while rising prices lower labor’s share since labor’s compensation normally lags behind inflation more than profits (Munley, 1981). Earlier studies have found that labor’s share in the USindeed declined throughout periods of economic expansion and job losses, while inflation has yielded conflicting results (Raffalovich, Leicht and Wallace, 1992; Wallace et al., 1999).