What killed real wage growth?

This is the Project Link chart that most startles me:

What killed the growth in real wages in the early 1970s? I've been trying to come up with an answer to this question, and I think I have one. I'm not entirely sure that it's the correct answer, but I think it's a plausible conjecture.

Of course, an obvious thing for a Canadian economist to do is to check what was going in in the US at the time:

 

Something happened to real wage growth in the US as well an in Canada in the early 1970s. What, though?

The first candidate I thought of for a negative shock that hit both the US and Canada at around that time is the oil shock that followed the October 1973 Yom Kippur War. The timing doesn't work though: US real wages peaked in October 1972, and the inflection point in Canadian real wages occurred in December 1976.

 

So the question now becomes 'Was there something that happened in the US a year before the oil shock, and that also happened in Canada three years after the oil shock?'

If your answer was "wage and price controls", then you are many, many months quicker on the uptake than I am. In both the US and Canada, real wage growth ended about one year after the imposition of wage and price controls. The Anti-Inflation Act was passed in December of 1974, and was in force until the period of decontrol started in April of 1978:

Most of what we remember of the August 1971 Nixon shock revolves around the abandonment of Bretton Woods and its implications for the international monetary system, but the New Economic Policy also included wage and price controls until April of 1974:

In both countries, the turning point arrived a year after the imposition of controls. But if wages are adjusted only once a year, and if the controls only applied for wage increases, then this sort of delay would be expected: it would take a year for the controls to be applied across the entire economy.

So the conjecture is that the imposition of wage an price controls put an end to real wage growth. Let's look at a couple of other countries, and we'll start with Australia. Australian wage data going back to the 1960s are hard to come by: the FRED series for hourly wages only goes back to 1976. After a bit of digging, I managed to unearth a series for Australian men's weekly earnings in a 1992 Australian Bureau of Statistics publication. There's no point in trying to splice men's weekly earnings to hourly wages for all workers, so I rescaled them so that the average over 1977-1986 is 100. The two series have similar properties over the years in which they overlap, so I think they basically tell the same story: real wage growth until 1975, a leveling-off (with a certain amount of volatility) between 1975 and 1985, a decline between 1985 and 1996, and finally a recovery in wage growth after 1996:

 

Australia also had wage and price controls, and the controls-kills-wage-growth conjecture fits this narrative surprisingly well:

You can even see a bit of a spike in the brief interval 1981-82 when there were no systemic controls, and a sharp drop during the wage freeze that was imposed between December 1982 and September 1983. Real wage growth resumed after the newly-elected Howard government let the Prices and Incomes Accord lapse in 1996.

Finally, let's look at Germany, a country that didn't impose wage and price controls:

 

No controls, no structural break in wages.

Weak as this evidence is - we're only looking at four observations from a not-particularly-clean experiment - it offers some support for the hypothesis that the imposition of wage and price controls stops or reverses real wage growth.

At this point, you have to ask yourself "what were they thinking?" From the perspective of 2019, wage and price controls were and are a pointless measure if monetary policy is too loose. But at the time, Milton Friedman's clam that "inflation is always and everywhere a monetary phenomenon" had not yet achieved the consensus it now enjoys. The idea behind 'incomes policies' - the popular euphemism for wage and price controls - was that corporations and unions had become so powerful that they could extract wage and price increases that were in excess of what a competitive market would predict. In the case of wages, the argument was that wages had been increasing faster than what could be justified by increases in worker productivity.

There any number of theoretical problems with using market power to explain inflation, but what did the data say? I'm still researching this episode, but I've yet to find an empirical basis for the conclusion that wages had been increasing faster than what productivity growth would justify. argument. Certainly I  don't see where that claim comes from.

It's perhaps unfair to apply the data we have now to the question, so I decided to look at the data in Historical Statistics of Canada. It's important to remember that in the standard model of the firm, the relevant price for calculating real wages is the output price, not the CPI. This is a point that is still not well-understood, so it's understandable that the analyses of the time used the CPI instead of the output deflator. Understandable, but still wrong.

When you use output prices to calculate real wages, you get a series that tracks productivity quite closely. If anything, real compensation lagged productivity during 1972-74, the data people would have been looking at at the time. It's true that workers' purchasing power - as measured by the CPI - increased faster than productivity over the years leading up to the Anti-Inflation Act, but that was due to an increase in the labour terms of trade. The prices of the goods and services they were producing were growing faster than the consumer goods they were buying.

Incidentally, real wages - as calculated using the output deflator - also tracked productivity in the US:

The difference there was that US workers were seeing a deteriorating labour terms of trade, so purchasing power lagged productivity.

This is a rough first pass at the question, of course: it uses average productivity instead of marginal productivity. This is fine if you made the standard assumption of Cobb-Douglas technology,  but it may be that a more sophisticated treatment might find something else. The point here is that the data support the vanilla model of competitive labour markets; there's no obvious evidence of workers using their bargaining power to extract increasingly larger rents.

I'm not sure why I haven't come across the link between wage and price controls and a structural break in real wage growth. The only reference I've come across that makes the connection is here, and Google Scholar says it has been only cited once, in a thesis submitted to the Federal University of Rio de Janiero. I suppose the obvious reason is that reducing real wages was a stated goal of wage and price controls - after all, the starting point was a claim that wages were 'too high'. Supporters of controls would greet such a finding with satisfaction. And since the most vocal opponents to controls were monetarists (who, with no small amount of justification, thought controls were as dumb as a sack of hammers) and labour unions (who were, with no small amount of justification, outraged at the adoption of a policy designed to reduce workers' buying power), it was easy for supporters of controls to tell themselves that theirs was a prudent, centrist option.

What I still don't understand is how the effect of controls on real wages could have persisted for so long after they removed. There's some sort of hysteresis story there, and it remains to be told.

 

 

 

 

 

Comments

[Julien]

What about Germany's stronger unions? Could they have resisted wage and price control and at the same time haulted the decline in real wages?

[Julien]

In any case, I suspect a confounding variable. Hysteresis seems like a reach for a such a short run policy.

[tkuipers]

Would significant increases in non-wage benefits and things like payroll taxes contribute? UI rates more than doubled in 10 years from 1973 to 1983 (Lin (2001) - https://www150.statcan.gc.ca/n1/en/pub/11f0019m/11f0019m2001149-eng.pdf). CPP contribution rates almost doubled from the 80s to the 90s and have continued to rise. QHSF rates increased by a factor of 7 in the 25 years from 1970. Somewhere in there (best guess, anecdotally, without stats the early-80s) health, savings matching, and other similar benefits began to rise. Lin (2001) notes that payroll tax rates, in aggregate, started to flatten out in the 90s, which might explain the rise in wages until it was arrested by the 2000 crash. Just guesses.

PS, thanks very much for the reminder of the Project Link data. It's pretty fabulous.

[Nick Rowe]

Great post!

I'm not convinced though. OK, it's not implausible that wage & price controls might have reduced real wage growth. But when the controls were lifted, wouldn't real wages (or real wage growth rates) soon return to normal? Why would temporary controls have such a long-lasting effect?

But I don't have a better explanation. (I once thought it might be married women's labour supply shift, but IIRC the timing just doesn't work for that story.)

Minor theoretical point: consider for example a simple model with equal degrees of monopoly power in both labour and product markets. In equilibrium, both W and P are "too high" relative to AD (or Ms), so U* is too high, and Y* is too low, but W/P is exactly the same as it would be in a competitive model, because both W and P are equally too high.

[Chris]

Is it possible wage and price controls precipitated a shift to in-kind compensation? I think this conversation typically centers around the supposed decoupling of wages from productivity, but that relation reappears (in the US) when looking at total compensation! (https://www.aei.org/wp-content/uploads/2019/02/The-Link-Between-Wages-and-Productivity-is-Strong.pdf)

[Peter Dorman]

(1) Wage and price controls are not exogenous. (2) From a comparative institutional perspective, three English-speaking economies and one German one are less than four independent observations; the US, Canada and Australia are clustered. (3) Causal explanation needs a mechanism, not only because it does, but also because you can test directly for its presence and operation.

[Philip J Cowan]

Hysteresis: great word. Wage inflation got a cold and was never the same again.

[Nick Rowe]

Stuck in the spam folder, Phil Koop says:

"Most of what we remember of the August 1971 Nixon shock revolves around the abandonment of Bretton Woods and its implications for the international monetary system, but the New Economic Policy also included wage and price controls until April of 1974 As it happens, David Glasner recently reposted his anniversary post on Nixon's imposition of wage and price controls. He puts the case a little stronger than you do:Interestingly, the current (August 13th, 2011) Economist (Buttonwood column) and Forbes (Charles Kadlec op-ed) and today’s Wall Street Journal (Lewis Lehrman op-ed) mark the anniversary with critical commentaries on Nixon’s action ruefully focusing on the baleful consequences of breaking the link to gold, while barely mentioning the 90-day freeze that became the prelude to the comprehensive wage and price controls imposed after the freeze expired. Of the two events, the wage and price freeze and subsequent controls had by far the more adverse consequences, the closing of the gold window merely ratifying the demise of a gold standard that long since had ceased to function as it had for much of the 19th and early 20th centuries. In contrast to the final break with gold, no economic necessity or even a coherent economic argument on the merits lay behind the decision to impose a wage and price freeze, notwithstanding the ex-post rationalizations offered by Nixon’s economic advisers, including such estimable figures as Herbert Stein, Paul McKracken, and George Schultz, who surely knew better, but somehow were persuaded to fall into line behind a policy of massive, breathtaking, intervention into private market transactions. Glasner doesn't make the link you do between wage and price controls and a structural break in real wage growth though. Why should the apparent effects be so long-lived? One possibility is that the imposition of controls mark a downward shift in market power of labour. It is doubtful, though, that controls were the cause of this shift. It is more plausible that they were instead the effect of a common cause, that being a downward shift in labour's political power."

The above was by Phil Koop.

[john]

Don't your charts then show the real curiosity is the kink in productivity?

[Phil Koop]

Thanks Nick. I see that my ratio of block-quoting to typing triggered your spam filter. I'll remember next time.

[Farid Elwailly]

"There's some sort of hysteresis story there, and it remains to be told."

Wage controls may have enabled the beginning of the crusade to neuter unions. Workers in general realize that "large" wage demands may lead to a new round of wage controls and find they don't have a desire to fight for the large increases going forward. This diminishes the value of unions.

[Lord]

Or a kink in inflation measurement. Could it have been significantly underestimated earlier? Seems unlikely without a change, whether in measurement or what is measured or whose is measured.

[Harlan Green]

"There's some sort of hysteresis story there, and it remains to be told."..Really?? How about establishment of the Business Round Table (BRT) in the 1970s? It was originally set up to oppose outsize construction union power, but morphed into corporate opposition to trade unions in general, first in public view with PATCO strike that caused Prez Reagan to fire Air Traffic Controllers. That, and panic over soaring inflation gave BRT ammunition it needed to espouse Paul Volker and Fed policy of making 2% inflation target; and therefore raise interest rates every time wages began to creep over 2%! Workers' wages became the bogeyman for inflation, in other words, hence the "Great Moderation" intoned by Krugman and others. Fed policy really became a 'Great Moderation' in wage increases, and wholesale transfer of wealth to the renters.

[L Mimas]

Automation. There was middle managment in many industries. I am most familiar with warehouse middle management and the nice 1950s-era apartments still good for Gen-Z. It has produced some good effects as there are fewer line-ups and more spare time to learn stuff. Selkirk taught me the Scottish were shock troops who received subsidized housing by London. BC had Keynesian spending and won more war medals than MB in WWII.
For this reason if wages are stagnant even after the Cold War is over, I suggest the ability to learn a living be made up for in some other way. Subsidized Letters of Permission courses and a fifth of a degree as a hiring credential into some hard to crack occupation is one possibility. A tenth of the Yang GAI would be enough to start up a Rendering Engine biz or a neuro-imaging sensor biz with microfinance.
My quick math got me a job until a Cup Final that is now not there due to better software. Think back to bank line-ups. I would trade wage growth for a candidate that wants to export his tech problem solutions.

[dilbert dogbert]

Thanks for the reminder of Nixon's w&p controls. My brother in law was affected. His company just reclassified some personnel to a higher wage. Presto no problem.

[David Martin]

Nicely represented! Great use of maps. Good Work!