Against anti-deflationism
A common critique of Bitcoin (and other cryptocurrencies) is that it is “deflationary”. Usually what people mean when they say that a currency is deflationary is that its supply can’t be tweaked in order to meet changes in the aggregate demand for money. A recent example of this critique:
In this post, I want to push back against the anti-deflationist claim that “fixed supply currencies are inevitable to end in mass unemployment”. It’s a common critique of Bitcoin and I’ve yet to find a good rebuttal from Bitcoiners, so I figured why not give it a try myself.
There is no such thing as fixed supply money
My first reaction to the anti-deflationist claim is that there is actually no such thing as fixed supply money, because of the existence of credit money. Base money may have a fixed supply (e.g. Bitcoin), but more flexible substitutes can be built on top of it. Those money substitutes (e.g. bank notes), redeemable for base money, become themselves money. They are “credit money”, with a more elastic supply. This is akin to the concepts of M0 and M1, etc.
How much supply elasticity can we afford via higher layers of money? This mostly depends on (1) the supply elasticity of the base money, and (2) the convenience of third-party custody.
The more elastic the supply of base money, the more elastic the credit money built on top of it can be. While the supply of gold was fairly inelastic in the short term, it was highly elastic in the long run. This means that, while the total value of credit money today may be higher than the total value of base money backing it, the stock of base money tomorrow could eventually catch up. As such, credit money could credibly expand in anticipation.
How does the convenience of third-party custody influence the extent of credit money? While in theory any individual can decide to issue their own substitutes for the base money, in reality this is a specialized business that historically was practiced by money custodians, which eventually became banks. By pooling base money from many people, banks reap the benefits of scale (uncorrelated redemptions) and can therefore engage in more “money creation” (add more money substitutes than they can redeem, because it’s unlikely everyone will redeem at the same time). And the more convenient third-party custody, the more people will entrust the custodians with their money, and therefore the greater the banks’ room for maneuver. “Convenience” here can mean various things, from security (in the case of Bitcoin, maybe you don’t trust yourself to not lose your keys) to the promise of a yield on deposits. Bitcoiners say with pride “not your keys, not your coins”, yes, but it’s pointless to keep literally 100% of your net worth under your mattress. It makes sense to lend some of it out via banks as intermediaries. This, in turn, allows banks to engage in money creation.
In short, credit money is fractional reserve banking. And the smaller the reserve in base money needed to back a given supply of money substitutes, the more elastic the overall money supply. Since Bitcoin is perfectly inelastic in the long run, we should expect the fractional reserves built on top of it to be much more constrained than under a Gold Standard. But we should not expect money creation to be nil, inasmuch as third-party custody is convenient. So we can indeed expect some elasticity, and therefore some ability in the short run to adjust to variations in the demand for money by a change in quantity rather than an adjustment of the price level.
Credit money accommodates short-run changes in money demand
What drives the demand for money? Some days, you need to carry out more transactions, so you need to load up your wallet with more money. Maybe you expect your income to dry up soon and want to build up a cushion. Maybe you got a raise and now you want to be a big spender! That sounds like a lot of variation in the demand for money! Crucially, though, what matters is not what happens at the individual level, but what happens at the aggregate level. Many of these individual variations in money demand cancel each other when aggregating. John may need more money today because it’s grocery shopping day while Paul may need less money today for whatever reason. The aggregate money demand is much less volatile than the individual money demand.
Why is this important? Because some people argue that adopting a fixed-supply money will cause variations in money demand to directly translate into variations in the price of all goods. Sounds chaotic! This, of course, is correct in the strict sense: if you assume that the quantity of money really cannot change (and if you assume interest rates are independent of any nominal phenomena), then necessarily an increase in the aggregate demand for money will bring down the price of all other goods. But my point is that variations in aggregate money demand are likely to be small enough to be accommodated by changes in the quantity of money via money substitutes, even in a world where the base money is completely inelastic in the long run. Therefore, the prices will remain stable in the short run.
For example, farmers typically have a seasonal variation in their money demand because their investments tend to be concentrated at specific times of the year. In the 19th century, this tended to drive the volume of bank notes in circulation throughout the year, illustrating how a change in money demand was accommodated by a change in its supply, even though the production of gold didn’t change. Money substitutes (bank notes) did the job.
Secular deflation isn’t bad
We’ve seen that short-term variations in aggregate money demand can likely be accommodated by changes in the quantity of money substitutes. What happens in the long run though? In the long run, aggregate money demand has an upward trend because it roughly follows income. Since in the long run, money supply cannot change, the increased demand for money will have to be accommodated by an improvement in its purchasing power. In other words, the price level will have to adjust downwards. Is that bad?
This is what I would call the broad version of the anti-deflationist claim: a fixed-supply currency leads to secular deflation and secular deflation is bad. The harm is usually spelled out in two flavors: (1) it makes investment more expensive because the real interest rate goes up (rate stickiness); (2) it creates unemployment because it increases real wages (wage stickiness).
Overestimating the expected inflation can raise the ex post real interest rate paid on a loan, but it can’t raise the real interest rate ex ante, and so it is not making investment more expensive at the time of signing the loan contract. Rate stickiness may happen when the anticipated inflation is such that the nominal rate would hit a lower bound (either zero or some nonzero “effective lower bound” if you subscribe to Keynes’s liquidity trap theory). That, however, never happened during the Gold Standard period. And an inflation of 0% (which is what Bitcoin is driving towards) still allows for positive nominal rates. Also, when the inflation is more or less guaranteed to be 0% (as per Bitcoin’s source code), errors in anticipation will be lower (the only uncertainty left is the rate at which coins get lost, which makes Bitcoin’s long-run inflation rate slightly negative).
Wage stickiness is mostly an artifact of the inflation regime of an economy. Historically, wage stickiness was lower when expected inflation was lower, although it was not zero even during long periods of near zero inflation rates during the Gold Standard. The degree of stickiness depends on people’s expectations, which in turn depend on the long-running monetary regime. For instance, this paper, published by the Bank of England, concludes that “there are theoretical models that predict that at positive rates of inflation, we are more rather than less likely to detect empirical relationships that reveal an apparent downward nominal rigidity” — hinting at the regime-dependence of wage stickiness. The paper also points to measurement challenges as employers may have different ways to cut effective nominal wages (e.g. overtime programs). They conclude that “the empirical evidence leaves the case for downward nominal rigidities at best unproven.” More anecdotally, it is common for tech workers nowadays to receive a substantial portion of their compensation in stocks. And when the stock has gone up significantly by the time their compensation is revised, it’s not uncommon to get less stock after renegotiation, which indicates a lack of downward nominal rigidity.
Secular deflation is harmless because prices fall as technology improves. If money demand roughly follows income, and income roughly follows productivity, then the resulting long-term price level adjustment required is that it falls at the same rate as the rate at which productivity grows. Which is the rate at which it naturally falls. Growth is not more difficult to sustain because nominal interest rates adjust. Real wages rise because goods become cheaper, reflecting the higher productivity of labor. At each point in time, there’s an ongoing monetary equilibrium with no excess demand for money and no excess supply of goods. This is also known as “good deflation”.
Even if the long-run aggregate demand for money outpaces productivity growth (say, due to population growth), this wouldn’t spell disaster. While this would require deflation above and beyond the “good deflation”, as long as rate and wage stickiness are contained, the real economy would remain unaffected. Evidence of the regime-dependence of stickiness gives us reasons to be confident.
Empirics seem to concur. Throughout the classical Gold Standard (roughly from 1790 to 1913), the supply of gold was a somewhat random process, albeit increasing in the long run. Because it was “unmanaged”, though, there was no guarantee that money supply would grow alongside money demand in the long run. In fact, long periods of deflation were not uncommon throughout the 19th century. Atkeson & Kehoe (2004) survey more than 100 years of economic activity across 17 countries and find no relationship between deflation (fall in prices) and depression (fall in output) — with the Great Depression as the sole exception (more on that later).
You can’t print your way out of a war
We’ve seen that routine variations in aggregate money demand can likely be absorbed by the elasticity of money substitutes in the short run. We’ve seen that in the long run money demand can be accommodated by secular deflation, and that secular deflation is not the calamity that people make it out to be. Now what about short-run spikes in aggregate money demand?
In that case, we no longer have a nice “ongoing monetary equilibrium”. If the spike in money demand is large enough, the production of money substitutes won’t be enough to meet the demand (because money creation is constrained), and people will start cutting spending to grow their money balances. This results in an excess supply of goods, in turn calling for a fall in prices and wages. Price and wage stickiness is likely real in the short term, so the fall in the price of goods and in nominal wages won’t be enough to clear the markets, and therefore there will be a need to adjust quantities (cut production and lay people off). We now have a recession on our hands. That sounds weird! Why would a simple change in money demand cause this much chaos? My needing more money to go to the grocery store should not be causing a recession!
It’s highly unlikely for the aggregate demand for money balances to suddenly jump that much for no good reason. Did everyone suddenly decide to go to the grocery store on a spending spree that cannot wait? Of course not. So most of the time such spikes in aggregate money demand correspond to real shocks. A few come to mind: war, pandemic, oil cartels. It’s normal for such shocks to cause depressions and bring prices down. If uncertainty is real, it’s normal for people to spend less and save more. It’s not necessarily an overreaction; there might be many reasons for doing that — for instance, save enough to physically move away from the shock.
You can’t really “print” your way out of a war, a pandemic, or an oil cartel. So why are anti-deflationists claiming base money expansion can help here? That’s because what they actually fear is the overreaction to these shocks. The panic. More specifically: the banking panics triggered by those real shocks. In fact, they argue that the shock itself need not be that big to trigger a banking panic, because it is the banking panic itself that drives the spike in aggregate money demand. And that’s what the money printer is really for, because that’s what it’s good at: stopping banking panics — not stopping pandemics.
Of course, printing money won’t solve wars, pandemics, or oil cartels, but it can short-circuit banking panics (maybe at the cost of some inflation and/or malinvestment). But the real question is: why are there banking panics in the first place? Is it an unavoidable feature of banking? Do the benefits of money creation via money substitutes necessarily come at the cost of banking panics every now and then? If it was so, this would be a good argument for a controllable base money supply.
Banking is not inherently fragile
I contend that banking is not inherently fragile. Whatever fragility banking is found to suffer from is, in fact, an artifact of managed money regimes.
Surveys of free banking episodes around the world provide striking evidence. What we find is that free banking worked fine, actually. (Free banking is roughly defined as (unmanaged) commodity money with a decentralized system of credit money on top.) Free banking systems naturally develop fast feedback loops and mutual monitoring that preempt massive crises. Indeed, banking crises during free banking eras and in free banking regions were less frequent and less intense than in central banking eras and regions.
In fact, the root cause of almost all free banking financial crises could be traced back not to animal spirits but to misguided regulation. The so-called free banking era in the United States (roughly the 19th century) was in fact far from free. Perhaps the two most consequential pieces of misguided regulation were anti-branching laws and bond collateralization requirements, which together were responsible for much of the instability characterizing the US “free” banking era.
Many free banking systems actually naturally developed their own homegrown, private lender of last resort (LOLR) in the form of clearinghouses (themselves developed to efficiently redeem each other’s bank notes). I’ll note that an unmanaged money regime does not prevent a government from acting as the LOLR. (I happen to believe that government LOLR is a bad idea, I’m just saying that, if you think that’s good policy, you can still pursue it in an unmanaged money regime.)
The fragility of the banking system, far from inherent to it, was foisted upon it by the ever greater role of government in the management of money. In this piece, I try to retrace some of the steps of this evolution, from the mandate of par acceptance and unconditional redemption, to the nationalization of note issuance, tender laws, deposit insurance, and the institution of the lender of last resort. That narrative can be summarized as the “fiscal theory” of the evolution from unmanaged money to managed money. It is a revisionist narrative that goes against the more common “market failure theory” of the government’s role in money.
The fiscal theory roughly points to a combination of firefighter arsonist syndrome and warfare imperatives as the twin impetus for increasing the management of the monetary system. Early interventions caused future problems, in turn justifying further interventions, etc., with the crises tending to increase in magnitude at each step. James Madison himself had warned that “one legislative interference is but the first link of a long chain of repetitions, every subsequent interference being naturally produced by the effects of the preceding.”
The final boss: the Great Depression
By far the largest banking crises occurred in eras and regions outside free banking — the most famous of all being the Great Depression. Indeed, the classical Gold Standard and free banking era had long been dead when the Great Depression happened. Regardless, most anti-deflationists will point to the Great Depression as their Exhibit A for Why Unmanaged Money Is Bad. So I cannot rest my case without addressing the Great Depression.
The standard account goes something like this: a banking panic seized the population in 1929, leading to a surge in demand for money balances and the ensuing fall in output as prices and wages failed to adjust. The strictures of the Gold Standard not only precluded the Fed from injecting new money into the system, it also served to spread the contagion to other countries, as the whole world scrambled for gold.
So what do I think is wrong with this account? My broad point is that the Great Depression had nothing to do with an inflexible, unmanaged supply of base money. It had everything to do with the mismanagement of money. First, we need to go back to World War I. The key monetary event of WW1 was the suspension of the classical Gold Standard by most belligerents, followed by massive money printing and the instauration of a planned war economy. Suspension and printing was functionally equivalent to massive taxation, and it was made possible by the fact that most gold was concentrated and therefore seizable, so that paper money could be foisted onto the population.
In particular, money printing and planned economy are likely to lead to malinvestment. A case in point was the overinvestment in agriculture by US farming in order to feed Europe (where farmers were fighting in the trenches rather than tilling the soil). US banks were of course involved in this malinvestment and restrictions on branching made them geographically underdiversified and therefore more prone to bust.
After WW1, with a massively inflated money supply, nations essentially had to choose between deflation or devaluation, if they were going to reinstate the Gold Standard. Because the volume of paper money was completely out of proportion with gold reserves, the gold parity of national currency had to go up to reflect the debasement of the national currency. The alternative was to somehow remove paper currency from circulation and for prices to fall massively (in proportion to their increase during the war). Most nations chose the latter (with the exception of France). In particular, Great Britain famously attempted deflation.
Without devaluation, nations were likely to compete hard for gold reserves, so the interwar international community jerry-rigged a proto-Bretton Woods but with the British pound in lieu of the US dollar: central banks agreed to use the British pound (as well as Federal Reserve promises) as their primary reserve and settlement medium. This system is also known as the “gold exchange standard”, and, crucially, is very different from the classical, market-based gold standard. Indeed, while the classical gold standard functioned “automatically”, the gold exchange standard depended heavily on cooperation between central banks.
There was now a strong pressure to defect from the gold exchange standard, and that is just what France ended up doing in the spring of 1927 by converting its sterling balances into gold, in turn triggering speculative attacks on the British pound. Meanwhile, between June 1927 and December 1928, the Fed expanded credit to avoid deflation, and likely accentuated malinvestment. When in early 1929, the Fed reverses course and tightens credit, the stage is set for deflation in late 1929. From 1930 to 1932, as deflation deepens, the Fed sits idly by and neglects to expand credit, even though its gold reserves are plenty enough to support expansion. It actually chooses to sterilize gold inflows that could have helped bring the price level up.
This is a very incomplete account of the Great Depression, but my main objective was to highlight that it had everything to do with mismanagement rather than unmanagement of money. Wartime money printing and economy planning set the stage for malinvestment. Refusal to devalue after the conflict led to a scramble for gold and a fragile gold exchange standard system. Easy monetary policy in the late 1920s likely aggravated malinvestment. Banking regulation made banks artificially underdiversified. Sterilization after the deflation precluded relief from gold inflows.
In a way, my above account of the Great Depression illustrates the firefighter arsonist syndrome mentioned earlier. Indeed, many of these powers (nationalization of banknotes, sterilization of gold inflows, payment suspension, etc.) had been legislated into existence with the establishment of the Fed in 1914. Somehow, the legislator felt the need then to boost the State’s managing powers over the money (some say the crisis of 1907 may have been a catalyzer), and I argue that the exercise of those greater powers were not unrelated to the precipitation of the historic catastrophe.
Is unmanaged money inevitable?
This post is agnostic on whether we’ll ever transition back to an unmanaged money regime. My goal was to make the case that, if it were to happen, it wouldn’t lead to a disaster. Short-run variations in money demand can be accommodated by money creation via money substitutes. The long-run rise in money demand can be accommodated by secular deflation without trouble. And free banking is not inherently fragile, so that endogenous spikes in aggregate money demand are unlikely in an unmanaged money regime.
This is not to say an unmanaged money is perfect. It has flaws. Banking panics are possible in free banking. They could lead to spikes in money demand above and beyond what the triggering shock would call for. It is possible that such spikes in money demand would overflow the money creation capacity of banks and lead to recessions. It’s also possible that, even in an unmanaged money regime, there would still be residual wage and rate stickiness that prevent deflation above and beyond “good deflation” and slow down economic activity as a result. But only dummies expect perfection in human affairs. It only needs to be better than the alternative. And I claim the alternative is not better (although I did not argue why in this piece).
Notwithstanding, if pressed, I’d say it’s unlikely that we will transition back to an unmanaged money regime. This would require unfathomably deep changes in the nature of the State. Most likely, Bitcoin won’t become money, in the sense of a generally accepted medium of exchange. Bitcoin, at this point, and for the foreseeable future, will remain just a store a value, mainly because the network effects for the incumbent money are huge and really hard to overcome. That’s ok though. The very existence of Bitcoin as a potential alternative money has some disciplining effects on the incumbent money. And even just as a high-quality store of value, Bitcoin can serve as a democratic tool to protect oneself against the abuses of managed money: it is an inflation- and censorship-resistant asset with no investment threshold or other forms of gatekeeping that many other assets have (it’s much easier to stack sats than to buy a house).
