The Special Theory of Employment, Interest and Money

I recently reread John Cochrane’s now infamous 2009 attack on fiscal stimulus for about the 4th time, and it’s not getting any better.  Oddly, he ends up arguing that we need more money and T-securities, which is precisely Brad DeLong’s argument.  But he seems confused about the Keynesian model, arguing that the Keynesians favor more consumption, not more investment.

I have some sympathy of Cochrane, as I also find Keynesianism very confusing.  You often hear Keynesians warn about too much saving.  Since it’s basically a closed economy model where S=I, that would seem to imply they fear too much investment.  Yet they don’t really worry about too much saving, rather they worry about too little saving triggered by an attempt by the public to save more at any given interest rate, which sets in motion a series of events that lead to less saving.  Or at least I think that’s what they assume (in accelerator models), but perhaps I’m just as wrong as Cochrane.

In my view the basic problem with the Keynesian model is not that it’s “wrong” (how could something be completely wrong and yet accepted by so many brilliant people?) but rather that it’s right, when it’s right, for peculiar and unreliable reasons.  Consider the effect of an increase in the propensity to save under the following three monetary regimes:

1.  Gold standard

2.  Interest rate targeting

3.  Inflation targeting

In the first two cases the attempt to save more will depress NGDP, and if wages and prices are sticky, will cause a recession.  But not in the third.  It’s interesting that the Keynesian model was developed during the first policy regime, achieved its greatest popularity under the second, and started to fade under the third.  Cochrane’s right that by the 1980s and 1990s the new Keynesians no longer talked much about the paradox of thrift, or fiscal stimulus.

Why did it work under the gold standard?  That’s easy.  An attempt to save more (or less investment caused by weaker animal spirits) will depress market interest rates.  This increases the demand for gold by reducing the opportunity cost of holding gold, thus increasing the value of gold.  Since gold is the medium of account, an increase in the value of gold depresses prices and NGDP.  (Recall that the supply of gold is nearly fixed in the short run.)  If wages and prices are sticky you get a recession.   During a recession people are poorer and may well end up saving less.

Why did it work under interest rate pegging?  That’s easy.  An attempt to save more (or less investment caused by weaker animal spirits) will depress market interest rates.  To prevent interest rates from falling the central bank must reduce the money supply.  That tight money policy will depress prices and NGDP.  If wages and prices are sticky you get a recession.   During a recession people are poorer and may well end up saving less.

Why doesn’t it work under inflation targeting?  That’s easy.  The central bank will adjust the money supply to keep prices on target, which means offsetting any factor that depresses AD.

When viewed from this perspective the Keynesian model seems very peculiar.  If the problem is the central bank reducing the money supply, then it would seem more natural to directly focus on central bank policy as the cause of changing NGDP.  If the problem is more demand for money, then why obsess so much over saving?  As Nick Rowe likes to point out, the problem isn’t too much saving, it’s too much money hoarding.  Attempts by drug dealers to hold more Federal Reserve Notes are just as contractionary as a higher propensity to save.  At that’s still true if the drug dealers don’t try to save any more, but merely convert bonds to cash.

The Keynesian model gives us the right answer in two of the three cases.  In contrast, the monetary perspective is always right, because NGDP is basically a monetary phenomenon.  Even worse, I suspect that many less sophisticated Keynesians start to take their model literally.  The sophisticated ones like Krugman and DeLong understand that attempts to increase saving trigger too much demand for base money (or too little supply.)  But others think that if the public attempts to save more it directly reduces output, because “people aren’t buying things.”  This is of course a basic fallacy, as saving equals investment in closed economy models.  More saving means they are buying more investment goods.  The problem, if there is a problem, is that attempts to save more may trigger more demand for the medium of account.  But if the central bank targets inflation, then we don’t need to worry about this increased propensity to save depressing AD.  In the same way, currency hoarding by drug dealers is contractionary under a gold standard or interest rate targeting, but not under inflation targeting.

Update: Commenter Tommy pointed out I erred in the previous sentence.  Drug dealer hoarding is contractionary under a gold standard but not necessarily under interest rate pegging.

I think Cochrane’s basic mistake was in attacking the simplistic, naive, common-sense version of the Keynesian model, and not the actual Keynesian model.  He thinks Keynesians believe that more saving is a bad thing, whereas they actually worry about a higher propensity to save causing lower interest rates and more money demand, or the central bank preventing a fall in interest rates by depressing the money supply.

If we want a general theory of NGDP, we need to model money.  There is no other way.   Instead, the dominant model is a roundabout way of looking at one particular cause of fluctuations in the money market, and hence in NGDP, which is completely conditional on the type of monetary regime.  Even worse, from a common sense perspective this model seems to argue that changes in propensity to spend have a direct impact on real purchases and real output, not merely an indirect impact through changes in the market for money.

I’d recommend that Cochrane focus more on NGDP determination, not the determination of P and Y.  It would make it easier to see where the Keynesians are coming from.  They are trying to explain NGDP, and then assume that changes in NGDP affect RGDP via sticky wages and prices.  When viewed this way, Cochrane would need to either criticize the sticky wage/price assumption, or criticize the assumption that fiscal stimulus can boost either M or V.  Instead he meanders all over the place, and criticizes the Keynesian model as it’s interpreted by its most naive proponents.

I should add that I find Keynesianism to be mind-numbingly illogical, so it’s quite possible I got Cochrane wrong, or DeLong/Krugman wrong, or the Keynesian model wrong.  If I did, there’s only one person I trust to set me straight, and he lives in Canada.  (Actually two, but both live in Canada.)



26 Responses to “The Special Theory of Employment, Interest and Money”

  1. Gravatar of marcus nunes marcus nunes
    10. January 2012 at 13:53

    Again I´m reminded of your early 2009 Gross Deceptive Product. That sums up the confusing Keynesian model very well!

  2. Gravatar of Adam Adam
    10. January 2012 at 13:56

    Excellent post (at least to this lay reader who only partially understands).

    How are things complicated by the fact that we do not really have inflation targeting, but rather have central banks (at least in the U.S. or Europe) that seem only interested in capping inflation at 2%? Wouldn’t that mean the central bank only partially offsets factors that depress AD and we’re partially in world #2?

  3. Gravatar of John John
    10. January 2012 at 14:14


    I’m not sure why you attack Keynesianism when you seem to have embraced so much of its worldview; especially the misleading aggregate thinking. For instance, in an earlier blog you commented that you think crowding out is unlikely at the zero lower bound. Thinking about it closely, that view seems strikingly naive. You’re viewing a micro phenomenon (crowding out) through a macro lens.

    Imagine that the government spends a couple million on a bridge project in a small county. Suppose the county has 200 people that could work construction; 125 of which are currently working. If the bridge project hires 100 construction workers, they’ve crowded out construction in that county. People who want to build will have to either out bid the government for workers, pay more for existing ones, postpone their projects, or get by with less people. Looking at the real world from the point of view of individuals involved in economic action, crowding out appears very possible, indeed likely, even with high levels of unemployment and low interest rates.

    If you’re referring to crowding out in the bond markets where government spending drives up interest rates, the logic of crowding out as raising interest rates doesn’t really apply. The Fed can create enough reserves in the banking system so that people can borrow at near zero virtually regardless of government spending. That’s one of the “benefits” of going off the gold standard.

    Still, the bottom line is that government spending diverts resources from private uses. The only way it wouldn’t is if they were able to specifically target all the idle resources and workers and there is simply no way to do this. Stop looking at capital and labor as homogenous units.

  4. Gravatar of ssumner ssumner
    10. January 2012 at 14:44

    marcus, Yes, that was my first on this topic. Interested readers can google it.

    Adam, It would make some difference, but I don’t think that’s entirely accurate. The Fed is still probably targeting inflation at 2%. They don’t hit that target precisely; sometimes they are too high and sometimes too low. But they clearly pull out policies like QE every time they see inflation falling below a critical level (about 1%.)

    But I agree that your assumption makes some difference.

    John, I look at aggregates because I’m a macroeconomist. Macroeconomics is the study of aggregates, and all macroeconomists (including Austrians) study aggregates.

    I agree that some crowding out occurs at the zero bound. If I said otherwise I misspoke–I meant you don’t get complete 100% crowding out at the zero bound. But I agree with your example.

  5. Gravatar of Mike Sax Mike Sax
    10. January 2012 at 14:48

    I never really get the distinction between money hoarding and saving-not saying it doesn’t exist just I never quite wrap my head around it.

    In Keynesianism investment and saving isn’t the same thing. Saving is basically hoarding, that amount of money that is neither spent on coonsumption or investment.

  6. Gravatar of Tommy Dorsett Tommy Dorsett
    10. January 2012 at 14:53

    Wouldn’t a rise in base money demand be accommodated under a rate-targeting regime without an adverse impact on the NGDP/trend level gap absent the target rate being set above/below Wicksellian natural rate?

  7. Gravatar of PrometheeFeu PrometheeFeu
    10. January 2012 at 15:01

    @Mike Sax:

    That’s because it’s weird and confusing and complicated. As I understand it, money hoarding just means people are willing to accept a lower rate of interest and still save the same amount. If the interest rate doesn’t move, that means they switch over to safer investments. Since T-bills are basically Fed Notes++, it’s money hoarding. Macroeconomists are confusing people who don’t like to speak in clear language.

  8. Gravatar of Nick Rowe Nick Rowe
    10. January 2012 at 15:08

    Mike Sax: “saving” (as defined in standard National income Accounting) is by far the most confusing term in macroeconomics. Desired saving is defined as: income from the sale of newly-produced goods minus taxes minus desired purchases of newly-produced consumption goods.

    It’s not a thing; it’s a non-thing. It’s a residual.

    If you wish to spend part of your income on:
    a house, and old painting, a bond, antique furniture, land, an old car, stocks, a new computer for your business, lend it to someone, etc. etc., that is ALL desired saving.

    If you take income in the form of money and stick it under the mattress, or in your chequing account, and leave it there for a bit, that is hoarding.

  9. Gravatar of Mike Sax Mike Sax
    10. January 2012 at 15:16

    Ok Nick. So when I look at desired saving that sounds to me like consumption or investment-or in one case lending-which of course is stimulative as it leads to consumption; it’s also kind of an investment.

    Hoarding sounds like saving iin my mind. Is it possible that the confusion with Keynesianism is partly a different vocabulary on this?

  10. Gravatar of PrometheeFeu PrometheeFeu
    10. January 2012 at 15:22

    @Nick Rowe:

    Setting aside the cash under the mattress. Don’t banks buy stuff with money in your non-interest bearing checking account?

  11. Gravatar of dwb dwb
    10. January 2012 at 15:28

    great post.

  12. Gravatar of Nick Rowe Nick Rowe
    10. January 2012 at 15:45

    Mike Sax: “Hoarding sounds like saving iin my mind. Is it possible that the confusion with Keynesianism is partly a different vocabulary on this?”

    The vocabulary I was using (for saving) *is* the standard Keynesian vocabulary.

    Prometheefeu: depends. (Sorry, my minds not up to giving you a better answer at the moment.)

  13. Gravatar of PrometheeFeu PrometheeFeu
    10. January 2012 at 15:57

    @Nick Rowe:

    That’s more or less my point. I think it makes more sense to think of money hoarding as a downward shift of the interest rate v quantity saved curve (same thing as a flight to safety basically) than a shift from a categorical savings bucket to a categorical money hoarding bucket.

    I’m sorry your “minds” are not up to the current task. :P

  14. Gravatar of marcus nunes marcus nunes
    10. January 2012 at 17:35

    It gets even more muddled:

  15. Gravatar of Nick Rowe Nick Rowe
    10. January 2012 at 17:53

    marcus: given the recent “burden of the debt on our children” issue, one might think about stones and glass houses…

    I’m curiously waiting for Scott’s thoughts on the burden of the debt.

  16. Gravatar of StatsGuy StatsGuy
    10. January 2012 at 18:02

    Keynesianism does need to deal better with saving/hoarding…

    I think the key to understanding this is in Nick’s observation long ago that in a barter economy, there’s no liquidity trap.

    Hoarding is a monetary phenomenon. Hoarding requires a store of value that exists in a social domain – essentially money, which is a measure of social obligation (defined either by formal debt or cultural attachment to something like gold).

    In the keynesian model, hoarding is not the implicit problem. The implicit problem is a desire to save and a lack of real opportunities to save (in the real economy sense) due to excessively low rates of real risk-adjusted return in the current equilibrium. The solution is to increase AD, which also shifts up the rate of real risk-adjusted return until there is an equilibrium that is above the liquidity trap threshold. At that point, hoarding ceases to exist.

    But the keynesian dilemma at the zero bound can’t exist unless there is a situation such that the best way for individuals to preserve their own future consumption is by “saving” in a non-productive manner.

    Keynesianism essentially says: There is a point where more consumption demand ==> more real “savings”. That is the fundamental mind-twist. That “point” is the zero bound, because anywhere below the zero bound more consumption ==> higher real rate of return ==> higher “real” savings/investment because there is no tradeoff (in the real resource sense) between savings and investment because there is excess real capacity – aka, the SRAS is FLAT.

    So what does it mean that SRAS is flat? It means that people are hoarding. If they were NOT hoarding, they would be “saving” everything they didn’t consume, and the tradeoff between consuming/saving would be complete, and SRAS would not be flat.

    The Keynesian zero bound is thus a monetary phenomenon – which Krugman circa 1999 understood very well.

  17. Gravatar of Full Employment Hawk Full Employment Hawk
    10. January 2012 at 20:51

    ” It’s interesting that the Keynesian model was developed during the first policy regime”

    Not so. Britain went off the gold standard in 1931. The GENERAL THEORY was published in 1936. (I believe that it acually was publisHed in Britain in late 1935, but am not sure of this.)

  18. Gravatar of dwb dwb
    11. January 2012 at 04:36

    The Fed is still probably targeting inflation at 2%. They don’t hit that target precisely; sometimes they are too high and sometimes too low. But they clearly pull out policies like QE every time they see inflation falling below a critical level (about 1%.)

    I looked through some stuff recently and i was reminded that there is a meaningful upward bias in inflation measures, maybe as large as 1.5%. I think the Fed itself is not consistent about usage: when the Fed says they are “targeting” 2% for (say) PCE (substitute your favorite Fed target here), i think they are really targeting 1% “actual” inflation. Yet when they say they are avoiding “deflation” they actually mean a decline in PCE. But they pull out QE at 1%. I get lazy and forget the upwards bias…but it raises the question, what the heck are they targeting with 2% “inflation” and what is the estimated bias these days? Are they targeting 2% actual inflation, or 2% PCE (which really means, maybe 1% actual inflation, so at 1% they are only pulling out QE for “actual” deflation)?? Used to be you would see Fed members discuss and debate this. I like them to actually discuss and debate this, along with what kind of inflation they should be targeting (supply, demand) vs what they have control over.

  19. Gravatar of TG TG
    11. January 2012 at 06:38


    It seems to me that you are forgetting about equilibrium: yes it is true in a closed economy that S=I, but that is an equilibrium outcome. It is not the case that consumers can increase S and therefore I rises by the same amount. Consumers can raise the *demand* for S (i.e. the supply of investment funds), which will cause the interest rate to drop until the demand for S equals the demand for I.

    Oh, but, whoops: we hit the zero lower bound because the demand (curve) for S is so high and the demand (curve) for I is so low. Now people don’t want to spend, banks can sit on the money because they don’t have to pay interest on it and so no-one spends it – depressing output, earnings and causing more demand for saving and even less demand for I.

    Now under inflation targeting the Fed has to resort to unconventional monetary policy to keep inflation up – but it can’t (I read somewhere, though I can’t find the reference, that one speaker at the AEA argued that a reasonable Fed would be causing inflation, and since it is reasonable but isn’t raising inflation it must mean that it can’t).

    Therefore at present no one wants to spend (since they are worried about their existing debt and/or their future earnings), no one wants to invest (since they can’t see when spending will rise) and so the government stepping in and spending is the only way that AD is likely to go up in the short term.

  20. Gravatar of ssumner ssumner
    11. January 2012 at 06:48

    Mike Sax, That’s wrong, saving is defined as the funds that flow into investment.

    Tommy, That’s right, I should correct that with an update.

    Thanks dwb.

    Thanks Marcus, I have a new post.

    Nick, At first I misunderstood your point about the debt, but now I see it. You are right. As a practical matter I think the biggest burdens are the effect of deficits on investment and the deadweight loss of higher future taxes. But the diversion of consumption from the unborn to the living is also a burden.

    Statsguy, I don’t think you should equate zero bound with flat SRAS. I don’t think anyone would claim the SRAS is flat in Japan (unemployment is 4%) but they are at the zero bound. The flat SRAS has to do with sticky wages and prices plus economic slack. The zero bound has to do with deflation and low equilibrium real interest rates.

    Full Employment Hawk, Yes, but that was viewed as temporary. The gold standard world was Keynes’s economic framework. The largest market economy in the world, by far, was still on gold in 1935. So were major European powers like France. It definitely had an impact on Keynes’s thinking. I published a paper in Economic Inquiry around 1999 documenting the way in which the gold standard underlies key parts of the Keynesian model.

    dwb, I don’t see the bias as being that large, but also think the question is entirely subjective. Indeed I’ve never even seen a persuasive definition of the term ‘inflation.’ What is it supposed to mean?

  21. Gravatar of ssumner ssumner
    11. January 2012 at 06:54

    TG, That “someone” at the AEA meetings was Robert Hall. And I haven’t found anyone who can make heads or tails of his argument. I did a post on that a couple days ago. Also check out Ryan Avent and Arnold Kling, who were equally confused.

    Let me know if you can figure out Hall’s logic. Meanwhile Bernanke insists the Fed has plenty of ammunition, and the asset markets strongly agree.

  22. Gravatar of StatsGuy StatsGuy
    11. January 2012 at 07:18

    Scott: “Statsguy, I don’t think you should equate zero bound with flat SRAS. I don’t think anyone would claim the SRAS is flat in Japan (unemployment is 4%) but they are at the zero bound.”

    Good point about Japan, but I think the relationship (hoarding = below zero bound = flat SRAS in keynesian model) still holds. There are a few things to note with Japan:

    1) Consider the possibility that 0.5% interest IS the point at which keynesians see savings/consumption hitting equillibrium. There’s no ability to increase the real return by stimulating consumption, since resources are fully allocated.

    Phrased a different way, if unemployment is only 4% (the natural rate in Japan???), then why is no NGDP growth a bad thing in real terms? In other words, if the economy is at capacity (is it?), then lifting NGDP just means more inflation. So who cares? (Arguably there are second order effects of stabilizing future consumption expectations and improving quality of investment…)

    There are two answers – either 4% unemployment is a bad proxy for lack of economic slack (people are employed wastefully), or there are structural impediments to real growth.

    2) The core keynesian model is a closed economy, but Japan is anything but that… savings can be allocated abroad. What is TRULY incredible about Japan is that the country has achieved near-zero inflation EVEN AS cost of input goods has skyrocketed. If we think the US is dependent on imports of foreign raw materials, we’ve got nothing on Japan. Nothing. They are net importers of ore, food, fuel, fiber… Extraordinary, really.

  23. Gravatar of Adam Adam
    11. January 2012 at 08:18

    Scott – right, those are the partial adjustments. But if inflation comes in at, say 1.4%, they aren’t going to do anything stimulative.

    And aren’t we heading into our fourth year of only missing on the minus side?

  24. Gravatar of Mike Sax Mike Sax
    11. January 2012 at 08:51

    ” saving is defined as the funds that flow into investment.”

    That may be the way that you define savings, but the question is that the way Keynesianism defines savings?

    Your fellow Market MOnetarist Nick Rowe had a different gloss.

  25. Gravatar of bill woolsey bill woolsey
    11. January 2012 at 11:08


    Rowe said “desired saving.”

    There is an accounting identity that says saving always equals investment. I don’t know about saving as funds that flow into investment, because it includes unplanned inventory investment–goods that are produced for sale but not sold.

    Then, there is a Keynesian equilibrium condition that says that the level of output will adust to the amount firms think they can sell (so that there is no unplanned inventory investment.) Output generates an equal income, and saving is positively related ot income. So, if saving is greater than planned investment, output, income, and saving all fall. When income is low enough, then saving falls enought to equal planned investment.

    Finally, there is the traditional approach where “the” interest rate adjusts to that desired saving at a level of income consistent with full employment of resources equals planned investment (like buying newly produced capital goods.)

    Leaving aside “hoarding” of money, when people desire to save they do something, and these things raise asset prices and reduce yields. They lower “the” interest rate. (They can even choose to save by purchasing new capital goods.) Anyway, this looks a bit like saving flowing into investmnet. Like I buy an old bond from someone who then buys a new bond which is issued by a firm to fund the purchase of an old capital good the seller of which buys a new capital good. The saving flowed into investment by a series of steps.

    Finally, when you earn income (getting check) and you leave it in your checking account (and hoard) the person who paid you (like your employer) has less money in his checking account. The total amount in checking accounts is the same, and so the amount of lending funded by the banks is the same. Leaving money you received in your checking account doesn’t increase total lending by the banking system.

    Free bankers argue that if there are no reserve requirements, banks will increase the total amount of money in checking accounts in this circumstance and expand lending. But leaving your money in the bank doesn’t do it directly.

    The person who would have received the money in your checking account if you spent it is earning less income. Only if the banks create new money and lend it out does the person who sells to the new borrower have added income that matches the income that would have been created if you spent the money instead.

    To sum up, if people choose to accumulate larger money balancs (demand for money,) and the quantity of money is the same, the result is reduced nominal expenditue on output and reduced income. This happens whether or not it is hand-to-hand currency or “money in the bank.” For nominal expenditure to remain the same, the quantity of money needs to rise to match the increase in the demand to hold money.

  26. Gravatar of ssumner ssumner
    12. January 2012 at 12:53

    Statsguy, It’s a matter of degree. The SRAS in Japan is neither flat nor vertical, but somewhere in between.

    Adam, Inflation was above target this year. They’ve tightened twice since 2008, (after QE1 and QE2) and each time had to loosen again later. They are very much steering the ship.

    Mike Sax, All the textbooks agree with me, as does Nick (I think.)

    He talks about how individuals can save by buying antique furniture, but the seller dissaves equally. I am talking about aggregate S and I. I’m not claiming they are the same for every individual.

    Also, I agree that planned saving and planned investment may differ.

Leave a Reply