Winner of the New Statesman SPERI Prize in Political Economy 2016

Wednesday, 14 August 2013

Why the Pigou Effect does not get you out of a liquidity trap

For macroeconomists

This issue has surfaced again (see Krugman and Rowe). I wrote a post awhile back on this, but it was quite difficult (for me at least!), so here is an attempt to restate the key conclusions more directly. Ashok Rao has a post covering some of the same themes as my earlier post. The key point here is that I am going to follow Nick in saying that money is different from bonds because money is irredeemable, but even then the Pigou effect is not a magic bullet that gets us out of a liquidity trap.

How is the Pigou effect supposed to get you out of a liquidity trap? In a liquidity trap nominal interest rates are at zero (ZLB). However pretty well everyone agrees that if by some means the monetary authority could induce higher inflation expectations, then the ZLB could be overcome, because real interest rates would fall, stimulating demand. That is a real interest rate effect. It is what some people think Friedman had in mind when he was so critical of Fed policy in the Great Depression. (I have no idea if this is true.) It is what Michael Woodford argues the Fed should now promise to mitigate the impact of the ZLB. It is what Paul Krugman recommended Japan do to get out of the lost decade. But none of these things is the Pigou effect.

The Pigou effect is when the authorities keep the current stock of money constant, and falling prices mean that its real value increases. The idea is that at some point people feel sufficiently wealthier that they spend more, which adds to demand. For this to work, we have to assume that the nominal stock of money will remain unchanged, unaffected by falling prices. Now you might say fine, let’s assume that. But if you do, you might also agree that the fall in prices is temporary. Simple neutrality implies that if you hold the money stock constant, falling prices today will mean higher prices tomorrow. But we have already established that in that case you do not need a Pigou effect, because higher inflation tomorrow at the ZLB will mean lower real interest rates, and you get the demand stimulus the good old real interest rate route. Furthermore, if people understand that prices will rise, they are not really wealthier in an intertemporal sense, because their extra real money balances will be inflated away. If you like, they save their extra real money balances today to pay for future inflation taxes. [1]

The alternative case is where future inflation does not increase as current prices fall - as would happen if the monetary authority targeted future inflation for example, and did not raise that target as prices fell. That would imply that the current nominal money stock was not fixed, because to prevent future inflation rising, the monetary authority must at some stage reduce the nominal stock of money - long run neutrality again. How does it do that without raising interest rates? It could raise taxes. But if it did that, then Ricardian consumers would not think of their higher real balances today as wealth, because this would be offset by future tax increases.

The central bank could reduce the money stock by selling some of its government debt. But under the conditions that Ricardian Equivalence holds, that has the same effect. Now the government will have to raise taxes to pay the interest on that debt, whereas before any interest they did pay came straight back via the central bank.

We can sum this up rather neatly, as Willem Buiter did here with the aid of lots of maths, by saying that what matters is the terminal stock of money, not its current value. The government can only make people feel wealthier by printing money if people believe that the increase in its real value is permanent.

We can apply the same reasoning to a helicopter drop. The first issue is whether issuing money to pay for a tax cut is any different from issuing bonds, and in particular does Ricardian Equivalence apply? Now macroeconomists are confused on this (see my earlier post), but here I’m happy to follow Nick and agree that money financing is different, because money is not redeemable. So a permanent helicopter drop of money will tend to increase consumption. To put it another way, the Ricardian Equivalence mechanism does not apply to a helicopter drop.

However there is another, more economy wide mechanism. If long run neutrality holds, and if people understand this, they will realise that their extra wealth will eventually be inflated away, so they are no better off. (Equivalently, their tax gain today will be offset by a higher inflation tax at some point.) But those expectations of higher inflation, if we are stuck in a liquidity trap, will shift consumption to the present, so the helicopter drop increases demand through a real interest rate mechanism.

The bottom line is that we can forget about the Pigou effect as a way out of the liquidity trap, at least in what is now our baseline macro model. What is important for the liquidity trap is expectations about future monetary policy. If monetary policy allows future inflation to rise, and expectations are rational, we can get out of the trap. If they do not, then we stay in the trap until some other force gets us out. That force will not be the Pigou effect.

[1] What if neutrality does not hold? Neutrality is pretty basic, but for the sake of argument let’s briefly consider this. Consumers are now wealthier, because they have more real money with no future costs to come. However I have the following problem if we stick with intertemporal consumers of the Ricardian type, who only consume the annuity value of any increase in their wealth. When do these agents consume their new found wealth? Any answer except never appears to violate consumption smoothing.


22 comments:

  1. You're right that if helicopter drops work, the real balance effect isn't a primary reason – I think Rowe (like myself) was reacting to the idea that it's fundamentally contrary to Ricardian equivalence.

    As far as expectations are concerned, I think think learning (as emphasized in your previous post) is the best way to think about it. If the central bank followed a rule broadly defined as "always drop a lot of money at the ZLB, and never contract" rational agents expect the NPV of terminal stock to be infinite, which defies the Laws of Physics.

    Therefore, agents know that there will be some contraction at some point. Each sign of easing will revise their priors that this point will be far enough to the future, and each to the contrary will tighten the recovery – a plausible explanation for the rapid reaction to "Taper Talk" in the US economy.

    As I wrote, though, I think there may be game theoretic reasons why even rational agents will spend more money today: so long as they expect others to, they know the future purchasing power will fall, and they'd rather spend now for a higher real value than later – not to mention a personal discount rate.

    ReplyDelete
  2. Furthermore, if people understand that prices will rise, they are not really wealthier in an intertemporal sense, because their extra real money balances will be inflated away. If you like, they save their extra real money balances today to pay for future inflation taxes.

    I don't think that this is correct. The inflation tax makes it more costly for agents to hold money. It's questionable whether it even applies in a liquidity trap, because the nominal interest rate is zero, so inflation equivalently affects bonds (i.e., it just reduces the real interest rate on saving, whether in the form of money or bonds).

    What you seem to be saying is that if people wish to have the same resources available in the future they must save more, and the amount they must save is the newly created money. The first point would apply in the absence of a helicopter drop: if real interest rates fall we must save more to achieve a given amount of future income. But there's no reason why that should be the objective: we would expect to consume more today in response to lower interest rates. With a helicopter drop, we are unambiguously better off today, thus one would expect higher consumption. There is no need to save to pay the inflation tax: that's just the price you pay for not saving.

    The only complications are the fact that spending money today will mean less money available for transactions in the future, and the effects of consumption smoothing. Since the newly created money is not destroyed, next period's money supply is unaffected by the decision to consume. As for consumption smoothing, it seems likely that consumption is temporarily depressed due to the liquidity trap, so households should welcome the opportunity to raise consumption close to the long-run average.

    ReplyDelete
  3. A helicopter drop reduces the central bank's capital. This doesn't necessarily have any effect until the capital is depleted. Then what? In the case of redeemable money, a run begins. To stop the run, the bank devalues. In the non-redeemable case, there's no run, but people know that the bank's current price level target isn't sustainable.

    ReplyDelete
  4. Here's perhaps a simpler way of putting it: banks do "helicopter drops" all the time, only we don't call it that. We call it paying dividends, and the helicopters selectively drop the money on the owners of the bank.

    ReplyDelete
  5. At the end of the day no single measure will save the day as there are simply unbalances (and other issues) galore.
    What things do is make a partial positive adjustment.

    To take your Dutch example of last post it is hard to see how a move away from austerity (whatever it might be)would get the economy back on track.
    There are next to a stimulus (again whatever one might think of that) the following issues:
    -Real estate bubble from overborrowing (part of the air is out but there is till hot air in it):
    -real estate bubble caused by tax policies. Part is out part still in.
    -uncertainty about tax policies re the above;
    -bankingsector still pretty unhealthy;
    -in the EZ mess as partly liable and as its biggest market;
    -worry about long term sustainability of pensions and possible pensioncuts and higher premiums;
    -overexporting (even when economy as a whole is running backwards);
    -inflation over 3% will growth is negative;
    -but somewhat longer term too high wagescosts level, while spendable income is on the way down;
    -overall debt level especially private is worrying;
    -inefficient government 90% excluding statepensions receive some sort of entitlement (austerity is tax increases, the government seems unable to get its spending properly under control);
    -unable to get grip om healthcare costs;
    -housing market is a mess also for other reasons (social housing and rental sector as well).
    Well that is about it.

    Starting from the stimulus side it is hard to see how an economy with consumers dead worried about their financial future can get back on track by what will likely be 1% or so (extra) stimulus via government spending. And subsequently debt.
    As mentioned earlier any stimulus will imho be largely rhetorical. The present austerity is no austerity (it is mainly reversing long term unsustainable overborrowing from the pre-crisis era). And so anti-austerity definitely will not be a stimulus of the size required. Look at the EU youth unemployment plan 6 Bn for the whole thing a complete joke).

    ReplyDelete
  6. Part 2

    From there it is also hardly likely that a stimulus will put the major underlying problem (huge lack of consumerconfidence) right and subsequently investing. There will not be proper level investment if the consumer remains depressed.

    You need next to stimulus also consumerconfidence building measures. And as it is clear that people are not buying the bullmanure rhetoric it will have to be put your money where your mouth is.

    So back to Holland. They need to put a bottom in the RRE market. Now the Dutch are 65 Bn under water on RRE (substantially more than 10% of GDP and all caused by the crisis). And on top of possibly further decreases; worries about pensions; rising unemployment and lower real wages. As well as further drops in RRE prices.
    It simply is not going to work with stimuli of the size politically feasible.
    Simon in his work imho completely underestimates the effect of the above and overestimates stimuli that would politically be possible.
    BTW thier CPB (hope I got it right) suggest abolishing the tax deductions all together. I simply think they havenot got a clue.

    What could they do without further financial spending on that.
    Taxes on housing should be made a point of less worry. Consensus political legislation that further cuts in interest deductions (a horror in its own right) will only be done if RRE prices go upwards again. You simply reduce downside risk for a big part. Highly important for investment decisions.
    Combined with putting the banks under pressure to finance housing at normal European spreads and not at the inflated ones used at present. 0.5-1.0% margin in there seen the present rates, the tax bubble and seen the further drop expected could be just enough to put the RE market back on track again.

    Stimulus probably best via more spendable income. Now so called austerity is more than 2/3 tax increases.

    Pensions will be a difficult one as they are already too late and would simply mean dumping the cost with the youth who subsequently would be worried and show that in consumption etc etc.

    Last remarks:
    The Hols' problems are basically similar in a lot of other countries. EG Belgium and France are still having their full blown RE bubble intact. If that goes very probably with rising interest rates it will be much worse than in Holland as their economy is much less competitive.
    Overall if consumer confidence doesnot get back on track stimulus will likely fail.
    Stimulus will be small and no way in line with the gap to be closed. Macro folks will likely get the blame when things donot work out. Even though it was clear from the start. As soon as the rhetoric goes on stimulus you put your reputation on the table. Just ask the reform/austerity folks hard to see that this was they had in mind when structural reforms were discussed, still they get it.
    The problem is the same political margins are very small either way. Basically no room for real structural reforms at the moment and no room for the style of stimulus required to get things back on track. Will be marginal and rhetoric and in an enviroment that sucks in many ways on the raod to fail (if we ever get there anyway).
    It is nearly impossible to clean up the mess and get it back on track. There a huge numbers of unbalances and Co and all over the place. Pre crisis policies would have been required to tackle these much earlier. Water under the bridge only a lesson for the future.

    ReplyDelete
  7. The Pigou effect is when the authorities keep the current stock of money constant, and falling prices mean that its real value increases. The idea is that at some point people feel sufficiently wealthier that they spend more, which adds to demand.

    I believe there is a more direct reason why the Pigou Effect is immaterial in the current situation. With wages as chronically depressed as they are, and job security/prospects as similarly low, it doesn't matter what the real value of money is. If a significant number of consumers are living paycheck-to-paycheck, with little to no savings available, they're going to spend what they have to spend to maintain food & shelter. The value of money doesn't matter if you have no disposable cash at the end of the pay period...

    ReplyDelete
    Replies
    1. Not everyone, and I doubt even the majority are in the Dickensian situation you suggest. That is the majority have savings of some sort, and if the value of those savings rises, that encourages spending.

      Delete
  8. Simon: "The Pigou effect is when the authorities keep the current stock of money constant, and falling prices mean that its real value increases. The idea is that at some point people feel sufficiently wealthier that they spend more, which adds to demand. For this to work, we have to assume that the nominal stock of money will remain unchanged, unaffected by falling prices. Now you might say fine, let’s assume that. But if you do, you might also agree that the fall in prices is temporary. Simple neutrality implies that if you hold the money stock constant, falling prices today will mean higher prices tomorrow."

    No I wouldn't agree. "Simple neutrality" implies that *if you permanently reduce the stock of money* then the *equilibrium* price level will fall both today *and tomorrow*. But, to use Don Patinkin's terminology, when we talk about the Pigou effect we are talking about a "stability experiment" and not an "equilibrium experiment", and money neutrality is an equilibrium experiment. We are talking about an economy that is out of equilibrium, and whether a fall in the price level will increase net wealth and return it to equilibrium.

    More on this later. I gotta go do stuff.

    ReplyDelete
    Replies
    1. I might be really really that wrong, but if you "permanently reduce the stock of money" then you cannot assume everything else holds constant. So I do not go past that.

      Delete
  9. Assume that recessions are causes entirely by increased demand to hold money in a sticky price world. Assume also that people attempt to consumption smooth between the current and future periods.

    In this case one reason people may increase cash holding in the present is because they think there income in the future may be lowered.

    If the CB attempts to remedy this by injecting new money then the affect it will have in the present will depend upon people's expectations about future CB actions. If the CB targets inflation then people will know that as soon as people start spending the new money then prices will start to increase and the CB will start withdrawing it. This will make people more likely to hoard it.

    Under (credible) NGDPLT however people know that the money will not be withdrawn until NGDP is back to the desired trend. When they get the new money they will know that either 1) there will higher inflation next period or 2) RGDP will actually have recovered.

    In either case they will be more likely to spend the new money than under IT.

    ReplyDelete
    Replies
    1. “If the CB targets inflation then people will know that as soon as people start spending the new money then prices will start to increase and the CB will start withdrawing it.” Nope. Under IT, CBs don’t withdraw money simply because inflation rises: they withdraw money if inflation looks like becoming EXCESSIVE. But NGDP targeting doesn’t aim for EXCESSIVE inflation either, so there’s not much difference between the two, seems to me.

      Delete
    2. In my simple example I was assuming that the CB was itting its IT even during the recession.

      NGDPT doesn't aim for any specific level of inflation but the fact the lower the rate of RGDP growth the higher the inflation rate seems to act as an automatic stabilize that might help in a liquidity trap.

      Delete
  10. Simon,

    Agree, 100%. I mean, maybe the economy is dynamically inefficient, or whatever. In which case, I suppose there are potential free lunches all over the place, including from free money. There really isn't much to say about that, except if there's free lunch (arbitrage) you should eat it. (And there's probably a million ways to eat it, but by all means, try the heli drop.)

    But since it's pretty damn tough to know if the arbitrage is really there (it's an intertemporal arbitrage that involves knowing stuff about asymptotic limits of macro variables), we usually go with the no-arbitrage assumption in devising economic policy. And in that universe, it seems to me that Eggertsson and Woodford 2003, is a pretty good place to start if we want to answer the question of what monetary expansions can or cannot do.

    Here's what they say:

    "This proposition implies that neither the extent to which quantitative easing is employed when the zero bound binds, nor the nature of the assets that the central bank may purchase through open-market operations, has any effect on whether a deflationary price-level path will represent a rational-expectations equilibrium. Hence the notion that expansions of the monetary base represent an additional tool of policy, independent of the specification of the rule for adjusting short-term nominal interest rates, is not supported by our general-equilibrium analysis of inflation and output determination."

    That's pretty damning for the "Pigou effect." And for those who claim that the effect only applies to heli drops, those are equivalent to OMOs plus fiscal transfers. Fiscal policy is interesting, in its own right, but if we are talking about the effects of fiscal policy, lets not attribute it some magical power of monetary policy.

    ReplyDelete
  11. Prof Simon,

    You assume in your 3rd paragraph that the private sector might assume that currently falling prices will be replace in the near future with rising prices, and hence that the Pigou effect doesn’t work. That’s not a rational assumption for the private sector to make. A more rational one is as follows.

    Assume to keep it simple that AD falls on Jan 1st, and thereafter all factors influencing AD remain constant for a long time. Prices will fall which will raise the real value of money savings which in turn induce the private sector to spend more. And it will spend more till some equilibrium is reached where prices and AD remain constant. I.e. your RISING PRICES assumption is not a very rational one.

    You then say “But we have already established that in that case you do not need a Pigou effect, because higher inflation tomorrow at the ZLB will mean lower real interest rates…” The fact that you don’t “need” an effect doesn’t have much to do with whether the effect is there or not. E.g. I don’t need the sun to warm my house, but if the sun is shining, that does warm my house.

    In your fourth paragraph you make an alternative assumption, namely that after the price fall, prices do not subsequently rise. Thus to prevent excessive AD, government claws back some of the extra money savings in private sector hands. OK but if government is rational, it will only claw back enough to prevent excess inflation. I.e. that point doesn’t scupper the Pigou effect.

    Next, you say in the case of helicopter drops that people “will realise that their extra wealth will eventually be inflated away, so they are no better off.” That’s not an assumption people would make if they were rational. Assuming the private sector’s stock of money influences spending, and assuming government aims to use that effect to control AD, I as a “person” would assume government would aim to keep that stock at whatever level kept unemployment at NAIRU. For example if government was aiming at 2% inflation, all else equal, I’d assume government would aim to expand the stock by 2% nominal (0% real) per year.

    Incidentally MMTers attach a fair amount of importance to the Pigou effect, though they don’t call it that. They claim that what they call “private sector net financial assets” influences private sector spending. But it’s essentially the same thing.

    And finally, I think the idea that private sector entities engage in the above sort of convoluted reasoning is totally unrealistic. Same goes for Ricardian equivalence. Bill Mitchell of Billyblog fame described Ricardian equivalence as an idea from la-la land. I rather agree. I.e. I suspect the main effect here is along the lines of: “if households find they’ve more money than expected, they’ll blow half of it within six months and won’t think for a moment about inflation in a year or two”. And if that’s what households do, then the Pigou effect works.


    ReplyDelete
    Replies
    1. " And it will spend more till some equilibrium is reached where prices and AD remain constant. I.e. your RISING PRICES assumption is not a very rational one"
      First of all, "some equilibrium" with prices and AD constant would necessarily imply less AS so bigger output gap. No sense in that.

      Another thing: what I think is said is not that we need or do not need the Pigou effect, it is that it is basically the same as the expectations theory, given that if consumers thought prices were "ever falling" they would not consume, they would wait for tomorrow to consume at lower prices, and this forever: they only due because there are certain inflation expectations.

      Finally, as for “if households find they’ve more money than expected, they’ll blow half of it within six months and won’t think for a moment about inflation in a year or two”, you could probably be right, except that we have an output gap for some reason. So, not right.

      Delete
    2. Apt, Thanks for your comment. I’m baffled as to why what I said implies less AS. The scenario I envisaged was thus. AD is far too low relative to AS. AS equals stock of capital equipment plus size and skills of the workforce, and that remains constant. Ergo prices fall because of deficient AD. Those falling prices raise AD to the point where prices are constant. AS remains constant throughout.

      Didn’t understand your 2nd para.

      In your 3rd para you argue that because we have an “output gap”, therefor the Pigou effect doesn’t work too well, which is evidence in support of Simon’s argument. My answer to that the Pigou effect is clearly near useless in the REAL WORLD because prices just don’t fall in a recession and that’s why the Pigou effect hasn't dealt with the output gap. But that’s not an argument against the THEORETICAL merits or demerits of the Pigou effect (or, and far more relevance, the arguments for and against helicopter drops).

      Delete
    3. I think the problem with your assumption is that there is, as proven, downward rigidity and for that reason you cannot get those low prices (precisely what you refer in your third paragraph).

      About the second paragraph: what I am arguing is that the reason behind the Pigou effect would not apply if prices were not to recover and would fall forever, because consumers would never consume and would indefinitly wait for lower prices. Therefore, the point is always that, in the future, prices will have to higher, and that is precisely the point when talking about expectations for inflation.

      About the third, agree, and my point is that it should be kept out of discussions. What is the point of any kind of knowledge and any theory if it does not fit the real world? Even in theory, what I think holds is precisely what I said above. People only spend if tomorrow things will become more expensive; if prices would continue to fall, no one would buy anything today, and would wait for ("a never coming") tomorrow.

      Delete
  12. Over 90% of the population does not understand economics. So the assumption that they all expect the central bank to compensate in the future does not hold. Hardly anybody takes future tax increases that have not yet been announced into account. There are also big uncertainties what the future will hold for everybody. People make decisions ad hoc based on feelings and are not all ways rational.

    If prices fall and people can buy more with there wages, they might buy more products. But they also might expect the prices to fall even further and save their money to buy in the future.

    ReplyDelete
  13. I'm a keynesian of 1 stripe or another, so don't usually think in terms of the quantity equation (any of them, actually), so I may be totally confused in what follows. That said...

    MV=PY.

    Suppose, holding M constant, Y is not limited by supply (i.e., we are in a recession or depression) and P drops in response. Until and unless Y rises, V drops, pretty much at the same rate as P. My understanding of the Pigou effect is that eventually V begins to rise, or to fall less slowly than P does. Until Y rises so far that it hits supply constraints, P does not start to rise.

    I don't see where expected inflation comes into the argument.

    ReplyDelete
    Replies
    1. When you say that "eventually V begins to rise", my question is, "ok, but how"? Expected inflation is precisely the mechanism that, in this model, would offset that.

      Delete
    2. "V begins to rise, or to fall less slowly than P does" because with the fall in P, real balances increase, total real wealth increases, and people spend out of that. No inflation anywhere, until output rises enough to become supply constrained.

      NB: I am not arguing for the actual existence or for the policy usefulness of the Pigou effect. I am only trying to articulate the logic underlying it.

      Delete

Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time.