Beyond BLS

January 2016

The laboring labor share of income: the “miracle” ends

Brian I. Baker

Economists classify the income earned by U.S. households as either labor earnings or capital earnings. From about 1950 to 2001, the labor share of income never wavered far from 62 percent, even as the nation underwent a transformation from an economy based primarily on manufacturing to one grounded chiefly in services. Earlier, the economist John Maynard Keynes called this kind of behavior—in which two opposing forces pull the labor force in opposite directions, yet manage to cancel out, so that the labor share remains stable—“a bit of a miracle.” As applied to the 1950–2001 period, the “miracle” inhered in the singular coincidence of a falling labor share in manufacturing industries and a rising one in service-providing industries balancing each other out, resulting in no change in the overall labor share. However, Keynes’s “bit of a miracle” eventually ended: beginning in 2001 through the present, the labor share slowly fell, dropping to about 56 percent of income by 2014. But why did the “miracle” end? That question is what Roc Armenter seeks to answer in “A bit of a miracle no more: the decline of the labor share,” published in the Federal Reserve Bank of Philadelphia Business Review for the third quarter of 2015.

After acknowledging a number of difficulties associated with measuring the U.S. labor share of income, Armenter presents three alternative measures that attempt to deal with one particular difficulty: what economists call proprietor’s income, the income of self-employed individuals working either as sole proprietors or in partnerships. The problem with proprietor’s income is that, although the self-employed earn both labor income and capital income, there is no need for them to distinguish between the two, so their records do not reflect the difference. Therefore, economists struggle to decompose proprietor’s income into how much is labor income and how much is capital income—ergo the three alternative measures. The main measure, developed by the U.S. Bureau of Labor Statistics (BLS) estimates proprietors’ labor compensation by assuming that proprietors earn the average hourly compensation that the employees working in the same sector earn. The BLS-estimated proprietors’ labor compensation is added to the employees’ labor compensation, and their sum is divided by output to yield the BLS headline labor income share. This methodology is the same for the entire range of the BLS labor share series, currently from 1947 to 2016. A charting of the BLS labor share series shows the labor share’s steady behavior at about 62 percent from 1950 to 2001 and falloff from then on. An alternative measure, formulated to take account of a decline in the BLS-estimated proprietors’ labor compensation as a proportion of proprietors’ total income, shows that at least one-third, and possibly half, of the drop in the headline labor income share is due to the BLS methodology. The other half to two-thirds of the drop is real, however, and like the overall trend, the alternative measure’s labor share of income declined after 2000. Finally, the third alternative measure ignores proprietor’s income altogether. Armenter shows that this measure, too, exhibits a stable trend from 1950 to about 2001 and then drops off significantly.

Following a discussion showing that the chief culprit in ending the “bit of a miracle” was wage stagnation in the manufacturing industry, Armenter takes up the question of what the “ultimate cause” is behind the decline of the U.S. labor share of income. Conceding that “economists have several hypotheses but no definite answer yet,” he cites the three leading hypotheses purporting to explain the decline.

First, there is the hypothesis of capital deepening: technological innovations of various kinds produce better and cheaper equipment—robots, ATM machines, powerful software—that replaces workers and redistributes income from labor to capital. But this hypothesis is challenged by a logical consequence of the combination of economic theory and empirical fact: even when innovation replaces workers with equipment, the remaining workers should be more productive; that is, innovation accelerates labor productivity rather than making wages stagnate. But this is not what happened in the manufacturing sector over the period examined.

Second is the hypothesis of inequality. According to this hypothesis, technological innovation augments productivity more for highly skilled workers than for low-skilled workers, making the low-skilled workers redundant and thereby dispensable. As a result, wage inequality between workers increases. Consequently, the capital share of income increases and the labor share decreases. The principal challenge to this hypothesis is that empirical evidence appears to show that the rise in wage inequality—particularly the increase in top wages—has actually sustained the labor share, not diminished it.

Finally, the globalization hypothesis links that phenomenon to the falling labor share of income: U.S. industries that are more labor intensive will outsource their work to countries with cheap labor while industries that are more capital intensive will remain in the United States. The result is an increase in the capital share of income and a decrease in the labor share. The trouble with this hypothesis, according to Armenter, is that there is little evidence to support it: the hypothesis predicts that the nation will import goods produced with a lot of labor and export goods produced with a lot of capital, whereas U.S. international trade involves exchanging goods that are quite similar, such as cars.

Download PDF »