If the economy keeps growing, why don’t paychecks feel like it?

Growing on paper, flat in the pocket

The news reports that the economy grew again last quarter, and the sentence lands on most people as a kind of rumour about somewhere else. The figure is real, the growth did happen, and still the household budget feels no looser than it did a year ago, or five, or ten. This is one of the most common suspicions a person can have about the economy: that the good numbers are true for someone, just never for you. It sits behind the sense that the game is rigged, and it deserves a careful answer, because the two easy answers, that it is all a lie or that pay simply stopped rising fifty years ago, are both wrong, and each is wrong in an instructive way.

The fuller answer is that several separate things stand between the growth of the economy and the money in your account, and they compound. Some of them are matters of measurement, where the gap is smaller than it first appears once you compare like with like. One of them is not a measurement artifact at all, and it is the largest: a rising share of what the economy produces has gone to the people already at the top, so that the worker in the middle has genuinely been left behind by the average, even where the average is real. Both parts are true at once, and holding them together is the whole task of this chapter. The place to start is the single chart that sits at the center of how people argue about this.

The chart at the center of the argument

For the three decades after the Second World War, the productivity of the American economy and the pay of a typical worker rose together, almost in lockstep, each roughly doubling. Then, some time around 1979, they came apart. Output per hour of work kept climbing; the pay of the worker in the middle slowed to a crawl. Draw the two lines on one chart and you get a widening pair of scissors, productivity pulling away above, typical pay trailing far below. The Economic Policy Institute, which maintains the best-known version, puts the most recent gap starkly, with net productivity growing several times faster than the pay of the typical worker since 1979. The chart is often offered as proof, on its own, that workers have been robbed of the fruits of their own rising output.

It is a real divergence, and it is also a chart built from a great many choices, each of which can be made differently. What counts as pay, wages alone or wages plus benefits. Which price index converts old dollars into new ones. Whether productivity is measured before or after the wear on machines is subtracted. Whether the pay in question is the average or the median. None of these choices is a trick, and none of them makes the gap vanish, but each moves it, and taking them one at a time is how to see what the chart is really made of. The economists who study it have done exactly this: watch the gap shrink as each adjustment is applied, and then see what is left when the adjustments run out.

Take the wedges in turn, without ranking them, because the point is that they coexist. The first is the price index. Productivity is naturally measured against the prices of everything the economy produces, including machines and exports; pay is naturally measured against the prices consumers actually face. Those two baskets have drifted apart, and a slice of the apparent gap is simply this difference in yardsticks rather than any shortfall in real pay. The second wedge is the distinction between wages and total compensation. A worker’s cash wage grew slower than the cost of employing that worker, because a growing part of compensation went to benefits, chiefly the rising cost of health insurance, as the previous chapter described. Compare productivity to wages alone and the gap looks wider than if you compare it to wages plus benefits. The third is the wear on capital: some of what is produced each year merely replaces machines that wore out, and is not available to pay anyone, so productivity measured net of that depreciation grows a touch slower than the headline figure. None of these three is a swindle, and together they trim the gap appreciably.

Then the adjustments run out, and something large is still standing. Even after the deflator, the benefits, and the depreciation are all accounted for, the pay of the average worker rose far faster than the pay of the median worker, because the average is dragged upward by enormous gains at the very top while the middle barely moves. This is the wedge that is not measurement. In the published decomposition it is the single biggest piece of the gap over the full period, larger than the deflator and the labour-share effects combined. After 2000 the mix shifted, the fall in labour’s share doing slightly more of the work than the mean-median gap, and between them the two inequality wedges came to account for roughly nine-tenths of the divergence, the price wedge for less than a tenth. The productivity was real, in other words, and so was much of the pay it generated; the pay simply went disproportionately to people who were already highly paid, and passed the typical worker by. That is why the chart cannot be dismissed as an artifact, and also why it cannot be read as a simple story of stagnant productivity or of pay that stopped rising. Pay rose. It simply rose fastest for those already paid the most.

What the economists actually disagree about

This is contested terrain, and the disagreement is worth stating precisely, because it is narrower than the shouting suggests. On one side, the Economic Policy Institute economists Josh Bivens and Lawrence Mishel, whose chart the figure above draws on, argue that the surviving gap is real, is driven by rising inequality and the erosion of labour’s bargaining power, and reflects deliberate policy choices rather than any natural law. On the other, Robert Lawrence has argued that once productivity is measured properly, net and at consumer prices, it tracked compensation reasonably well until around 2001, so that much of the headline gap is the measurement wedges rather than a broken link between output and pay. Between them, Anna Stansbury and Lawrence Summers tested whether the link itself had snapped and found that it had not: across the period, each added point of productivity growth still came with something like 0.7 to a full point of growth in the typical worker’s pay. Productivity, on their finding, still lifts pay; other forces have simply been pushing the other way harder.

What all three camps accept is the decomposition itself. They agree the gap can be split into the deflator piece, the labour-share piece, and the inequality piece; they differ on which deserves the emphasis and on what it implies for policy. The figure above takes no side in that argument. It reports the sizes each piece has in the published accounting and leaves the weighing to the reader, which is what is left to do when a literature has settled the arithmetic and not the meaning. The one conclusion the arithmetic forces on everyone is the double one this chapter began with: the famous chart overstates the gap between what the economy produced and what workers were paid, and a large, real remainder survives the overstatement, and that remainder is inequality landing below the median.

Why even the raise that came does not feel like one

Suppose your pay did rise, modestly, over the years. It can still fail to feel like a gain, for two reasons the growth headline never mentions. The first is that the headline is an average, and an average can climb while the middle stalls. One way to see the difference is to watch the mean and the median of household income move apart: when the mean pulls away from the median, the extra income is landing above the middle, so the typical household’s experience diverges from the figure the news reports. The second reason is that some of the prices which rose fastest are the ones you cannot avoid. Shelter and healthcare have outrun the general price level for four decades, and the essentials as a group take the biggest bite from the households with the least, so the same measured income stretches less far the lower down you are.

The two panels are the felt half of the story. The growth headline is an average, and the average has been rising faster than the median because the top pulled it up, which is the same inequality wedge that survived the decomposition of the productivity chart, now showing itself in household income. And the money that did reach the middle is pressed hardest by what a household cannot do without: the essentials take nearly three-quarters of the poorest fifth’s spending, and the price of shelter, the largest of them, has outrun the general price level for four decades, with healthcare climbing faster still. A raise that is eaten by rent and a premium does not register as a raise. So a person can be told, truthfully, that the economy grew and that their own income edged up, and still be right that they do not feel it, because the average was not theirs and the prices that mattered most rose fastest.

Where this leaves the paycheck

The paycheck that does not keep pace with the headline is not an illusion, and it is not a simple theft either. It is the compound of several things: a real gap between output and pay that is partly measurement and largely inequality, an average that has outrun the middle, and a basket of essentials that has outrun the average. The previous chapter explained why the wage is set where it is, by leverage rather than by worth. This one has shown what happens to that wage over time when the leverage runs steadily one way: the economy can grow, the productivity can be real, and the typical worker can still be handed a shrinking share of it. The question that follows is what, if anything, is done about the slack in the labour market that keeps that leverage tilted, and why governments treat a certain amount of unemployment as a price worth paying rather than a failure to be fixed. That is the machinery the next chapter takes up, and it is the frame behind every downturn a worker lives through.

And the long view from inside a household — what pay actually bought at each documented stop from 1900 to today, in the money of the day and in the hours of work it took to earn — is set out era by era in the Ledger’s American-household edition.