The tortured intellectual history of the price-specie-flow mechanism (PSFM), which received its classic exposition in an essay (“Of the Balance of Trade”) by David Hume about 275 years ago is not a history that, properly understood, provides solid grounds for optimism about the chances for progress in what we, somewhat credulously, call economic science. In brief, the price-specie-flow mechanism asserts that, under a gold or commodity standard, deviations between the price levels of those countries on the gold standard induce gold to be shipped from countries where prices are relatively high to countries where prices are relatively low, the gold flows continuing until price levels are equalized. Hence, the compound adjective “price-specie-flow,” signifying that the mechanism is set in motion by price-level differences that induce gold (specie) flows.
The PSFM is thus premised on a version of the quantity theory of money in which price levels in each country on the gold standard are determined by the quantity of money circulating in that country. In his account, Hume assumed that money consists entirely of gold, so that he could present a scenario of disturbance and re-equilibration strictly in terms of changes in the amount of gold circulating in each country. Inasmuch as Hume held a deeply hostile attitude toward banks, believing them to be essentially inflationary engines of financial disorder, subsequent interpretations of the PSFM had to struggle to formulate a more general theoretical account of international monetary adjustment to accommodate the presence of the fractional-reserve banking so detested by Hume and to devise an institutional framework that would facilitate operation of the adjustment mechanism under a fractional-reserve-banking system.
In previous posts on this blog (e.g., here, here and here) a recent article on the history of the (misconceived) distinction between rules and discretion, I’ve discussed the role played by the PSFM in one not very successful attempt at monetary reform, the English Bank Charter Act of 1844. The Bank Charter Act was intended to ensure the maintenance of monetary equilibrium by reforming the English banking system so that it would operate the way Hume described it in his account of the PSFM. However, despite the failings of the Bank Charter Act, the general confusion about monetary theory and policy that has beset economic theory for over two centuries has allowed PSFM to retain an almost canonical status, so that it continues to be widely regarded as the basic positive and normative model of how the classical gold standard operated. Using the PSFM as their normative model, monetary “experts” came up with the idea that, in countries with gold inflows, monetary authorities should reduce interest rates (i.e., lending rates to the banking system) causing monetary expansion through the banking system, and, in countries losing gold, the monetary authorities should do the opposite. These vague maxims described as the “rules of the game,” gave only directional guidance about how to respond to an increase or decrease in gold reserves, thereby avoiding the strict numerical rules, and resulting financial malfunctions, prescribed by the Bank Charter Act.
In his 1932 defense of the insane gold-accumulation policy of the Bank of France, Hayek posited an interpretation of what the rules of the game required that oddly mirrored the strict numerical rules of the Bank Charter Act, insisting that, having increased the quantity of banknotes by about as much its gold reserves had increased after restoration of the gold convertibility of the franc, the Bank of France had done all that the “rules of the game” required it to do. In fairness to Hayek, I should note that decades after his misguided defense of the Bank of France, he was sharply critical of the Bank Charter Act. At any rate, the episode indicates how indefinite the “rules of the game” actually were as a guide to policy. And, for that reason alone, it is not surprising that evidence that the rules of the game were followed during the heyday of the gold standard (roughly 1880 to 1914) is so meager. But the main reason for the lack of evidence that the rules of the game were actually followed is that the PSFM, whose implementation the rules of the game were supposed to guarantee, was a theoretically flawed misrepresentation of the international-adjustment mechanism under the gold standard.
Until my second year of graduate school (1971-72), I had accepted the PSFM as a straightforward implication of the quantity theory of money, endorsed by such luminaries as Hayek, Friedman and Jacob Viner. I had taken Axel Leijonhufvud’s graduate macro class in my first year, so in my second year I audited Earl Thompson’s graduate macro class in which he expounded his own unique approach to macroeconomics. One of the first eye-opening arguments that Thompson made was to deny that the quantity theory of money is relevant to an economy on the gold standard, the kind of economy (allowing for silver and bimetallic standards as well) that classical economics, for the most part, dealt with. It was only after the Great Depression that fiat money was widely accepted as a viable system for the long-term rather than a mere temporary wartime expedient.
What determines the price level for a gold-standard economy? Thompson’s argument was simple. The value of gold is determined relative to every other good in the economy by exactly the same forces of supply and demand that determine relative prices for every other real good. If gold is the standard, or numeraire, in terms of which all prices are quoted, then the nominal price of gold is one (the relative price of gold in terms of itself). A unit of currency is specified as a certain quantity of gold, so the price level measure in terms of the currency unit varies inversely with the value of gold. The amount of money in such an economy will correspond to the amount of gold, or, more precisely, to the amount of gold that people want to devote to monetary, as opposed to real (non-monetary), uses. But financial intermediaries (banks) will offer to exchange IOUs convertible on demand into gold for IOUs of individual agents. The IOUs of banks have the property that they are accepted in exchange, unlike the IOUs of individual agents which are not accepted in exchange (not strictly true as bills of exchange have in the past been widely accepted in exchange). Thus, the amount of money (IOUs payable on demand) issued by the banking system depends on how much money, given the value of gold, the public wants to hold; whenever people want to hold more money than they have on hand, they obtain additional money by exchanging their own IOUs – not accepted in payment — with a bank for a corresponding amount of the bank’s IOUs – which are accepted in payment.
Thus, the simple monetary theory that corresponds to a gold standard starts with a value of gold determined by real factors. Given the public’s demand to hold money, the banking system supplies whatever quantity of money is demanded by the public at a price level corresponding to the real value of gold. This monetary theory is a theory of an ideal banking system producing a competitive supply of money. It is the basic monetary paradigm of Adam Smith and a significant group of subsequent monetary theorists who formed the Banking School (and also the Free Banking School) that opposed the Currency School doctrine that provided the rationale for the Bank Charter Act. The model is highly simplified and based on assumptions that aren’t necessarily fulfilled always or even at all in the real world. The same qualification applies to all economic models, but the realism of the monetary model is certainly open to question.
So under the ideal gold-standard model described by Thompson, what was the mechanism of international monetary adjustment? All countries on the gold standard shared a common price level, because, under competitive conditions, prices for any tradable good at any two points in space can deviate by no more than the cost of transporting that product from one point to the other. If geographic price differences are constrained by transportation costs, then the price effects of an increased quantity of gold at any location cannot be confined to prices at that location; arbitrage spreads the price effect at one location across the whole world. So the basic premise underlying the PSFM — that price differences across space resulting from any disturbance to the equilibrium distribution of gold would trigger equilibrating gold shipments to equalize prices — is untenable; price differences between any two points are always constrained by the cost of transportation between those points, whatever the geographic distribution of gold happens to be.
Aside from the theoretical point that there is a single world price level – actually it’s more correct to call it a price band reflecting the range of local price differences consistent with arbitrage — that exists under the gold standard, so that the idea that local prices vary in proportion to the local money stock is inconsistent with standard price theory, Thompson also provided an empirical refutation of the PSFM. According to the PSFM, when gold is flowing into one country and out of another, the price levels in the two countries should move in opposite directions. But the evidence shows that price-level changes in gold-standard countries were highly correlated even when gold flows were in the opposite direction. Similarly, if PSFM were correct, cyclical changes in output and employment should have been correlated with gold flows, but no such correlation between cyclical movements and gold flows is observed in the data. It was on this theoretical foundation that Thompson built a novel — except that Hawtrey and Cassel had anticipated him by about 50 years — interpretation of the Great Depression as a deflationary episode caused by a massive increase in the demand for gold between 1929 and 1933, in contrast to Milton Friedman’s narrative that explained the Great Depression in terms of massive contraction in the US money stock between 1929 and 1933.
Thompson’s ideas about the gold standard, which he had been working on for years before I encountered them, were in the air, and it wasn’t long before I encountered them in the work of Harry Johnson, Bob Mundell, Jacob Frenkel and others at the University of Chicago who were then developing what came to be known as the monetary approach to the balance of payments. Not long after leaving UCLA in 1976 for my first teaching job, I picked up a volume edited by Johnson and Frenkel with the catchy title The Monetary Approach to the Balance of Payments. I studied many of the papers in the volume, but only two made a lasting impression, the first by Johnson and Frenkel “The Monetary Approach to the Balance of Payments: Essential Concepts and Historical Origins,” and the last by McCloskey and Zecher, “How the Gold Standard Really Worked.” Reinforcing what I had learned from Thompson, the papers provided a deeper understanding of the relevant history of thought on the international-monetary-adjustment mechanism, and the important empirical and historical evidence that contradicts the PSFM. I also owe my interest in Hawtrey to the Johnson and Frenkel paper which cites Hawtrey repeatedly for many of the basic concepts of the monetary approach, especially the existence of a single arbitrage-constrained international price level under the gold standard.
When I attended the History of Economics Society Meeting in Toronto a couple of weeks ago, I had the pleasure of meeting Deirdre McCloskey for the first time. Anticipating that we would have a chance to chat, I reread the 1976 paper in the Johnson and Frenkel volume and a follow-up paper by McCloskey and Zecher (“The Success of Purchasing Power Parity: Historical Evidence and Its Implications for Macroeconomics“) that appeared in a volume edited by Michael Bordo and Anna Schwartz, A Retrospective on the Classical Gold Standard. We did have a chance to chat and she did attend the session at which I talked about Friedman and the gold standard, but regrettably the chat was not a long one, so I am going to try to keep the conversation going with this post, and the next one in which I will discuss the two McCloskey and Zecher papers and especially the printed comment to the later paper that Milton Friedman presented at the conference for which the paper was written. So stay tuned.
PS Here is are links to Thompson’s essential papers on monetary theory, “The Theory of Money and Income Consistent with Orthodox Value Theory” and “A Reformulation of Macroeconomic Theory” about which I have written several posts in the past. And here is a link to my paper “A Reinterpretation of Classical Monetary Theory” showing that Earl’s ideas actually captured much of what classical monetary theory was all about.
Uneasy Money 
Great post that I need to read again to fully digest.
I’m not clear on ‘ According to the PSFM, when gold is flowing into one country and out of another, the price levels in the two countries should move in opposite directions. But the evidence shows that price-level changes in gold-standard countries were highly correlated even when gold flows were in the opposite direction.’
PSFM in its simplest form seems to assume that demand for holding gold stays constant. If you relax that assumption and assume it can vary, then (starting from a point where PSFT has led to a consistent set of world prices) if demand to hold gold rises then wouldn’t you expect to see that prices would rise in all countries and this tendency for prices to move in the same direction globally would outweigh the relatively minor movements in gold and relative global prices needed to maintain equilibrium under PSFM ?
LikeLike
David: very interesting post, as always, but again I disagree. Here’s a simple model:
Gold is medium of account, but is not used as medium of exchange. Gold is used for jewelry only. The demand for gold jewelry is proportional to nominal income: G = gPY. Paper money is used as medium of exchange. The supply of paper money is proportional to nominal income M = kPY. Assume Y is exogenous.
Gold has zero transportation costs, but all other goods have “Sailing Ships” transport costs. Ships are costly to build, and have finite lives, but zero operating costs.
Suppose half the gold in England (a small country) is accidentally destroyed. The price of goods in England halves. The supply of paper money in England halves too. People now build sailing ships to transport goods from England to France, with gold flowing the other way in the otherwise empty ships. The annual rent on a ship equals the price differential times the volume of goods the ship can carry per year. The Expected Present Value of that rent equals the cost of building a ship. Only a finite number of ships are built. So a finite inflow of gold slowly returns the stock of gold in England to its original level, along with the price level.
LikeLike
As I understand it, the rejection of the PSFM amounts to an explanation of what doesn’t determine the international flow of gold under a gold standard (price level differences).
So what is the companion explanation for what does determine the international flow of gold?
LikeLike
Hi David,
This is a wonderful post. I like that in your blog you focus on quality over quantity. How much time does a post like this take you to write?
I have always had trouble understanding the gold standard. There are so many more factors involved than under a simple fiat system. What readings would you recommend to help to develop a solid grounding in it? I’m thinking of going through Leland Yeager’s textbook “International Monetary Relations”
LikeLike
Rob, Thanks an increase in the demand to hold gold under the gold standard causes the value of gold to rise, and prices in terms of gold to fall. In other words an increase in the demand for gold is deflationary under the gold standard. I agree that an increase in the total world demand for gold would cause prices to fall everywhere so that would account for the observation that prices were falling everywhere even if minimal gold flows were causing some second-order effects on relative price levels among countries. So that could account for the failure to observe correlations between price level changes and gold flows, but if there such a second-order effect, you would expect to see some evidence that gold flows between countries had a statistical affect on price-level changes. I don’t think that there is any evidence that price level changes were correlated with the direction of gold flows between countries.
Nick. Are you assuming that, before the shock to the English gold stock, there was trade or that there was not trade between England and France?
JKH, The flow of gold depends on changes in the geographic distribution of changes in the stock and changes in the demand for gold. Gold is produced in a few locations and exported to other places from the gold producing countries. Demand for gold increases in various countries by different amount for a variety of reasons. Gold flows from countries where there is an excess supply of gold to countries where there is an excess demand.
Ilya, Thanks so much. I actually worked on this one for over a week, but I was also doing some other stuff as well. Let’s see how long it takes for me to crank out part 2. I haven’t read Yeager’s book so I can’t give you a positive or negative recommendation. In general, I have a high regard for Yeager as an economist, but if you google Glasner and Yeager, you will find that I have been very critical of his treatment of Tobin’s wonderful essay “Commercial Banks as Creators of Money.” I think Nick Rowe is awesome, but he is much too attached to the quantity theory of money. I think Yeager has the same problem, and he not nearly as much fun to read as Nick.
LikeLike
David,
That strikes me as a marginal explanation (in response to my question above), in that it seems like a supply/demand ex post truism within what a full answer must be. For example, from previous post’s quote of Hawtrey:
“In Ricardo’s day, relatively considerable differences of price were possible between distant centres … Nevertheless it was fallacious to explain the adjustment wholly in terms of the price level. There was, even at that time, an approximation to a world price. When the difference of price level attracted goods from one country to another, the effect was to diminish the difference of price level, and probably after an interval to eliminate it altogether (apart from cost of transport). When that occurred, the importing country was suffering an adverse balance, not on account of an excess price level, but on account of an excess demand at the world price level. Whether there be a difference of price level or not, it is this difference of demand that is the fundamental factor.”
In that description, the flow of gold is based on trade deficits due to excess import demand for goods at the world price level. So my follow up question is – is it even possible to separate the answer to my first question from such a trade-based explanation? Because you seem to have framed your response above in that way.
And my third related question is – how do interest rates factor into a complete answer?
(This sort of thing is what I find frustrating about the subject matter. More generally, it seems difficult to find explanations that integrate all factors. Its like wandering through thickets of marginality trying to understand this stuff.)
LikeLike
JKH, You are right to say that my response to your earlier question left out a lot. One reason for the terseness of the response is that the reasons for shifts in the demand for gold are very numerous, so it would not be easy to provide a comprehensive list of all or even most of the reasons for fluctuations in the demand for gold. But you may have been expecting to hear more about why there could be changes in the monetary demand for gold. The monetary demand for gold by a central bank can be thought of as dependent on two factors a) the total demand by a central bank for foreign exchange reserves and b) the fraction of its foreign exchange reserves that it wants to hold as gold rather than claims to gold or to other foreign currencies which may or may not be convertible into gold. The way in which a central bank controls its holdings of foreign exchange reserves is by the lending rate it charges banks. When a central bank wants to increase its holdings of reserves it typically raises its lending rate, thereby reducing lending by the banking system which reduces domestic spending. For any given holding of foreign reserves, the central bank also decides how much to hold as gold and how much to hold in alternative, presumably higher-yielding, assets. If a central bank decides it wants to hold more gold relative to claims to gold, it can demand that those issuing the claims convert their liabilities into gold causing gold to flow from the central bank issuing the claims into the central bank cashing in the claims. Does that help?
LikeLike
Ilya,
Yeager’s textbook is IMO of much higher quality than his essays, although his essays are quite good. I recommend getting the second edition if you don’t already have a copy: The first half is theory and the second half is historical; the second edition’s history is much less outdated. (2nd ed was nonetheless written in the late 70s, I believe.)
If you (reasonably) don’t want to take the word of a random commentator, Kurt Schuler opines similiarly, in the reading list and comments.
https://www.alt-m.org/2013/06/30/monetary-economics-a-reading-list/
I highly recommend the book, though I admit I haven’t read it start to finish.
LikeLike
David:
1. The simplest model is where England and France produce the same good, and there is no trade and no ships before England accidentally loses half its gold.
2. A slightly more complex model would assume England and France produce different goods, and there is trade between them initially, with ships carrying full loads in both directions across the Channel. We still observe PSFM if England loses half its gold, because more ships will need to be built.
3. More complex still would be if English wheat exports mean ships are full sailing South, but return half empty carrying wine (which takes up less space per unit value). Again we observe PSFM,
4. Finally, if the ships sailing north were full, but half empty sailing south, then what I say in my first comment would not be true. In this 4th case, we would not observe PSFM if England lost half its gold (but would observe PSFM if France lost half its gold). Because transport costs would be zero going south, but positive going north.
LikeLike
5. Maybe a simpler model would be to assume that “ships” are specialised, and can only carry one particular good. This may be realistic, if we think of “sailing ships” as just a metaphor for the sunk costs of setting up a network/advertising etc. to export your goods.
(By the way, thank you very much for saying that. But there are lots of things I am bad at doing that you are excellent at doing. Like all the history in this post. I think I remember David Laidler saying, when teaching Monetary Theory of Balance of Payments, and Hume, that Harry Johnson had shown that PSFM was wrong.)
LikeLike
David,
“If a central bank decides it wants to hold more gold relative to claims to gold, it can demand that those issuing the claims convert their liabilities into gold causing gold to flow from the central bank issuing the claims into the central bank cashing in the claims.”
Can you be more specific about exactly what kind of “claim” is issued by the issuing central bank in this case? What does it look like on its balance sheet? i.e. – what is it called (apart from being categorized as a “claim”) ? What is it exactly ?
And how does the other central bank get a hold of that kind of claim on gold in the first instance? Where does it come from? i.e. – who was the counterparty from whom that central bank obtained that claim as an asset in the first instance?
LikeLike
separate point
I noticed this comment from Nick Rowe on one of your other posts. I think its rather nifty:
“Arbitrage” is just another name for a very fast and powerful price-specie flow mechanism.
LikeLike
If the demand for gold jewelry also depends on the opportunity cost (the nominal interest rate, which depends on the expected inflation rate), that will weaken the PSFM. The price level in England will still fall, but by less than half. when half the gold is lost.
LikeLike
Apologies to all for not responding till now.
Nick, The nominal price of gold is fixed (actually not quite true given the limits imposed by the gold points), so how does a trader make a profit from his gold transaction without actually using the gold to buy commodities which he can’t consume or resell at a profit without importing the goods back to his home country? I don’t see how gold arbitrage works unless real commodities are actually being shipped? Read Hume’s essay. He is talking about international competition to see tradable commodities.
JKH, I’m talking about short-term interest-bearing foreign debts held by central banks. I’m talking about whatever instruments central banks hold that are classified as foreign exchange reserves. I assume those instruments, whatever they are, are not actual foreign currency, but are interest-bearing claims to foreign currency that can be cashed in on short-notice or sold in secondary markets
LikeLike
I think the discussion around the subject is a bit off the mark, clearly Hume’s gold standard is the modern day flexible currency exchange rates then the theory of what happens to the economic activity applies as Hume postulates except of course if you have the government meddling as you now have
LikeLike