The Share

We have followed the wage down to its smallest unit, the bargain between one worker and one employer, and traced the forces that decide it. Skill changes what the worker can do; risk changes the conditions on which it is done; discrimination changes who is admitted to the bargain and on what terms. The employer's alternatives meet the worker's alternatives, with the union, the law, and the provision for a life outside employment altering the strength of each. The machine changes the work itself. One question remains, and it is the largest. Add the payments for labor together and place them beside the income generated by production: how much of the whole has reached the people who work? Labor's share brings the wage into that wider frame. It gathers millions of payments into a fraction whose movement can reveal a change too gradual to be visible in any one pay packet. But the fraction is an account of the outcome. To read the contest that produced it, we must know what has been counted, how the parts have moved, and which mechanism connects the movement to the wage.

The account begins with compensation, which includes the benefits an employer provides as well as wages and salaries, divided by the corresponding value added. A corporate-sector measure and an economy-wide measure do not describe exactly the same territory. The income of someone working in a business they own presents a particular difficulty: part rewards labor and part rewards the ownership of capital, though the accounts may record a single income. The method used to divide it can alter the measured trend. In their 2013 study of the United States, Michael Elsby, Bart Hobijn, and Ayşegül Şahin attributed about a third of the decline in the headline measure they examined to the imputation of self-employed labor income; a substantial decline in the payroll share remained. Gross and net measures ask different questions too, because gross value added includes the capital consumed in production. A payment needed to replace worn-out equipment cannot all be treated as an owner's spendable gain. The boundary of the account is therefore part of the argument. The figure can become more revealing when its components are separated, showing which part of a fall reaches the division we want to explain.¹

The connection to productivity is exact when the units are kept together. Divide compensation and value added by the same number of hours: labor's share is average compensation per hour divided by nominal value added per hour. If output per hour rises and average compensation rises in the same proportion, measured in the same prices, the share holds steady. The pay of the typical worker introduces a second division. A median can fall behind the average when compensation becomes more unequal within labor, even if labor's total share is unchanged. Cash wages also omit benefits, while a consumer-price index measures what income buys and an output-price index measures the prices of what the sector produces. Both questions matter, but they are different questions. A graph placing typical purchasing power beside average productivity can contain the movement of the labor share, the distribution of pay among workers, and the relative prices of consumption and output. The widening of those lines calls for an explanation of each component. Calling the entire gap a transfer to owners would assign to the boundary between labor and capital changes that may have occurred within labor itself.²

The apparent stability of the share once gave growth theory a fixed point around which to build. A familiar production function, Cobb–Douglas, can reproduce that stability: with fixed exponents, constant returns, competitive factor payment, and the other conditions of the benchmark, labor receives a constant fraction of output. The constancy follows from the specification. Competitive labor markets alone do not impose it on every possible technology. Change the way capital and labor can substitute for one another, change the tasks assigned to them, or change the composition of production, and the share can move while factors still receive their competitive returns. This matters for the force of the evidence. A falling share cannot by itself measure the distance between a wage and a worker's marginal product. It can nevertheless unsettle a theory that treated stability as a fact requiring little further explanation. Karabarbounis and Neiman's international study, published in 2014, documented a broad decline in corporate labor shares over the preceding decades. The movement reopened the question the convenient constant had allowed economists to put aside: what determines the division, and what had changed in its determinants?³

The machine supplies one possible answer, but even here the mechanism has to be specified. If cheaper capital induces a sufficiently strong substitution away from labor, labor's fraction of income can fall; the required response depends on how readily the two inputs can substitute. Karabarbounis and Neiman estimated that cheaper investment accounted for roughly half the global corporate decline in their sample, within a model and an estimated elasticity that made that response possible. A task-based account instead follows which operations move from workers to capital and which new operations return to labor. Trade adds another route, as the location and composition of production change. Elsby, Hobijn, and Şahin found offshoring a leading potential explanation in the U.S. evidence they examined, while their evidence on deunionization was inconclusive. These accounts invite comparisons among sectors, firms, countries, and periods that can distinguish their predictions. The same movement of a national fraction can be reached by different paths. Identifying the path matters for a worker deciding where to invest in skill, and for a society deciding whether the response belongs in training, bargaining institutions, trade adjustment, or the rules governing technology and ownership.⁴

The composition of firms supplies a further answer that a single average can conceal. Suppose firms paying a smaller fraction of their value added to labor gain a larger share of the market. The aggregate labor share can then decline even if each firm's own share is unchanged. The superstar-firm account developed by Autor and his colleagues connects such reallocation to productive firms with high markups and low labor shares. Its evidence asks whether concentration, productivity, markups, and the movement of activity among firms follow the pattern the account predicts. The mechanism also draws a necessary distinction between two forms of market power encountered elsewhere in the table. A firm's power over its selling price concerns the product market; its power to set a wage below the worker's marginal revenue product concerns the labor market. One can coexist with the other, but their coexistence must be established. The superstar model can generate the falling share with competitive factor markets and imperfect product markets. A high profit margin and a small payroll fraction therefore sharpen the inquiry into power without identifying, on their own, a particular employer's wage markdown.⁵

The share returns us to the volume's opening question with a larger field of evidence and a more exact task. The institutions that shape an individual bargain can also shape the distribution of income, but a national fraction cannot tell us by itself which institution changed, who gained, or which worker lost. A household may receive both wages and income from ownership; workers may gain unequally even when labor's total claim grows. To judge the division requires following those claims through to the people who hold them. The lesson of the wage remains consequential: productive capacity, available alternatives, the terms of competition, and the organized power to bargain all belong in the explanation of what work pays. Their relative force is a matter for inquiry, while the standard by which a society judges the result is a question it cannot delegate to an accounting identity. Labor's share is the wage seen whole, provided we keep looking inside the whole. The separate volume on The Share takes up that inquiry. This volume closes with the worker still at the gate, asking what an hour of life can command and what would have to change for it to command more.