This archived content originally appeared at, the predecessor site to, and does not carry the sponsorship of the Cato Institute.

Scott Sumner Has Got Me all Confused

To be specific, he confuses me by his response to Ezra Klein's column quoting former FRB vice chairman Donald Kohn's opinion that “It’s difficult, if not impossible, to create persistent inflation without demand exceeding potential supply over an extended period” along with Lawrence Summers' view that "inflation is mostly driven by demand, and when you increase demand, you increase inflation. And if you don’t increase demand, you don’t increase inflation. But if you’ve solved demand, you’ve solved your problem.” To the question implicit in such remarks, regarding why anyone would expect the Fed to succeed in raising the current rate of inflation despite being incapable of increasing the current growth rate of aggregate demand, Scott's response is "Um, maybe because the Fed promised a more expansionary policy in the future?"

I realize that Scott wants his readers to imagine Kohn and Summers, upon hearing this reply, smacking their lower palms against their foreheads while saying "Dope!" But at the risk of appearing to be a dope myself, I confess that I believe that the statements by Kohn and Summers make perfect sense, whereas Sumner's rhetorical response does not.

It doesn't make sense, first of all, because, assuming a non-declining AS schedule, an increase in the rate at which prices increase that is not accompanied by a corresponding increase in aggregate spending must be accompanied by ever-declining sales and ever-worsening unemployment, and is for that reason unlikely to persist. So it would seem to be true after all that a persistent increase in the rate of inflation requires a persistent increase in the growth rate of aggregate demand relative to that of aggregate supply.

Of course I don't doubt for a moment that Scott understands this last point perfectly well. Unless I'm mistaken, what bothers him about the opinions expressed by Kohn and Summers is their implicit failure to appreciate the possibility that, when it raises its announced inflation target, the Fed also increases the expected rate of inflation, and with it the velocity of money. Consequently the Fed is able to boost aggregate demand and to combat recession without resorting solely to the usual, more direct but less reliable means of more aggressive monetary expansion, and its higher announced inflation target becomes in this respect at least partly self-fulfilling.

But even this more sophisticated objection to Kohn and Summers, understood (as I understand it) to imply that those experts have overlooked a potentially effective means for combating recession, seems wrong to me. True, if prospective buyers expect prices to increase, that's a reason for them to spend more now. But if prospective sellers expect consumers to spend more, that is a reason for them to start raising prices now. So while a higher announced inflation target might be self-fulfilling, there's no reason to suppose that by announcing such a target the Fed can achieve anything other than a higher rate of inflation.

Inflation expectations, in other words, inform the positions and rate of change of both demand and supply schedules–as should be especially obvious to anyone familiar with Wicksteed's famous exposition in which the latter schedules are nothing other than flipped-over portions of total ("communal") demand schedules. Changes in inflation expectations will, in still other words, tend to affect in the same manner the decisions of both buyers and sellers. Consequently, if sellers' expectations have been excessively rosy, so that their pricing decisions have resulted in disappointing sales, there's no reason to suppose that an announced increase in the inflation target won't cause them to become rosier still, ceteris paribus. Expectations are a double-edged sword that policy tends to sharpen on both sides, or not at all.

None of this contradicts the view, which I share with market monetarists, that an increase in aggregate demand that is not merely the result of an increase in the expected rate of inflation can be effective in reducing unemployment. For in that case, and again assuming that sellers' expectations have been excessively rosy, the increase serves, not to boost those expectations further, but to bring reality closer to them. To my way of thinking this difference between a policy that works by fulfilling established demand expectations that have been overly-optimistic, and one that seeks to boost demand by raising the expected rate of inflation, is absolutely crucial. If an economy is depressed because its rate of NGDP growth falls short of sellers' expectations, then surely the best way to close the gap is by raising the actual rate of NGDP growth only, while either leaving NGDP growth and inflation expectations alone or, were it possible to do so, lowering those expectations. I'm not saying that doing this is easy, by means of monetary expansion or otherwise. But it is nonetheless what needs to be achieved, and it is unlikely to be achieved by raising the Fed's inflation target.

It seems to me that monetary economists who overlook this obvious truth risk adding to rather than subtracting from our current monetary troubles. Maybe Scott isn't among them; maybe I've misunderstood him. Or maybe my own reasoning is all wrong. Like I said, I'm confused. But whatever the reason for my confusion I hope that Scott, or someone, will help me out of it.

Addendum (July 24): As the argument of this post is not the easiest to get across, I hope I may be pardoned for adding, by way of clarification, the observation that believers in NGDP targeting who also endorse raising the target (and thus the expected) rate of inflation as a means to end recession appear to subscribe to view, which I think badly mistaken, that, if economic recovery can be encouraged by providing for adequate (but not inflation-enhancing) NGDP growth, then it can be encouraged still further by promising a rate of NGDP growth such as would serve to actually increase the equilibrium inflation rate.

  • Rob R.

    I have a question: If in the present NGDP is growing slower than expectations then this may well cause a recession. Boosting NGDP growth would seem likely to boost output and address this. But how will reducing expectations to be line with actual NGDP cure the recession ? I can see that having expected and actual NGDP growth matching seems like it would lead to a kind of equilibrium, but if prices are sticky and this NGDP level is too low for full employment then by what mechanism will suppliers be motivated to increase production?

    • George Selgin

      It's important to remember that it is not prices themselves that are "sticky" (as any look at recent data will show) but the expectations upon which price (and corresponding supply function) settings are based that are so. Consequently, the decline in the level and growth rate of AD fails to be followed, or followed quickly enough, by an offsetting downward adjustment of AS or its rate of growth. Ideally, one wants to get AD up, thus boosting actual sales for any given setting of AS, without sponsoring any corresponding expectations-based upward shift in AS. One wants, in other words, to sponsor a move along the upward-sloping AS schedule, and perhaps a downward shift in that schedule based on lowered (and more realistic) demand expectations, either of which brings things closer to N*. But raising inflation expectations means raising AS, which can result in a move the other way.

      Another way of putting this is to observe that one cannot consistently attribute recession to a lack of NGDP, and hence to P>P*, and insist that the problem can be solved by convincing people that P is about to be rise further. (One may modify the language as one likes to allow for positive growth of both P and P*.)

  • BillWoolsey


    I agree. I always tell Scott, "stay on message." It is nominal GDP not inflation.

    However, Scott sometimes wants to talk the talk of the dominant inflation targeters.

    Here is the translation. As you say, higher inflation is only possible with more rapid growth in aggregate demand.

    Further, the only way a monetary authority can impact inflation is through more rapid growth in aggregate demand.

    A central bank promising higher inflation in the future _is_ promising more rapid growth in aggregate demand in the future.

    And here is the key–more rapid growth of aggregate demand in the future leads to more rapid growth in aggregate demand now.

    And one other element of Scott's emphasis is that the key problem is sticky wages and too little employment rather than sticky final goods prices. And so, higher inflation lowers real wages, increases labor demand, and increases employment. Yes, this only works if the higher inflation is due to more rapid growth in aggregate demand.

    The adverse impact on short run aggregate supply that you (and I) would associate with increased inflation expectations would be associated in Sumner's view more closely with more rapid growth in nominal wages. I don't know that this has much substantive impact, but it explains his rhetoric a bit.

  • Gonzalo R. Moya V.

    Mr. Selgin, very good article, but there is indeed a contradiction when you say: "there's no reason to suppose that an announced increase in the inflation target won't cause [sellers' expectations] to become rosier still" and then "assuming that sellers' expectations have been excessively rosy, the increase [in the expected rate of inflation] serves, not to boost those expectations further, but to bring reality closer to them".

  • George Selgin

    GRMV, I believe that the apparent contradiction stems from your having truncated the expression that properly belongs within your second set of brackets. The full expression is "an increase in aggregate demand that is not merely the result of an increase in the expected rate of inflation" (emphasis added). Ideally, as I say elsewhere, one wants to boost demand without also boosting sellers' inflation expectations.

    • Gonzalo R. Moya V.

      "Closing the gap between the rate of NGDP growth and sellers expectations […] is what needs to be achieved." "Raising the actual rate of NGDP growth only, while leaving inflation expectations alone, […] is unlikely to be achieved by raising the Fed's inflation target." Sorry Mr. Selgin, but the contradiction persists even by including more excerpts from your article.

      • George Selgin

        I'm afraid then that I'm unable to see the contradiction. In recession we have NGDP(A) P*. Conventional expansionary policy seeks to make reality conform to those predictions. If the Fed instead announces a new P target P' > P(A) >> P*, it can make matters worse to the extent that sellers respond by adjusting actual prices upward.

        • Gonzalo R. Moya V.

          I believe you misunderstood me, Mr. Selgin, I wasn´t advocating for the usefulness of varying the inflation target as an instrument of economic policy for the monetary authority. On the contrary, my point was that changes to it are ineffective either for contractionary or expansionary purposes, which is also your standing, but without exceptions, for the same reasons that you give: Suppliers adjust their expectations when the shock is announced, rendering it sterile regardless of its channel of action. Your exception lies ultimately on whether or not the suppliers have an excess of inventories, as if their desire to clear them out would make them keep their prices down until it gets done. That where we disagree, and what I pointed out as an inconsistency.

  • Another possible flaw in the idea that increasing inflationary expectations increases demand is that consumers or the private sector generally may aim to hold some specific level of monetary savings measured in REAL TERMS. Or more generally, the aim could to be hold what MMTers call a given stock of “net financial assets” (valued again in real terms).

    I’d guess the above phenomenon certainly exists at least to a limited extent. But it could be a MORE POWERFUL effect than the “increased inflation causes more spending” effect. In which case, far from raised inflationary expectations leading to increased spending, raised inflationary expectations would lead to more SAVING.

    • Gonzalo R. Moya V.

      Mr. Musgrave, although it is true that people switch their cash holdings for other assets that are unaffected by high inflation in order to fullfil the "store of value" function lost in the process (whether with real goods that are non-perishable, foreign currency or other financial intruments, if accesible), raised inflationary expectations lead in the net to increased spending, not saving. You could see it as a two-step conversion: first, from cash into durable goods that are marketable (i.e. liquid); then, from these durable goods hoarded to standard consumer goods as they are being needed.