My views on money and banking

Money and banking is one of those big picture topics that are endlessly debated. Although it’s a pretty complex topic — or perhaps because it’s a complex topic — I have often found that people are satisfied with lazily picking a view without diving too deeply into it. And yet, your views on money and banking color so many other opinions about history and policy, because the monetary system is upstream of so many social phenomena. I think it’s important to get it right. In this post, I try to articulate my own views as a way to clarify them to myself (and to whoever may be interested in my views).

I first lay out my criteria for what I consider “good money”. I then go through a historical whirlwind tour explaining how and why we gravitated away from the ideal I laid out. At each step, I try to highlight why these changes led to the fragilization of the monetary and banking system. At the end of the post, I try to bring it all together and speculate on what the future may have in store. Many thinkers have influenced my views, which I try to acknowledge throughout the post, as well as in the references at the end of the article. That said, I do not pretend to have done an exhaustive literature review, or even an unbiased one.

It seems like I can’t help writing massive, rambling essays lately, so I apologize for yet another meandering one. Thankfully I mostly write those for myself, but I’m happy if there’s someone out there with whom it may resonate. As always, I’d be delighted to hear any feedback. Don’t be fooled by my sometimes overly assured tone — we’re all learning here. I see tranchantness as a device for eliciting feedback more than anything else. My goal with these is always to ever so slightly update my way toward greater wisdom. The tranchantness reflects a methodological device, not an ideological commitment. Opining based on limited information is inevitable, and is paid for by the willingness to change one’s mind.

Good money

Like any other good, the value of money is a function of demand for it and its supply. Good money is money whose value is highly predictable. On the supply side, inflation or deflation is neutral to economic growth, as long as they can be anticipated. This highlights the importance of a predictable supply schedule, which allows prudent agents to set prices and wages appropriately in long-term contracts. On the demand side, there are various reasons people want to hold money instead of other assets: (1) for daily transactions (velocity); (2) as a precaution for unexpected expenses; (3) for speculation (when you think other assets might go down in value or you are very risk averse); (4) your income is increasing (just as wealthier people demand bigger houses and bigger cars, they tend to demand more money); (5) ceteris paribus, if the price level is high, you will need more money; (6) the higher the cost of converting non-money assets into money, the more money you will want to hold.

Historically, the best way to make money supply predictable has been to rely on some “natural process” of money production. Famously, metal extraction is such a “natural process”. It is predictable because gold production is likely to track growth of the overall economy: gold mining is a technological problem and it is expected to improve at the same rate as general technological progress (which ultimately underpins economic growth). On the demand side, of the six determinants I identified above, (1) and (4) are probably the ones most likely to change. Changes in money velocity (item 1) tend to be seasonal. There are periods where many transactions need to happen or investments need to be made, and demand for money rises. For instance, a yearly crop harvest, or a new technological development inspiring a spur of investment activity. The impact of rising income on demand is more subtle. If the income elasticity of the demand for money is unitary (i.e. an increase in income leads to a proportional increase in demand for money balances), then there is no need for the supply of money to accommodate the increase in demand. Assuming inelastic long-term supply, prices will come down at the rate of economic growth and the purchasing power of money will rise accordingly, satisfying the increased demand for money balances. There are good reasons to expect the elasticity of the demand for money to be unitary.

Generally speaking, therefore, you want money supply to be short-term elastic (to accommodate temporary changes in velocity) and long-term inelastic (unitary income elasticity of demand is self-accommodating). As for any other scarce good, demand and supply of money are equilibrated by adjusting its quantity or its “price”. As a medium of exchange, there is a regular, short-run demand for money, which can fluctuate either seasonally or throughout a natural business cycle. It’s useful for prices in the short run to be stable, therefore it makes sense for short-run money demand fluctuations to be managed via an adjustment in quantity. Such changes in the quantity of money are useful for facilitating exchanges in the short term, but they cannot persist long-term lest they would undermine money as a store of value and as a unit of account for contracts. The solution to this tension is credit money. Short-run fluctuations in demand for money can be accommodated through promissory notes, which have greater velocity and whose supply can be elastic via fractional-reserve banking. When such demand for money subsides (e.g., we are past harvest season), the notes are redeemed. In contrast, long-term changes in the demand for money (eg, due to incomes rising) should be adjusted via changes in its purchasing power (aka adjusting its “price”).

The long-run inelasticity of money supply also means prices, including perhaps those of production factors, are to be free to move downwards, as is bound to happen if technological progress increases the efficiency of production. This does not necessarily mean wages go down, as technological progress also increases productivity.

Two important remarks on credit money. First, you want the value of whatever is backing credit money to be as stable as possible, because instability in the value of base money will be magnified by the credit creation process. Second, the credit creation process needs to be self-limiting. Historically, such self-limitation was implemented by making notes redeemable for “hard money”. Overissuance will thus cause a feedback loop via adversarial clearing of the excess of notes, leading to an outflow of hard money for the overly enthusiastic issuer. Historically, this fast-acting self-regulating mechanism has been a staple of any free banking system. In free banking, banks have a natural incentive to accept each other’s notes in deposit, with or without a discount, because it brings more business. If a merchant sells goods in exchange for notes from Bank A and then deposits the proceeds in Bank B, it makes sense for Bank B to accept the deposit in the form of Bank A notes. That said, banks also have an incentive to redeem each other’s notes, also known as clearing, and settle any balance with specie. In a free market, banks will naturally form a consortium to establish a common clearinghouse — improving the efficiency of adverse clearing.

The reason banks are incentivized to establish a clearing mechanism is that they compete, among other things, on how many notes they issue, and how they secure those notes. In the eyes of the issuer, a bank note is a liability, in the sense that it is on the hook to redeem it for specie. Liabilities typically come with an asset to secure them, known as a collateral. It is the thing the creditor can seize if the debtor defaults. For example, a house loan is secured, or collateralized, by the title of the house. If the borrower cannot repay, the lender takes possession of the house. Similarly, bank notes need collateral, and that’s the capital of the bank, which (simplifying slightly) is basically the amount of specie it holds. In other words, banks compete in how much and with what they collateralize their notes, making them more or less secure. You don’t want to hold on for too long to a note whose security you’re not so sure of. You’ll go ahead and clear it. In particular, say Bank A receives, from a customer, notes from Bank B. If Bank A is not sure about the security of those notes, it will be motivated to quickly redeem them for specie, or clear them, at Bank B.

Adverse clearing just means that the bank that happens to extend a lot of credit by issuing a lot of notes will quickly see all the other banks redeem those notes, resulting in an outflow of specie from the aggressive issuer’s capital. To maintain a desired reserve ratio, that bank will need to curb its credit, lest it suffers what is effectively a speculative attack from its competitors. This is known as the marginal liquidity cost of issuing notes, referring to the outflow of specie at the margin when issuing one more note. As interbank clearing typically would happen on a daily basis, this fast-acting disciplinary mechanism is what prevents overextension of credit in a free banking system. Competition ensures banks remain honest and any overissue is self-abortive. This picture also preempts a common critique of plural issue, that dealing with multiple notes entails high transaction costs (discounting notes) and monitoring costs (disciplining overissue). The public would typically delegate these two tasks to the banks themselves, who have a strong incentive to institute par acceptance and to check overexpansion, and this is indeed what we observe in the historical record.

Those are the general principles of good money. Over the past two centuries we have significantly deviated from those principles, for a variety of complex reasons. I argue a series of developments have made the system artificially more fragile. The developments are:

• Par acceptance
• Unconditional redeemability
• Nationalization of note issuance
• Tender laws
• Lender of last resort
• Deposit insurance

I’ll go through each of those.

Gone bad

Mandates and assurances

Bank notes used to be more or less discounted depending on the perceived solvency of their issuer. This served the dual purpose of (1) financing redemption, and (2) producing valuable market information about a bank’s health (or, in other words, an effective wildcat radar). As I’ve mentioned, incentivizing redemption is a crucial part of credit self-regulation. But this was somehow considered “messy” and par acceptance was eventually mandated. This has the side effect of dismantling the wildcat radar, making it harder to know how some bank’s difficulties are affecting other banks, and creating a climate of general suspicion in hard times.

At the same time, unconditional redeemability was also mandated. Until then, it was routine for banks to design contingent-redeemability contracts that would suspend cash payments in cases of force majeure, with proper compensation for the client, and for mutual benefit. Mutual benefit indeed, because, even if the bank ended up being liquidated, organized suspension of convertibility would ensure an ordered liquidation rather than a firesale which could in turn affect proceeds. With par acceptance, you remove a useful signal of banking health and create uncertainty. With unconditional redeemability, you incentivize panic. Later, deposit insurance was added to the mix, further reducing depositors’ incentives to check solvency and evaluate trust, leading to more careless entanglements than would otherwise be.

All these mandates and assurances result in a system that now has acquired nonlinear dynamics. Of course, regulators then attempt to address all those side effects with various rules and constraints on bank behavior, but an artificial game of cat and mouse between profit-seeking banks and regulators has started, and complexity inevitably piles up. This increased uncertainty affects both the supply of money and the demand for money. We have now introduced the possibility for “bad deflation”, which happens when prices decline across the board without an improvement in the economy’s productivity. Bad deflation typically is sharp and severe, hard to predict, and originates in a drop in demand, while good deflation is gradual and originates in technological change or general productivity improvements in supply. Later, we will see which other developments added to these nonlinear dynamics.

Ironically, those uncertainty-inducing institutional changes were motivated by what seemed like a somewhat “chaotic” financial system in 19th century America. But not only have we seen that what to the naive observer seems “messy” may in fact be highly orderly (eg, note discounting), in addition whatever real instability existed in the system was more often than not caused by misguided regulation. I’ll just mention two examples: (1) anti-branching laws; (2) bond collateralization requirements. Selgin (2017) does a great job detailing both.

The rationale for anti-branching laws was to avoid consolidation of the banking industry into a small number of very large banks. The resulting banking landscape was that of tens of thousands of “unit banks” with only a local presence, as banks were disallowed from opening branches in other locations. In contrast, Canada next door had no such restrictions and several dozens of banks were operating nationally — a much smaller number, but still plenty for healthy competition. The fragmentation of the American banking system had tremendous consequences, first and foremost a disproportionate exposure of banks to local risks. A second consequence was to weaken the interbank clearing mechanism, which, as we saw, is a crucial feature of free banking by checking overexpansion via adverse clearing. Indeed, anti-branching made it difficult to redeem remote notes and to organize interbank clearing amongst tens of thousands of banks. Finally, the inability to open branches meant that each bank had to establish a network of “correspondent banks” in other locations to accommodate their customers’ out-of-town withdrawals. Those correspondent banks themselves had correspondent banks, etc. Additional regulation allowed banks to count as reserves their cash held at correspondent banks. But it also allowed the correspondent banks to count the same cash as reserve, thereby unwittingly creating a fragile pyramid where the same dollar was counted as reserve by multiple parties.

While you still would have had centralizing tendencies without anti-branching, you wouldn’t have had this pyramiding phenomenon. The pyramid typically ended at New York banks, where most reserves naturally gravitated. In turn, New York banks would routinely loan those reserves out for stock market gambling. Risk was heavily centralized as a result, purely as a consequence of ill-advised regulation. Correspondent bank balances served the big New York banks well, though, creating a natural alliance with the small bank lobby (who feared getting gobbled up without anti-branching laws) to keep the status quo.

The second major restriction on banking activity was that “free” banks were not in fact free to secure their notes as they saw fit. They were mandated to use specific collateral, from a narrow list that included state and federal government bonds, as well as those of some select railroad companies. This list was further narrowed to a single asset during the Civil War when the National Bank Act was passed, which required that notes issued by federally-chartered banks be collateralized by Treasury bonds only. In fact they had to be overcollateralized: for every $100 in notes issued, the bank had to acquire $110 in federal government bonds — a convenient way to force public debt down people’s throats in times of war.

The setup survived long after the war, and the unintended consequences reached far. This rule in effect tied overall credit money supply to the amount of public debt, such that the amount of inflation or deflation of the money supply was driven by whether the government was putting out more debt or paying it back, rather than by the actual demand for money. In fact, during the second half of the 19th century, the US government tended to bring down the national debt, thereby artificially shrinking the money base and increasing the fragility of the banking system as a result. That this collateral requirement led to unnatural movements in the supply of money is made evident by comparing the US and Canada over the same period. The differences are striking. While the Canadian money stock shows clear seasonal patterns with spikes during harvest season, the US exhibits no such pattern. This lack of elasticity of the US money supply was a direct consequence of the bond collateralization requirement. This artificial money scarcity might even explain why people sometimes had to knowingly resort to accepting notes from shady banks in order to trade.

Another subtle but important side effect of bond collateralization is that it homogenizes notes. Since notes are all backed similarly, banks will tend a bit more to treat every other bank’s notes as their own. As a result, a bank receiving another bank’s notes, instead of redeeming them, will just reissue them, a further blow to the crucial mechanism of adverse clearing. This, combined with the fact that the Federal government accepted national bank notes at par, conspired to make national bank notes almost de facto high-powered money, which in turn could be used to back further monetary aggregates.

The rise of central banking

We’ve seen thus far how misguided regulation led to more misguided regulation to fix the side effects of the former. The story was mostly US-focused. In Europe, another key development was happening much earlier than in America: the rise of central banking, which would have deep ramifications for the fragility of the banking system.

In 1694, the Bank of England was founded to provide “privileged lending” to the British government, which Bagehot (1873) depicts as being “in desperate want of money” at the time. This marks the beginning of the public debt era, which changed everything in the competition amongst the great European powers. Ferguson (2001) argues that the development of a system of national debt went hand in hand with the establishment of central banks. Beyond anticipating future tax revenues, a public debt gave government flexibility. This in turn tended to improve a state’s chances of survival in military competition, especially when borrowing at below-market rates (the whole point of privileged lending). This is what leads Ferguson to argue that “military expenditures have been the principal cause of fiscal innovation for most of history.”

How could privileged lending be sustained in the long term? Wouldn’t the lending terms naturally gravitate back to the natural rate? The trick was to nationalize note issuance. No one else but the central bank is allowed to issue notes. Combined with tender laws, this has the effect of turning central bank notes into base money. They’re basically made as good as gold by decree, artificially increasing demand for it, thereby enabling the central bank to print more of them than would happen in a free banking context. And sure enough, in 1742, the British government made the Bank of England the sole issuer of bank notes. And by 1833, those notes became legal tender. By the mid-19th century, the Bank had acquired most of the attributes of a modern central bank. The only fundamental thing still distinguishing it from today’s central banks was that it had to honor the conversion of its notes into precious metal. Notwithstanding, the pound sterling experienced many crises that temporarily suspended convertibility (sometimes for up to two decades) until its outright abandonment in 1931.

Why do the nationalization of note issuance and the establishment of tender laws constitute such a crucial evolution? First of all, nationalization means abrogating the adverse clearing mechanism. Second of all, tender laws make bank notes as good as gold. How so? Tender laws stipulate what can be used to settle debts. Without such laws, it’d be up to whatever the contract between two or more parties stipulates: it could be gold or number of sheep or literally anything. But the introduction of a legal tender can override such private agreements. If you look at a dollar bill, you’ll notice it reads “This note is legal tender for all debts, public and private”. The heading of the note signals that it’s issued by the Federal Reserve. The fact that it’s legal tender means you can force someone to accept it in repayment of debt derived from a private contract, and is valid for final settlement, even if the private contract stipulated payment in sheep. This is why this little verbiage on a Federal Reserve note means it’s as good as gold.

Modern seigniorage

You may start to see how all this enables privileged lending. The reason the central bank is able to lend cheaply is because it can extract seigniorage via money printing, which is old-fashioned coin debasement on steroids. Issuance of bank notes beyond what would have been possible in a free banking market is the modern form of seigniorage. Thus privileged lending is itself an artifact of seigniorage.

And just like with coin debasement, money printing is most egregious in times of war. For instance, Friedman and Schwartz (1962) showed that 75% of the increase in the US money stock between 1916 and 1920 went toward war financing. It’s worth pausing on that for a moment. The alternative to financing war through inflation would be to finance it through either taxes or bond sales. The former need consent from the people in a representative government; the latter needs to be voluntarily subscribed by the public. In a free banking system undergirded by specie, bonds need to eventually be repaid in specie and government borrowing is constrained by what can credibly be repaid by future budget surpluses. Inflation, on the other hand, can be done with or without the consent of the people. While there are limits to how much inflation can be generated without causing an internal revolt, the surreptitious nature of inflation gives it considerable runway. As such, it allows a state to finance war much beyond what would be supported through taxes and bonds. We therefore expect wars to become more destructive as the state acquires more control over the money stock. It’s no surprise that Napoléon Bonaparte deliberately nationalized note issuance to finance his wars.

Generally speaking, the immense fiscal benefits of central banking are the reason for its spread to most countries. Seigniorage profits were the main motivation behind the establishment of central banks in countries like Sweden, Italy, Portugal, Spain, Brazil, and China (White, 2014). The last industrialized countries to adopt central banking were Canada and New Zealand, and they adopted it by 1934. Only later, once Great Power war became less likely in the nuclear era, did the rationale for central banking switch from warfare to welfare, and from military supremacy to activist and discretionary monetary and fiscal policy along Keynesian lines.

Monopolistic money printing has distinct advantages for the state over coin debasement. The latter introduces heterogeneity in the coins and leads to culling good ones and passing on bad ones, as per Gresham’s law, which states that “bad money drives out good”. In effect, and perhaps counterintuitively, debasement, while it obviously causes an inflation of bad money, also causes a deflation of good money, with unclear net effect. Indeed, in the face of an incoming tide of bad money, people will tend to hoard the good money, thereby taking it out of circulation, or export it to some other country, where, because it won’t be devalued there, they can get more for their good money. To avoid this strategic reaction from the current money holders, the state mint would have to offer a recoinage of old coins, but that’s costly and kind of an admission of guilt, so it’s rare in history. In contrast to these consequential side effects of coin debasement, printing more bank notes does not result in heterogeneity and therefore bears no risk of deflation; the new bank notes are indistinguishable from the old ones. We have a pure inflation tax.

Furthermore, as we saw, nationalization of note issuance means the central bank is isolated from the negative feedback loop of adverse clearing, removing a crucial natural barrier to overissuance. In turn, tender laws nudge commercial banks to use central bank notes as collateral on the same footing as specie. Credit expansion by the central bank will thus get compounded by credit expansion by commercial banks themselves. In other words, making a bank note legal tender artificially increases demand for it, which otherwise would have been constrained by trust in the issuer. In fact, commercial banks will prefer to hold notes as reserves over cash because it’s more convenient. Specie, in turn, will tend to find its way to the central bank and become centralized there, further shrinking the number of nodes in which base money resides, and increasing the chances of cooptation.

While in a free banking system every note issuer is constrained by adverse clearing, a central banking system is essentially only constrained by the central bank’s reserves. Such a system, mostly relying on self-discipline in lieu of market discipline, will tend to inflate the overall money supply and cause greater instability of the financial system. As the money supply inflates, the rise in domestic prices leads people to look for cheaper goods elsewhere. Since international trade in those times was settled in specie, you’d start to see a drain of specie to foreign banks, increasing the central bank’s risk of default. The central bank would then suddenly contract credit to stop the hemorrhaging, which in turn would trigger a contraction by the commercial banks as well, leading to a generalized credit crunch. By substituting the fast-acting feedback loop of free banking’s adverse clearing mechanism with the sluggish correction from an adverse international balance of payments event, we allowed credit expansion to go unchecked for longer, leading to more dramatic corrections and thus more marked boom-and-bust cycles.

Lender of last resort

With the rise of central banking, another development became inevitable. When banks no longer monitor each other as closely and when the system as whole has been made more fragile by unconstrained note issuance, the crises that become possible acquire unprecedented scale and contagion, hence the need for the central bank to act as the “lender of last resort”. This was well understood by Walter Bagehot, who, while he is often remembered as the first to theorize the lender of last resort, actually always considered central banking second-best compared to free banking, and lamented it as much. In Bagehot’s eyes, last resort lending was the duty of a central bank for the privilege of note issuance.

Of course, a lender of last resort runs the risk of creating moral hazard. Bagehot knew as much, which is why he was so adamant that, yes, the central bank should lend “freely and vigorously” in times of crisis, but only to solvent banks, and at punishing rates, in order to avoid moral hazard. This requires great restraint, and naturally central banks often fail to act accordingly. A guessing game of who will be bailed out has been added to the system.

A central bank has two main tools to adjust the money supply: (1) “open market operations” (buying or selling assets), and (2) the “discount window” (lending to financial institutions). Only a select set of banks (“primary dealers”) can participate in the former, while any bank can participate in the latter. Open market operations thus rely heavily on those primary dealers to spread the liquidity. The assets involved in open market operations are typically limited to government bonds but it can also involve other securities, particularly in times of crisis. The discount window is useful when the open market for some reason freezes up (primary dealers no longer dealing) and needs bypassing. During the 2008 financial crisis, new tools were created beyond these two, but they are basically just variations around these two fundamental approaches: either work through a limited number of primary dealers and rely on the market to spread liquidity, or spread liquidity yourself directly.

In Bagehot’s vision, open market operations would be preferred. The central bank does not have an informational advantage that would justify it lending directly to institutions via the discount window facility. In a time of crisis, when liquidity has dried up, the central bank may inject some via open market operations and let the private market lend to each other and figure out who shall be saved. Combined with quick unwinding processes for failing institutions (“living wills”), this may be the best one could achieve in this fragilized system. Of course, as Bagehot feared, this is not what typically happens in times of crisis. Central banks routinely extend credit to insolvent institutions. This was most obvious during the Great Financial Crisis of 2008 when the Fed deliberately allowed banks to post toxic assets as collateral for loans without an appropriately high interest rate. The irony is that the scale of the modern crises is used to justify ignoring Bagehot’s advice, in turn fueling greater crises down the road. And for some reason we remain blind to the actual source of the system’s fragility.

In fact, free banking does naturally evolve something akin to a lender of last resort — but one that functions much more closely to Bagehot’s vision. We’ve seen how reliance on adverse clearing drives banks to form a common clearinghouse. That clearinghouse in turn can act (and has historically acted) as a lender of last resort in hard times, but it is much more able to distinguish between insolvency and illiquidity, and much more likely to exert restraint.

Irreducible uncertainty

I’ve argued that free banking naturally leads to fast feedback loops and incentivizes mutual monitoring by banks — generally preventing the development of insidious systemic risk or what I loosely called “nonlinear dynamics”. Those emerge as a result of generations of intervention, whose history I retraced in the above. Fanciful mandates and assurances first anesthetized the checks and balances naturally built into a free banking system (adverse clearing, note discounting, conditional redeemability, etc.). The nationalization of bank issuance and the institution of the lender of last resort further increased the system’s fragility. The instability of the banking system is therefore largely man-made, not natural.

And yet there is a persistent belief that the banking system is special and, if left to its own devices, inherently fragile, thereby requiring some kind of state intervention. While most people have by now recognized that central planning doesn’t work, we nevertheless still accept it for money production and finance. Central banking is the last, but triumphant, remnant of high-modernist social philosophy.

It’s worth exploring, then, the source of this belief in the specialness of money and banking — and for that I turn to John Maynard Keynes. Indeed, the belief in the natural instability of the banking system is closely linked to Keynes’ famous notion of “irreducible uncertainty”, which underpins most of his thought. Essentially, Keynes believed the world is too inherently uncertain for people to take on significant initiative. In the banking universe, this uncertainty translates into instability, regular panics, etc. Let’s go through Keynes’ reasoning.

Entrepreneurs need people to spend their money so their goods and services get sold. But they also need people to lend them some of their money, so that they can build the capital necessary to make those goods and services. So some amount of spending and some amount of lending are both good news for entrepreneurs. What’s bad news is when people hoard their money. Keynes’s key point is that that actually happens more often than we think.

Keynes argued that interest rates do not in fact serve to equilibrate the supply of loanable funds and the demand for funds. Interest rates, he says, are just the reward for parting with one’s money, which we have a tendency to want to hoard. He calls that tendency “liquidity preference”. The higher the interest rate, the more likely savers are to part with their money and lend it out; but also the less likely entrepreneurs are to think they can come up with a venture that would justify borrowing at such high rates. What if the interest rate is low? This is where things get interesting. Keynes argues that, because people have this fundamental liquidity preference, if the interest rate gets too low, they just won’t part with their money at all, or very minimally. So interest rates won’t go below that; there’s a floor. In technical terms, still according to Keynes, the elasticity of the demand for money becomes very high at sufficiently low interest rates. So interest rates are too low for people to part with their money.

But won’t entrepreneurs’ demand for funds increase when the interest rates are low, thereby driving the rates up? This is where Keynes introduces his notion of irreducible uncertainty. Keynes says that the future is just too damn uncertain for entrepreneurs to form enough long-term plans to match the amount of loanable funds available. Saving, in other words, has a tendency to run ahead of investment. This should theoretically drive rates down, but we’ve seen that there is a floor to them. So we’re in a situation where the rates are too low for lending because of “liquidity preference” and too high for investing because of “irreducible uncertainty”. That’s what Keynes calls the liquidity trap. Interest rates remain higher than needed to equalize saving and investment at full employment. Equilibrium, then, must be reached by a fall in income, which will reduce people’s propensity to save. This typically means unemployment. Alongside the problem of the structural mismatch between saving and investment, Keynes emphasizes the additional problem that irreducible uncertainty also tends to reduce spending. And as we saw, entrepreneurs (i.e., investment) need both money to be lent to build up their capital and money to be spent to get a return on that capital. Thus, not only might interest rates remain too high for entrepreneurs to take risks, spending might also be too depressed for those who did to turn a profit.

This is a pretty bleak state of affairs. Keynes was obsessed with this idea that the future is too inherently uncertain. It was essentially the bedrock of most of his intellectual edifice. Any deviation from this grim baseline was deemed a temporary exuberance that couldn’t be relied upon. As Skidelsky (2010) puts it, “Keynes believed that, under laissez-faire, full-employment levels of investment were achieved only in moments of excitement strong enough to overcome the uncertainty normally attaching to estimates of future returns. The normal tendency was for the propensity to save to be stronger than the inducement to invest.”

From this, Keynes comes to the conclusion that only the public sector has the requisite “diamond hands” to face true uncertainty, which he takes care to distinguish from “risk”. For Keynes, risk is mathematically tractable and can be left for the market to handle: entrepreneurs can be trusted to bet on such risks, and profit or suffer from those bets. In contrast, there are some truly uncertain activities, with large impacts, that are not tractable by the market. Those activities should be controlled by the state in the public interest. How to make this distinction between risk and uncertainty in practice is the key challenge of political leaders.

And because there is a floor to interest rates due to liquidity preference, expanding the money supply alone won’t necessarily stimulate more investing and spending, according to Keynes. The money just goes into the liquidity trap. Keynes argues that monetary policy is powerless lest it be combined with fiscal policy to overcome the “irreducible uncertainty” plaguing man’s plans. And indeed, fiscal expansion often ends up being accommodated by the central bank.

Although the technical details of Keynes’s reasoning have been criticized and improved upon since then, Keynes’s core message and key takeaways are to this day fairly popular among economists, especially government-affiliated economists. For Keynesians, money hoarding and drops in consumption and investment are “natural”. Bad deflation, thus, is always around the corner, and something to be intently feared and avoided at all cost. Recall: deflation is bad when, at a given price level, there is either excess money demand or excess supply of goods. Deflation is good where there is neither excesses and productivity simply goes up.

The final key tenet of Keynes’ thought is the notion of wage stickiness. If there is a spike in money demand, people will cut their spending to grow their money balances, leading to unsold inventories and rising unemployment, requiring a decrease of prices and wages. Nominal wage has to come down because real wage is going up, but it’s not going up because of a productivity increase; it’s going up because of a fear-driven spike in money demand. Keynes argues that that necessary adjustment won’t happen because people are reluctant to change their nominal wages (“wage stickiness”). In contrast, in good deflation, prices move together with real income: prices fall at the same rate real income rises. In good deflation, nominal wage is fixed so real wage goes up, but that’s okay because it goes up by the amount of productivity growth.

Wage stickiness is another argument that is often invoked to justify fiscal and monetary expansion. But the phenomenon of wage stickiness is closely tied to the currency regime. Wage stickiness naturally emerges in a regime of inflationary currency: expectations regarding the inflation rate changes the degree of price and wage stickiness. If you expect the price level to generally go up, why would you agree to a wage cut? In a regime of predictably deflationary currency (which arguably has never really been tried), it’s unclear wage stickiness would develop.

Of course, the Great Depression often serves as Exhibit A in support of Keynes’ views. That said, Keynes’ analysis of the depression is still hotly debated, and the fiscal solution he proposed is nowadays generally considered a failure by most economists. While I cannot do full justice to the complexity of the topic, I’ll say a few words. Europe’s throes with postwar runaway inflation made price stability a priority. In particular, in the US, through the 1920s, the Fed aimed to maintain the price level, even though it was a period of high productivity growth that would have naturally led to “good deflation” (more on that later). The Fed did so by increasing the money supply, fueling company profits, which rose faster than production factor prices (wages and interest rates basically). This in turn stoked unrealistic expectations in the stock market, and when factor prices started catching up, a severe correction became inevitable. Without going into too much detail, the high-level point here is that the Great Depression is a shining example of what I mean by systemic risks slowly building up because the fast feedback loops have been incapacitated, and then quickly unraveling, to considerable damage.

Freeish banking

And yet, is it true that entrepreneurs are regularly mired in the fog of uncertainty? Is it true that, left to their own devices, fear will periodically grip market participants, leading to bad deflation? How does free banking actually fare?

True free banking systems have been rare in history. In fact, there scarcely were any true “free banking” episodes, merely “lightly regulated banking” ones. The US banking system in the 19th century is often touted as an example of free banking, but we’ve seen that it was far from that. And yet, it keeps coming up as a cautionary tale. In particular, commentators will emphasize the antebellum period (before the Civil War), when private state banks were free to issue notes, a period frequently associated with so-called wildcat banking, or the fraudulent practice of incorporating a bank in remote places (where the wildcats are) and issuing worthless bank notes, backed by nothing, and hard to redeem precisely because of the remoteness of the issuer. This, however, is a clear misread of history: whatever financial instability existed then, it was often attributable to misguided regulation — chiefly the anti-branching laws and the bond collateralization mandate.

Scotland between 1716 and 1844 is perhaps the most famous historical example of freeish banking, and it’s generally considered a success, with a fairly competitive landscape that persisted throughout the period, well-capitalized banks, extensively branched, a narrow spread between borrowing and lending rates (indicating efficient matching of savings and investments), mutual par acceptance, absence of major crises, and virtually no inflation.

Dowd (1992) surveys most other episodes of free banking around the world and throughout history and concludes that “most if not all can be considered as reasonably successful, sometimes quite remarkably so.” Besides Scotland, Canada and Australia constitute two other interesting case studies in freeish banking. In particular, Canada offers a striking contrast to the troubles that plagued American banks in the late 19th century. Unlike their neighbors, Canadian banks suffered no financial panics, and none went under even during the Great Depression.

Australia is an interesting case in point; much like Scotland, its banks operated relatively freely, exhibited widespread branching, economies of scale without a natural monopoly, and mutual par acceptance. True, Australian banks did experience, between 1891 and 1893, one of the worst financial panics of any free banking era when, at the tail end of a 10-year real-estate boom, “some building societies and land banks failed, after which 13 of 26 trading banks suspended payments in early 1893” (White, 2014). Scholars, though, agree that there is no evidence of Australian banks being undercapitalized or overexposed during the boom, which was mostly financed by British funds. Those capital inflows in turn “suddenly stopped after the Baring crisis of 1890” (White, 2014). According to Dowd (1992), it was then the “misguided government intervention” that followed that led to a series of bank failures.

Another way to look at instability is GDP growth and volatility and the stability of the price level, and compare the two in different banking regimes. There is some evidence that GDP volatility was higher during the freeish banking era, or market-based Gold Standard era (1790-1913), than during the central banking era (since 1914), or more specifically the fiat standard era (since 1971). That said, GDP growth was also higher during the freeish banking era, and Romer (1986) has criticized the quality of the price index used to calculate GDP volatility during the 19th century. Following her corrections, the differences in volatility appear smaller. In terms of the price level, it is crucial to distinguish between volatility and autocorrelation. Volatility captures the short-term predictability of the price level, whereas autocorrelation captures its long-term predictability. Autocorrelation, unlike volatility, tells us if price changes build on each other and “compound”; it reflects the persistence of variance. And the data indicates that, while the price level was more volatile during the market-based Gold Standard era than during the fiat standard era, it was much less autocorrelated (Earle et al., 2021). In other words, near-term inflation has become more predictable under the Fed, but the long-run price level has become less predictable. Indeed, small differences in inflation rates, if autocorrelated, lead to very different outcomes in terms of price levels over time. The more autocorrelation, the more difficult it is to predict future price levels.

Takeaways

I’ve argued that good money is short-term elastic, long-term inelastic. This can be achieved with a layered monetary system: at the base layer, we have hard base money; at the second layer we have credit money. Credit money is a natural and desirable phenomenon in a world of short-term changes in the demand for money, which are more appropriately accommodated by a change in the quantity of money than a change in its purchasing power. In turn, the latter is more appropriate for long-term changes: the inelasticity of money supply in the long run enhances the predictability of the price level over long time horizons, which is crucial for long-term contracting — itself crucial for prosperity, investment, and a culture of low time preference.

Free banking has been relatively rare throughout history, but its instantiations have been quite successful. I argued that the deep reason for that is that free banking naturally evolves quick feedback loops that prevent systemic risks from building up. The US so-called free-banking era, often used as a cautionary tale, was in fact far from free, and most of its fragility can be traced back to misguided regulation. A combination of fireman arsonist syndrome and warfare imperatives favored the advent of central banking, first in Europe, later in the US. As White (2014) puts it, “central banking legislation often arose […] from attempts to remedy weaknesses due to earlier legal restrictions on banking by imposing a further layer of intervention.” The fiscal advantages of central banking further served to guarantee the unstoppable rise of central banking.

And yet, behind the apparent “messiness” of free banking, with its note discounting, reliance on the “barbarous relic” of metallic money, and higher volatility of both GDP and price level, a messiness often invoked to justify the State’s meddlesomeness, behind all that lies in fact great orderliness. As Taleb (2012) has shown, short-run volatility often is the surest guarantee of avoiding sudden extinction, or in other words existential systemic risk.

Going back to base money, advocating for long-term inelasticity in money means advocating for a deflationary currency: a currency whose purchasing power rises as the real economy grows. While the Gold Standard era is the closest historical example of deflationary money, it’s worth emphasizing that gold is not fundamentally deflationary: its issuance schedule, while predictable, is unbounded. In that sense, a truly deflationary currency has never really been tried, and most experts will balk at it. I argue that this reticence is rooted in a belief in Keynes’ irreducible uncertainty as the source of bad deflation. But if you grant me the point that uncertainty and panic-proneness are largely man-made, bad deflation becomes much less likely. As the Lucas critique goes, you can’t simply assume that a phenomenon observed under one policy regime will exist under a different policy regime.

Cryptocurrencies are, then, the first truly deflationary base monies. It’s an ongoing experiment. So far, the primary critique leveled at cryptocurrencies, beyond their deflationary nature, is that their volatility makes them poor numeraires and media of exchange. This critique, though, is only valid because we are in a fiat regime. In a fiat regime, where coercion and network effects conspire to maintain the preeminence of fiat currencies, cryptocurrencies like Bitcoin are not really monetized, and mostly function as speculative vehicles and inflation-hedging devices — much like gold has become. In a world where a cryptocurrency like Bitcoin is fully monetized, we can expect its value to be much more stable, because demand would be much more stable. Indeed, the value of gold was much more stable when it was monetized. In addition, the autocorrelation of inflation rates under a fiat standard regime leads to greater unpredictability of the long-term price level. This in turn adds to the instability of inflation-hedging devices such as cryptocurrencies.

Finally, a key element that enabled the rise of central banking was the relative ease with which a system based on metallic money can be co-opted. Most monetary gold was custodied by relatively few banks that ultimately the State could boss around. This makes it much easier to, first, issue mandates and assurances, and later, nationalize bank note issuance, with all the concomitant side effects. The ethos of cryptocurrencies, in contrast, is one of self-custody, due to its ease of transportation (speed of light) and storage (seed phrases). As such, it has natural affinity with a free banking regime, and thus one can hope for a rediscovery of the virtue of the latter sometime down the road, and high-modernism’s last breath. That said, the jury is still out on whether cryptocurrencies will serve us well. Their path to monetization is far from guaranteed. Whatever may be, exciting times are ahead of us.

References

Bagehot, Walter, Lombard Street: A Description of the Money Market, 1873

Carter, Zachary D., The Price of Peace: Money, Democracy, and the Life of John Maynard Keynes, 2019

Dowd, Kevin, The Experience of Free Banking, London, Routledge, 1992

Earle, Peter C., and Luther, William J., The Gold Standard: Retrospect and Prospect, American Institute for Economic Research, 2021

Ferguson, Niall, The Cash Nexus: Money and Power in the Modern World (1700-2000), Basic Books, 2001

Menand, Lev, The Fed Unbound: Central Banking in a Time of Crisis, Columbia Global Reports, 2022

Selgin, George, Money Free and Unfree, Cato Institute, 2017

Skidelsky, Robert, Keynes: The Return of the Master, PublicAffairs, 2010

Taleb, Nassim Nicholas, Antifragile: Things that Gain from Disorder, Random House, 2012

White, Lawrence H., Free Banking in History and Theory, GMU Working Paper in Economics, 2014

White, Lawrence H., Competition and Currency, NYU Press, 1989