Trade unionism — or its absence — profoundly shapes inflation. This fact first dawned on John Maynard Keynes and other leading economists in the 1920s, and it was taken for granted in much of the mainstream analysis of inflation at least through the 1980s. Some faint, residual awareness of the connection is still ambient in the discourse, but by the time COVID-era inflation arrived, understanding of the idea, both in the broad public and among experts, had become murky and confused.

In a labor market dominated by strong collective bargaining, the process of wage-setting differs from that of an atomistic labor market not just quantitatively — in the sense that unions give workers more leverage and help them capture a bigger share of the economy’s proceeds — but qualitatively.

In a nonunion context, individual workers can exert leverage over their employers — at best — via the threat (explicit or implicit) of quitting. When a worker quits, the employer incurs the cost of filling the resulting vacancy: the costs of search, training, and a period of dampened productivity in the interim. To avert a quit, a rational and informed employer may be willing to concede higher wages. But the amount conceded should, in principle, be no greater than the expected cost of filling the vacancy.

In a unionized context, workers exert leverage over employers not by threatening to quit, but by threatening to strike. A strike is like a quit in that labor is withdrawn from the firm in a way that imposes costs on the employer. But that’s where the similarity ends.

Unlike a quit, the aim of a strike is generally to shut down production entirely. And if successful, the cost it imposes on the firm amounts to the entire value of the firm’s output for the duration of the strike. Obviously, any firm that tries to get away with paying “too low” wages risks incurring a vastly greater penalty in a strongly unionized labor market than in an atomized one.

In fact, the penalty is so great, compared to the relatively paltry cost of vacancy-filling, that the real puzzle is why workers in a strongly unionized economy don’t capture an incomparably greater proportion of the output than they actually do, or than their counterparts in a nonunion economy do.

The reason they don’t is that, in a strike, workers have to bear a heavy cost, too: to prevail, they have to be able and willing to hold out against the financial strain of foregone pay longer than the employer can hold out against foregone revenue.

This “worker resistance” variable, though partly a function of objective factors like the size of a union’s strike fund, is also influenced by a welter of subjective factors — political, sociological, psychological — that can generally be filed under the heading of “militancy” or “morale.” Of course, similar psychological factors influence the employer side, as well: probably most firms, for example, would view wage concessions granted under duress to be more costly, in the long run, than concessions offered voluntarily, even when the dollar amounts are the same.

The upshot of all this is that, unlike an atomized labor market, wage-setting in a unionized labor market takes the form of a strategic interaction between employers and workers. As in a war, or interstate rivalry — but unlike an impersonal market — each side decides its next move on the basis of its beliefs about how the other side will respond. This, inevitably, introduces a large measure of indeterminacy to the outcome of unionized wage bargaining.

At first glance, you might think the indeterminacy would at least be kept within bounds by the fact that a given employer can only afford to pay so much: there’s always some hypothetical wage level that is so high that a firm’s profit would fall to zero (or lower) — or its sales would be wiped out from the need to raise prices — if it had to pay it. Barring extraordinary circumstances, no employer, presumably, would ever agree to a wage too close to that level.

But here we get to the crux of the twentieth-century inflation problem: whatever wage level is agreed upon in each industry will go on to affect the terms of wage negotiations in all the other industries. If workers and employers in some initial group of industries settle, for whatever reason, on a relatively high wage level, the overall level of money income in the economy will increase, and that will increase the prices that every other firm can get away with charging for their products without losing sales.

And if the new, higher wage level in the first group of industries also results in higher prices for the products of those industries (as typically happens in such circumstances), the cost of living will rise for all workers, which will strengthen the determination of workers in all other industries to obtain comparable increases in their own wages when their time to bargain comes around.

It doesn’t take much imagination to see how — given the right conditions — even a relatively modest uptick in wage pressure across industries could develop a self-reinforcing dynamic, with higher wage agreements and higher prices in one set of industries causing a jump in wage settlements and prices in other industries, which then feed back on wage dynamics in the first set of industries, and on and on.

Although the now-familiar term for this pattern — “the wage-price spiral” — was coined sometime around 1941, the phenomenon to which it refers was first noted during World War I, when it was given its maiden name: “the vicious spiral,” a term that, as far as I can tell, was first used in a 1916 editorial in the Spectator lamenting a wartime rail strike that had ended in a big pay increase for the railway workers while the country was still at war:

There is a further defect in this method of dealing with the problem which is perhaps even more fundamental. It is this: that it starts a vicious circle, or, shall we say, a vicious spiral, which leads progressively upwards to an ever-increasing aggravation of the difficulty which the original step was intended to cure. If railway-men’s wages are raised, the railway companies must raise rates for goods and passengers. That means an increase of the cost of commodities and an increase in the cost of living for all persons who have to travel backwards and forwards to their work. The same consideration applies, of course, to dockers’ wages and to the wages of carters . . . agricultural labourers and milkmen . . .

A true wage-price spiral is an inherently explosive process. Unless it’s stopped by some outside force — normally monetary tightening engineered or allowed by the central bank — the price level will rise continually, eventually at an accelerating rate.

Joan Robinson forcefully stated the case in her 1937 essay “Full Employment,” a summary and synthesis of the General Theory’s labor market analysis that aimed, with Keynes’s approval, to correct some of the book’s wobblier points. In that essay, Robinson went so far as to write that, absent a countervailing tightening of monetary conditions:

the point of full employment, so far from being an equilibrium resting place, appears to be a precipice over which, once it has reached the edge, the value of money must plunge into a bottomless abyss.

Now, throughout that essay, Robinson simply assumes a unionized labor market. Never in the text does she relax, or even address, the assumption in any explicit way. She takes it for granted — as did Keynes in most of his contemporaneous writings — that the supply of labor is (or might as well be treated as if it were) almost purely and simply a matter of “trade union policy” or “trade union psychology.”

This habit of analyzing inflation without explicitly specifying, or even considering, the range of institutional contexts to which the analysis could apply is the norm in economic discourse. It’s understandable: most of the time institutions change very slowly — at a tempo of decades rather than years — while we are usually most concerned with the problems of the immediate present.

But it’s a habit that can create enormous confusion. Certainly, it’s done so in the COVID-era inflation debate.

It’s in those rare moments when institutions are changing rapidly that the issue is most likely to be forthrightly addressed. Thus in 1944, just as a consensus was crystallizing throughout the capitalist world in favor of a permanent postwar commitment to full employment, we find an apprehensive Keynes writing to a colleague that “a serious problem will arise as to how wages are to be restrained when we have a combination of collective bargaining and full employment.”

If there is one common charge against Keynes that has always been wrong, it’s the claim that he was blind to the risks of inflation. More than most, he saw the potential for the postwar economy to be chronically beset by the plague of rapidly rising prices.