When economists tick off the reasons for our current wage problems, we usually include “slow productivity growth” on the list. Well, somebody asked me the reasonable question: Why should that matter?

Productivity, or output per hour, is growing slowly, but it’s still up, so why should it explain stagnant real wages? And at a more intuitive level, suppose a business figures out how to produce more efficiently and its productivity goes up. Where is it written that it must share that extra income with its workforce? Isn’t a big chunk of the inequality problem precisely that it doesn’t?

The thrust of these questions is perfectly sensible, making this one of those rare occasions (snark alert) when economists and normal people talk past each other.

Here, using truly simple arithmetic, is the economists’ model:

w/h = w/Y  *  Y/h

… where w is all the money paid in labor compensation (\$10.8 trillion last quarter, if you’re counting), h is all the hours worked in the job market, and Y is output, or GDP. Therefore, w/h is just the hourly wage, w/Y is labor’s share of total income, and Y/h is output per hour, or … productivity!

This helps to show you where we’re coming from: Productivity growth clearly feeds into wage growth. But, as is always the case in economics, there are many more moving parts to consider.

First, the model refers exclusively to average wages — not the more relevant metric, when we’re talking about workers’ living standards, of the median, or mid-level wage. When inequality is high, the average, by definition, diverges from the median. Thus, in our labor market, the average is an increasingly uninformative statistic about middle-class living standards. For that, you need to look at median compensation, which the figure below does by tracking productivity growth and median compensation.

As you see, after growing together for decades, the two lines began diverging in the mid-1970s. Since then, productivity growth has outpaced real, median compensation by a factor of six.

So, point No. 1: The simple model holds for the average, not for the median, and in the age of inequality, it’s the median that answers the question: Is economic growth being broadly shared?

To be clear, productivity growth is far from irrelevant to middle-class and poor workers. Faster output per hour is necessary (with a caveat I’ll get to next) to generate wage gains, but it is clearly not sufficient.

If you look back at the model, you will see that there’s another potential source of wage growth: Labor’s share of total income can go up, even if productivity plods along at the same pace. The figure below shows this share, which, at about 62 percent, is pretty low right now, relative to both its average since 1959 and even more so, where it was the last time that unemployment was as low as it is now, at the end of the 1990s expansion. That four-percentage-point difference in labor’s share of income between then and now amounts to \$690 billion, or more than \$4,000 per worker.

This, too, like the median/average divergence, is a matter of inequality. Thus, point No. 2 is: If middle- and low-wage workers had more bargaining clout, they’d be able to push for more output flowing to their paychecks as opposed to profits, dividends and share buybacks, all of which are highly concentrated forms of income compared with wages.

Finally, while productivity is growing slowly, it’s still growing, while mid-level, real wage growth has been flat. Since 2016, productivity’s annual growth rate is about 1 percent compared with zero for mid-level hourly wages. Slow productivity can explain slow wage growth. But point No. 3 is that it shouldn’t be invoked to explain no wage growth.

As I’ve written, current wage stagnation is partly a function of high energy prices, and, especially if the jobless rate keeps falling, I expect faster nominal wage growth and perhaps slower top-line inflation. So, we may see mid-level paychecks post gains in line with productivity in coming quarters.

But the question is: Will they be sustained? Tight labor markets help, to be sure. But persistent structural forces — wage inequality; weak worker bargaining power; a deeply ingrained reluctance among powerful employers to share the benefits of productivity growth with middle-wage workers; a hostile politics to workers, unions and labor standards — are all pushing hard the other way. Until there’s a countervailing force, productivity growth and the pay of too many workers will continue to diverge.