Causes of the Decline of the US Wage Share 1960-2015

Deon Gibson
dfgibson79@hotmail.com

Abstract

This paper examines the decline of the US wage share over the period 1960-2015. It uses a model of the rate of change of the wage share to identify the variables most responsible for the wage share’s decline in the US. The model used in this paper is based in classical political economy as it assumes a negative relationship between the wage share and unemployment. This paper demonstrates that the impact of unemployment on the wage share increased as the US economy transitioned into the Neo-Liberal Era, that technological change is the variable with the most influence on the wage share, and that financialization and globalization have small significant effects on the wage share in the US. These results indicate a reduction in the bargaining power of workers as the main cause of the decline of the wage share in the US, as the combined effect of unemployment, declining size of government and degree of unionization, increases in technological change; weakens the bargaining position of workers.

Introduction and Rationale

Studies of the functional distribution of income[i] have shown a decline in the share of national income paid to labor. These studies[ii] identify several factors that have contributed to the decline of the labor income share or wage share in recent decades. For example in the mainstream literature largely dominated by the neoclassical traditions it has been argued that increases in the rate of technological change is the main cause of this decline (IMF 2007, EC 2007, Adih & Danninger 2017). While papers based in alternative perspectives have demonstrated that the decline in the wage share is primarily the result of declining unionization, government size, globalization and financialization (Jayadev 2007, Stockhammer 2009, 2013).

In the literature the labor income share and the wage share have been used interchangeably (Engelbert and Stockhammer 2012, Onaran and Galanis, 2012, ILO/OECD, 2015 Granados and Francese, 2015 and Lawrence 2015).  Figure (1) shows the labor income share and wage share for the US during the period 1940-2015, which appears to follow a similar time trend. This decline has had very serious economic consequences in a wide variety of countries. Atkinson (2009) explains that in countries that have experienced significant growth, improvements in household personal income have halted due to declining labor income shares. Piketty (2013) shows that higher capital shares (lower labor shares) are associated with higher inequality in the distribution of wealth over time and across many countries. Jacobson and Occhino (2012) demonstrate that declining labor income share has contributed to income inequality in the past and will likely worsen inequality in the future. Studies by the International Labor Organization (ILO) (2012) and Wolf (2014) demonstrate that downward trends in the labor share of income has had negative effects on macroeconomic aggregates, which in turn has led to a slowing of economic growth. Abdih and Danninger (2017) argue that the decline has been largest in countries that have experienced a high degree of technological change which has led to “routinizable” occupations[iii], steep declines in unionization, and a high level of competition from imports and foreign input usage.

Within the dominant literature two curves have emerged that defines the wage-unemployment relationship: the Phillips Curve (Phillips, 1958) and the Wage Curve (Blanchflower and Oswald, 1995, 2005). A.W. Phillips (1958) originally tested the relationship between the rate of change in the money wages and the unemployment rate, while Blanchflower and Oswald (1995, 2005) tested the relationship between the level of pay and the unemployment rate. Despite differences in underlying methodology, both curves show that there is a negative relationship between unemployment and the earnings of workers. Recent work by Shaikh (2013, 2016) describes a “classical” Wage-Share Curve, based in classical political economy, which estimates a negative relationship between the rate of change in the wage-share and an index of unemployment intensity on US data for the period 1948-2011.

The intention of this paper is to investigate the causes of the decline in the US wage share after 1970 by using the “classical” Wage-Share Curve. This curve will be estimated on US data from 1960-2015 and will include proxies for unemployment; unionization and the size of government, financialization, globalization and technological change to determine which of these variables are most responsible. The “classical” Wage-Share Curve is an empirical tool based on the wage share-unemployment relationship theorized in Marx (1867), formalized in Goodwin (1967) and developed further in Shaikh (2013, 2016). It is based on the hypothesis that there exists a negative relationship between the rates of change of the ratio of real wages to productivity (the wage share) and unemployment.

This paper is divided into four sections: Section (I) causes of the decline of the wage share, Section (II) methodological approach to studying the decline of the wage share, Section (III) results of estimating the rate of change of the wage share for the US 1960-2015; and Section (IV) conclusions and implications.

Causes of the Decline in the Wage Share: Declining Worker Strength, Financialization, Globalization and Technological Change

Several papers have offered explanations for the declining labor share of income observed in different countries. These papers (Engelbert and Stockhammer 2012, Onaran and Galanis, 2012 and ILO/OECD 2015), maintain that the decline of worker strength, globalization, financialization and technological change are the main factors causing the decline of the wage share.

(i) Decline of Worker Strength

In political economy as well as mainstream economics one explanation advanced for the decline of the wage share is the decline of worker bargaining strength. Marxian economic theory allows for the gradual degradation of the worker’s ability to negotiate wages leading to a decline in the wage share. Several studies in both political science and economics have attempted to explain the impact of declining bargaining power of workers on the distribution of income. These studies have either focused on the size of the welfare state or weakening of labor market institutions in particular unions. In the neoclassical literature it is argued that a higher bargaining power of workers will lead to increases in wages and therefore an increase in the wage share if the demand for labor is relatively inelastic. This perspective inspires much of the literature in political science and economics on welfare state retrenchment see (Pierson 1994, Korpi and Palmer 2003). Harrison (2002) and Jayadev (2007) use the share of government spending in GDP to track welfare state retrenchment and find a strong negative relationship. Recent empirical research has been more political economic in nature, focusing on the effect of labor market institutions on unemployment and the resulting changes in the wage share. These Studies of the labor income share conducted by the IMF (2007) and EC (2007) include union density, employment protection legislation, unemployment benefits and the tax wedge as institutional variables that affect the level and direction of wages. Both studies found that union density had a small positive effect on the distribution of income, while in the case of the IMF study union density and the tax wedge were the only variables found statistically significant. In this paper union density (UND) and the Government Share (GS) government expenditure as percentage of GDP, will be used to proxy worker strength. Fig. (2) shows the decline of union density in the US over the period 1960-2014, indicating a fall in the strength of workers.

(ii) Financialization

Since the mid-1970s the economies of industrialized countries have experienced a transformation that has resulted in an increase in financial activity and increased importance of financial institutions. Erturk (2008) and Stockhamer (2010) refer to these changes as financialization and maintain that it is characterized by increased household indebtedness, increased volatility in financial markets and short-termism of financial institutions. A number of papers have sort to discuss the impact of financialization on the functional distribution of income. Hein & Schoder (2011) and Omaran (2011) have shown that there have been sharp increases in dividend payouts and interest payments by non-financial institutions. Empirical work on the declining labor share of income (Rodrik, 1998 and Harrison, 2002) has used measures of capital controls and capital mobility, while an IMF (2007) study of the distribution of personal income used foreign direct investment to proxy financialization. Power (2003) calculates the capital gains on financial assets and demonstrates that this variable grew dramatically after 1994. An ILO (2008) study using total financial assets and foreign direct investment as a percentage of GDP, argues that financial globalization has resulted in the depression of the labor income share.

Financialization of the economy has had two effects that have weakened the bargaining position of labor. Firstly, firms have more options for investment activities, they can choose between increasing quantities and varieties of financial assets, as well as domestic and foreign assets. Secondly shareholders and management interests are now more closely aligned, which has increased the profitability of investment. These changes have reduced the firm’s reliance on physical assets and production as sources of income. When pressure by workers threatens profits, firms may now switch their investment activities into financial assets. Fig. (3) shows the assets of non-bank financial institutions to GDP (FIN) for the US 1960-2014 used in this paper to proxy financialization. As can be seen the variable has fluctuated over the period while displaying an upward trend, rising from just under 40% in 1960 to over 60% by 2008. The decline of this variable is indicative of the dot.com and housing bubbles that precipitated two major financial crises in the US economy.

(iii) Globalization

There are two general approaches to explaining the role that globalization plays in the decline of the wage share. Classical trade theory based on the Stopler-Samuelson theorem, maintains that the abundant factor will benefit from international trade. Therefore capital being the abundant factor benefits in advanced economies and experiences an increase in earnings in developed countries while labor’s earnings will increase in less advanced, developing countries. This conclusion is based on the assumptions of full employment and the immobility of capital and labor, which is not consistent with recent trends in capital mobility and both the importation of workers and exportation of jobs abroad (outsourcing of labor[iv]) that lead to unemployment (Stockhammer, 2013). These trends are consistent with the Political Economy approach to international trade that focuses on changes in the bargaining position of capital and labor as production becomes more globalized and the workforce more educated. Despite these limitations it is still used to argue that globalization will hurt workers in advanced economies and benefit workers in poor countries. Studies of the functional distribution of income (Goldberg and Pavcnik, 2007) have shown that the less skilled relatively abundant workforce are not better off due to globalization when compared to workers with greater levels of skills and education. Feenstra and Hanson (1997, 1999), show that while globalization will have an adverse effect on low skilled workers, skilled workers may in fact gain. Jobs exported to less advanced countries will negatively affect unskilled workers in advanced economies while having positive effects on skilled workers in less advanced economies.

As referenced above the political economy of trade approach maintains that globalization affects the distribution of income through its effect on the relative bargaining power of capitalists and workers. Rodrik (1997) uses a bargaining framework to argue that trade liberalization will benefit the mobile factor more; typically this is capital. Epstein and Burke (2001) show that due to threat effects income redistribution will take place without changes in the location of production. Empirical studies of the effect of globalization (Rodrik 1997, 1998, Harrison 2002, EC 2007 and IMF 2007) employ several measures of globalization including openness, terms of trade and measures of offshoring, immigration and capital account liberalization. These studies all find that globalization has had substantial effects on the functional distribution of income. For example, Rodrik (1998), Harrison (2002) and Jayadev (2007) find that increased trade has had a negative effect on the wage share. An IMF 2007 study concludes that globalization is one of several factors that acted to reduce the share of income accruing to labor in advanced economies. Fig. (4) shows OPEN (the ratio of exports plus imports to GDP) or trade openness for the US for the period 1948-2014. The data shows that in the post WWII period US trade openness (OPEN) was relatively stable up to the 1970s when it began to rise dramatically in the 1980s after the liberalization policies of Reagan-Bush Administrations began to take effect.

(iv)Technological Change

In both mainstream and political economy technological change is offered as a key factor contributing to the decline of worker earnings. In Marxian political economy for example, an increase in the organic composition capital and the rate of capital accumulation decreases labor force participation rates and increases the size of the reserve army of labor, causing the wage share to decline. At the core of the neoclassical theory of income distribution are the assumptions of perfect competition, full employment and stable production. Given these assumptions, the distribution of wages among skilled and unskilled workers is determined by skilled biased technological change. For example the increased use of Information and Communication Technology (ICT) leads to increased demand for skilled labor, which increases the wage of skilled workers while decreasing the wage for unskilled workers. A number of empirical studies on the US have been motivated by this theory, Autor, Katz and Krueger (1999), Card and Di Nardo (2002), for example argue that sharp increases in personal income inequality can be explained by skill-biased technological change[v].

Inequality in the functional distribution of income can also be explained by changes in technology, as technological change is viewed as being capital augmenting rather than labor augmenting. As argued above increased use of technology increases the demand for skilled workers while lowering the demand for unskilled workers. This results in higher wages for skilled workers and lower wages for unskilled workers. However, as unskilled workers acquire the in-demand skills the supply of skilled workers increase, applying downward pressure on the wage. These shifts cause capitalist firms to experience increased profitability (IMF 2007 and EC 2007).  The impact of technological change on the earnings of workers has been often studied by examining changes in the Capital Labor ratio (CLR), Information Communications Technology, Ellis and Smith (2007), Guscina (2006), Bentolila and Saint-Paul (2003), IMF (2007) and EC (2007). These studies all conclude that technological change has had the strongest negative impact on the wage share since the mid-1980s. In this paper several variables will proxy technological change including CLR, ICT and Capital intensity. In Fig. (5) CLR in the US is shown to have grown steadily over the period 1960-2015.

This paper investigates the decline of the wage share in the US over the period 1960-2015 by estimating a model of the rate of change of the wage share. Proxies for the strength of labor, growth of the financial sector, globalization and technological change are used in the model to discuss the causes of the decline of the wage share. Several conclusions are tested including the assertion that technological change has had the strongest impact on the decline of worker earnings.

Methodological Approach

The methodological approach used in this paper can be traced back to Marx (1867) discussion of the power struggle inherent in economic relationships. This paper argues that the declining share of productivity received by workers and the resulting worsening of income inequality result from the power struggle between workers and firms. It relies on Shaikh’s (2013, 2016) discussion of a “classical” Wage-Share Curve, which estimates the relationship between the rate of change of the wage share and an index of unemployment intensity.  In selecting an equation for the “classical” Wage-Share Curve, Shaikh begins with the Phillips Curve, stating that it focused on nominal wages, not prices. Phillips found that there was a negative relationship between the rate of change of money wages and unemployment. Friedman and Phelps later argued that the focus of worker struggle is for improvements in their real wage not money wages. So that, from their perspective a “Phillips-type” relationship examines the rate of change of real wages and unemployment. Following the classical traditions Goodwin (1967) demonstrates that in the struggle for improvements in the real wage the overall level of productivity must be considered. So that a “Phillips-type” relationship based on Goodwin’s discussion would examine the relationship between the rate of change of the ratio of real wages to productivity and unemployment.

Shaikh (2013, 2016) based his discussion of a “classical” Wage-Share Curve on Goodwin’s arguments. He uses an index of unemployment intensity that combines the unemployment rate and the average duration of unemployment because it captures the cumulative effect of the unemployment path. The “classical” Wage-Share Curve assumes that the rate of change in the wage share is a function of the unemployment intensity index, with the inflation rate and the growth rate of productivity as shift factors. This section will describe the derivation of the “classical” Wage-Share Curve, discuss the unemployment intensity index as an alternative measure of unemployment, and outlines it’s connections to classical political economy.

(i) The “classical” Wage-Share Curve

In the classical literature competition plays a principal role in the distribution of income. Adam Smith argued that competition would lead to a reduction of income inequality as increased demand for workers caused real wages to rise. Marx demonstrated that workers faced competition in the labor market from the presence of a reserve army of unemployed workers, which allowed their capitalist employers to reduce the value of their labor power in production. So that in the classical literature competition is viewed as a major determinant of the share of productivity received by workers in the form of earnings. Shaikh (2013, 2016) outlines a “classical” Wage-Share Curve as a basis for macroeconomic estimation of the wage-unemployment relationship, an empirical tool that demonstrates that the wage share is a negative function of unemployment.

This “classical” Wage-Share Curve is significant because it is based on the classical notion of competition within the labor market resulting in downward pressure on real wages relative to productivity. To see how this works if we let yr, wr, and ml represent the productivity of workers, real wages and profit per worker respectively. Then an equation representing the distribution of income from production is

yr = wr + ml (1) and,

ml= yr – wr (2)

the profit per worker equation implies a negative relationship between profit per worker and real wages. So that for a given level of productivity, to extract more profit from the production process the firm must cause the real wage of workers to decline. Equation (2) implies that profits and real wages are determined by a socio-historical struggle faced by workers and firms in the process of distributing productivity and changes in the relative strength of either party will result in either higher real wages or profits. In Marx this is referred to as vertical competition.

Using equation (1) the wage share (ws), can be written as

ws = wr/yr (3)

which implies that changes in the wage share are the direct result of real wages and productivity. For a given level of productivity the wage share increases as real wages increase, so that if higher real wages are the result of increases in worker strength, increases in the wage share must also be the result of worker strength. As the strength of workers within the labor market declines so will their real wages and wage share. If the amount of unemployment in the labor market determines the relative strength of workers then unemployment determines real wages and therefore the wage share. For example, in a “tight” labor market in which unemployment is relatively low workers are in a position to raise their real wage relative to productivity and therefore their wage share. When unemployment is high or the labor market is “loose” worker strength would be significantly weakened therefore real wages will decrease relative to productivity causing the wage share to fall. Therefore, the rate of change of the wage share can be viewed as a negative function of worker strength and by extension unemployment (the “classical” Wage-Share Curve). This is referred to as horizontal competition, as downward pressure on wage share is the result of competition faced by workers in a “loose” labor market.

Shaikh (2016) defines the distribution of productivity over time as t (4)

where  is productivity,  is real wages and  is profits at time t. From equation (4) an equation for real wages at time t can be derived

wrt= yrt – mlt (5),

which implies that productivity is the upper limit to the real wage. He defines the historical link between real wages and productivity as

??? = ?? ∙??? (6)

where ?? is the social-historical level of labor strength. Furthermore the rate of change of the wage share

??= ??̇? + ???̇??? (7)

is dependent on changes in labor strength and productivity where the rate of change labor strength can be expressed as

??̇?? = ? (??∗ −??),?́>0 (8),

??∗ and ?? are the critical and actual level of unemployment respectively.

Equation (8) demonstrates that changes in worker strength depend on whether the labor market is “tight” or “loose”, or the relative size of and . For example a “tight” labor market is one for which  which will result in an increase in . Equations (7) and (8) together imply that the rate of change of the wage share is a negative function of unemployment relative to some critical level of unemployment

(ii) The Unemployment Intensity Index (UI)

This section describes the index of unemployment intensity as an alternative measure of unemployment. The unemployment intensity index assumes that worker earnings will be highly sensitive to both the unemployment rate and the average duration of unemployment. The amount of people unemployed as well as the length of time that these people remain without employment affects real wages and productivity, and therefore the wage share. This index was first used by Shaikh in his 2013 paper entitled “Wages, Unemployment and Social Structures: A New Phillips Curve”, in which he estimates the relationship between the wage share and unemployment.

Howell (2010) argues that the present indicators of the underutilization of labor used by the Bureau of Labor Statistics (BLS) are inadequate as they fail to account for millions of displaced workers. In an attempt to address this issue the BLS has developed an alternative more inclusive measure of unemployment (U6), which is calculated as the sum of those counted as unemployed, marginally attached workers and employed part time for economic reason as a percentage of the civilian labor force and marginally attached workers. However, data on U6 is available only for years 1994-2017, and so the unemployment intensity index is used as an alternative index that extends back to the post war period. To demonstrate the strong relationship between UI and U6 both indices are graphed in Fig. (6).  US data for the period 1948-2017 was collected from the Bureau of Labor Statistics (BLS) website[vi]. The unemployment intensity index (UI) was calculated using data on the annual unemployment rate (UR) and average annual duration of unemployment (UD) from BLS tables LNS14000000Q and LNS13008275 respectively, while U6 is collected from BLS tables LNS13327709. As can be seen in the chart both variables exhibit a similar time trend and display a cyclical movement characteristic of unemployment. The correlation of both variables over the period 1994-2016 indicates that UI is a good proxy for U6.

(iii) Political Economy and the “classical” Wage-Share Curve

Marx maintained that the dynamics of capitalist accumulation results in a pool of involuntarily unemployed workers that keeps wages low by maintaining competition for available jobs. In response workers must organize forming labor governments and unions to advocate on their behalf and negotiate for higher wages. Shaikh (2013) demonstrates that in the classical tradition real wages depends on the strength of labor relative to capital as well as the general level of development. The basis of the “classical” Wage-Share Curve is derived from Marx’s argument that capitalism creates and maintains a persistent pool of involuntarily, unemployed workers (Shaikh, 2013, pg.13).

Marx took a dynamic approach[vii] to the analysis of the capitalist mode of production, and it is through this approach that he uncovered the regulating forces of the aggregate labor market; the maintenance of a reserve army of labor and the continual accumulation of capital. Capital accumulation leads to the creation of a reserve army of labor as changes in the rate of accumulation, the organic composition of capital and the labor force participation rate leave workers unemployed. The rate of capital accumulation is the speed at which the capitalist increases in size through reinvestment. This rate determines; the demand for labor, the reserve army of unemployed workers and the wage share. For example, a fall in the rate of accumulation will reduce the demand for labor and will increase the reserve army of workers, which will apply downward pressure on the wage share.

The organic composition of capital influences the wage share by causing changes in the demand and supply of labor. Mechanization of the production process increases the organic composition of capital, which leads to a reduction in the demand for unskilled workers and an increase in the demand for skilled workers, which increases the reserve army of labor until these workers can become retrained. Consequently, as the demand for skilled workers increases, the wage share of these workers will temporarily increase through increases in their real wage. However, as workers in the reserve army acquire the skills that are in demand by employers and re-enter the market, the real wage and the wage share will reverse and begin to decline. Therefore, increases in the wage rate that is accompanied by increases in both the rate of accumulation and the organic composition of capital is mitigated by dynamic adjustments in the labor market overtime, as well as the capitalist need to maintain profitability. When wages increase to the point where it threatens profitability, capitalists will slow the rate of accumulation and the organic composition to protect their profitability.

Marx argued that capitalist accumulation would have a profound effect on labor force participation rates. As the growth of capital and monopolization of the means of production forced more sectors of the population into the labor market, the commodification of the domestic economy[viii], mobilization of the working class and state imposed limits on the use of child labor and the length of the work day, as well as the continued encroachment of large-scale enterprise on petty commodity production would lead to increases in labor force participation. Consequently real wages would be depressed and forced below the value of labor power causing the wage share to fall.

So that the growth and maintenance of a reserve army of unemployed workers both creates competition among workers and is used as an instrument in the competitive struggle between workers and their capitalist employers. A growing reserve army of unemployed workers weakens the bargaining position workers and leads to a decline of the wage share.

Estimating the “classical” Wage-Share Curve

This section describes the model of the rate of change of the wage share used in this first chapter to estimate the “classical” Wage-Share Curve. It presents the results of three (3) experiments performed by estimating the model on US data for the period 1960-2015. This section is divided into two subsections, which provide a description of the model, and a discussion of the results of its estimation.

(i) The Model

The basis of the model of the rate of change in the wage share is the assumption that the ratio of real wages to productivity (the wage share) is a negative function of unemployment. The model includes variables from within the literature that have been determined to be the cause of the decline of the wage share. Such variables include: the degree of unionization, the size of government, technological change, globalization and growth in the financial sector (financialization). To estimate the model, an unemployment intensity index[ix] (UI) is calculated for the US and used as a proxy for unemployment following recent work by Shaikh (2016). Union density (UD) and the Government Share[x] (GS) are used to proxy the degree of unionization and the size of government respectively. The ratio of the Assets of non-bank financial institutions to GDP (FIN) will be used to trace changes in the financial sector (financialization), while Openness (OPEN) the ratio of the sum of imports and exports to GDP, Import Share (IS) the ratio of imports to GDP and the KOF index[xi] are used to proxy globalization. Three variables: the Capital Labor ratio (CLR), Information Communication Technology (ICT) and Capital Intensity (CapInt) will proxy technological change. Data for all variables were extracted from the Federal Reserve of St. Louis, Federal Reserve Economic Data (FRED) website[xii]and the European Commission Economic and Financial Affairs AMECO data set[xiii].

The model will first be used to provide evidence in support of the claim that due to Neo-Liberalism[xiv] the wage share has become more sensitive to changes in unemployment (experiment 1), as well as to identify the main drivers of changes in the wage share (experiment 2). In studies by IMF (2017) and Adih & Danninger (2017) technological change has been identified as the main cause of the decline of the wage share. Other studies have listed the declining the size of government and unionization; globalization and financialization as main drivers of changes in the wage share Jayadev (2007), Erturk (2008), Stockhammer (2009) and Levoie & Stockhammer (2012). Recent studies by Stockhammer (2015) Guschanski & Onaran (2017), Doan & Wan (2017) and Young and Tackett (2017) have found that globalization is the main driving force behind the decline of the wage share in advanced and emerging economies. A similar report prepared by the G20 Working Group (2015) demonstrated that decline of the wage share is explained primarily by the decline in labor market institutions, financialization and globalization with technological change having a marginal effect. Rapid growth of the financial sector in the US and increased openness of the economy may have interacted to have a negative impact on the wage share so that a third experiment (experiment 3) will be performed to determine whether financial-globalization has a significant impact on the wage share. Experiment (3) will use interaction terms created from the variables representing globalization and financialization.

 

(ii) Estimating the model

As is customary when modeling time series data tests of stationarity (Dickey-Fuller Unit Root test) and co-integration (Johansen Co-integration test) are performed and the variables are determined by a mixture of I(1) and I(0) processes and are co-integrated. An Autoregressive Distributed Lagged model (ARDL) will be used to estimate the rate of change of the wage share. This estimation method includes lagged versions of the dependent and independent variables as explanatory variables in the equation. An ARDL model is used when the deterministic trend cannot be removed from all of the variables included in the model, but the included variables are co-integrated. Before estimating the model all variables are HP-filtered with a default parameter of 100. After estimating the model a Bound’s test is performed to confirm that a long-run co-integrating relationship exists between the rate of change in the wage share and the explanatory variables.  The model is estimated using a difference-log equation, in which signs of the estimated coefficients indicate whether each variable quickens or slows the change of the wage share, while the absolute size of each coefficient indicates the relative impact of each variable on the wage share.

Experiment (1)

Shaikh (2016) demonstrates a break in the “classical” Wage-Share Curve as the US economy transitions from the Golden age of unionism to the Neo-liberal era of the 1980s.  Fig. (7) reproduced from Shaikh’s book graphs the rate of change of the wage share in relation to unemployment intensity for the period 1949-2012. Shaikh uses this figure to discuss the political economic changes that occurred during this period. The nine (9) year period 1984-1993 signifies a shift downward of the “classical” Wage-Share Curve, where the curve flattens, indicating that the wage share became more sensitive to changes in unemployment after this period. This Shaikh identifies as the end of the Golden Era of unionism and the beginning of the Neo-liberal period.

Experiment (1) estimates the model of the rate of change of the wage share to test the hypothesis that the “classical” Wage-Share Curve flattened. To perform this experiment the data is split into two periods 1960-1980 and 1981-2015, and the estimated coefficient of unemployment evaluated for changes in the size. Table (1) shows the results of estimating the model on the two periods. As can be seen a doubling of unemployment intensity reduces the growth rate of the wage share by 2.6% in 1960-1980 and 3.6% in 1981-2016. These estimates indicate that the wage share became more sensitive to changes in unemployment. The Neo-Liberal Era is associated with an erosion of the bargaining position of labor. Evidence of the decline in worker strength can be seen in the declines in the coefficients of union density and the government share.

The remaining three (3) variables provide some additional explanation for the decline of the wage share. The coefficient of Assets of Non-Financial Institutions to GDP became negative, indicating that growth in the financial sector began to slow the change of the wage share. Globalization had a negative impact on the wage share in both periods, but in the Neo-liberal period the negative impact became stronger as the coefficient of the Import Share became larger in absolute terms. The coefficient of the capital-labor ratio decreased between the periods, indicating a reduction of the impact of technological change on the wage share. This first experiment provides evidence and some explanation for the flattening of the “classical” Wage-Share Curve in Fig. (7), experiments (2) and (3) will expand the discussion to identify the variables with the greatest impact and to determine if financial- globalization has a significant impact on changes in the wage share.

Experiment (2)

Table (2) shows the result of estimating the model using the Capital Labor ratio as a proxy for technological change, and Open, the Import Share and the KOF index as proxies for globalization. Coefficients estimates are at the top of each row followed by standard errors in parentheses and p-values to assess statistical significance. For all three equations the coefficient of unemployment intensity confirms that the wage share is a negative function of unemployment. Based on these estimates the rate of change of the wage slows by approximately 2% in response to a doubling of unemployment intensity. The results of all three equations indicate that the variables with the strongest impact on the change of wage share are technological change, unionization, size of government and globalization, with financialization having a small significant impact and technological change being the most influential. So that these results confirm the conclusions of the earlier mentioned studies that technological change is the variable primarily driving the decline of the wage share at least in the US.

To test this conclusion the Capital Labor Ratio is replaced by the Information Communications Technology (ICT) and the Capital Intensity index, and the results presented in Tables (3) and (4) respectively. In table (3) equations (4) through (6) confirm that a negative long run relationship exists between the wage share and unemployment, as indicated by the coefficient of unemployment intensity. A doubling of unemployment intensity decreases the growth rate of the wage share by approximately 3% in all three equations. Using ICT as the proxy for technological change leads to different results regarding the impact of technological change. As can be seen from these results technological change is no longer the variable with the largest coefficient and therefore the most influential. In these equations unionization and the size of government have relatively larger coefficients. The impact of financialization is once again small but significant.

In table (4) equations (7) through (9) list the results of estimating the model with Capital Intensity as a proxy for technological change. The coefficients on unemployment intensity confirm the negative long run relationship between unemployment and the wage share. In these equations technological change is once again the variable with the strongest impact on the wage share, followed by the size of government, union density and globalization, with financialization having a small significant effect on wage share.

The results reported in tables 2 through 4 provide some evidence in support of the hypothesis that technological change is the variable most responsible for the decline of the wage share. As with the exception of ICT, when the Capital labor ratio and Capital intensity are used as proxies for technological change the coefficients are highly significant and relatively large. Union density, the size of government and globalization also appear to be relatively influential in determining the rate of growth of the wage share, while financialization was shown to have a relatively small but significant impact. The results of experiment (2) indicate that the conclusion that technological change is the variable with the most influence on the wage share is sensitive to the choice of proxy.

Experiment (3)

Experiment (3) tests the hypothesis that financialization and globalization may have interacted to cause the wage share to decline. To test this hypothesis each equation is estimated with an interaction term. Table (5) summarizes the results of estimating the model with the interaction term. In Equation (1) financial-globalization is proxy using Open and the Assets of Non-Bank Financial institutions, while equation (2) uses the Import Share and Assets of Non-Bank Financial institutions and equation (3) combines the KOF index with Assets of Non-bank Financial institutions.

All three equations indicate a relatively consistent long run negative relationship between unemployment and the wage share. The financial globalization variable displays a small but significant influence over the wage share in equation (3) but remains small and insignificant in equations (1) and (2). In equations (1) and (2) the interaction variable was constructed using the standard proxies for globalization; OPENNESS and the Import Share. The KOF index was used to construct the interaction term in equation (3), which is composed of many other variables that may interact more strongly with financialization to influence the wage share. This may explain why only equation (3) generates a small significant coefficient.  The results in table (5) provide little evidence that financial-globalization influenced the wage share over the entire period but perhaps by studying the Neo-liberal Eras more closely more information maybe revealed.

Palley (2007) identifies the period 1979-2005 as the Financialization Era within the United States. He writes that this period was characterized by a growth in importance of the financial sector to the U.S. economy. During this time the contributions of the finance, insurance and real estate sectors to U.S. GDP rose from 15.2 to 20.4 percent, while employment as a share of total private sector employment rose from 6.6 to 7.3 percent. Furthermore he identifies the 1980s as the period of rapid growth in the financial sector (Palley 2007 pg.8). So that perhaps by focusing on the Neo-Liberal period more information can be obtained for the impact of increases in the financial sector can be obtained.

In table (6), Equations 1-3 show the results of estimating the model of the rate of change of the wage share on US data for the period 1983-2015. In equations 1, 2 and 3 the interaction term representing financial-globalization was IS-Fin, Open-Fin and Kof-Fin respectively. The data shows statistically significant estimates for the impact of financial-globalization on the change of the wage share. During the period 1983-2015, a 100% increased in the financial sector of the economy slowed the change in the wage share by between 3.4 and 4.7 percent depending on the index of financial-globalization. So that while over the entire period estimates of the impact of financial-globalization was statistically insignificant, a closer examination of the financialization era does generate statistically significant estimate.

Taken together these three (3) experiments provide evidence that the “classical” Wage-Share Curve can be used to identify the main drivers of the decline of the wage share. At the very least these experiments confirm that the most influential determinants of the wage share are technological change, the size of government and the degree of unionization.

Conclusions and Implications

This paper investigates the decline of the wage share in the US. It builds on previous work completed by Shaikh (2013, 2016) and references many of the recent empirical studies in this line of investigation. This study contributes to the discussion of the decline of the wage share by attempting to identify the variables with the strongest impact. It tests previous conclusions that technological change is the single most important determinant of the decline of the wage share and provides support for the argument that the impact of technological change is sensitive to the variable used as its proxy. In this regard three variables are used; Capital Labor ratio, Information Communication Technology and Capital Intensity. It was shown that the estimated coefficients were largest when the Capital Labor ratio and Capital Intensity were used but when ICT was used the impact of technological change was shown to be small and insignificant.

Building on previous empirical work, some evidence was obtained in support of the argument that the size of government and degree of unionization are significant determinants of worker earnings. In all versions of the model estimated the coefficients of these variables were relatively large and highly statistically significant. These findings lend support to the political economic and institutional theories of wages that maintain the importance of worker strength in determining worker earnings. These findings in particular highlight the need for strong and stable labor-governments to defend and maintain the rights of the working class through public policy and advocacy and the establishment and promotion of institutions that promote the well being of workers. Quite surprisingly growth in the financial sector and globalization trends for the US economy did not emerge as particularly relevant determinants of the decline of the wage share. Despite the potential for increased unemployment and downward pressure on worker earnings that is associated with structural adjustments to the economy, competition from increased trade, migration of capital and “off-shoring” of labor. In this model the coefficient of the variable representing financialization was relatively small, which maybe the result of a variable selection bias or interactions within the model. It was determined that financialization and globalization variables were highly correlated and therefore an interaction term representing financial-globalization was introduced. This variable was shown to be relatively insignificant over the period 1960-2015 but became significant when the period of study was narrowed to 1983-2015, which includes the time period identified as the financialization era.

Three measures of globalization were used in this model; Open, the Import share and the KOF index of globalization. In the case of Open and the Import share it was theorized that the impact of globalization would be reflected in changes in the exports and imports relative to GDP. These variables are expected to reflect the impact of volume and international price competition on the wage share of workers. While the KOF globalization index is composed of twenty-three economic, social and political factors that reflect the pace of globalization. Though the globalization variables impact were many times larger than financialization, the coefficients were relatively small as compared to the size of government, union density and technological change. The only exception was in the KOF index, as when this variable was used as the proxy, its coefficient was relatively large and significant.

Three experiments are conducted; firstly that the wage share-unemployment relationship changed in the Neo-liberal era, secondly that technological change is the most influential variable in the decline of the wage share and, finally that financial globalization is an important determinant of the decline of the wage share. A model of the rate of change of the wage share was constructed so that in addition to the unemployment intensity index, included: proxies for the size of government and union density, as well as growth of the financial sector, globalization and technological change. These experiments lead to three conclusions: firstly that the influence of unemployment on the wage share increased as the US economy transitioned into the Neo-liberal Era, secondly that technological change is the most influential determinant of the wage share but this conclusion is sensitive to the proxy used, and thirdly that financial-globalization has a small significant impact on the wage share. These results are taken as further evidence of the existence of the “classical Wage-Share Curve.

These findings imply that in the case of the US, a reversal of the decline of the wage share may be achieved through more progressive governance. Policies and programs that protect and promote the rights of workers may lead to increases in the wage share. Such policies include; a more progressive minimum wage program implemented at the federal level, employment protection programs, and Public and Private sector partnerships in providing retraining for workers displaced by technological change, globalization and growth in the financial sectors.

References

Abdih Y. and Danninger S. (2017), “What Explains the Decline of the U.S. Labor Share of Income? An Analysis of State and Industry Level Data” IMF Working Paper, July 2017

 

Autor D., Katz L. and Krueger A. B. (1999), “Computing Inequality: Have Computers Changed the Labor Market?” Quarterly Journal of Economics 113 1169–1214

Atkinson A. (2009), “Factor shares: The principal problem of political economy?” in Oxford Review of Economic Policy, Vol. 25, No. 1, pp. 3–16.

Bentolila S. and Saint-Paul G. (2003), “Explaining movements in the labor share Contribution to Macroeconomics” Vol. 3, Issue 1 (Berkeley, California Berkeley Press).

 

Blanchflower D. and Oswald A. (1995), The Wage Curve, Cambridge, Massachusetts, MIT Press

 

Blanchflower D. and Oswald A. (2005), “The Wage Curve Reloaded”, IZA Discussion Paper Series No. 1665

 

Botwinick H. (1993) “Persistent Inequality: Wage Disparity Under Capitalist Competition”

Card D. and DiNardo J. E. (2002), “Skill-Based Technological Change And Rising Wage Inequality: Some Problems And Puzzles,” Journal of Labor Economics, 20 (4), 733-783

 

Doan H. and Wan G. (2017) “Globalization and the Labor Share in National Income”, ADBI Working Paper 639

Ellis L. and Smith K. (2007), “The global upward trend in the profit share”, BIS Working Paper No. 231

European Commission (2007), “The labour income share in the European Union”, Chapter 5 of: Employment in Europe

 

Feenstra R. C. and Hanson G. H. (1997), “Foreign direct investment and relative wages: evidence from Mexico’s maquiladoras”, Journal of International Economics 42 (3–4), pp. 371–93.

 

Feenstra R. C. and Hanson G. H. (1999), “The impact of outsourcing and high-technology capital on wages: estimates for the United States, 1979-1990”, Quarterly Journal of Economics, 114 (3),pp. 907-940

 

Goodwin R.M. (1967), A Growth Cycle, in Socialism, Capitalism and Economic Growth, ed. C.H. Feinstein, Cambridge University Press, London

 

Goldberg P. and Pavcnik N. (2007) “Distributional Effects of Globalization in Developing Countries”, Journalof Economic Literature, 45(1), 39-82

Guschanski A. and Onaran O. (2017) “The political economy of income distribution: industry level evidence from 14 OECD countries” No 17518, Greenwich Papers in Political Economy, University of Greenwich, Greenwich Political Economy Research Centre

Guscina, A. (2006), “Effects of globalization on labor’s share in national income”, IMF Working Paper 06/294

Harrison A. (2002) “Has globalization eroded labor’s share? Some cross-country evidence”, Mimeo, UC Berkeley.

Howell D. (2010) “Undercounting the Underemployed: How Official Indicators Have Missed Millions of Underutilized Workers”, Schwartz Center for Economic Policy Analysis Series.

ILO (2011), “The labour share of income: determinants and potential contribution to exiting the financial crisis”, Chapter 3 of World of Work Report 2011: Making Markets Work for Jobs. Geneva: ILO

ILO (2012), “Global Wage Report 2012/13: Wages and equitable growth” (Geneva)

IMF (2017) Annual Report, “How to Jump Start Growth”

IMF (2007a), “The globalization of labor”, Chapter 5 of World Economics Outlook, April 2007 Washington: IMF

Jacobson M. and Occhino F. (2012), “ Labor’s Declining Share of Income and Rising Inequality”, Economic Commentary No. 2012-13, September 2012

Jayadev A. (2007) “Capital account openness and the labour share of inomce. Cambridge Journal of Economics 31, 423-443

Lavoie M., Stockhammer E. (2012), “Wage-led Growth: Concept, Theories and Policies”’ Conditions of Work and Employment Series No.41, ILO Geneva

Marx K. (1867) Capital: A Critique of Political Economy, Volume I, Ch. XXV

Palley T. (2007) “Financialization: What It Is and Why It Matters”, Working Papers wp153, Political Economy Research Institute, University of Massachusetts at Amherst

Piketty T. (2013), Le Capital  au XXIe siècle (Paris: Seuil)

Phillips A. W. (1958), “The Relation between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom, 1861-1957”, Economica, New Series, Vol. 25, No. 100, 283-299

 

Power, D, Epstein, G, Abrena, M, (2003) “Trends in the rentier income share in OECD countries 1960-2000”, PERI Working Paper 58a

 

Rodrik D. (1997), Has Globalization Gone Too Far? Washington: Institute of International Economics

 

Rodrik D. (1998), “Capital mobility and labor”, Manuscript, http://ksghome.harvard.edu/drodrik/capitalism.pdf

 

Shaikh A. (2013), “WAGES, UNEMPLOYMENT AND SOCIAL STRUCTURE: A NEW PHILLIPS CURVE” G. & L. E. R. Volume 17 No.2

Shaikh A. (2016), Capitalism: Competition, Conflict, Crisis, New York, New York, Oxford University Press

Stockhammer E. (2009), “Determinants of functional income distribution in OECD countries”, IMK Studies 05-2009

Stockhammer E. (2013) “ Why have wage shares fallen? A panel analysis of the determinants of functional income distribution” ILO Conditions of Work and Employment Series No. 35

Stockhammer E. (2015) “Rising inequality as a cause of the present crisis”, Cambridge Journal of Economics, Vol. 39 Issue 3 Pg. 935-958

Wolf M. (2014), The shifts and the shocks: What we’ve learned – and have still to learn – from the financial crisis (London, Penguin).

Young A. and Tackett M. (2017) “Globalization and the Decline of the Labor Shares: Exploring the relationship between Trade and Financial Flows, European Journal of Political Economy

[i] The functional distribution of income is the division of national income into the income paid to capital and labor. The labor share is the share of national income paid to workers while the capital share is the share of national income paid to capital.

[ii] Harrison 2002, Guscina 2006, Jaydev 2007, EC 2007, Ellis and Smith 2007, ILO/IILS 2011, 2012, Stockhammer 2009, 2013

[iii] A kind of deskilling of labor by reducing the effort required to perform tasks, thus reducing the value of labor power in the production process.

[iv] Outsourcing of labor occurs when firms use subcontractors to carry out specific tasks rather than hiring internal employees. Many US companies have used subcontractors in developing countries due to lower labor costs, a process commonly referred to as “off-shoring”.

[v] Skill-Biased Technical Change is a shift in the production technology that favors skilled over unskilled labor by increasing its relative productivity and, therefore, its relative demand.

[vi] www.bls.com

[vii] Orthodox theory takes a static approach to the analysis of markets i.e. markets move towards a single balancing point, whereas, with dynamic analysis equilibrium is defined by a natural tendency.

[viii] Commodification of the Domestic Economy is a term used in Labor and Gender economic theory to refer to the provision of goods that reduce the time and effort to complete household chores.

[ix] The product of the unemployment rate and unemployment duration

[x] The ratio of government expenditure to GDP

[xi] The KOF index is an index of globalization based on 23 variables in three dimensions; economic, political and social globalization. Source the KOF Swiss Economic Institute website www.kof.ethz.ch

[xii] fred.stlouisfed.org/

[xiii] ec.europa.eu/economy_finance/ameco

[xiv] The ideology and policy model that emphasizes the value of the free market and competition. The Neo-Liberal Era in the US is believed to have begun in the 1980s.