WARNING: if you are not a macroeconomist you may not understand this post, even if you think you do. This is especially true if you are not a macroeconomist but think you know something about "political economy". (When I hear the words "political economy" I usually reach for my shovel.*) This post is an experiment. You, the reader, may be part of that experiment.
This post is partly about mutual incomprehension between macroeconomists and non-economists. But it's written mainly for an audience of macroeconomists. So stop and think before you come out swinging wildly in the comments.
This post is in two (OK, three) parts. The second part is about likely reactions to the first part, and is the main part of this post.
1. Consider this policy proposal:
"I think the Bank of Canada should switch from targeting CPI inflation to targeting wage inflation. I'm not hung up on the exact rate of wage inflation the Bank should target. My guess is that something like 2.5% wage inflation would be roughly right, and would give us roughly the same 2% CPI inflation in future. But if you want to argue for a higher or lower target rate of wage inflation, I don't really care a lot. So if wages start to increase faster/slower than 2.5%, the Bank of Canada should raise/lower interest rates, reducing/increasing demand for goods and labour, which would put downward/upward pressure on wage increases."
(BTW, I'm not actually proposing that, though it's not a bad policy, and is worth considering. And the merits or demerits of that proposal is not the point of this post.)
2. Reactions.
2a Macroeconomists. Any New Keynesian (for example) macroeconomist would react to the above policy proposal like this:
"Ho hum. Nothing new here. Nick hasn't even given us any reasons why targeting wage inflation would be better than targeting CPI inflation. I could build a model where one would be better in some cases, and the other would be better in other cases. It all depends on: whether wages are stickier than prices; on the source of the shocks; the exact specification of the model and its parameter values; and stuff like that. It might matter in the short run, but won't matter much if at all in the long run (unless better performance in the face of short run shocks leads to a higher growth rate).
Is Nick's policy "anti-labour"?? Don't be silly. Of course not. You would get exactly the same average levels of real wages and employment with either target, unless there are some sort of weird non-linearities in the model, or if better short run performance in the face of short run shocks leads to a higher growth rate. It's got nothing to do with being pro- or anti-labour."
2b. Non-economists. I'm guessing here:
"Nick's policy is anti-labour. He doesn't want workers' wages to increase. If he sees workers getting a better deal, he wants the Bank of Canada to tighten monetary policy to increase unemployment to weaken their bargaining power to force wages down again. Nick's policy isn't just anti-labour, it's deeply immoral in using unemployment to fight against workers' interests."
Roughly right?
2c. This one's trickier. Suppose you were a labour union leader who also believed in (something like) New Keynesian macroeconomics. In other words, you had the macroeconomics of (say) Paul Krugman, but your job was to run a labour union.
Even if you thought, as a good New Keynesian macroeconomist, that the policy might either be better or worse than inflation targeting, depending on the precise model, you would be very wary of the policy.
First, a lot of your members would be against it, for the reasons given in 2b above.
Second, if the wage targeting policy were implemented, and even if you knew, as a good New Keynesian macroeconomist, that it would make little or no difference to real wages and employment, which is what you care about, and that it made no difference to whether unions are good or bad, it might make a massive difference to how your job was perceived.
Suppose you managed to get an above-target 3.5% wage increase for your members. Everyone might then say: "Oh, that's great for your group of workers, but it simply means that some other group of workers are going to have to accept a 1.5% wage increase, to keep the average at 2.5%, and the Bank of Canada is going to have to make sure there's enough unemployment to force some other group of workers to only get 1.5%!".
And you would be left desperately trying to argue: "No, it doesn't mean that, because our wage increase could mean that average real wages would go up, even though nominal wages can't, because prices could come down as a result of our wages going up…..because,….er,…, well , it's all to do with the long-run neutrality of money, and the difference between real and nominal wages, and…..err….let me try to explain to you how this works…..brothers and sisters."
Would you want your job to be perceived like that? Would you like your job to be to have to try to explain that an increase in your wages might mean a fall in prices and an increase in average real wages even though average nominal wages were fixed by the Bank of Canada? Even though your argument for the benefits of union power would be equally valid (or equally invalid) whether the Bank were targeting CPI or nominal wage inflation?
Of course you wouldn't. I wouldn't wish that job on my worst enemy. Instead, you would say "Oh, to hell with trying to explain all that in New Keynesian macro, let's just blame the Bank of Canada for not letting us increase wages without creating unemployment! I can explain that to my members."
3. Coda. Back to macroeconomists: given the likely plausible reactions in 2b and especially 2c above, I wonder if your 2a is the correct answer?
* "Political economy" has two (modern) meanings. It can mean political scientists bullshitting about economics. Or it can mean economists bullshitting about politics. Economists tend to use the words in the second sense. Non-economists tend to use the words in the first sense.
The original is actually the rather clever: "when I hear of culture… I unlock (entsichere) my Browning". Browning being both a poet and a make of gun.
Worthwhile Canadian Initiative 
Nick — Ceteris paribus, if the cost of conflict is greater for one party than for another, the party with the lower cost of conflict is more likely to get its way. But ceteris is not always paribus. Maybe the party that bears the higher cost from one round of conflict also has more to gain or less to lose long-term than the other party, and so endures. Maybe the conflict is sufficiently costly that neither party engages, so the relative cost of conflict is irrelevant. Maybe one party is ideological or irrational, willing to make sacrifices what we would impute as large costs because their cause is divine or they are spiteful or simply nuts.
A central bank that tries to employ a “grim trigger” strategy by imposing recessions on restive labor is not preordained to win. I’m certainly not claiming central bank omnipotence in this or any other matter. But, all else equal, the ability to credibly threaten a recession to parties who stand to lose more than you do is a powerful bargaining chip.
Again, the story of declining labor share in the US is likely more due to globalization and technology than any central bank strategy. But expectations about central bank reactions do affect and condition the behavior, of ordinary people and of policymakers. Magnitudes are uncertain. Perhaps there was no material influence. But the direction of the effect at the margin was pretty visible, and engendered a fair amount of hostility.
Steve Randy Waldman has put the essence of my criticism of NGDP targeting rather well and I think I’ll leave it to him.
I will however disagree with this, or rather, the conclusions that follow from it:
Central banks can’t affect labour’s share in the long run, precisely because it is a real variable. They can only affect the variance of labour’s share.
This is far from a trivial issue. Among other things, I would expect that a high-variance economy would exhibit a very high savings rate and depressed aggregate demand as people self-insure against the inherent uncertainty. I wonder if that is actually rather what is going on in places like China?
Alex: I was wondering about that myself, if you and Steve might be on about the same thing.
“This [variance] is far from a trivial issue.”
Indeed. This to me, and to most macroeconomists, is a very big issue indeed. It’s why we argue about inflation targeting vs NGDP targeting etc. Which one best stabilises the economy? Which one minimises the variance? (Though we have to be careful how precisely we define “variance”, because some types of variance may be a good thing.)
Steve: I think the model you have in mind is where there’s One Big Union in a game against the Central Bank. 2 player game. And we don’t know whether they will play Nash, or whether the OBU or the CB will be the Stackelberg leader. And a switch from (say) inflation targeting to nominal wage targeting (or NGDP targeting, or whatever) is like a change in the strategy space and can affect the equilibrium.
People have modeled a 2 person game between the CB and the fiscal authority. (Did it myself once, but I can’t immediately recall the model.)
There’s also that literature on “Corporatism”, (Dennis Snower?) where you sometimes get better results with OBU than with lots of little unions. And examples (like Ireland or the UK) where you try to get an agreement between the OBU and the government in which the OBU keeps wage inflation low while the government increases AD. Those models never really made it across the Atlantic (though I did one once) because unions were never that centralised here. They died out in the 1990’s AFAIK.
Alex: “I wonder if that is actually rather what is going on in places like China?”
I don’t think so. More likely that people in China are saving to insure against individual specific shocks (with no social insurance). Or against political risk, rather than AD shocks per se, which might be small in comparison.
Steve and Alex may also be referring to the empirical evidence regarding how central banks actually implement their stated target, e.g. as in http://utip.gov.utexas.edu/papers/utip_42.pdf
And examples (like Ireland or the UK) where you try to get an agreement between the OBU and the government in which the OBU keeps wage inflation low while the government increases AD
Germany in the GFC. Lohnzuruckhaltung in full effect, jobs protected via short-time working with partial compensation, both agreed through the tripartite system, and a substantial fiscal stimulus (the biggest increase in public debt as a % of GDP for discretionary stimulus, according to the IMF). Seemed to work pretty well, although hard to distinguish good structure from good policy.
2 player game, between OBU, which sets W, and CB, which sets M. Both have preferences defined over W and M.
1. Nash equilibrium with simultaneous moves: just like Kydland and Prescott, Barro and Gordon. High inflation natural rate equilibrium.
2. OBU moves first, and CB moves second. Corporatist solution, because OBU knows that the LRPC is vertical, so it keeps W low. Unemployment at natural rate, low inflation. “Monetary policy by the trades Unions”.
3. Suppose the CB can make binding conditional commitments. CB announces a threat for M, conditional on W. The CB can reward or punish OBU behaviour. The CB can get the unemployment rate below the natural rate, in equilibrium?
aelilia: I briefly skimmed that Jamie Galbraith article. Here’s my take.
Just assume that his VAR results are correct. OK, so the Fed has a rather weird kinked Taylor Rule. It tightens in response to low unemployment, and loosens in response to low inflation, rather than responding symmetrically to both.
If you put that instrument rule into a standard NK model, it would (almost certainly) not be optimal. It would only be optimal (I guess) in some equally weird kinked NK model. (But maybe the real world is kinked, or at least non-linear, so a non-linear instrument rule would be best).
It would affect the variance of unemployment, but not affect the average rate of unemployment (if you put that kinked Taylor Rule into a standard linear NK model).
That’s why central bank presidents, boards, etc, need to be popularly elected.
When I read this two weeks ago I thought there was something missing, but I couldn’t quite see what it was. I was just reminded to come back and look again by @stephenfgordon and @mattyglesias on twitter.
I’ve realised that I didn’t instinctively see the proposal as being anti-labour, or even as looking anti-labour to voters. A more sellable (or in Nick’s term political-economy-compatible) proposal might be the following.
Target a wage increase of not 2.5% but 4%.
Based on expected productivity increase of 1.5-2% per year, this should be long-term equivalent to a CPI increase of 2-2.5%. Make sure CPI figures are still published so that people can see the difference between the two.
Sell it to voters as a real-terms annual increase of 1.5-2% (plus whatever you get from being promoted or gaining more experience or seniority in the workplace – which every voter, being overconfident, will expect to happen disproportionately to them personally)
As Nick hinted, it might seem to a macroeconomist like this is effectively equivalent to his proposal, but I think to a voter it’s not. The 4% increase is better compared to pre-existing anchors than 2.5%. And a focus on productivity and showing the difference between CPI and wages would help to make the idea more popular.
Am I wrong on the productivity figures? In the long run will it matter what the actual number is, once this expectation gets baked in? I’m not sure, but I think this proposal would be more politically achievable than Nick’s, just because of that change in number.