Monday, August 7, 2023

Blinder, supply shocks, and nominal anchors

An a recent WSJ oped (which I will post here when 30 days have passed), I criticized the "supply shock" theory of our current inflation. Alan Blinder responds in WSJ letters 

First, Mr. Cochrane claims, the supply-shock theory is about relative prices (that’s true), and that a rise in some relative price (e.g., energy) “can’t make the price of everything go up.” This is an old argument that monetarists started making a half-century ago, when the energy and food shocks struck. It has been debunked early and often. All that needs to happen is that when energy-related prices rise, many other prices, being sticky downward, don’t fall. That is what happened in the 1970s, 1980s and 2020s.

Second, Mr. Cochrane claims, the supply-shock theory “predicts that the price level, not the inflation rate, will return to where it came from—that any inflation should be followed by a period of deflation.” No. Not unless the prices of the goods afflicted by supply shocks return to the status quo ante and persistent inflation doesn’t creep into other prices. Neither has happened in this episode.

When economists disagree about fairly basic propositions, there must be an unstated assumption about which they disagree. If we figure out what it is, we can think more productively about who is right. 

I think the answer here is simple: To Blinder there is no "nominal anchor." In my analysis, there is. This is a question about which one can honorably disagree. (WSJ opeds have a hard word limit, so I did not have room for nuance on this issue.) 

Suppose there are two goods, TVs and hamburgers. Chip production problems make TVs temporarily scarce.  We agree, the relative price of TVs must rise. TVs could go up, hamburgers could go down, or half of both. When the chip shortage eases, the relative price goes back to normal. TVs could come down, or hamburgers go up. 

In Blinder's view, TV prices go up now, and hamburger prices catch up when the chip shortage eases. ("Inflation creeps in to other prices.") The reason, true enough, is that prices are stickier downward than upward. 

As Blinder writes, a half century ago monetarists started making an "old argument" against this analysis. Informally, they pointed out that people have to have enough money to buy TVs and hamburgers at higher prices. If they don't, the price level cannot permanently rise. Maybe TVs go up temporarily, but if people don't have enough money to pay the higher prices, those downwardly sticky prices eventually drift down. 

A bit more formally, monetarists start with MV=PY, money times velocity equals overall price level (TVs and hamburgers) times overall quantity. Without more M, PY cannot go up. In the short run with sticky downward prices, we'll have less Y. But lower demand for hamburgers eventually causes sellers to lower their prices, and when the chip shortage eases lower demand also brings back the price of TVs. Without more M there is, eventually, no more P. 

So to monetarists, a "supply shock" would indeed first raise the price of TVs, and thus the measured price level. But lower demand for TVs and hamburgers means the price level eventually comes back to where it was, as I had stated. 

MV=PY is a "nominal anchor." It is a force that determines the overall price level. Relative prices changes cannot change this overall price level, once we get past the price stickiness that sends some demand into output. 

My analysis also has a nominal anchor, fiscal theory, that nails down the overall price level, while allowing some stickiness in the short run. The difference between fiscal theory and monetarism or some other theory does not matter for today's discussion. (Exercise for the reader: do new-keynesian models have a nominal anchor, and if so what is it?) 

Now, how does Blinder avoid this logic? Simple. In his view there is no nominal anchor

There are two kinds of economic analysis that removes the nominal anchor. First, the nominal anchor simply may be absent. Standard 1970s ISLM analysis, the subject of an eloquent elegy (eulogy?) in Blinder's recent book, does not have a nominal anchor.  The price level today is whatever the price level was yesterday plus whatever supply and demand shocks move it around today. The MV=PY constraint on the overall price level is simply absent. Given Blinder's writing in favor of ISLM and against everything that has happened since, I think this is the basic disagreement in our analysis. 

So you have it: If you think there is a nominal anchor, then you're with me: Supply shocks without more demand (more M, more debt), cannot permanently raise the price level. If you think there is no nominal anchor, and prices are whatever they were yesterday plus shocks, then Blinder's analysis that supply shocks can set off inflation that does not reverse, and permanently raise the price level, is possible. 

A second possibility: There is a nominal anchor, but it is passive. In a monetarist analysis, a supply shock comes, that raises the price of TVs because hamburgers are sticky downwards. The Fed, not wishing to see a recession, "accommodates" this supply shock by printing more M, so the price level goes up. And in monetarist analysis an interest rate target automatically provides that sort of accommodation, which is why they don't like interest rate targets. 

This is a standard analysis of the 1970s supply shocks. They didn't cause inflation directly, but they did induce the Fed to accommodate, to print up more money, which allowed broad-based inflation. Indeed, that might be a good fiscal theory story for recent inflation, merging supply shocks and fiscal theory. Why did the  government send $5 trillion to people? Exactly so they could pay the higher prices, and no price had to fall. It worked like gangbusters. Facing the energy shortages after the Ukraine war, Europe subsidized demand with fiscal transfers to much the same effect. 

I don't think Blinder is making this argument, though. If that's the argument, the "supply shock" really still  isn't the important driver of inflation. The inflation would not happen without the Fed (or fiscal policy) giving people the money to pay the higher prices. It's a supply-induced demand shock. The emergence of inflation is still 100% tied to loose monetary and fiscal policy. It's just making a plea that this inflationary policy was a good idea, avoiding painful price declines in some goods and wages. 

A minor rhetorical complaint: "This is an old argument that monetarists started making a half-century ago," Old arguments aren't necessarily bad. Adam Smith made the argument that free trade is good 250 years ago!  And if "old" is an insult, I might point out that ISLM hasn't been publishable since about 1980, while "new" Keynesian models, which work in a completely different way, have been the rage. 


I emailed Alan, who answered in part, 

The Fed’s reaction function is the nominal anchor, but for the inflation rate more than the price level.

This is an interesting and important comment, and I think it is representative of the standard view. It's  halfway between the two views I articulated. There is no natural nominal anchor – no M in MV=PY or no B in B/P = expected present value of surpluses (fiscal theory alternative) – that constrains the price level. The price level is wherever the Fed drives it by interest rate policy (the "standard view" of "interest rates and inflation part 1" But for a supply shock to cause inflation in this view, there must be monetary accommodation. The Fed could run a price level target. It  would raise interest rates to keep the price level from rising, or to bring it back after a shock. The Fed could run an inflation target. It would raise interest rates to keep the price level from rising. So inflation does not really emerge as a pure “supply shock.” It is supply shock + demand accommodation, via the interest rate policy.  

 Alan added 

Back when OPEC I struck in 1973-74, many monetarists denied that it could possibly cause inflation without a rise in M.

The argument has indeed been around for a while. Did M rise? I'll leave monetarists to the history on that one. 

To be clear in this sequence, I used MV=PY as an easy illustrative example. I think there is a nominal anchor, but it is fiscal theory not monetarism. And I acknowledge that Blinder's "standard view" is most prevalent in policy commentary.  


A correspondent reminds me of basic IS/LM AS/AD. Do not confuse a "supply shock" -- a relative price movement -- with an "aggregate supply shock" -- a shock that makes the overall price level, and wages, rise for a given level of output. The latter is a Phillips curve phenomenon. They may be related in the depths of Keynesian thinking, but they are not the same, and it is a bit of a mistake to jump from one to another.  


  1. Couldn't this debate be easily resolved just by opening FRED and observing that GDP correlates very well with the GDP deflator going back all the way to 1947, which means it a demand shock caused by monetary policy.

  2. The nominal anchor in NK models would be the Gov't budget constraint?

    1. In my version of the NK model, yes, though I call it the "government debt valuation equation." price = present value of dividends is not a "budget constraint." In standard NK models it is the equilibrium-selection policy of the central bank which promises hyperinflation or deflation to select one of many multiple equilibria.

    2. If the ‘debt valuation equation’ provides the nominal anchor’, what has produced the shift from the low and stable to the high and variable inflation regime? Have the markets ceased to believe that governments will ‘repay’ their debt? Should this apply to all governments in the industrial world? If not, should not we expect persistent inflation divergences?

    3. well actually, if we want to be a bit pedantic price=PV of dividends is a budget constraint. The equity is issued to fund investment spending and it's the PV of future dividends that constrains how much can be raised and spent.

      I like to the think of the FTPL as simply saying the government has no nominal budget constraint but it does have a real one. For a company issuing equity it's basically the same, there's no constraint on how many shares they can issue, there is on how much the total issuance is actually worth.

    4. Price = PV of dividends, like real value of nominal debt = PV of surpluses is NOT a "budget constraint." This little bit of verbal pickiness is very important, and has caused endless confusion. A budget constraint must hold for all prices, and out of equilibrium. When the "auctioneer" calls out a price vector, you respond with your supply/demand, and that supply demand must satisfy the budget constraint. Prices do not adjust to make budget constraints hold. To say that the price in Price = PV of dividends or real value of debt = PV of surpluses adjusts to make an intertemporal budget constraint hold is nonsense. (It would be novel economics if true, as claimed by some adherents and as criticized rightly by critics.) Both equations are equilibrium conditions. They include budget constraints, but also include supply=demand conditions. I call them "valuation formulas." Call them what you want, but do not call them budget constraints.

  3. The conversation would be more productive if we had a micro-founded model behind the idea that inflation results from the interplay of "aggregate demand" and "aggregate supply". Is there any? (adaptive expectations + passive fiscal/monetary doesn't seem to be what they have in mind, like JC said)

  4. Geez, IS-LM analysis takes me way back in time to Gordon Macro book and Chandler Money and Banking text...
    I subscribe based on a reading of history that all inflations come from government so fiscal theory resonates a bit, but unclear how strong it really is given the year after year of US deficits and enormous and still growing US debt picture and yet markets have not yet pushed back in a meaningful other than via the $5T-$6T spend monetized by the Fed giving people lots of spending power(demand) vs the supply of goods out there(supply shock or not) and hence inflation.

  5. Can we simplify a little and explicify a little?

    For MV = PY, with V fixed à la monetarists, a supply shock is a reduction in Y, real output. Given MV, P must rise.

    In IS-LM, real money supply must equal real money demand, but that's now a function of the nominal interest rate and real income, M/P = m(i,Y). Here, too, Y falls on account of the supply shock. So, the demand for money falls. Thus, i must fall to increase the demand for money and/or P must rise to reduce the real money supply.

    I think all the policy conclusions follow.

    Ah, those were the days! :-)

  6. The correlation between GDP and the deflator is negative at business cycle frequencies from 1948-2022, and has been shown to be unstable across subperiods. It doesn't convey any information at all.

  7. I'm of the belief the Lucas Island model/ thought experiment concretely teases out the problem when inflation surprises rational economic agents.

  8. Re: Exercise

    There is not necessarily a nominal anchor in a New Keynesian Model. A good example would be a zero net debt equilibrium. Such a model would have an indeterminate price level.

    A good way to understand whether inflation is coming from a supply or demand shock. Holding monetary policy constant do real interest rates rise or fall? Assuming prices are sticky, in a supply shock real interest rates rise above the steady state as agents attempt to smooth consumption. In a fiscal policy induced demand shock real interest rates are pushed lower than the stead state as inflation expectations rise to revalue government debt. A chart of 5y TIPS yields will lead you to a clear conclusion.

  9. This comment has been removed by the author.

  10. (Exercise for the reader: do new-keynesian models have a nominal anchor, and if so what is it?) --JHC.

    1) The basic two-equation new-Keynesian DSGE model.
    This nK-DSGE model lacks an explicit nominal anchor. The nominal interest rate is given ( i = i(t) ). The current period consumption ( x(t) ) and the current period inflation rate ( pi(t) ) are chosen by the representative household and the representative producer firm, respectively, on the basis of expected consumption and production in the next period. The implicit nominal anchors are 'habitual consumption across time periods' (rep. hshld.) and 'stable profit growth rates' (rep. producer). Monetary policy is a given.

    2) The basic three-equation new-Keynesian DSGE model.
    This n-K DSGE model has an explicit nominal anchor. The third equation in this model is usu. either the Type-1 Taylor Equation, or the Type-2 Taylor Equation. In either case, the explicit nominal anchor is the target rate of inflation ( pi*(t) ).

    3) The basic FTPL-adapted new-Keynesian DSGE model.
    This model will have one of two explicit nominal anchors, depending on the whether the fiscal authority pursues a 'Ricardian' or 'non-Ricardian' fiscal (budget) policy. If the fiscal authority pursues a 'Ricardian' budgetary policy, then the central bank (monetary authority) will pursue an inflation target policy via aTaylor Rule, or other similar variant, where the overnight interest rate is manipulated to effect conformance to the target inflation rate, i(t).

    If the fiscal authority pursues a 'non-Ricardian' fiscal budget policy, i.e., budget deficits and borrowing to fund current expenditures without an explicit nominal anchor (e.g., a politically effective binding balanced-budget commitment) the central bank (monetary policy authority) is either passive (monetizes the government debt) or, if active, then its explicit nominal anchor is ineffective (be it an interest rate target, inflation rate target, or a foreign exchange rate target) in the face of the fiscal authority's budget deficit funded expenditures.

    Alan Blinder's response to your query, 'the nominal anchor is the central bank's response function', is a 'marshmallow'. The nominal anchors that fit 'the central bank's response function' could be (1) the foreign exchange rate target, (2) the money aggregate (M1, M2, base money, etc.) target, (3) the interest rate target, (4) the inflation rate target, (4) the unemployment rate target, etc. Take your pick.

    1. During the decade before 1965 the annual compounded rate of increase in our means-of-payment money supply was about 2 per cent. During the same period, the annual transactions velocity of money increased from c. 21 to 31. In ten in-year period since 1964, the money stock grew at an annual compounded rate of c. 6.5 percent and the transactions velocity of money reached an average level of 70 in 1973.
      Velocity continued to increase to over 80 during 1974. Both money and velocity figures were taken directly from the Federal Reserve Bulletin. The impact of both an accelerated increase in the volume and velocity of money on prices is made even more evident if the rate of increase in aggregate monetary demand (money time’s velocity) is examined.
      During the decade ending in 1964, money flows increased at an annual compounded rate of about 6 percent. In the nine years since 1964, the increase was more than 13 per cent, and in 1972-73, nearly 30 percent. Because R-gDp and presumably, the volume of goods and services offered in the markets, was increasing at a rate of less than 5 per cent, it should have been no surprise that there was an intensification of our chronic rates of inflation to devastating levels.

    2. John, constant velocity seems to be assumed implicitly but without justification

    3. What do you assume about acceleration?

  11. "... a supply shock comes, that raises the price of TVs because hamburgers are sticky downwards. The Fed, not wishing to see a recession, "accommodates" this supply shock by printing more M, so the price level goes up."

    "...that might be a good fiscal theory story for recent inflation, merging supply shocks and fiscal theory. Why did the government send $5 trillion to people? Exactly so they could pay the higher prices, and no price had to fall."

    Putting these quotes, and actually the whole couple paragraphs I took them from, together would appear to suggest the choice was inflation or recession, you couldn't avoid one of the two possible outcomes (or maybe taken some of each). Which option is worse is certainly open for debate and opinions can differ.

    However, we should be careful not to sound as though we're saying policy made any sort of obvious mistake in causing/allowing the inflation. In this case I think that's certainly not true. If the choice is between recession or a permanently higher price level choosing the later is a reasonable choice.

  12. How about Y going down and P going up in equation MV=PY ? In case of goods being necessities and/or market entry frictions due to regulations (monopolies in an extreme case) and/or prices of raw commodities going up across their wide spectrum. Potato price during the Great Famine in Ireland is a classical (and tragic) example.

  13. Explain how you can use statistics when the FED covers its elephant tracks?

  14. The increase in housing price is misleading. It is the result of the way house purchases are financed, namely long term mortgages. When interest rates rise dramatically it effectsboth supply and demand. Because owners are locked in to good mortgages the supply is constrained. Ask a real estate agent and they will tell you that they have lots of potential buyers, but few houses on the market. Contrast this to England where most mortgages are short term. Their demand dominates and prices are falling. Vic Goldberg


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.