My last two blogs have revisited the issue of whether or not base money is a liability of the state. The issue played a central role in a discussion over the future of monetary policy at the Brussels think-tank, Bruegel, earlier this month.
“Why does it matter?”
That was Ben Bernanke’s question, when we briefly discussed the subject in Washington several years ago.
“It matters because if base money is not a liability of the state, the net government debt declines by around 30% and there has been no significant increase in government debt since the financial crisis.”
As it happens, I also briefly discussed MMT with Bernanke, at least the parts I think are sensible. He’d never heard of it. His comment:
“Isn’t that just economics?”
Bernanke’s smart. He’s also read Friedman and Tobin and (I bet) Lerner.
Many readers are probably also wondering why I am spending so much time on MMT. I’m not really. I think a misunderstanding of the value and power of money lies at the heart of policy impotence, as the debate in Brussels and the Peterson Institute reveals.
Macroeconomic policy failures are a very significant factor behind huge potential political turmoil in Europe. If money was understood better, we would be redesigning our central banks and our societies would be functioning a lot better. In extremis, I am an economic determinist.
Also MMTers are not alone in believing that accounting convention is the truth. In central bank accounting, base money is treated as a liability. I have argued (as have many – Willem Buiter, Simon Wren-Lewis, Nick Rowe, and others), that this convention is false. In simple terms, no one owes you anything for your $10 bill. Now, many mainstream economists have been confounded by the payment of interest on reserves in the US. So even though you are not owed anything for a dollar bill, banks now get paid for holding reserves at the central bank. “Paying interest” on reserves makes them look like liabilities of the central bank (although less so when you realise the banks can’t return the reserves if they don’t like the interest rate). Treating reserves as central bank liabilities because of the IOR allows John Cochrane and Chris Sims to resuscitate the fiscal theory of the price level, because money leaves their model and everything is a government debt .
So on the one hand we have conventional economists arguing that money is a state liability (debt), because ‘interest is paid’ on reserves and unconventional economists (MMT) arguing it is a liability because we use to it to pay our taxes (and other payments to governments such as fines, utility bills etc.).
MMT’s position is counterintuitive, at best. I think I have shown it is worse than that: MMT systematically adopts an obvious linguistic sleight of hand – it uses the same terms to identify two distinct phenomena, attempting to prove that different things are the same. Randall Wray has elevated this ‘skill’ to the level of genius – hence the Randall Wray Fallacy (RWF).
So how do MMT argue that by paying taxes (etc) we render money a state liability? Randal Wray’s clearest exposition uses the example of airline miles. I think we can all agree that some estimate of outstanding air miles should be treated as a liability of the airlines. There is at least some reputational or brand risk if an airline makes them too hard to use, or repeatedly changes the terms of the programme. However, even though air miles are a liability, the book value of this liability would not be the face value of outstanding miles, because a reasonably predictable percentage of air miles are never used, availability is also targeted at inventory that will otherwise not be fully utilised, and the airline can change the terms.
Wray tries to apply the same argument to money issued by the state: the government creates money, and money is used to pay taxes, fines, and other services provided by the state, so money is a liability of the government.
So we have two arguments as to why money is a liability of the state: 1) reserves pay interest, 2) governments issue money and they receive money as payment for taxes.
(1) is superficially convincing, although by implication reserves and notes & coins are different – because there is currently no interest rate on notes and coins. And so is (2), if one buys the analogy with air miles.
Let’s start with some obvious objections and then try to explain the remaining conundrums.
The obvious problem is that I am still not owed anything for a $10 bill. I am not ‘owed’ flights, because I can spend $10 on anything. But I am owed something by GE if I own a GE bond. And my neighbour owes me money if I lend her $10. Notes and coins are assets we use to pay for things, liabilities are things people owe. So money and debt are distinct phenomena. As I have said before, debts have and still are often used as money. But this doesn’t make money a debt. Cigarettes are used as money in prisons, this does not make money a cigarette. In fact, as soon as a debt is used as money, it has an independent value and the market value of the debt should trade at a premium to the book value. If that makes no sense, think of it like this. If everyone else starts using the air miles as money – i.e. storing them as wealth, exchanging them for other goods etc., the actual liability to the airline starts to decline. They could sell $1000 of air miles, and receive $1000 in cash, perhaps only $500 are used for flights, and the cash generates a return. Warren Buffett makes a similar argument about the accounting treatment of insurance float, which I have discussed before. (Unfortunately, many of those, including Randall Wray, who are dismissive of Buffett, haven’t read him. Buffett does not advocate changing the accounting treatment, he just argues why it is under-estimating value. As with many things, understanding a liability actually requires thought).
There’s another fundamental difference between money and air miles. An airline can over-produce miles relative to the amount of flights it can provide. In which case it will not honour the miles. But the government can’t over-produce money relative to taxes. Governments can only over-produce money relative to the supply of total goods and services in the economy. Yes, too much money causes inflation. Too many liabilities result in default. This is true even in the case of government liabilities. If the government can only honour its liabilities by producing too much money and creating inflation, it is still too much money that is creating the inflation, and the probability of default rises, because the government at some point might decide that the cost of inflation outweighs the cost of debt restructuring. This can happen under any institutional structure, as we saw in Russia in 1998.
If you are still unsure about the difference between air miles and money, ask yourself the following question: what would you require to take on the ‘liability’? If an airline approached another firm and said ‘we have these outstanding air miles would you honour them?’, the firm would require payment. An airline would have to pay another airline to transfer this obligation to provide flights. That is because a liability or a debt has a negative present value to the debtor. But if the government approached you and said, please accept the money that we have issued as payment for work, goods and services, you would reply, ‘that’s what I aready do!’.
My synthesis of MMT, which I think most people agree with, is that taxes have been used historically to establish the dominance of a monetary standard. That is what the wealth of historical evidence shows. But this does not make money a liability of the state, nor does it mean that once a dominant network is establish continued payment of taxes or other levies is essential to the value of the money. In my view, this is the position of smart chartalism/MMT – it is a theory which helps explain network dominance.
So a key difference between money and air miles, is that money issued by the central bank is used to settle payment for everything, not just government services. Perhaps if the government issued vouchers which could only be used to pay for government services (as has been discussed in Italy), they would be state liabilities. In other words, imagine we use bitcoin to pay for everything including government services, but the government sells non-transferable vouchers to all citizens which can be used to pay for taxes or any government-related service. We can also buy these vouchers with bitcoin. The government could raise money – bitcoin – by selling vouchers, and the vouchers would be a government liability, because it has to either provide a service or cancel a tax bill every time a citizen shows up with a voucher.
I think these vouchers are government liabilities, but they differ fundamentally from money, and it’s worth thinking why. The crux is that the government has to sell them, it can’t use them to buy things at will. If we already have enough vouchers to buy things from the government, we may not accept any more, and the government will not be able to sell vouchers to raise revenue, or it will have to sell them at a discount. Immediately, you can see why money has a value to the government in a way that non-transferable tax or government service vouchers do not.
For the same reason the IOR is not really an interest rate. The government doesn’t create money by selling it into the market; it creates money by spending it, by buying assets, or by giving it away. Interest rates are the price paid in a market to induce lenders to lend or an investor to buy a bond or lend money. The ‘interest’ on reserves is really a transfer payment to support the profitability of banks. It will also influence the price at which they are willing to lend reserves. But paying no IOR, or setting an arbitrary IOR to either shrink or expand base money, in no way constrains the government from creating money.
Finally, a couple of additional MMT idiosyncrasies are worth debunking. Using reverse reasoning, MMTers will argue that the government can ‘default’ on money. I put the word ‘default’ in inverted commas because this is a classic RWF. They do not mean ‘default’, they mean refuse payment. So the government ‘defaults’ by refusing to accept money as payment for taxes. In the past there are examples where governments have changed the means of payment eligible for tax payment. It is clear that refusing to accept payment is not a default. It is almost always the opposite. If I refuse to accept your repayment of a loan, I would be forcing you to default. In fact, Randall Wray says as much at one point:
“ … referring to the exchequer’s specific promise to accept his own coins in tax payment. If he refuses to do so, you cannot redeem yourself.”
Odd. But MMT say the government is breaching a promise if it refuses to accept money as payment for taxes. Now I don’t know if it is the case everywhere that it is legally binding for governments to accept state issued money to take payment for taxes. It’s an unnecessary law in most well-ordered societies. But breaking a commitment, refusing to take payment and breaking the law are not defaults. They are only defaults if they involve failing to honour a liability. But refusing payment is not failing to honour a liability. As I have said (and Randall Wray implies) it may cause someone else to default.
There are lots of pitfalls with this convoluted argument. What if the CB creates the money, and the Treasury raises taxes? Is money a liability of the Treasury or the CB, or the ‘state’? What if the state raises all its revenues through land sales and the land sale department uses commercial banks to receive payments? What if the CB is privately owned, as in Switzerland? What if the state decides to finance all its activities through money printing and uses no taxes or other forms of revenue? What if the state decides to sell its tax receivables to the private sector, which it is not compelled to by law, and the private sector collects the taxes? Would money stop being a liability?
Accounting is not economic truth, nor is law, nor is prevailing convention.
In conclusion, the denial of money’s uniqueness has a long and recurring history in economics. It likely betrays the counter-intuitive nature of network effects. There are many misunderstandings around the nature of money – it is a property of things, which can disappear; it does have intrinsic value, but often no value in alternative use, and no physical value; it is not a liability or a debt, even though debts are frequently used as money in order to establish the network. The economics of language teaches you more about money than studying central banks’ balance sheets. Understanding the true nature of money is hard, but important. Most importantly, when it is properly understood we recognise that permitting sustained shortfalls in demand, deflation, and high unemployment in deregulated free-market economies is always and everywhere unforgivable.
 Curiously, Sims provides a definition of ‘default’ which he correctly distinguishes from ‘inflation’: “Economists and journalists sometimes treat inflation as a form of default, but it is not. Default is a situation where the contracted payments cannot be delivered, and the contract does not specify what happens in that eventuality. For private firms, this leads to renegotiation and/or court proceedings. There can be a long period in which investors cannot get access to their investments and the amount that will be returned to them remains unknown. Creditors holding different maturities or types of debt may suffer different degrees of loss, and the allocation of losses across creditors may be uncertain. For example, a minor default may involve a modest delay in returning principal of a short term debt. Other creditors may be unaffected, or, if the holder of the short debt goes to court, all debtors may find themselves impaired. Similar, or perhaps more severe, uncertainties surround sovereign default.” Oddly, he persists in describing base money as a liability. Nowhere, however, does he explain how, if the government cannot default on money, it can be a liability.