"You're working harder, but your paycheck isn't going up." So claims a widely shared chart from the Hillary Clinton campaign, which shows the pay of American workers lagging far behind productivity growth after the 1970s. The idea that rising inequality has cheated most Americans out of wage gains has become the cornerstone of the liberal narrative about today's economy. It is a vivid example of the kind of start-with-the-conclusion reasoning that Brookings Institution scholar Richard Reeves has dubbed "policy-based evidence-making."

No one supplies liberal-friendly evidence-making as well as the Economic Policy Institute, whose researchers originated Clinton's chart of doom. Just one problem — the chart does not show that workers' pay has failed to keep up with their value to employers.

(Screenshot courtesy of the Hillary for America website)

The latest iteration of EPI's chart comes from a report by Josh Bivens and Larry Mishel. Bivens and Mishel show national productivity — the value of what American workers produce per hour of labor — rising by 72 percent from 1973 to 2014, while hourly compensation rose by just 9 percent. They (and Clinton) assume that worker pay should have also risen by 72 percent.

To explain this divergence, Bivens and Mishel invoke three "wedges" that split the two trends. The first is a "terms of trade wedge," which reflects the fact that the prices of what Americans buy have risen more than the prices of what they produce. But this is irrelevant to the question of whether workers are being paid fairly in accord with their value to firms. If gas prices spike, employers of secretaries should not be expected to increase their pay accordingly.

Second is the "loss of labor's share wedge." They argue that over time, more national income has flowed to shareholders than to workers. This claim is simply wrong. As I pointed out nearly a year ago — reinforced since by several studies — labor's share only looks like it has fallen because of improper methodological decisions.

The inflation adjustment is just one problem. Bivens and Mishel show productivity growth for the entire nation, but their trend for pay excludes managers and many salaried workers. They also include as an output in their productivity measure the benefits accruing to homeowners by not having to pay rent to a landlord; this "output" is not produced by workers at all, so including it within productivity for the purposes of evaluating worker pay makes no sense.

The Bureau of Labor Statistics has long recognized these issues, which is why it typically compares productivity and pay not for the national economy but for the "nonfarm business sector," excluding the housing sector and including managers. Under this methodology, labor's share of income was largely unchanged over time until the Great Recession. Since labor's share usually falls during recessions, there is no reason to think of this as a permanent redistribution between workers and firm owners.

The fact that labor's share of income has been flat also means that worker pay in the aggregate has kept up with productivity growth. By my calculations, after adjusting for inflation, productivity rose 76 percent from 1973 to 2008 and hourly compensation rose 74 percent. In contrast, Bivens and Mishel show national productivity rising 62 percent and hourly compensation rising by 39 percent.

To be sure, the increase in pay has not been evenly spread across workers, which points to EPI's third wedge — the "compensation inequality wedge." Since pay has risen much more at the top than among typical workers, the pay of the latter has not increased as much as overall productivity. Bivens and Mishel see this as evidence that their pay should have risen much more than it did.

But we lack data on the productivity of the typical worker. If the productivity of non-managers rose by less than that of managers, then we should not expect that employers would give them pay increases in line with those of managers. New analyses by economic blogger Evan Soltas show that the relationship across industries between productivity and pay is quite strong. Barring a stronger case from EPI, the Clinton campaign's main economic theme is resting on a bed of sand.

This piece originally appeared in Washington Examiner