Skip to content

Implications of Friedman’s Commodity-Reserve Currency for Partial-Reserve Banking, Bitcoin, and Better Currency Designs.

Ideally a currency would have price stability and avoid the risk of the issuer inflating the supply to their benefit. One proposed remedy for this is the idea of a commodity-reserve currency, where a currency is backed by commodities (e.g. gold, oil, wheat etc.).

In 1951, economist Milton Friedman provided a critique of commodity-reserve currencies versus a) a gold standard and b) fiat (government backed currency).

The paper (Commodity-Reserve Currency) is available on JSTOR and may be read online by creating a free account. I came across the paper as it was referenced in Tyler Cowen’s recent book, EconGOAT, covering great economists. I saw the article was paywalled, but then realised it was accessible via JSTOR.

Friedman argues a commodity-reserve currency suffers from the same disadvantages of gold, namely, price volatility and long-term price deflation. A commodity-reserve currency would further lack the historical precedence and acceptance of a gold standard (even if that precedence is irrational). Therefore:

  • If the choice is to avoid risks of issuer/government driven inflation, a gold standard is superior to a a commodity reserve currency.
  • If the choice is to optimise for price stability, a fiat standard is superior to a commodity reserve currency.

Friedman proceeds to offer a specific recommendation for a fiat currency whereby:

a) The money supply is allowed to grow in a manner that matches long term economic growth – thereby holding prices roughly constant.*

b) Growth in the money supply is achieved by allowing the government to spend an amount in excess of tax revenue, and vice versa.

c) The banking system would be a “full reserve” as opposed to “partial reserve” system – meaning that banks must have deposits available for withdrawal.

In practise, there have been times where the relationship between money supply and prices has been unclear. Specifically, following 2008, there was a large increase in money supply without significant increases in prices. In other words, implementing this strategy for a fiat currency may not work as stated.

I recommend reading the full paper, but here I’ll describe some Friedman’s key arguments against commodity-backed currencies and discuss then the implications for:

  1. How the 2023 banking crises might be resolved in the medium to long run by keeping a fiat currency but moving from a partial to full reserve banking system.
  2. The weaknesses of Bitcoin as a currency OR as the basis for a commodity reserve currency because Bitcoin suffers the problem of price volatility and long term price deflation that is suffered by of gold or a commodity-reserve currency.

In a brief appendix, I give short consideration to currencies that are backed by assets whose value derives from their future cash flows (e.g. stocks, bonds and real estate).

The Benefits of Commodity-Backed Currencies

Ideally, a commodity-backed currency would be backed by all commodities and even all services in the economy. This broad base would allow the price to move along with the “average price” of goods and services in the economy. This average price would stay in line with the long run growth of the economy. The first proposed benefit of a commodity-reserve currency is therefore price stability.

From Friedman’s 1951 paper. Symmetallism refers to a currency backed by a combination of silver and gold.

With each unit of this currency backed by a full basket of commodities and services, the issuer – typically a government – would not be in a position to cause inflation. This is the same argument that is given in favour of gold. The second proposed benefit of a commodity-reserve currency is therefore robustness to inflation by its issuer.

Later, I will describe Friedman’s arguments as to why a commodity-reserve currency – in practise – is likely to be less stable than fiat currency and susceptible to inflation by its issuer.

Practical Issues with Commodity Reserve Currencies I: Price Instability

In theory it desirable to include every good and service as backing for a currency. In practise, this is difficult.

  • Commodities must be liquid (bought and sold regularly in large amounts), traded at competitive prices and, ideally, storable**. Already, this greatly narrows the potential goods that could be included in the basket.
  • There are few liquid markets for services, and services are typically not storable**. This makes it hard to include services as a backing for a commodity-reserve currency.

In 1951, Milton Friedman estimated that goods that could realistically serve as backing for a currency (including goods such as gold, silver, oil and coal) would account for – at most – 3-6% of the economy’s output. As such, there would always be a very large divergence between the price of the goods in that basket and the price of the average good and service in the economy as a whole. In short, while a theoretical basket of goods and services might provide price stability, a practical basket of goods could not come close.

Yes, a basket accounting for 3-6% of the economy would provide better price stability than gold alone. On the other hand, there is a tradition – albeit perhaps an irrational one – for gold backed currencies. Friedman saw the track record of a gold standard as an overwhelming advantage over the small improvement in price stability offered by a commodity-reserve standard. Further, neither a gold-standard nor a commodity-backed standard could come close to the stability of prices offered by a fiat approach.

**Were futures markets to be more prevalent and liquid, perhaps this would somewhat relax the requirement that goods or services be storable. The contracts would be sold before delivery/expiry of the goods/services. By their nature, services tend not to be fungible, there tend not to be large liquid marketplaces for services, and there are even fewer (if any) marketplaces for futures contracts in services. So, I don’t rule out that an increase in markets and futures for goods and services could increase Friedman’s 3-6% to a larger number. I expect there is much more thought on this area that I have not yet read. Friedman devotes some coverage to the use of futures as envisaged by Graham and also Hayek. Setting aside issues of prevalence and liquity of futures markets, there is perhaps the more important issue that involving a futures market makes the whole currency contingent on the design and robustness of the futures market. As Friedman puts the matter – it is a shift from a backing by warehouse certificates (i.e. physical goods) to debt certificates (reliant on the design and operation of futures markets – that inherently bear risks associated with bad debt and often leverage).

Friedman’s 1951 paper on a futures based device as a form of backing.

Practical Issues with Commodity Reserve Currencies II: Lack of Robustness to Inflation by the Issuer

A commodity reserve currency in its practical form faces the same problem (or, perhaps, problem and feature, as a gold-backed currency): long-run price deflation.

By definition, with an incomplete basket of goods and services, there will be divergence between the price level of the overall economy and that basket of goods. If the basket is small*** – which it likely is – the divergence will be large.

With technological improvements and economic growth, there will be an increase in the value produced by the economy, that is not matched by a consistent increase in the supply of gold. This is the same problem, whether you use gold, or whether you use a variety of commodities that make up a minority percentage of the economy’s output. Therefore, the only way to maintain longer-term stable prices, or slightly inflationary prices, would be to introduce, on top of the commodity backing, a level of government printing of money every year. If the country needs to increase the supply of currency by 2% to account for economic growth the government could choose to print – without any commodity backing – a 2% increase in the supply and immediately use those funds to pay down government debt or pay for government purchases. In this way, the supply of the currency would be increased to the government’s benefit and to the benefit of price stability. Thus, a means of inflation by the currency’s issuer would be introduced. Absent norms and traditions governing the controlled increase in monetary supply, the benefit of robustness to inflation would be negated.

***It is insufficient to simply index to a larger basket of goods. Indexing is not equivalent to backing. Ultimately, the growth in supply of a collateralised currency is controlled by its backing. Friedman points out how the use of a partial reserve (e.g. part gold, part faith/credit of the country) neither side-steps nor even relaxes the requirements for collateral. It appears to me that the same argument applies for over-collateralised currencies (e.g. a US dollar indexed currency backed by gold, or, for that matter, Bitcoin or Ether). Under- or over-collateralised currencies are both bounded in supply by the supply of their collateral. The rate at which the supply of US dollars can be changed by issuing new dollars backed by gold is inherently government by the supply-demand characteristics of gold.

Friedman on the question of whether a partial gold standard (i.e. gold + government backed) can economise on the amount of gold required as backing.

A Summary of Gold Versus Commodity Reserve Currency

In short, Milton Friedman argues that a commodity reserve currency would largely have all of the disadvantages of gold, but without the historical precedence or acceptance for gold as a means of backing a currency. Furthermore, the expected advantage in theory of a commodity reserve currency, namely of stable prices, and of zero government influence would be negated because of the inability to achieve a basket that represents the overall economy and, further, the inability to have a currency that is stable or slightly inflationary without having some government discretion over how the monetary base is increased in size every year.

Commodity Reserve Currencies Versus Fiat Currencies

Friedman makes the very interesting interpretation that a fiat currency is, in effect, a way to achieve an underlying basket of currencies that is an almost perfect representation of the economic output of the country, including both goods and services. The reason for this is because a fiat currency could be underpinned by the taxation policy of that nation.

Friedman’s characterisation of the collateral or “base” backing a fiat currency

Friedman proposes that the government issue a currency that is not backed by any commodities, namely, a fiat currency. The government would further manage its budget such that its revenues from taxation of income and other sources would roughly balance out with the government’s expenditures. In order to maintain a currency that is stable in price, the government would allow spending to be greater – in the long run – than income from its tax revenues. This difference would offset the increase in production of the economy as a whole – what Friedman calls secular growth.

So while this fiat system – unlike a pure gold standard – does involve government intervention, and is at risk of misuse by the governing authority, it has the significant advantage over a gold (or commodity-backed) standard that its price will be much more stable. In essence, the base backing the currency is the taxation of production within that jurisdiction.

The US dollar today is not entirely unlike Friedman’s proposal for a fiat currency. Governments today control the supply of money through a) controlling the reserve ratio that banks must hold i.e., how much they must hold on deposit relative to the amount they lend out, and b) control the overnight interest rate paid on deposits at the Federal Reserve, and also c) affect the overall supply of money by buying and selling bonds on the open market. In practice, reserve requirements at banks have been relaxed and are not a key determinant of the money supply in recent decades. The main effect on the supply of money is through the overnight interest rate, and through these open market operations conducted by the Federal Reserve.

By contrast, in Friedman’s system, government rules on taxation/income versus revenue would be more programmatic and banks would not be able to lend out deposits, i.e. banking would operate as a full- rather than a partial-reserve system.

Implications of Friedman’s Fiat Currency Proposal for the SVB and First Republic Banking Crisis of 2023

In the run up to 2023, the federal reserves increases in interest rates led to significant devaluation in US government bonds. This led to a reduction in the market value (although, perhaps strangely, not the book value) of reserves held by banks. A large sale of bonds by Silicon Valley Bank to generate cash for increasing withdrawals by depositors served as a cautionary signal to the market. A run on the bank ensured, including withdrawals by depositors holding over $250,000 (the limit under the federal deposit insurance scheme) in their accounts.

This led to the failure of SVB, Silvergate and Signature banks as well as First Republic going into receivership and being acquired. To stabilise the markets, government institutions in the US and internationally stepped in to provide deposit insurance in all but name (i.e. via liquidity injections to banks).

The core issue at hand with the SVB and First Republic banking crisis is that certain banks became illiquid i.e., they did not have enough cash on hand when customers began to withdraw deposits. The system of partial reserve banking contains fragility. That fragility was delayed via the deposit insurance scheme that provided confidence up to a limit of $250,000 per account. The guarantee level has proven insufficient and a full guarantee is now (indirectly) provided by government institutions. In essence:

  • Monetary policy shoulders the burden of deposit guarantees.
  • While the business of banks, historically, has sometimes been of loan underwriting, the risk of underwriting has been shifted to monetary policy.

Building on Milton Friedman’s proposal, it is possible to maintain a fiat standard (even with today’s means of controlling supply via overnight rates and open market operations), but without the fragility of the fractional reserve system. In other words, a fiat currency could be maintained while removing the fragility that is present in the system due to our system of fractional reserves. Simultaneously, the risk (rather than primarily the rewards) of underwriting might be restored to banks.

What are the Pros and Cons of a move to a fractional reserve system?

If the case for fractional reserve banking is to vary the money supply through varying reserve ratios, but that tool is no longer used, there is less to be lost from retiring that tool. Yet, there would perhaps be value lost in that optionality.

If the case for fraction reserve banking is to effectively extend credit into the economy, it is questionable whether using banks as a conduit is effective when the risks of underwriting flows through to the government. Why not have the government directly extend credit?

On the other hand, many banks do have deep community and business relationships and are effective at underwriting (although surely with skewed incentives, given the arguments above). Foregoing those relationships and connections would be a loss. Those relationships and lending need not be foregone in a full-reserve system. Lending would continue, but the lending would not be from the deposits.

Perhaps the strongest case against a move to a full reserve system is that banking relationships today with customers are underpinned by a type of “freemium” model, whereby services such as deposit safeguarding, ATMs and fraud protection are provided “free of charge” in return for the net interest rate earned by the bank between their lending and what banks pay customers on deposits.

There are strong and many complaints about this type of freemium model – many of which mirror the arguments against advertising based models prevalent in social media. Yet, behaviourally these systems are preferred by consumers (perhaps only for short term reasons?) and economically these business models create large consumer surplus. This, perhaps, and the inertia and strength of banking relationships merits the continuation of the partial reserve system of banking.

I am certain that I lack the necessary context to judge whether full reserve banking provides a solution or even an improvement over the current fragilities. The question deserves much deeper consideration than I am capable of providing here.

Implications of commodity-reserve currency theory for Bitcoin

The use of Bitcoin as a currency is similar to the use of gold in the sense that it represents a very small amount of the world economy’s production. As such, it does not provide a basis for stable prices. This makes Bitcoin similar to a gold-backed currency or a commodity reserve currency in its practical form – with the important difference that Bitcoin bears much less of a track record than gold. This might change, but Bitcoin will likely retain the problem of being a deflationary currency – making it difficult for use as it medium of exchange. Expressed in weaker form, Bitcoin does not provide a better unit of account than is provided by a well managed fiat currency. At the same time gold (and maybe Bitcoin) offers a better unit of account than a very badly managed fiat currency.

Bitcoin/gold are strong to the degree a fiat-issuing jurisdiction is weak and untrusted. Conversely, Bitcoin and gold are weak to the degree there is a fiat-issuing jurisdiction that is strong and trusted.

In times of strong and trusted government, the price stability afforded by fiat currency is far superior to what can be offered by a practical commodity reserve currency. As such, the utility of practical commodity reserve currencies or of gold as currency is as a last resort in the case of government weakness/mistrust and inflation.

Appendix A: A Note on Asset Backed Currencies

A further idea to achieve both price stability and robustness to government-driven inflation is the use of asset backed currencies – i.e. backed by a basket of stocks, bonds, real estate and other such assets.

The key issue I see here is that these assets represent the discounted present value of their cash flows. This value depends on future expectations. As such, there is no reason to believe a basket of stocks, bonds and/or real estate can provide a representation for an average price level in an economy.

With ETFs and digital assets, it may be possible to achieve – with reasonable precision – a close representation of the value of all assets. This is not equivalent to representing the price level of an economy that is representative of production (not discounted present value).

Appendix B: A local approach to commodity-reserve currencies

One might argue that it is a mistake to consider the overall price level of an economy as relevant to individuals. It is obviously true that my personal basket of goods and services only partially overlaps with those of the economy as a whole. As discussed above, practical commodity backed currencies are contingent on liquid and competitively priced markets for goods in the practical basket. As such, to the degree one considers a basket that is more personalised, one also considers a commodity-backed currency that is relevant to fewer people. The more widespread the use/acceptance of a currency, the more useful it is to hold and use that currency.

Therefore, local commodity-reserve currencies face two issues:

  1. They face the same issue as any commodity-reserve currency with price volatility. There is nothing inherent about local reserve currencies that allow for increased liquidity of underlying commodities.
  2. If anything, at a local scale, liquidity will be worse – both for commodities and for the currency itself.

As two asides:

  • The more robust local solution is perhaps one focused on credit, perhaps in the style of Sardex type exchanges and obligation clearing.
  • In the vein of Friedman’s fiat currency design, I would expect future currencies to come from large entities that a) are capable of taxation and b) incorporate a wide variety of products and services. This can include countries but also companies (i.e. Apple, Amazon etc.).

Leave a Reply