Did Free Banking Stabilize Canadian NGDP?

Canadian banking system, base money, free banking, economic history, monetary base
By Adrian Raeside http://raesidecartoon.com/vault/canada-economy-united-states-usa-economy/

Canadian banking system, base money, free banking, economic history, monetary baseAbout a month ago, a Facebook post drew my attention to an attempt, by Casey Pender of Prague's CEVRO Institute, to test my thesis that free banking contributes to NGDP stability using statistical evidence from Canada, which had a relatively free banking system between 1867, the year of Canada's confederation, and 1935, when the Bank of Canada was established.

In "Some Odd Data on Free Banking in Canada," a blog post discussing his preliminary findings, Pender reports that he had hoped to

be able to show that Canada, from 1867-1935, had a more stable NGDP percent change from year to year than the U.S. And I thought this would be an easy and quick historical example that I could use to bolster my underlying theory. But things seem like they just ain’t so.

Instead, in comparing the fluctuations of Canadian and U.S. NGDP using data from the Macrohistory database, Pender found that Canadian NGDP was not less but more volatile. Moreover that conclusion held not just for the full 1870-1935 sample period, but also for the sub-period 1870-1914, which omits various extraordinary Canadian government interventions during WWI and the Great Depression.

Here is Pender's chart showing his results from the full sample period:

Having now been made privy to these findings, I suppose that you are looking forward to seeing ol' Doc Selgin eat humble pie. Well, you can quit holding your breath 'cause that won't be happening anytime soon. In fact, for the moment at least, I remain thoroughly impenitent.

How come? First of all, I never claimed that free banking alone could achieve any particular degree of absolute stability of nominal spending. What I have claimed is that, by tending to offset changes in the velocity of money with opposite changes in its quantity, free banking makes for a more stable relationship between the level of overall spending and the available quantity of base or high-powered money than might exist otherwise. To the extent that the available quantity of base money itself changes, however, the quantity of money will also tend to change independently of its velocity. Total spending will then tend to vary also.

What's more, under an international gold standard regime like the one in place during Canada's free banking episode, the amount of base money in any particular gold standard country was hardly likely to remain stable or grow at a very steady rate. On the contrary: changes in trade patterns and international capital flows were bound to routinely alter the distribution of gold among gold standard countries, just as changes in trade patterns and capital flows within individual countries are bound to alter the distribution of gold among those countries' various regions. To the extent that it consisted of holdings of monetary gold, Canada's monetary base was no less subject to change than the monetary base of, say, Nova Scotia.

In fact during the free-banking era Canada's base money consisted of both monetary gold and paper "Dominion notes" first issued by the Canadian government in 1866. While some portion of these had to be fully backed by gold, there was also an un-backed or fiduciary component, the quantity of which rose from just $8 million in 1868 to $30 million by the outbreak of World War I. Until 1885 or so at least, those fiduciary issues, instead of being linked even loosely to gold flows, or otherwise regulated for the sake of economic stability, varied according to the Canadian governments' fiscal needs. Consequently Dominion note issues tended to be an additional cause of irregular changes to Canadian NGDP.

A proper test of the theory that free banking helped to stabilize Canadian NGDP must therefore consider, not just fluctuations in Canadian NDGP as such, but the relationship between those fluctuations and underlying changes in Canada's monetary base. So long as Canada's NGDP fluctuations were driven by underlying changes in Canada's stock of base money, instead of being independent of such changes, the fluctuations were perfectly consistent with the theory.

So, what does the record have to say about it? To find out, I had Tyler Whirty, the CMFA's trusty RA, run some simple regressions for me, with Canadian NGDP as their dependent variable, and Canada's stock of base money (M0) as the independent variable. The NGDP estimates are the same ones employed by Pender, from the Macrohistory database[1], while the monetary base numbers are from Metcalf, Redish, and Shearer (1998). So far we've looked only at the pre-WWI record, as that era is most aptly described as one of relatively unblemished free banking. We also ran the regressions both using raw data and after taking logs. I shall report only the log results, as those fit the data best; but the general conclusions to be reported here don't depend in the least on the log transformations.

The results, shown in the next figure, are not just consistent but remarkably consistent with the theory, and especially so given that Canada's arrangement involved some not entirely trivial departures from the theory's assumptions, one of which is that paper money is supplied by commercial banks alone. (In fact, Canada's banks were prohibited from issuing notes worth less than $5.) The regression R-squared is .980254, while the coefficient on M0, about .80, seems quite reasonable allowing for the fact that some Canadian base money, including smaller Dominion notes, circulated instead of serving as bank reserves.

Although the Canadian results seem perfectly consistent with the theory, we still have to compare them with evidence from the U.S. as a step toward establishing that the stable relationship they point to might be attributable to Canada's having had a free banking system: were a similarly close relationship to exist in the U.S. data, that would suggest that Canada's distinct institutional arrangements, including free banking, didn't really make any difference.

For our base U.S. regression, we drew again upon the Macrohistory database, using its U.S. NGDP estimates (which were originally developed by Louis Johnston and Samuel H. Williamson)[2]. For the U.S. monetary base, we subtracted national bank notes from Philip Cagan's monetary base figures: while Cagan himself justified including those banknotes by noting that they were practically claims against the U.S. Treasury, we exclude them because, despite that, they were neither legal tender nor capable of satisfying banks' legal reserve requirements.

Because Cagan's numbers only start in 1875, the U.S. regression covers a somewhat shorter period than the Canadian one.

While the results of the U.S. regression suggest a relationship between U.S. NGDP and M0 that's broadly similar to the Canadian one (M0 coefficient .825), that relationship, as depicted in the next figure, is considerably less tight, with an R-squared of .908. That the Canadian relation should be so much tighter now seems all the more remarkable, given both the relatively small size and openness of Canada's economy, and its heavy dependence upon U.S. exports to the U.S.

To check the robustness of these U.S. results, we performed the same regression using Angus Maddison's NGDP estimates in place of the Johnston-Williamson numbers. Although the R-squared, of about .935, is a bit higher in that case, the coefficient (.769) is similar:

In brief, both our Canadian regression results themselves, and a comparison of those results with results using U.S. data, seem fully consistent with the theory that free banking helps to stabilize the relationship between NGDP and the monetary base.

Does that mean they confirm the theory? Alas, it doesn't. Freedom in banking is but one of many differences between the pre-WWI Canadian system and its U.S. counterpart. Furthermore, even if Canada's more stable NGDP-M0 relationship were in fact due to its having had a relatively free banking system, it wouldn't follow that my theory is correct. Free banking could well have contributed to the stable relationship in question for reasons apart from the one my theory points to. We know, for example, that branch banking — itself an element of free banking — made Canada's system less fragile, and therefore less vulnerable to financial crises, than the U.S. system. We also know that financial crises tend to involve a collapse in bank credit and spending. So the relative stability of the Canadian NGDP-M0 relationship, instead of reflecting a tendency for changes in M to offset opposite changes in V, may instead simply have reflected a relative lack of banking crises and associated increases in the ratio of bank reserves to bank credit.  Although all this is still good news for fans of free banking, it leaves my particular hypothesis unproven.

In short, while my theory has yet to be discredited, it also has yet to be confirmed. I hope that either Mr. Pender or some other enterprising econometrician will eventually settle the matter, one way or the other.

[1] Òscar Jordà, Moritz Schularick, and Alan M. Taylor, “Macrofinancial History and the New Business Cycle Facts,” in NBER Macroeconomics Annual 2016, volume 31, edited by Martin Eichenbaum and Jonathan A. Parker. 2017. Chicago: University of Chicago Press.

[2] Louis Johnston and Samuel H. Williamson, "What Was the U.S. GDP Then?" MeasuringWorth, 2017.

  • Ray Lopez

    Lord Keynes in an alt-universe: "When the facts change, I change my mind. What does George Selgin do? He doubles down!". LOL. Another thesis to fit the data is that money is, by and large, and outside of hyperinflation, largely neutral. Relax and learn to love fiat money and inflation. Most of the time it does not really matter.

    • George Selgin

      Ray, no claims have been here made about real variables, because the question isn't about them. It is whether free banking make for a tighter relation between M0 and NGDP, as (my) theory suggests, or not. That there's bound to be SOME positive relation between these over time few people would doubt. So the R-squared differences matter. Do they settle things? As I said, they do not. Far from it being the case that I "double down" when the facts change, I do no more than point out that as yet the facts don't seem to refute the theory. If you have a better test or data to offer, propose it–as I also encouraged anyone to do. I will happily consider the results, as I considered Casey Pender's.

      • Ray Lopez

        OK thanks professor for that clarification. But I think that "whether free banking makes for a tighter relation between M0 and NGDP, as (my) theory suggests, or not" is exaggerating the statistics ( 0.980254 for Canada vs the fit of 0.908 for the USA proves nothing, they are both very close to 1, i.e. a perfect fit).

        Regarding money supply and NGDP, I suggest by analogy (since it deals with money supply vs inflation, not NGDP, though arguably if money is neutral as I claim, it also has a bearing on NGDP, since NGDP = RGDP + inflation) you study this graph: https://en.wikipedia.org/wiki/Money_supply#/media/File:Money_supply_growth_vs_inflation_rates.png

        1. Consider these countries: US, UK, Thailand, Denmark, Poland; then Indonesia, Brazil, South Africa, Chile, Mexico; then Singapore, Austria, France, Italy; then India, Columbia, China.

        2. What is your conclusion for 2014? Mine is that money creation does not really necessarily impact inflation, or else these 'spreads' would be much closer.

        Ergo, not only is money neutral, but arguably even base money creation, the raison d'etre of central banks, does not affect inflation.

        PS–another graph summed over many years not just one that shows we are arguing over nothing, debating money creation is about as relevant as memorizing railroad timetables for the ordinary person, is this, for the USA from 1980 to present, tracking inflation and M2: https://en.wikipedia.org/wiki/Money_supply#/media/File:CPI_vs_M2_money_supply_increases.png (note after 1995 no proportional increase in CPI with M2 monetary increases, this effect most pronounced after 2007)

        • "What is your conclusion for 2014? Mine is that money creation does not really necessarily impact inflation…"

          Money creation IS monetary inflation. The quite purposeful substitution of CPI measures, or consumer price increases, for an increase in the quantity of money, by central banks and nation-state governments enables a far more rosy assessment of the current economy, provides cover for the propping up of asset values, and dramatically reduces the real cost of state government entitlements. The price increases, from monetary inflation, obviously show up in assets, not just consumer prices.

          If we measured nominal GDP against the annual increase in the money supply, true inflation, the stark reality would emerge, that the major western developed economies are not growing at low single digits, they are declining by several percentage points in real terms. How is this possible? Very few are working, and many that do work are employed in national and state governments, are complying with government regulations, or are otherwise participating in unproductive or counter-productive activities. The small portion of the populations that are actually producing something can no longer bear the weight. Those not working are increasingly dependent on national governments to steal from those working through taxation or price inflation of their pensions and "investments". The non-working entitlement recipients and asset holders are the beneficiaries of ill-gotten gains that enable them to go out and compete for, and outbid, the largely working poor on products and services.

          While money is neutral in the sense that printing or mining more cannot feed, cloth, transport, entertain, house, etc., us, it is anything but neutral in the ultimate allocation of real wealth. If money were truly neutral, nation states would have no interest in monopolizing it.

          • George Selgin

            I can only say that I completely reject this way of thinking, which essentially throws all monetary economics (or the good sort, anyway) out the window. Having a well-working monetary system in a modern economy is not a simple matter of maintaining a constant money stock. Would that it were so simple! And not every received tenet of modern mainstream economics is part of a central bankers' conspiracy.

            Frankly the tendency to treat all money creation as "bad" on ethical grounds is no less wrongheaded than the tendency, exhibited every time there's a hurricane or other natural disaster, of many to treat any raising of prices similarly.

          • My analysis of the situation is 100% correct and I am not saying that all money creation is bad. And, I fail to see how monetary economics must be moth balled without a central bank or state involvement in money or banking. Quite the contrary. Finance departments of public corporations the world over are practicing monetary economics when they "increase their balance sheet" or "reduce their balance sheet" by issuing and buying back their security issues. Practically, there is no fundamental difference in what a public company does and what central banks do. The huge difference, of course, is the voluntary association and alignment of incentives, benefits and risks in public companies that does exist for central banks.

          • Aaron Cuevas

            Milton, the word "inflation" shouldn't even be use among careful students of economics. It's an irrational term.

            It's better to use the term "purchasing power of money" rather than using the word "inflation."

            Chacanosky writes: "The term inflation is not a praxeological term coming from economics, but an idea that comes from political or popular use. For this reason, the word inflation lacks the accuracy expected to be found in a theoretical term, becoming a tricky and sloppy concept. Mises writes in 1949 (p. 442):

            'The notions of inflation and deflation are not praxeological concepts. They were not created by economists, but by the mundane speech of the public and of politicians.'"


          • Well, perhaps we should not use it but it is used in good measure so for now we are stuck with it. So long as one defines how they are using it, and I have, we can at least be clear on what we are discussing or debating. My point stands that by excluding asset price inflation, the "official" "inflation" numbers are vastly misreported and, more importantly, the conclusions drawn from them are highly misleading. Further, we can skip all together the farce of coming up with baskets of consumer goods, assets, their prices, etc., by simply utilizing the known quantity, the increase in the supply of money.

          • Aaron Cuevas

            Then you would focus in the increases in prices of financial assets, what negative consequences that could bring, and what it's cause is, so as to prevent the negative consequences.

            Trying to use the word "inflation" in a way that is not popularly used in Academia nor among the masses, it's just bad communcation skills.

            All you have to say is "sure, consusmer prices are not wild, but asset prices are. That will create problems and so the caused has to be adressed, central banking increases in the money supply."

            We don't even need to use the word "inflation" to adressed the problems with central banking.

          • The following piece speaks to at least part of the points I am trying to make:


            An excerpt:

            Inflation, a term that first referred to
            a condition of the currency and later
            to a condition of money, is now commonly
            used to describe prices. This shift
            in meaning seems to have originated in
            an unfortunate—but perhaps inevitable
            —sequence of events. By referring to
            inflation as a condition of “too much
            money,” economists were forced to
            struggle with the operational issue of
            “how much is too much?”

            LOL… For the nation as a whole, economists were NOT forced to struggle with the operational issue of "how much is too much?"!!! Over a series of decades the central government nationalized money and banking through their own choice, and by force. Independent private suppliers/issuers of base money and banks were more than capable of producing money without the central state government or a central bank.

            Had academic economists not been charged with management of the national money supply, by the state, the definition would never have changed. In fact, without state or central bank monopoly privilege or control of base money and bank-produced money, there would never have been any discussion of inflation, using either the original, or most recent definitions, in the first place.

            In a free competitive market (and absent natural disasters, wars, or the like) for food, shoes, cars, houses, etc., no one need ever discuss, nor need worry about, any private producer, fully accountable to the marketplace, going off the rails and producing far more, or too little, food, shoes, cars or houses, than demanded by the that market. Multiple producers subject to potential losses, and immediate feedback from customers ensures as close to an optimal quantity of production as possible. With regards to this matter, money is no different than any other good.

            The Soviet Union, incidentally, due to monopolization/cartelization/central planning of nearly all facets of production, DID have to be concerned with quantity inflation, or deflation, for goods such as food, shoes, cars and houses, and there are many examples, quite tragic, of their failure to properly manage the supply of goods. In the absence of prices, profits and losses, central planning was a complete failure in the USSR the same way central planning of money is a complete failure in the United States.

          • Aaron Cuevas

            and so we need to use a term that is understood today, by whomever is our audience.

            And then be consistent among audiences so there is no confusion.

            Purchasing power of money has that sociological quality. There is no reason to fight over definitions that are not used any more.

          • George Selgin

            It isn't true that people only had reason to concern themselves with the problem of an excessive or deficient money supply only once governments began to deliberately manage that supply. In the U.S. the "money question" long predated the Fed's establishment. And your statement about free markets producing monetary stability is too strong: certainly it isn't self-evidently true as you seem to suggest.

            At least, I hope it isn't, or else I have wasted a lot of effort trying to establish, using empirical evidence and theory, the (limited) stabilizing properties of unregulated monetary systems!

          • George Selgin

            I don't see any harm in using the word inflation just as most economists use it, viz: to refer to an ongoing, positive rate of change of the general level of prices, which means the same thing as a negative rate of change in money's "purchasing power" (or what Mises sometimes referred to as its "outer objective exchange value"). Of course, measuring either that purchasing power or its reciprocal (the "price level") is fraught with well-known dangers. But that's no reason to resist either concept.

            Generally, one cannot get far in forwarding economic knowledge if one insists on defining terms differently than most other authorities. When it comes to how commonly-employed terms should be understood, Mises is not the relevant authority to defer to. That would be a fellow named Webster.

            For some of my thoughts concerning Mises' dislike of attempts to measure the price level, see http://citeseerx.ist.psu.edu/viewdoc/download?doi=

          • Ray Lopez

            @miltonchurchill:disqus "Money creation IS monetary inflation" – in theory, but the data I linked to in the two charts shows in practice it is not. I am not doubting that there's an inflation tax, but in these days of inflation adjusted securities that tax is very small, not to mention actual inflation in western countries is less than 2%. Do you get excited over 2% Milton? You're easily excitable if so.

          • It should be obvious that I do not submit to the CPI = total price inflation theory. Nor do central bankers by the way. The Fed, in particular, has made very public statements regarding their desire to create a "wealth effect". Now, they call it appreciation, but we all know that what they have done is generate nominal financial asset price increases, not increase real wealth. Asset holders realize an increase in their claims to consumer, producer and real goods at the expense of non asset holders.

  • Aaron Cuevas

    I had this question last week. I am glad you addressed it.

  • The problem is that state central planning and targeting of NGDP, assuming a central bank base money regime, necessarily requires a violation of the liberty and property of individuals so subjected to the central authority. Individuals, whether acting alone, or as part of voluntarily associations, should be free to chart their own course, free from interference from anyone. In short, it's nobody's business, including politicians and state economists. Whatever the NGDP turns out to be, that's what it is. Assuming the legality of chattel slavery, we might also argue for or against state regulation of agriculture and whether or not crop output would be better with, or without, the central authority. Of course, the question is mute when the entire foundation of the inquiry runs counter to natural rights.

    • George Selgin

      Milton, it should be clear from the post that NGDP stability here is treated as a theoretical ideal, which as such says nothing about the pros and cons of particular means for achieving it, including central banking, which I never even mention, let alone endorse, in this post. On the contrary: all I do say is the Canada's experience isn't inconsistent with the claim that free banking contributes to NGDP stability.

      In any case it is as improper to suppose that favoring stable NGDP as a theoretical ideal _means_ favoring central banking of the money supply as a means toward that end as it is to suppose that favoring having price system that accurately reflects the state of goods' relative scarcities means favoring having prices set by a central planning bureau.

      • Hi George, I am certainly in no way, at least intentionally, insinuating that favoring stable NGDP as an ideal equates to support for central banking. I believe I understand the context of this post and your prior one referenced. Your analysis seems spot on to me and at a minimum we can conclude that, violations to liberty and property rights aside, free banking is no more destabilizing than regulated banking and on the whole there is no proof that a central bank or state-produced base money versus a free-market-determined money, and base for banking, produces more stable results.

        I concur with your theoretical support of stable NGDP, and believe the best way to achieve this result is through the absence of state involvement period. All I am trying to point out, is that supporters of state interference must justify theft and violations of free contractual exchange even if they could prove that monopolization of base money and state regulated banking did produce more stable or even greater NGDP or wealth, for that matter. But, of course, they cannot prove this empirically, or otherwise. They are left supporting theft and interference with people's rights to trade as they please, without a shred of evidence to support their thesis.

        • George Selgin

          Milton, you should consider that many supporters of NGDP targeting take existing institutional arrangements for granted, not necessarily because they regard them as ideal, but because their concern is to recommend immediately-implementable changes that can improve things somewhat, without ruling-out more fundamental, future reforms.

          • Matthias Görgens

            If anything, the quip is that NGDP level targeting makes macroeconomics approximately classical. Things like Say's Theorem will just work.

            A stable economy is one where voters will have an easier time stomaching neoliberal reforms—getting us closer to some more libertarian ideals.

  • Mattyoung

    An expanding steam railroad economy across the great northern expanse. Cash balances were telegraphed, rail arrivals may be once a week. I doubt the central bankers could have done it. What happens in 1912? Suddenly all the high powered money has personal telephones in their homes. The banking system has to get the technology change right.

  • Aaron Cuevas

    Dr. Selgin, beyond the overall stability of NGDP, your free-banking theory (FB) for sure stabilizes the seasonal and regional money stream (Ms)* in the economy by responding to increases in currency demand (CD) with their liabilities.

    Even if other factors are capable of either stablizing or de-stablizing the annual and national Ms, it's logically unescapable that FB adds stability to the Ms by moderating wild fluctuations caused by high seasonal and regional CD. Changes in CD can't be the only factor that influences the total Ms. But FB reduces the severity of Ms fluctuations

    Plus, there is the added benefit that a much lower quantity of banks' financial assets have to be sold for currency if the banks are allowed to create their own currency. That stabilizes the prices of assets that are demanded by the savers and entrepreneurs, not the assets demanded by bureaucratic fiat.

    Your free-banking theory is much more rich and robust than annual/national NGDP stability. The empirical evidence that you are correct is that Canada didn't suffer the currency crisis that the USA suffered, independent of NDGP. It's logically unescapable

    • George Selgin

      Aaron, you absolutely are correct to point out that, whether it offsets V changes with opposite M changes or not, free banking definitely reduces, if it doesn't eliminate entirely, the tendency present in less free systems for seasonal and other fluctuations in the public's desired currency-to-deposit ration to cause M itself to fluctuate. Thanks for reminding me and other readers of this indisputable and important advantage of Canada's system.

      • Aaron Cuevas

        Late but important Dr. there is an important logical consequence of your free banking model.

        Since Canada's banking system had limited entry, during booms they had less adverse clearings (to slowdown NGDP) and less credit expansion during busts (speeding up NGDP and investment recovery).

        More liberalization of the banking system would have stabilized NGDP far more.

    • Aaron Cuevas

      *by the way, I think money stream is a better term than NGDP.

      • George Selgin

        Perhaps, Aaron; and your term is certainly more generic and general. I likewise often prefer to simply speak of total spending; but here again (as with the term "inflation," discussed in the comments below) I prefer to use terms that are already relatively popular. Moreover, the matter here is statistical. One cannot run a regression on "the money stream." One has to run it on some actual statistic that serves as a reasonable, specific and concrete representation of that notion.

  • Warren

    This is all well and good but how pure was their electrum? What are you hiding Selgin!?