Income inequality,managers’ compensation and the falling rate of profit: Reconciling the US evidence

Abstract

Evidence based on US government statistics indicates that the rate of profit fell and employees’ share of output was trendless between 1970 and the Great Recession. This paper defends that evidence by showing that it is compatible with phenomena that may initially seem to contradict it; that is, rising income inequality and rising managerial compensation. While some other studies claim that managers’ compensation and the ‘class’ rate of profit skyrocketed, this paper shows that these studies seriously overestimate managers’ pay, and the share of their pay that can be construed as capital income (disguised dividends) rather than as genuine labour compensation.

Keywords

Falling rate of profit income distribution income inequality managerial pay wishful thinking

Introduction

Kliman (2012) argues that US corporations’ rate of profit fell substantially between the mid-1950s and the early 1980s, and that it failed to recover in a sustained manner thereafter. Furthermore, he argues that this persistent fall in the rate of profit was a key factor that helped set the stage for the Great Recession and the malaise that continues long after its official end. It led, directly, to a long-term decline in the growth of productive investment, and, indirectly, to sluggish growth of output and income as well as rising debt burdens.

Four measures of the rate of profit are shown in Figure 1. Each expresses corporations’ profits as a percentage of their accumulated investment in fixed assets, net of depreciation measured at historical cost. None of the four rates rebounded in a sustained manner from the recessions of the mid-1970s and early 1980s. The rates of profit that employ the more inclusive measures of profit – property income and net operating surplus – continued to trend downward sharply during the quarter-century of ‘neoliberalism’ and ‘financialisation’ that preceded the Great Recession. During the same period, the rates that employ more restrictive definitions of profit – before- and after-tax profits – trended downward to a smaller extent.¹

Figure 1.

Rates of profit of US corporations, 1947-2007.

This evidence has seemed implausible to some people who are familiar – but not familiar enough – with data pertaining to the performance of US capitalism during the period of ‘neoliberalism.’ For example, sociology student and author Richard Seymour has alleged that the arguments in the book cited above:

often depend on statistical shenanigans …. For example, iirc [if I recall correctly], the figures he gives for U.S. profitability in his book on the crisis specifically exclude profits earned from overseas transactions. … those profit figures can’t be used to claim that U.S. capitalism was weak, in permanent crisis, etc., since 1973.²

The evidence that the rate of profit fell has also been dismissed because it is based partly on data (which will be discussed in detail below) that implies that the share of net output that workers received was stable. In light of the rise in US income inequality, it seems obvious to some people that workers’ share must have declined. For example, in a review of the same book, Wood (2013) writes,

We also learn that wages have been replaced by non-wage benefits such as healthcare, education and perhaps credit cards. Kliman maintains this in the face of even mainstream economists’ acceptance of the growth of inequality, the language of the 99 per cent, and the decline in things like trade union membership, labour rights, the growth of underemployment and casualization. [The book makes] the extremely dubious claim that American workers had, until the crash of 2007, never had it so good [… but] it is not just the ‘conventional left’ who are aware of labour’s worsening position relative to capital in the past thirty years.

The purpose of this paper is to defend the falling-rate-of-profit evidence, and the associated evidence that workers’ share of net output was stable, against such dismissals. It defends the evidence by showing that it can be reconciled with other evidence – pertaining to rising income inequality and an apparent increase in managerial compensation relative to average compensation – with which it may at first seem incompatible. It is important to defend the evidence, I believe, because due regard for it is crucial if we wish to understand the causes of the Great Recession and the continuing economic malaise, and if we wish to properly evaluate whether ‘neoliberalism’ was an economic success in its own terms.

The facts that will be presented in this paper raise many interesting and important theoretical questions that the paper does not attempt to answer. It is necessary, first, to be clear about the facts. The sole aim of the paper is to establish what the facts actually are, in order to defend the falling-profitability and stable-compensation-share evidence against unwarranted dismissals. To provide a theoretically satisfactory explanation of the facts would be impracticable here – it would require at least a whole additional paper – and it is unnecessary in order to defend the evidence that the paper aims to defend. It is sufficient to show that this evidence is compatible with other evidence that may seem to contradict it.

The next section of the paper focuses on the belief that rising inequality and stagnant ‘wages’ must have caused workers’ share of output to fall. It shows that this belief is based on several misunderstandings of the US income-inequality data. Specifically, it shows that neither the rise in interest payments and dividends as shares of personal income, nor the widely-reported ‘fact’ that productivity growth far outstripped the growth of workers’ pay, imply that profits rose at the expense of employee compensation. It also reminds the reader that inequality can increase because of rising inequality among workers, and documents instances in which this occurred. Finally, it reviews recent evidence suggesting that the degree to which income inequality increased has been greatly exaggerated, through the use of income measures that exclude in-kind income and that fail to control for the effects of taxation and changes in household size.

The remainder of the paper considers a different challenge to the evidence that the rate of profit fell, and that workers’ share of output was stable. Recent studies (Magdoff & Foster 2013; Mohun 2013, 2014) claim that the compensation received by US managers and supervisors skyrocketed, and that this caused compensation paid to ‘the working class’ (non-management/supervisory employees) to fall dramatically as a share of output. In addition, by treating the compensation of management/supervisory employees as capital income (disguised dividends) rather than as labour income and a cost to business, Mohun (2013) computes an alternative ‘class’ rate of profit based on this redefinition. In contrast to the rates of profit depicted in Figure 1, that paper’s redefined ‘rate of profit’ rose markedly during the quarter-century preceding the Great Recession.

However, I argue that these studies seriously overestimate both the share of compensation that management/supervisory employees receive and the share of their compensation that can legitimately be treated as capital income, rather than as labour income. I also argue that even in 2007, after several decades during which the capital-income component of compensation supposedly rose dramatically, it remained quite small. Hence, any such rise in the profit component of compensation must also have been quite small – too small to have had a sizable effect on the ‘true’ compensation share of output or the rate of profit.

In a recent paper that exposed egregious and fatal errors in a widely-cited body of research in the field of positive psychology, Brown, Sokal and Friedman (2013: 813) concluded with the ‘hope that other scholars will be encouraged to question, in public fora such as this one, other research that seems to require correction’. The present paper is written in that spirit. I hasten to add that it criticises research, not authors as persons. In this respect, too, Brown, Sokal and Friedman’s (2013: 813) clarification of the motivation behind their paper aptly expresses the motivation behind the present one:

Let us stress that our concern here is with the objective properties of published texts, not the subjective states of mind of the authors (which might, however, be of interest to philosophers, such as Frankfurt, 2005). We do not, for example, have an opinion about the degree to which excessive enthusiasm, sincere self-deception, or other motivations may have influenced Losada and colleagues when writing their articles. Our only interest here is to bring the fundamental errors in this widely-cited body of work to the attention of the scientific community, before its considerable influence can do any further damage to the cause of science in general and academic psychology in particular.

Rising income inequality and the stable compensation share of output

The evidence that income inequality has increased in the US is much better known than understood. Below, I discuss four of the most important misunderstandings, and explain why there is no contradiction between rising inequality and the fact that the compensation share of net output did not decline between 1970 and the Great Recession.

Distribution of profit to individuals

During the period in question, incomes of property owners rose as a percentage of total income received by people (personal income). Nonetheless, as Figure 2 shows, compensation of employees – wages and salaries plus nonwage compensation³ – was basically trendless, as a share of US corporations’ output (net value added).⁴ Figure 3 shows that the compensation share was also basically trendless in the business sector as a whole. This means that income was not redistributed from compensation of employees to profit.⁵

Figure 2.

Shares of US corporations’ net value added, 1947-2007.

Figure 3.

Compensation share of net value added, corporate and total business sectors, USA 1970-2007.

Thus, there is no contradiction between the fact that property owners’ share of income increased and the fact that workers’ share of output was stable. The two facts are compatible because property owners’ share of income can increase without raising profit, by reducing companies’ retained profits instead. For example, if corporations distribute a bigger share of profit as dividends to their shareholders, total profit is unchanged, the shareholders’ incomes rise, and retained profits fall. Similarly, if the share of profit that is used to pay interest to bondholders and other creditors rises, this boosts their income at the expense of retained profit without affecting total profit.

This is in fact what took place, as Figure 4 shows. As a share of US corporate profits (net operating surplus), interest payments spiked sharply at the start of the 1980s and remained very high for about a decade.⁶ Shortly before they subsided, dividend payments began to rise, and continued to rise during the next two decades. As a result, the average percentage of profit that was used to pay interest and dividends rose from 21 per cent between 1947 and 1968, to 30 per cent between 1969 and 1979, and to 47 per cent between 1980 and 2007. This rise came at the expense of the portion of profit that was retained by the corporations themselves, not at the expense of employee compensation.

Figure 4.

Shares of US corporations’ profit (net operating surplus), 1947-2007.

In principle, increases in capital gains can also lead to increased inequality without affecting profit. The US national accounts do not count companies’ income from capital gains as profit, and when assets are owned by individuals, the capital gains they receive from increases in the assets’ prices are not profits. However, Armour, Burkhauser and Larrimore (2013) have recently found that inclusion of accrued capital gains as income reduces measured inequality in the US between 1989 and 2007 (comparable data were not available for earlier years). This study will be discussed further below.

Lagging growth of ‘pay’ relative to ‘productivity’?

The stability of the compensation share implies that total output and total compensation increased at basically the same rate. It follows that output per labour-hour and compensation per labour-hour also increased at basically the same rate. And since, in studies that discuss the relationship between these variables, ‘productivity’ refers to output per labour-hour while ‘compensation’ refers to compensation per labour-hour, it also follows that productivity and hourly compensation increased at basically the same rate: the latter did not lag further and further behind the former. This conclusion necessarily holds true when the variables are expressed in nominal terms. It also holds true when they are expressed in real (inflation-adjusted) terms – provided that the same price index is used to ‘deflate’ compensation and output.

This puts in perspective the widely-publicised ‘fact’ that productivity has increased much faster than employees’ pay. Studies such as those by Mishel and Shierholz (2011) produce lopsided growth of productivity relative to compensation by deflating the variables by means of two different price indexes, a procedure that is rather like deflating the tires on the left side of your car by more than you deflate the tires on the right side, and then complaining that the dealership sold you a lopsided car. Moreover, it tells us nothing about the distribution of output between employees’ compensation and profit (see Kliman 2014a). When the output pie is divided into slices, the sum of the slices must clearly equal the whole pie, but the use of different price indexes violates this requirement.

A gap between productivity growth and growth of ‘pay’ has also been produced by using only wages and salaries in the narrow sense to measure pay. As Figure 2 shows, the wage-and-salary share of output has indeed declined, but this decline likewise tells us nothing about the distribution of output between employees’ compensation and profit. Total compensation did not decline as a share of output.⁷ Thus, the decline in the wage-and-salary share was associated with a rise in the share of output that workers receive in the form of nonwage compensation, not a rise in the profit share.⁸

Inequality among workers

Income inequality can also increase because of growing inequality among workers. This is a case in which employee compensation is redistributed from low-paid to more highly-paid workers, not one in which income is redistributed from compensation to profit. Figure 5, based on data reported in a 2011 US Congressional Budget Office study, presents the trend in the concentration index for labour income.⁹ (The concentration index is similar to a Gini coefficient; labour income as defined in that study is synonymous with employee compensation.) Concentration increased – i.e. the distribution of compensation became less equal – between 1979 and the Great Recession. The increase in concentration was particularly rapid between 1979 and 1988, after which it moderated substantially.

Figure 5.

Concentration of labour income, USA 1979-2007.

There is considerable debate among economists about what caused the distribution of compensation to become more unequal,¹⁰ but it is clear that this trend is closely associated with a rising pay differential between workers with more formal education and those with less. Data reported in another Congressional Budget Office study, summarised in Figure 6, shows that real wages (exclusive of non-cash benefits) rose substantially between 1979 and 2009 among women with at least some college education, and men with at least a four-year college degree. At the same time, the real wages of less-educated workers stagnated or even fell.¹¹ Since workers with more schooling were paid significantly higher wages already in 1979, the quicker growth of their pay during the three decades that followed is indicative of a rise in wage inequality.

Figure 6.

Real median hourly wage growth, by sex and educational attainment, USA 1979-2009.

Overestimation of rising inequality

If US income inequality had increased by as much as is often believed, it would indeed be difficult to reconcile the inequality data with the fall in the rate of profit and the stability of the compensation share of output. But the degree to which income inequality has increased is often greatly exaggerated. There is no single thing called ‘income’, only a great variety of things,¹² and the degree to which measured inequality increased is very sensitive to the particular measure employed, as we shall see. Studies that report the most dramatic (and most-widely publicised) increases in inequality are based on a narrow and rather peculiar definition of income.¹³

The work of Thomas Piketty and Emmanuel Saez (2003) overturned a prior consensus that the growth of income inequality in the US, which had been rapid during the 1980s, slowed down greatly after about 1993. They reported a shockingly large – and continuing – increase in the incomes of ‘the 1 per cent’ and other high-income groups relative to the rest of the population (see Figure 7, which presents data contained in Table A1 of Piketty and Saez [2013]). Some researchers suggested that the markedly different results Piketty and Saez obtained stemmed from the fact that they employed data from tax returns while prior research relied on inferior census data. (The US Census Bureau censors information on those with very high incomes in order to protect their anonymity, so researchers have to employ a good deal of guesswork in order to estimate top incomes.)

Figure 7.

Top tax units’ shares of taxable cash income, USA (not including capital gains), 1960-2007.

However, Burkhauser et al. (2012) have recently shown that one can obtain results from Census data that are very similar to those of Piketty and Saez, if one employs their definition of income and their income-sharing ‘units’. Thus, the reason why Piketty and Saez have come up with such shocking conclusions about rising inequality has little to do with the fact that they use a different data source. It has to do with the fact that they use a very different definition of income, and focus on very different income-sharing ‘units’. Prior to their work, research on US income inequality most commonly defined income so as to include cash transfer payments received from government – such as Social Security, unemployment and welfare benefits – and the units most commonly considered were size-adjusted households. Yet owing to the limitations of tax-return data, Piketty and Saez’s definition of income excludes transfer payments, and their units are ‘tax units,’ not adjusted for size. (In the US, a tax unit is either an unmarried individual or a married couple that files a joint tax return, plus [in both cases] any dependent children.)

Related papers by Burkhauser and colleagues have also shown that different definitions and choices of units have an enormous effect on the conclusions. The information summarised in Table 1 comes from Tables 1 and 3 of Armour, Burkhauser and Larrimore (2013). The top rows of panels A and B employ Piketty-Saez methods, while the middle rows employ methods that were most common prior to their research. The bottom rows are like the middle ones, except that they also consider the effects of taxes and in-kind income. Piketty-Saez methods suggest that income inequality increased markedly during the three decades preceding the Great Recession, and that the real incomes of lower- and middle-income groups fell or stagnated. However, if we consider more inclusive definitions of income and look at size-adjusted households, there is no stagnation of real income, and the rise in inequality is much more modest. Indeed, when the effects of taxes and in-kind income are factored in, there is a decline in inequality from 1989 onward.

Table 1.

Real growth of annual income in US, by Quintile.

Years	Income measure			Units		Includes annual accrued capital gains on			Real growth of annual income
	after-tax	Includes in-kind income	Includes transfer income		adjusted for size?	public investments	housing	privately-held businesses	Quintile
	after-tax	Includes in-kind income	Includes transfer income		adjusted for size?	public investments	housing	privately-held businesses	bottom	2nd	middle	4th	top
A. 1979–2007	no	no	no	tax units	no	no	no	no	–33.0	0.7	2.2	12.3	32.7
	no	no	yes	households	yes	no	no	no	9.9	15.6	22.8	29.2	42.0
	yes	yes	yes	households	yes	no	no	no	31.8	31.3	34.4	38.8	54.0
B. 1989–2007	no	no	no	tax units	no	no	no	no	–32.9	0.8	2.3	8.0	12.9
	no	no	yes	households	yes	no	no	no	9.9	10.8	12.6	14.4	15.1
	yes	yes	yes	households	yes	no	no	no	26.4	22.7	20.2	20.0	19.0
C. 1989–2007	yes	yes	yes	households	yes	yes	no	no	29.4	22.9	18.1	17.0	10.9
	yes	yes	yes	households	yes	yes	yes	no	12.9	8.5	6.1	5.5	−0.7
	yes	yes	yes	households	yes	yes	yes	yes	15.2	3.1	−0.5	−0.6	−14.5

Notes: Income is deflated by the CPI-U-RS price index. In-kind income is the value of medical insurance and other in-kind benefits received from government and employers. Transfer income consists of cash transfer payments (Social Security benefits, etc.) received from government. ‘Public investments’ include only stock holdings. Italicised rows report results obtained from definitions like those of Piketty and Saez.

The figures in panel C are like those in the bottom rows of panels A and B, but also include various kinds of annual accruals of capital gains. They likewise indicate that incomes were more equal in 2007 than in 1989, and the degree to which inequality went into reverse is far greater than when capital gains are excluded.¹⁴

The findings of Burkhauser and colleagues do not directly provide information on trends in compensation of employees. They pertain to size-adjusted households rather than employees, and to income from all sources rather than compensation alone. Moreover, the most striking findings pertain to after-tax income, while compensation paid by employers is of course a pre-tax figure. Yet these findings, in contrast to those of Piketty and Saez, certainly do suggest that income-inequality trends are compatible with the stability of the compensation share of output and the fall in the rate of profit. They also suggest that much of the apparent discrepancy has been produced by Piketty and Saez’s exclusion of non-wage compensation, and that their focus on tax units may also be a factor.¹⁵

The main conclusions of this section are that the compensation share of output was stable, and that this fact cannot properly be dismissed by pointing to rising income inequality and the supposed gap between productivity growth and compensation growth. Inequality increased because (1) interest and dividend income rose at the expense of other components of profit, not at the expense of employee compensation; and (2) there was a rise in inequality of compensation, not a fall in the compensation share. And given the extent to which inequality actually increased, and the inconsistent inflation-adjustment methods that create the gap between productivity growth and compensation growth, the stability of the compensation share is hardly an astonishing phenomenon.

Managers’ and supervisors’ share of compensation

Recent papers by Magdoff and Foster (2013) and Mohun (2013, 2014) attempt to estimate the compensation received by ‘the working-class’ – i.e. by workers other than management and supervisory employees – rather than by all employees. They note correctly that compensation data in the US National Income and Product Accounts overstate the pay received by the working class, because the pay of top executives is included in this data.

These papers then approximate the compensation received by the working class by means of data on the wages of ‘production and nonsupervisory’ (P&NS) workers. They find that the share of total compensation the US ‘working class’ receives has declined precipitously, and that it has fallen to a shockingly low level during the last several decades. According to Magdoff and Foster (2013, Chart 3), the ‘working class’ received only 55 per cent of total compensation in 2007, and Mohun’s estimate is even lower. In addition, after counting the compensation of non-P&NS workers as profit (capital income) rather than as a cost (labour income), Mohun (2013) finds that what he calls the ‘maximal class rate of profit’ experienced a massive rebound between the early 1980s and the Great Recession.

However, the estimates provided in these papers are not remotely plausible. They suffer from three major defects. First, the data for non-P&NS workers seriously overestimates the number of management and supervisory employees, and even more seriously overestimates their share of employees’ compensation. Second, management/supervisory employees’ compensation is further inflated by treating the difference between two datasets’ total compensation figures as if it were additional compensation received by this group. Third, even a reasonable estimate of the compensation actually received by management/supervisory employees vastly overcounts the portion of compensation that can be regarded as capital income (disguised dividends). Each of these problems is discussed in turn below.

Non-P&NS workers vs. management/supervisory employees

As noted above, Magdoff and Foster estimate that the ‘working class’ received 55 per cent of total compensation in 2007, which means that their proxy for management/supervisory employees, non-P&NS workers, received the remaining 45 per cent. However, the May 2007 Occupational Employment Statistics (OES), published by the US Bureau of Labor Statistics, indicate that the combined wages of employees in management occupations and ‘first-line supervisors/ managers’ – employees whose primary duty is to directly supervise and coordinate the activities of other workers – were only 16 per cent of total wages. This is less than three-eighths of Magdoff and Foster’s estimate of the share of compensation received by non-working-class employees.

The problem here is that the category of employees that Magdoff-Foster and Mohun exclude from the working class, non-P&NS employees, is far too broad. In 2007, 17.7 per cent of private-sector workers were in that category, while the OES data indicate that only 9.2 per cent of private-sector workers, barely half as many, were employed in management occupations or as first-line supervisors/managers.¹⁶ The authors do take note of this problem; for instance, Magdoff and Foster (2013) point out that their category ‘undoubtedly includes many employees who might well be considered part of the working class’. Yet the phrase ‘many employees’ seems to seriously understate the gap between non-P&NS employees’ and management/supervisory employees’ shares of employment.

One reason why the P&NS data fails to capture a large portion of the wages of non-management/supervisory employees is that the P&NS category excludes workers in goods-producing industries who are not directly engaged in ‘production’, but who also are not managers or supervisors – for example, clerical workers in manufacturing and construction firms.¹⁷ Even more importantly, the P&NS data also fail to capture the wages of millions of other workers, in both goods-producing and service-providing industries, because the P&NS category doesn’t make much sense to the people who answer the government’s survey questions. As the US Department of Labor (2005) noted, ‘the production and non-supervisory worker hours and payroll data have become increasingly difficult to collect, because these categorizations are not meaningful to survey respondents. Many survey respondents report that it is not possible to tabulate their payroll records based on the production/non-supervisory definitions.’¹⁸ Thus, such respondents either fail to provide the requested data or ignore the official definition of P&NS workers and instead provide data on other, often narrower categories of workers (e.g. those covered by the Fair Labor Standards Act, or paid on an hourly basis; see Abraham, Spletzer & Stewart [1998: 311–3]).

As a result, the P&NS category greatly undercounts the number of working-class employees. This is one reason why Magdoff-Foster and Mohun’s estimates of these employees’ share of total wages are far too low. Another reason is that the workers who should be counted, but aren’t (e.g. those not covered by the Fair Labor Standards Act or paid on an hourly basis) are undoubtedly the more highly paid workers, at least on average.¹⁹ If wages of more highly-paid workers are rising faster than wages of low-paid workers – as has occurred in the USA in recent decades – then, as time proceeds, data that wrongly excludes more highly-paid workers will understate the working-class share of total wages to a greater and greater extent.

This is one reason – perhaps the main reason – why Magdoff-Foster and Mohun find that the working class’s share of wages has fallen precipitously. It should also be noted that, since P&NS data is not based, in practice, on a standard definition – firms seem frequently to employ their ‘own’ definitions, and the firms included in the survey change over time – and since failure to respond to questions about P&NS workers has been a significant problem, reported trends in the wages and hours of these workers are questionable.

Imputation of ‘missing’ wages to management/supervisory employees

Another reason why the Magdoff-Foster and Mohun papers produce such gross overestimates of management/supervisory employees’ share of compensation is that they treat the discrepancy between two datasets’ total compensation figures as if it were additional compensation received by management/supervisory employees. One of these datasets, Current Employment Statistics (CES), is the source of their data on wages of private-sector P&NS workers. The other dataset, the National Income and Product Accounts (NIPAs), is the source of their total wage bill for all private-sector workers. The CES data indicates that P&NS workers received 67 per cent of total wages in 2007, $2908 billion out of $4348 billion. But Magdoff and Foster choose instead to express P&NS workers’ wages as a percentage of the total NIPA wage bill of $5326 billion.²⁰

There is a huge difference between the two total wage bills, of no less than $978 billion. It reflects the fact that the CES definition of wages is quite narrow, while the definition employed in the NIPAs is much more inclusive. Because Magdoff and Foster count none of the $978 billion difference as wages of P&NS workers that the CES data fail to capture, they improperly treat all of it as additional wages received by non-P&NS employees. This procedure artificially boosts non-P&NS employees’ share of wages from 33 per cent to 45 per cent, and artificially depresses P&NS workers’ share from 67 per cent to 55 per cent. With equal justification, one could treat the discrepancy between the CES and NIPA data as additional wages of P&NS workers, which would bring their share of total wages to 73 per cent.

The Magdoff-Foster paper does not explain why it treats the entire gap between NIPA and CES wages as additional wages received by non-P&NS employees. In contrast, Mohun (2014) contains an extended discussion of the reasons why the two datasets’ total wages differ. In the end, however, this paper likewise counts the entire gap between NIPA and CES wages as wages of non-P&NS workers! Mohun (2014) justifies this decision as follows:

[It] accounts for stock options, bonuses, retroactive pay and the like, … for these sorts of payments typically do not accrue to the working class. It is less reasonable to make such an allocation for irregular tips, free rent, fuel, meals, and other payment in kind. … It is evidently a major assumption to allocate all of the missing CES wages to supervisory workers, but it does not seem an unreasonable approximation.

However, in Appendix 2, below, I estimate the share of the discrepancy between CES wages and NIPA that can properly be allocated to management/supervisory employees, bending over backwards to refrain from underestimating their share. I estimate that no more than 37 per cent to 45 per cent of the discrepancy can be allocated to them, and that their share of NIPA wages was between 16.4 per cent and 20.6 per cent.

The share of capital income in the wage data

In addition to using the wages of non-P&NS workers as a proxy for management/supervisory employees’ compensation, Magdoff-Foster and Mohun use their wages to approximate the portion of compensation that is actually ‘capital income’ (dividends in disguise) rather than ‘labor income.’ Magdoff and Foster (2013) write that ‘the compensation going to CEOs and other upper-level management … ought to be counted as income to capital rather than labor’; Mohun (2013: 178) writes that compensation ‘include[s] the labour income of top echelons of management’, and is thus ‘too inclusive a category to be of use as a class category’. Although I largely agree with these statements,²¹ wages of non-P&NS workers are a woefully inappropriate proxy for disguised dividends.

Part of the problem is the one discussed above: the exclusion of a large and relatively highly paid segment of the working class from the P&NS data. But even a proper estimate of the share of wages actually received by management/supervisory employees (16 per cent of the total in 2007, according to OES data, not 45 per cent) vastly overstates the share of capital income (dividends) that is disguised as compensation. According to Magdoff-Foster and Mohun themselves, the capital-income component of employee compensation includes only the compensation received by ‘CEOs and other upper-level management’ (Magdoff and Foster 2013), the ‘top echelons of management’ (Mohun 2013: 178), not the total compensation received by management/ supervisory employees as a whole. The compensation received by first-line supervisors and most management employees is legitimately counted as labour income (payment for labour services) and as a cost to businesses rather than as disguised dividend income received by the owners of the businesses, even if these employees are not members of ‘the working class’ in some senses of that term.

‘Capital income’ and ‘labour income’ are standard concepts in mainstream economics; ‘disguised dividends’ or ‘constructive dividends’ is a concept of the Internal Revenue Service (IRS), the US federal tax authority (see USLegal.com, n.d.). The distinction between capital income and labour income is the distinction between income deriving from ownership and income deriving from the provision of labour services. This is a different distinction from that between property income (profit, rental income, interest income, etc.) and compensation of employees. For example, if the owner of a business works (provides labour services) in that business, some of her income is, and is typically treated as, labour income (see, e.g., Gomme & Rupert 2004). Conversely, if some of the compensation received by an employee such as a corporate executive derives from his de facto ownership interest in the corporation rather than from the performance of managerial work, that portion of his compensation is capital income.²²

The IRS concept of ‘disguised dividends’ refers to this portion of his compensation or something quite similar. For instance, the IRS determined that almost all of the compensation received in 1998 by John Menard, the CEO of and controlling shareholder in Menards, a chain of home-improvement centres, was a disguised dividend. A key criterion it employed was that his compensation exceeded the amount that would ‘ordinarily be paid for like services by like enterprises’. The Tax Court largely agreed; since the highest-paid CEO among Menards’ competitors received $6.1 million in compensation, it ruled that all but $7.1 million of the $20.6 million that Menard paid himself was implicitly a dividend rather than payment for services he rendered as CEO.²³

As this case suggests, the capital-income or disguised dividend component of compensation can be understood as compensation in excess of what the employer needs to pay to obtain the labour services that are performed. One indicator that a portion of compensation is capital income is thus the amount of compensation itself – relative to compensation paid in other years and to compensation paid ‘for like services by like enterprises’. Another indicator is that the compensation varies according to how well the company does, independently of how well the manager does his or her job.

Given these criteria, it seems clear that a sizable portion of the compensation paid in 2007 to the highest-paid executives of the largest US corporations consisted of disguised dividends. The data in Table 2 pertains to the highest-paid executives (generally, four or five per company) of the eight or nine thousand publicly-traded US corporations.²⁴ Almost three-fifths of their compensation in 2007 consisted of ownership rights in these corporations – stock grants and stock options; so when stock prices plummeted between 2007 and 2009, their total compensation plummeted by roughly the same percentage. The 31.5 per cent fall in their compensation suggests that a large portion of their compensation in 2007 should be understood as disguised dividends according to the first criterion. And the fact that the compensation they received was very sensitive to movements in stock prices suggests that much of it was disguised dividends according to the second criterion.

Table 2.

Compensation of highest-paid executives, publicly-traded US corporations.

	Component of compensation				S&P 500 Stock Index
	Salaries	Stock grants & options	Other	Total	S&P 500 Stock Index
Percentage change, 2007–2009	8.5%	−51.3%	−9.9%	−31.5%	−35.8%
Share of total compensation, 2007	15.8%	59.3%	24.8%	100.0%	N/A

However, it is extremely unlikely that first-line supervisors/managers or the bulk of management employees are recipients of dividends disguised as labour income, because it is extremely unlikely that they are paid significantly more than their employers need to pay to obtain the labour services they perform. OES data indicates that the average annual wage rate of first-line supervisors/managers was $47,370 in 2007, just 16 per cent more than the overall average wage ($40,690), and that even three-fourths of those in management occupations received an annual wage of $121,690 or less, which means that their wages were less than triple the overall average. Even if we increase these figures by 50 per cent or even 100 per cent in order to account for components of total compensation not captured by OES wages, such employees’ pay was only a small fraction of the compensation ($1.7 million, on average) received by the highest-paid executives of public corporations.

It would thus be unreasonable to count the wages of either the first-line supervisors/managers or of the bottom 75 per cent of management employees as capital income. Moreover, even the wages received by a large segment of ‘chief executives’ seem to be payment in exchange for their labour services rather than dividends in disguise: OES data indicate that 25 per cent of them received wages less than or equal to $97,960 in 2007, which was 2.4 times the overall average wage.

Thus, if we wish to estimate the share of wages or compensation that is actually capital income or disguised dividends, it seems advisable to base the estimate on wage levels that may be in excess of what employers need to pay for labour services, rather than on specific occupational categories like ‘chief executive’ or ‘CEO’. And it seems advisable to include, as recipients of dividends disguised as wages, only the small percentage of management employees who were paid well in excess of triple the average wage. Bending over backwards to avoid underestimating disguised dividends, I have chosen to include the most highly paid 10 per cent of employees in management occupations and to count all of their wages as capital income.²⁵

The OES dataset censors data on very high wages, but I was able to estimate wages of the top 10 per cent of management employees by using other data contained in the OES (see Appendix 3 for details). My estimates indicate that, to be part of the top 10 per cent in 2007, a manager needed an annual wage of at least $171,680, which was 4.2 times the overall average wage, and which probably placed him or her in the top 2 per cent or 3 per cent of all wage earners, but not the top 1 per cent.²⁶ The top 10 per cent of management employees received 23.34 per cent of the total wages received by those in management occupations, or 2.46 per cent of the total wages received by all employees. Since the top 10 per cent of managers were 0.45 per cent of all employees, their average wage was 2.46 per cent/0.45 per cent = 5.5 times the overall average wage.

Mohun’s (2013: 180, Table 10.1) estimates imply that the share of total compensation received by P&NS workers fell from 69 per cent in 1966 to 52 per cent in 2007. The fall in their share depicted in Magdoff and Foster’s (2013) Chart 3, from 76 per cent in 1965 to 55 per cent in 2007, is even more extreme. Although these authors are aware that the remaining share, wages of non-P&NS workers, is not a perfect measure of capital income disguised as wages, they do suggest that their computations tell us something meaningful about how the share of compensation paid to the ‘top echelons of management’ has increased at the expense of the working-class workers. My estimates – based on what I think are clearly better measures of the ‘top echelons of management’ and disguised dividends – indicate otherwise. Since the top 10 per cent of management employees received 2.46 per cent of total wages in 2007, their share of total wages could not have risen by more than 2.46 percentage points between 1965 and 2007. Thus, other employees’ wage share could not have fallen by more than 2.46 percentage points, which is less than one-eighth of the decline that Magdoff and Foster report for P&NS workers.

The OES data extends only as far back as 1997, so it is impossible to use them to directly estimate long-term trends. But backward simulation using the 2007 data is informative. As I have noted, my estimates indicate that the maximum possible rise in the share of wages received by the top 10 per cent of management employees was 2.46 percentage points. But their share could have risen by that much only if, in Marx’s (1991: 356) famous phrase, they were able to ‘live on air’ back in 1965, which was not the case. If they were paid no more than the average worker in 1965, then their share of wages rose by 2.02 percentage points, and if their average wage were originally double the overall average wage, then their share rose by 1.57 points (see Table 3).²⁷ But these cases also seem to be quite unlikely. The average wage of all management employees might well have been no more than double the overall average wage, but it is hard to believe that the top 10 per cent of them were paid so little.²⁸

Table 3.

Simulated changes in shares of wages and output, top 10 per cent of managers.

Original average wage rate (multiple of overall avg. wage rate)	0	1	2	3	4	5	5.5
Percentage-point rise through 2007
in share of total wages	2.46	2.02	1.57	1.12	0.68	0.23	0.00
in share of total net output	1.55	1.27	0.99	0.70	0.42	0.14	0.00

It is much more likely that their average wage was originally as least triple the overall average, which implies that their share of total wages rose by no more than 1.12 percentage points. And this, in turn, implies that any rise in their wage share that may have occurred led to a fall in other employees’ share of output (net value added) of only 0.7 percentage points between 1965 and 2007. This is one-fifteenth of the 10.3 percentage-point fall in P&NS workers’ wages as a share of GDP that Magdoff and Foster report.

Given the paucity of information that is available, I cannot prove that a rise in disguised dividends received by top management employees failed to depress other employees’ share of output by more than 0.7 percentage points.²⁹ Nor can I prove that the disguised-dividend share of compensation actually rose at all. I do suggest that the evidence provided here in support of my maximum estimate is considerably stronger than any evidence which challenges it. In particular, I caution against drawing conclusions about the top 10 per cent of management employees from compensation trends that pertain to quite different populations, such as the top 1 per cent of tax units or the 1500 highest-paid executives of the largest corporations – a mere 0.025 per cent of management employees – to whom the findings in Frydman and Saks (2008) apply.

Conclusion

US corporations’ rate of profit fell and the employee-compensation share of net output was trendless between 1970 and the Great Recession. This paper has shown that these facts can easily be reconciled with evidence that income inequality has increased in the USA. It has also responded to Magdoff and Foster (2013) and Mohun (2013, 2014), who have recently challenged the rate-of-profit and compensation-share evidence on different grounds. These contributions contend that non-management/supervisory workers’ share of employee compensation plummeted and that the ‘maximal class rate of profit’, which counts compensation of management/supervisory employees as profit, rose markedly. However, this paper has shown that the Magdoff-Foster and Mohun papers seriously overestimate management/supervisory employees’ compensation, as well as the share of compensation that is actually disguised dividends rather than payment for these employees’ labour services. If a rise in the disguised-dividend share of compensation occurred, it was almost certainly too small to have had a sizable effect on the ‘real’ compensation share of output or the rate of profit. The evidence that the rate of profit fell and that the compensation share of output was stable thus emerges unscathed.

Although this paper has refrained from trying to provide a theoretical explanation of the facts it has presented, it may be appropriate here to venture a few related hypotheses about why employees’ share of US business-sector output remained stable despite globalisation, the decline of unionisation, the erosion of labour rights, giveback contracts, and so on. First, the evidence suggests that ‘power’ relations may have played a smaller role in wage determination than is sometimes assumed. Second, only a small percentage of US workers belongs to unions, so phenomena that depressed their wages may not have had a strong direct effect on aggregate wages. Third, the fall in union workers’ wages may have had little or no spillover effect on other workers’ wages; this hypothesis is consistent with and seems to be suggested by theories of labour-market segmentation. Finally, it seems likely that market forces (such as rising revenues in the healthcare and financial services industries and skill-biased technological change generally) acted to raise the wages of some workers – especially more highly-paid ones – even as some combination of market forces and ‘power’ relations tended to depress the wages of other workers.

Footnotes

Appendices

Notes

Author biography

Andrew Kliman is a professor emeritus of economics at Pace University and the author of Reclaiming Marx’s ‘Capital’: A Refutation of the Myth of Inconsistency (Lexington, 2007) and The Failure of Capitalist Production: Underlying Causes of the Great Recession (Pluto, 2012). He works politically with the Marxist-Humanist Initative. His current projects include a book on US income inequality, and the editing of a collection of Raya Dunayevskaya’s writings on capitalism.

References

Abraham

Spletzer

Stewart

(1998) Divergent trends in alternative wage series. In Haltiwanger

Manser

Topel

(eds.) Labor Statistics Measurement Issues. Chicago: Univ. of Chicago Press.

Abraham

Spletzer

(2009) New evidence on the returns to job skills. American Economic Review 99(2): 52–7.

Abraham

Spletzer

(2010) Are the new jobs good jobs? In Abraham

Spletzer

Harper

(eds.) Labor in the New Economy. Chicago: Univ. of Chicago Press.

Armour

Burkhauser

Larrimore

(2013) Levels and trends in United States income and its distribution: A crosswalk from market income towards a comprehensive Haig-Simons income approach. NBER Working Paper No. 19110, June.

Bakija

Cole

Heim

(2012) Jobs and income growth of top earners and the causes of changing income inequality: Evidence from US tax return data, April. Online at <web.williams.edu>, accessed 26 January 2015.

Balsam

(2012) Taxes and executive compensation, Economic Policy Institute Briefing Paper #344. Online at <www.epi.org>, accessed 26 January 2015.

Bishow

(2009) A look at supplemental pay: Overtime pay, bonuses, and shift differentials. Bureau of Labor Statistics publication. Online at <www.bls.gov >, accessed 26 January 2015.

Brown

Sokal

Friedman

(2013) The complex dynamics of wishful thinking: The critical positivity ratio. American Psychologist 68(9): 801–13.

Burkhauser

Feng

Jenkins

Larrimore

(2012) Recent trends in top income shares in the United States: Reconciling estimates from March CPS and IRS tax return data. Review of Economics and Statistics 94(2): 371–88.

10.

Bureau of Economic Analysis, US (2012) Local Area Personal Income and Employment Methodology, April. Online at <www.bea.gov >, accessed 26 January 2015.

11.

Clementi

Gallegati

(2008) Pareto’s law of income distribution: Evidence for Germany, the United Kingdom, and the United States. Online at <ideas.repec.org >, accessed 26 January 2015.

12.

Congressional Budget Office. US (2009) Changes in the Distribution of Workers’ Annual Earnings between 1979 and 2007. Online at <www.cbo.gov>, accessed 26 January 2015.

13.

Congressional Budget Office (2011a) Changes in the Distribution of Workers’ Hourly Wages between 1979 and 2009. Online at <www.cbo.gov>, accessed 26 January 2015.

14.

Congressional Budget Office (2011b) Trends in the Distribution of Household Income between 1979 and 2007. Online at <www.cbo.gov>, accessed 26 January 2015.

15.

Department of Labor, US, Office of the Secretary (2005) Submission for OMB review: Comment request, 8 April. Online at <tinyurl.com/6bwrgt5>, accessed 26 January 2015.

16.

Diehm

Hall

(2013) Inequality in US is scarily high, rising. Huffington Post 12 June. Online at <www.huffingtonpost.com >, accessed 26 January 2015.

17.

Fisher

Houseworth

(2012) Occupation inflation in the Current Population Survey. US Census Bureau. Center for Economic Studies, Discussion Paper CES 12-26. Online at <www2.census.gov>, accessed 26 January 2015.

18.

Frankfurt

(2005) On Bullshit. Princeton: Princeton Univ. Press.

19.

Fry

Cohn

(2010) Women, Men and the New Economics of Marriage. Pew Research Center. Online at <pewsocialtrends.org>, accessed 26 January 2015.

20.

Frydman

Saks

(2008) Executive compensation: A new view from a long-term perspective, 1936-2005. NBER Working Paper No. 19110 June. Online at <www.nber.org>, accessed 26 January 2015.

21.

Gittleman

(1994) Earnings in the 1980s: An occupational perspective. Monthly Labor Review July: 16–27. Online at <www.bls.gov>, accessed 26 January 2015.

22.

Gomme

Rupert

(2004) Measuring labor’s share of income. Policy Discussion Papers, No. 7, Research Department of the Federal Reserve Bank of Cleveland. Online at <citeseerx.ist.psu.edu>, accessed 26 January 2015.

23.

Johnston

(2011) 9 things the rich don’t want you to know about taxes. Willamette Week 13 April. Online at <www.wweek.com>, accessed 26 January 2015.

24.

Kliman

(2012) The Failure of Capitalist Production: Underlying Causes of the Great Recession. London: Pluto Books.

25.

Kliman

(2014a) Are corporations really hogging workers’ wages? Truthdig, 9 April; online at <www.truthdig.com>, accessed 26 January 2015.

26.

Kliman

(2014b) Were top corporate executives really hogging workers’ wages? Truthdig, 18 September. Online at <www.truthdig.com>, accessed 26 January 2015.

27.

Lain

(2013) Diet Soap podcast #196: answering the potshots (interview with Andrew Kliman). Online at <dietsoap.podomatic.com>, accessed 26 January 2015.

28.

Magdoff

Foster

(2013) Class war and labor’s declining share. Monthly Review 64(10). Online at <monthlyreview.org>, accessed 26 January 2015.

29.

Marx

(1991 [1894]) Capital: A Critique of Political Economy, vol. 3. London: Penguin Books.

30.

Mishel

(2012) The wedges between productivity and median compensation growth. Economic Policy Institute Issue Brief #330. Online at <s4.epi.org>, accessed 26 January 2015.

31.

Mishel

Shierholz

(2011) The sad but true story of wages in America. Economic Policy Institute Issue Brief #297. Online at <epi.3cdn.net>, accessed 26 January 2015.

32.

Mohun

(2013) Rate of profit and crisis in the US economy: A class perspective. In Taylor

Rezai

Michl

(eds.) Social Fairness and Economics: Economic Essays in the Spirit of Duncan Foley. Abingdon, UK and New York: Routledge.

33.

Mohun

(2014) Unproductive labor in the US Economy 1964–2010. Review of Radical Political Economics 46(3): 355–79.

34.

Moylan

(2008) Employee stock options and the national economic accounts. Survey of Current Business Feb: 7–13. Online at <www.bea.gov>, accessed 26 January 2015.

35.

Piketty

Saez

(2003) Income inequality in the United States, 1913–1998. Quarterly Journal of Economics 118(1): 1–39.

36.

Piketty

Saez

(2013) Updated spreadsheet file accompanying ‘Income inequality in the United States, 1913-1998’. Online at <elsa.berkeley.edu>, accessed 26 January 2015.

37.

USLegal.com. (n.d.) Constructive or disguised dividends; online at <dividends.uslegal.com>, accessed 26 January 2015.

38.

Wood

(2013) Review of The Failure of Capitalist Production by Andrew Kliman. Marx & Philosophy Review of Books, online at <marxandphilosophy.org.uk>, accessed 26 January 2015.