Via Scott Sumner I found another of Nick Rowe’s remarkably thoughtful and thought-provoking posts about the foundations of monetary theory. The object – at least as I read him – of Nick’s post is to explain how and why money can (or must) be neutral. And Nick performs this little (or maybe not so little) feat by juxtaposing two giants in the history of monetary theory, David Hume from the eighteenth century and Don Patinkin from the twentieth. Both, it seems, were convinced of the theoretical, indeed logical, necessity of monetary neutrality, but both felt constrained by observational experience to acknowledge that money has real effects, which is just another of saying that money is not (or at least not always) neutral.
I am not going to discuss Nick’s post in detail. Instead, I want to question what I take to be an underlying premise of his post, that there is a theoretical presumption that money is neutral, at least in the long run. In questioning the neutrality of money, I do not mean that one cannot easily write down a model in which it is possible to derive the conclusion that a change in the quantity of money changes the equilibrium in that model by changing all money prices proportionately, leaving all relative prices and all real quantities of goods produced and consumed unchanged. What I assert is that the real world conditions under which this result would obtain do not exist, with the possible exception of a currency reform in which a new currency unit is introduced to replace the old unit at a defined rate between the new and old units. In such a case, but only in such a case, it is likely that the results of the change would be confined to money prices, with no effect on real quantities. (It is because of this trivial exception that I inserted asterisks after “always” and “everywhere” in the title of this post.)
Let me give a few, certainly not all, of the reasons why money is never neutral. First, most agree as David Hume explained over 250 years ago that changes in the quantity of money do have short-term real effects. The neutrality of money is thus usually presented as a proposition valid only in the long-run. But there is clearly no compelling reason to think that it is valid in the long run either, because, as Keynes recognized, the long run is a succession of short runs. But each short-run involves a variety of irreversible investments and irrevocable commitments, so that any deviation from the long-run equilibrium path one might have embarked on at time 0 will render it practically impossible to ever revert back to the long-run path from which one started. If money has real short-term effects, in an economy characterized by path dependence, money must have long-term effects. Real irreversible investments are just one example of such path dependencies. There are also path dependencies associated with investments in human capital or employment decisions. Indeed, path dependencies are inherent in any economy in which trading is allowed at disequilibrium prices, which is to say every economy that exists or ever existed.
If workers’ chances of being employed depend on their previous employment history, short-term increases in employment necessarily have long-term effects on the future employability of workers. Chronic high employment now degrades the quality of the labor force in the future. If arguments that potential GDP has fallen since 2008 have any validity, a powerful reason why potential GDP has fallen is surely the increase in chronic unemployment since 2008.
Another way to make this point is that the proposition of long-run neutrality presupposes that there is one and only one equilibrium time path for the economy. The economy is in equilibrium if and only if it is on that unique time path. Under long-run neutrality, you can deviate from that equilibrium time path for a while, but sooner or later you must get back on it. When you’re back on it, monetary neutrality has been restored. But if there is no single equilibrium time path, there is no presumption of neutrality in the short run or the long run.
Let me also mention another reason besides time dependence and irreversibility why it is a mistake to conceive of an economy as having a unique equilibrium time path. As I have observed in previous posts on this blog, every economic equilibrium is dependent on the expectations held by the agents. A change in expectations changes the equilibrium. Or, as I have expressed it previously, expectations are fundamental. If a change in monetary policy induces, or is associated with, a change in expectations, the economic equilibrium changes. So money can’t be neutral. Ever.
PS Let me just mention that I have drawn in this post on an unpublished paper by Richard Lipsey “The Neutrality of Money,” which he was kind enough to share with me. Lipsey particularly emphasized path dependence as a reason why money, as he put, “is an artifact of economic models,” not a universally correct prediction about the world. Lipsey developed the idea of path dependence more fully in another much longer paper co-athored with Kenneth Carlaw that he shared with me, “Does History Matter? Empirical Analysis of Evolutionary versus New Classical Views of the Economy” forthcoming in the Journal of Evolutionary Economics. Perhaps in a future post, I will discuss the Carlaw Lipsey paper at greater length.
Uneasy Money 
Terrific stuff. I look forward to more on the Carlaw / Lipsey work.
LikeLike
Very well argued. Oh! the power of words. So Fed policy is sure to take the economy on a very undesirable long run path. But we´re safe from inflation!
LikeLike
hmm. not sure i agree. If you look at some textbook examples, there are lots of countries/periods where money growth has been very rapid, real growth small, and inflation very high (relative to real growth). Maybe the “long run” really means over a period where aggregate supply is pretty inelastic. maybe neutral really means *mostly* neutral (i.e. 70% M growth, 10% real growth, 60% inflation over some period),and perfect neutrality is the limiting case. If i then pumped up the M supply to 140% would i get 20% real growth? probably not.
LikeLike
Greg Ransom,
You’re slipping. No Hayekian worth his salt would allow this line-
“as Keynes recognized, the long run is a succession of short runs”
– to go unchallenged. I’m amazed that this thread has been allowed to progress this far without at LEAST several reading recommendations from you. 😉
David Glasner,
Regarding that line, it seems to me that Keynes committed the fallacy of composition here. The fact that something is true of every part of an thing does not necessarily mean that it is true of the thing as a whole. Every cell of an elephant is small, but elephants are not small; individual savings are good, but net savings in a monetary economy can only occur through idleness etc.
Similarly, the fact that money is non-neutral over short periods of time does not entail that money is non-neutral in the long-run. The path dependency argument can’t work because it can lead to the opposite conclusion: every short-run is dependent on the preceding history of that country. Ahistoricity has been a flaw of many of the Keynesian arguments over that past few years e.g. “Nominal interest rates are near zero in the US/UK? Fiscal stimlus!” without a single thought about the debt history of the US or UK.
There are a lot of interesting a priori arguments for the long-run non-neutrality of money. (Sadly, most of them involve some straw-man definition of “neutrality”.) However, there are strong a priori arguments in the opposite direction. How do we resolve this difference? I hope through empirical studies. As far as I know, the weight of empirical evidence is in favour of long-run neutrality, but not superneutrality or short-run neutrality. So the historic difference in the quantity of money in Australia and Canada works out as a difference in price levels rather than standards of living and restrained money growth in the classical Gold Standard period was just as compatible with steady growth as rapid monetary expansion under a fiat currency system.
LikeLike
To pre-empt some definitional problems: what do we mean by “long-run” here? Do we mean when an economy reaches equilibrium or a long period of time e.g. a period of multiple business cycles? If the former, then Keynes’s statement is false as a matter of theory. If the latter, then the long-run non-neutrality of money is vulnerable to the all the empirical evidence against it.
LikeLike
David,
I have difficulty seeing the non-neutrality of money in the long run for aren’t there some “real anchors” to the path an economy can take?
Let me give you an example,
I. There are two agents, A1 & A2 who derive utility from the consumption of output.
II. A2 is the only agent capable of producing units output.
III. For each unit of output produced the agent receives a token that entitles her to a unit of output; no tokens no output entitlement.
IV. Markets always clear; i.e. A2 always produces a unit of output in exchange for a token.
V. A2 has N tokens at the start and keeps all tokens received.
VI. quantity demanded by A2 is always N.
If A1 then happens to obtain a token without producing a unit of output whilst A2 has received a token for each unit of output produced the quantity demanded goes up by 1. As a result of IV quantity supplied has to go up as well: i.e. demand shifts outward along a horizontal supply curve.
If equilibrium is to stay there the next period A1 will have to keep obtaining tokens each period otherwise the equilibrium will revert to the one where A1 has no tokens and quantity demanded and thus quantity supplied go down.
The “real anchor” here is the ability of the agents to produce output.
I do understand that there is path dependence in the economy, I just have difficulty seeing how that can be the result of long-run monetary non-neutrality without making certain assumptions about the productive ability of the agents and the market mechanism.
For it seems to me that for this that redistribution would have to give certain agents access to resources without which they would not be able to produce for a product that is in demand but not produced yet: i.e. allocation has failed due to barriers (e.g. high transaction costs) and redistribution effectuated through an expansion of the money supply could improve upon it. However once improved upon it we seem to be back in long run neutrality: i.e the redistribution is not sustainable, we prefer the previous allocation and we revert back to it and liquidate the capital goods devoted to the other allocation.
LikeLike
David
Sorry, but I think this post begs the question – what do you mean by money here? And what do people have expectations about? Nominal income? real income? Interest rates? Default? The subsuming of money under ‘monetary policy’ is not a move that can be made so seamlessly.
LikeLike
To me, long run neutrality of money means that the simple, common sense short run impacts of more rapid monetary growth, such as lower real interest rates, cannot be expected to persist.
Similarly, extra high levels of real output and employment (and lower unemployment) cannot be expected to persist.
That there will be some kind of permanent impact of a shift in money growth seems likely to me. I am not confident that I know what it is.
By the way, aren’t there all sorts of reasons to reject perfect superneutraility–at the very least the amount of segnorage and the demand for real balances?
So, it is only the one shot shifts in the quantity of money?
LikeLike
I don’t see why it matters whether money is neutral or not. It’s a medium of exchange. In terms of currency. Gold would be another case which has been money for well over 2000 years.
LikeLike
In some ways this reminds me of the Archibald and Lipsey vs Patinkin debate that came between the first and second editions of MIP. Except, IIRC, Lipsey is now arguing from the other side!
There was a discussion in the second edition of MIP on what preferences would have to look like for (say) a temporary change in the distribution of wealth (caused say by a distribution of new money balances that wasn’t in proportion to existing balances) to persist indefinitely. It depends on whether the subjective rate of time-preference proper would increase or decrease with wealth.
If the underlying growth model is an AK model, temporary shocks to Investment and the capital stock would permanently postpone the growth path of the economy. If instead it were a Solow growth model, the effects would be transitory, since the steady state is independent of starting conditions.
(Actually, I was thinking more about the transmission mechanism from M to NGDP when i wrote my post, rather than long run neutrality per se.)
LikeLike
The way I think about neutrality is like this: for any equilibrium time-path there exists a second equilibrium time-path where all nominal variables are doubled, and all real variables are the same, for all points along those two time paths. By choosing a nominal anchor the central banks chooses one of those time-paths.
But if the central bank changes its choice in real time, the economy will not jump immediately to that alternative time-path. And it is an open question whether the economy will ever converge exactly to that same alternative time-path.
LikeLike
Greg, Thanks. Will try to get around to Carlaw-Lipsey in the next few weeks.
Marcus, Thanks. The problem, as I see it, is that sometimes — not always, but sometimes — an economy can get stuck at a low level equilibrium and monetary stimulus is needed to get the economy out of the low-level equilibrium. Simply invoking long-run neutrality (“monetary policy is not a panacea”) in such circumstances is fatalism of the worst sort.
dwb, I think that you are misinterpreting what I said (or, alternatively, I did not make myself clear). My point was not that monetary stimulus always increases real GDP. There is obviously no empirical support for that proposition. But monetary neutrality means that money has no effect at all on relative prices and real quantities. That is an implication of a very narrow class of theoretical models similar to frictionless models in physics. Many physical phenomena can’t be understood in frictionless models and many economic phenomena can’t be understood in models with monetary neutrality. Why should we draw empirical propositions about the real world from a model that implies monetary neutrality? But recognizing that money is not neutral doesn’t mean that I think that there is a long-run tradeoff between inflation and unemployment that provides a set of alternative combinations of inflation and unemployment that policy makers could choose from. But it is also possible that there are circumstances under which monetary policy can reduce unemployment. That’s why Hawtrey called one of his great works The Art of Central Banking. There’s more to it than just mechanically applying a simple model.
W. Peden, I don’t mean to suggest that monetary policy can necessarily be used to reduce unemployment in the long run, because it can do so in the short run. That would be a fallacy of composition. But there is a theoretical argument that explains why using monetary policy to reduce unemployment in the short run will eventually produce effects that tend to increase unemployment. My point is that there is no theoretical argument that allows you to infer the long-run equilibrium is unaffected by the previous history of monetary policy. An expansionary monetary policy may fail to reduce unemployment over the long run, but it will not fail to affect the final equilibrium at the end of the period. The final equilibrium may even be worse than the equilibrium that would have been achieved with a less expansionary monetary policy, but it won’t be the same. It is also possible that the effects of different monetary policies, as you suggest, do not have major obvious differences on long-run real rates of growth. But that is not what what neutrality means. It just means that the real effects of monetary policy on real GDP are small relative to the effects of technological advances and increases in factor supplies.
Martin, Your example contains no intertemporal production processes using durable capital. If investments in fixed or other forms of durable (e.g., human) capital exist and the existence of such fixed and durable capital assets affects future production decisions, you get a different kind of path dependence. You are also assuming that there is only one equilibrium in this model. What would happen if there many possible equilibria?
Ritwik, People have expectations about future prices in nominal terms which depend on their expectations about monetary policy and expectations about their future real income. Obviously they have expectations about future interest nominal and implicitly real interest rates. What is the ambiguity that you perceive in my use of the term “money?”
Bill, It seems that people are using monetary neutrality in a very weak sense, much weaker than I am which is that monetary policy has little effect on real output and income. I certainly agree that this is sometimes, and in reasonably well-behaved economies probably usually, the case. But I am using the term in a stricter sense which is that money has no effect on relative prices and quantities. Money is a veil. I agree that just because money is not neutral doesn’t mean that the monetary authorities can tell in advance what sort of relative price and quantity effects their policies will have. I agree with you about super-neutrality.
Tas, It matters, because there is an extreme doctrine of monetary neutrality that denies that monetary policy can ever be used to improve the performance of the economy. It also matters just because economists like to think about the implications of the models that they use.
LikeLike
Nick, Yeah, I was also thinking about the Archibald and Lipsey paper criticizing Patinkin and his response, but I haven’t looked at it 30 years, so I can’t really remember what it was about. I do vaguely recall Patinkin’s discussion about what preferences would make distribution effects irrelevant to relative prices.
I accept your clarification, but I interpret your thought experiment in which “the central bank chooses one of two equilibrium time paths” to be akin to a currency reform, not the choice of an alternative monetary policy. A currency reform is a regime change and monetary policy is carried out within the regime. Regime changes are neutral; monetary policy is not neutral.
I’m going to be at Brock University over the weekend for the HES meeting. I didn’t see your name on the list of participants, so I guess I won’t be meeting you this time. Too bad. Maybe another time.
LikeLike
Good post.
Once one recognizes that “K” and “L” (and “Y” for that matter) are not homogeneous and that there are sunk or irreversible costs/decisions, one concludes, as you do, that SR non-neutrality dictates LR non-neutrality, other than in the weak sense that Bill Woolsey mentioned in which some of the distortions of monetary disequilibrium cannot be sustained.
Sounds rather Austrianish which of course is one of the reasons why Greg Ransom is happy, I assume. 🙂 Once one recognizes decisions being made regarding heterogeneous factors of production based on unsustainably distorted price signals, the potential for malinvestments is a short step.
Part of the confusion I think is that, at least in micro, the LR does not equal the sum of the SRs. In micro, the concepts of short and long run are meant to capture not the passage of time per se but an interval of time – i.e., the long run does not “follow” the short run. In micro, the longer interval translates into a greater degree of flexibility over the means of production (you can make more decisions simultaneously). Ignoring the heterogeneity of capital and labour, by allowing one to assume the potential for costless redeployment, is perhaps a way of carrying over into macro the “flexibility” necessary to the micro conception of LR.
LikeLike
” Your example contains no intertemporal production processes using durable capital. If investments in fixed or other forms of durable (e.g., human) capital exist and the existence of such fixed and durable capital assets affects future production decisions, you get a different kind of path dependence. You are also assuming that there is only one equilibrium in this model. What would happen if there many possible equilibria?”
David,
Thank you for your answer.
Investment is fixed in essence in the example, it’s fixed by the assumption that only A2 can produce. I agree with you that I have assumed that there is no process of capital accumulation and allocation that could be affected by an expansion of the money supply.The reason for this is twofold, one is to illustrate how previous investments (i.e. the ability to produce of A2) restricts how much a monetary expansion can change the real values and two is – and i realize that now – that I assumed that before substantial investments could be made that could change the real values (i.e. move the economy to a different equilibrium), people’s expectations would anticipate the continued expansion of the money supply based on the past (i.e. adaptive expectations) and incorporate that information in their choices.
Am I therefore correct, when I read you as stating that during a monetary expansion investments are made that are size-able enough as people don’t anticipate well enough to push the economy onto a different equilibrium path?
P.S.
I am tempted to think of the Central Bank and Government as being one; the tension then is that whilst I am inclined to think that sustained fiscal policy can push an economy to a different equilibrium, sustained but limited expansion will be anticipated. Further the reductio ad absurdam is of course that hyperinflations do illustrate that money can be non-neutral in the long run as they do change the equilibrium path of the economy I presume.
LikeLike
david, Thanks. Some might, understandably, take offense at being compared to the Austrian school. I don’t, though I do admit to a touch of embarrassment. But there is a valid core of Austrian theory that is not, or at least should not be, controversial. Austrians and I agree that money is not neutral. Where we disagree is that a priori belief that monetary expansion must always have unambiguously harmful effects. I think that the effects can go in either direction depending on circumstances. I am not sure that long run and short run have different meanings in micro and macro. It all depends on the problem. Short run implies something (factor deployment, expectations or something else that might change in response to a given parametric change) is held fixed.
Martin, I think that is a fair interpretation. Investments are irreversible, so if people are induced to change their actions in the short run, the effects of those changes cannot necessarily be undone in the long run. As Carlaw and Lipsey put it, history matters. Hyperinflations, which produce all kinds of long-term damage, are an extreme case of monetary non-neutrality. I have no problem treating the fiscal side and the monetary side as effectively a unified decision making unit. That seems like a pragmatic choice of modeling strategy.
LikeLike
” Some might, understandably, take offense at being compared to the Austrian school. I don’t… ”
Actually, I didn’t think you would. I was picking up more on the use of “path dependence”, which I gather is Lipsey’s term in any case, rather than what seems to be the more loaded “malinvestments”. Path dependence is a good term too in that it perhaps captures more than “malinvestments”. On the other hand, it lacks the implication of error that “malinvestments” does.
LikeLike
Horwitz, Hayek & White do not disagree with you on this.
Example — if there is a below productivity norm deflation.
Horwitz, Hayek & White all advocate a monetary expansion in such a situation, as good for the economy.
David writes,
“Where we disagree is that a priori belief that monetary expansion must always have unambiguously harmful effects.”
Hayek also argued that monetary expansion could have certain positive effects that were worth other economics costs because of their political benefits, or because of there benefits on particular classes, or because a sticky wages problem was too costly, etc.
LikeLike
David, it by some sort of magic “forced savings” via monetary expansion creates exactly the set of production goods across time that perfectly coordinate with goods and prices which would be in some sort of equilibrium _after_ the monetary expansion, then, yes, Kaldor would be right and that a monetary expansion would indeed have beneficial output effects.
But if you think about it, this would be magic on the order of fantastic miracles.
I would suggest it is just very hard for people to imagine coordination and discoordination on this scale and across all goods and potential goods and across time and an infinity of alternative possible futures.
LikeLike
“But monetary neutrality means that money has no effect at all on relative prices and real quantities. That is an implication of a very narrow class of theoretical models similar to frictionless models”
ok fair enough: you are using the term in a very strict technical sense while i am using it in the weak sense. I agree that in the strict sense its an implication of a very narrow class of models. Unfortunately, i think the original idea for those models was to narrow the assumptions for money neutrality so that we could convince ourselves that those narrow assumptions were clearly and obviously violated in the real world (how about, um, nominal debt contracts??) – and move on to gauge the relative size of different frictions. Unfortunately it seems to have gone the other way: now we have a narrow class of economists who believe those frictionless models are true.
LikeLike
david, Agreed, path dependence does not imply error or “unsustainability,” an undefined term that Austrians throw around as if it were totally self-evident and requiring no further demonstration. Austrian attempts to justify it with simple Robinson-Crusoe parables do not strike me as even remotely plausible.
Greg, Fair enough. There is a range of views among Austrians about the circumstances in which monetary expansion could have positive effects. However, I think that there is probably too strong an a priori bias against monetary expansion even among the less dogmatic Austrians, but this may be hair-splitting.
But in your next comment you undercut yourself, because you set up an unnecessarily stringent condition for forced savings to be sustainable. I am suggesting that monetary expansion doesn’t have to create “exactly the set of production goods across time that perfectly coordinate with goods and prices.” I am suggesting that there is a lot of flexibility in the uses of most capital goods so that even if expectations are not perfectly realized the capital invested based on those expectations is not rendered useless. No miracles are necessary. Markets adjust and resources are reallocated over time.
dwb, I agree the concept of monetary neutrality has been bent completely out of shape.
LikeLike