I’m writing about this subject with some trepidation. Many economists who have entered the fray have made fools of themselves; and many who triumphantly point out their errors usually go on to make some of their own. Several years ago, Paul Krugman fell into the former trap, with this uncharacteristically confused blog. But virtually all of his critics introduce new errors – usually, that bank reserves don’t matter, or relatedly, that a central bank can’t control the money supply – it can, just not in the way of the crude multiplier.
Which of these two camps I will end up falling into remains to be seen. Either way, what’s more important is to introduce something novel. I don’t really have a ‘general theory’ of banks and money, just a series of observations which lead me to an unlikely synthesis of many competing positions.
Terminology. I should get this out of the way up front. I’m only going to designate base money (bank reserves & physical cash) as ‘money’. Tobin thinks this is a clear distinction, if somewhat old school – so be it. From the perspective of semantic efficiency, there is nothing to be gained from calling a deposit ‘money’, just call it a ‘deposit’.
Banks and central banks create deposits, but only central banks create money. And money is very different to deposits. You can’t default on money. So money is not a liability. Bank reserves are the electronic equivalent of physical cash.
Central banks create deposits in the banking system and reserves when they buy assets. Banks create deposits when they make loans.
Deposits can be a ‘means of payment’, when they are held as on-demand loans we make to banks, and when banks have the liquidity to honour the loans. Deposits with longer maturities are term loans like any other.
Are banks unique? Yes, because they can create purchasing power at will. This is at the core of the dispute between Steve Keen and Krugman, which I will summarise, and try to clarify:
Sometimes leverage increases spending on goods and services (usually called ‘aggregate demand’), sometimes it doesn’t. For example, I could go straight to my bank now and take out a £10,000 loan, which gets credited to my bank account as an increase in deposits.
Let’s say I am purely motivated by proving a point. If I choose to spend that money on something I was not otherwise going to purchase, aggregate demand rises, if I choose to hold the deposit because I want to keep my options open, nothing happens to aggregate demand.
Only banks can create deposits like that. But Krugman also has a point, in one sense: let’s say I decide to ‘spend the money’ in one year’s time – then, in a year’s time there is an increase in demand without any change in leverage. So preferences are important, and leverage needs to be interpreted.
Banks, bubbles, and Minsky. James Tobin was clearly troubled by the following observation: when banks make new loans, they create new deposits (ignore the situation where the loan is immediately withdrawn as cash). We all agree on this. So deposit growth will occur if credit conditions are eased. Now this does not have to be caused by a decline in interest rates, it could be caused by a change in banks’ willingness to lend, or regulatory standards, or an increase in demand for loans. What troubled Tobin was the fact that a very large increase in deposits appears to occur without any increase in demand for deposits, but simply because of loans growth.
This does not fit with a view of households and firms holding portfolios of assets consistent with a set of preferences. It appears to be the case that deregulation of a mortgage market, for example, will trigger an increase in holdings of deposits. In fact this very phenomena was observed by the Bank of England during the housing boom.
Tobin tries to make peace with himself using the following line of reasoning: if there are ‘too many’ deposits relative to what people want, the yield on other assets will fall, and the returns to banks will decline – he doesn’t really detail precisely how, but the implication is that demand for bank loans will be substituted with alternative, now cheaper, forms of competing finance. Also, in the standard view of demand for deposits, lower interest rates increase demand – so falling yields on competing assets make deposits relatively more attractive. That at least is the theory behind some kind of equilibrium result of portfolio rebalancing – all in response to an increase in lending. It’s a shame Tobin didn’t detail the stages in this line of thinking. It leads to very interesting conclusions.
Let’s try and think about how this process might actually occur – it is a variant of the conventional view that Nick Rowe outlines, that the ‘demand for money’ is a function of the interest rate. This debate would also benefit from calling deposits ‘deposits’ and base money ‘money’, and specifying which interest-bearing assets we are talking about – because the apparently simple rebalancing of portfolios may not be so simple.
Consider the situation in the mid-2000s when young families in the UK were purchasing houses using large amounts of debt from much older home owners who had significant equity in their homes. In principle this did not need to affect aggregate demand – there’s a change of ownership of the housing stock, and the older generation transform their home equity into (at least initially) a bank deposit.
Now in Tobin’s ‘equilibrium’ framework this sudden increase of deposits in the share of wealth should result in some shift in portfolios which undermines the profitability of banks and self-regulates the system. This is totally unconvincing. First of all, let’s think about individuals’ portfolio preferences. Behavioural finance has a well established concept called the ‘endowment effect’. It’s another way of saying, people don’t have portfolio ‘preferences’, they simply hold whatever assets they ‘inherit’. If sellers of houses receive deposits in exchange for home equity, they hold … deposits.
The supply of deposits is starting to create its own demand. And it’s very easy to see where Keen/Minsky comes in (and Bernanke/Gertler, and Soros). The supply of housing has long lags. So an easing of credit conditions often causes house prices to appreciate rapidly. This has multiple self-reinforcing effects: rising prices drive up collateral values, which results in even easier credit conditions; this releases more equity than people thought they had; and because virtually no one buying a house knows how to ‘value’ it, people use adaptive expectations which creates speculative demand for property, which drives prices up further. None of this necessarily causes bank profits to fall, or results in a substitution of the role of equity or other assets in finance. No: it causes bubbles, rapid growth in deposits, and then busts – with highly unpredictable effects on the yields of other assets – and banks are at the heart of the process.
Matters are made even worse if banks respond to declining returns on assets by increasing their leverage – a perceptive observation which dates back to Hayek.
So I hate to be so reasonable, but Tobin, Keen (Minsky), and Krugman are all in some sense right – Keen a lot more so than Tobin and Krugman.
CBs can control deposit growth with bank reserves. Let me now address this uncontroversial topic. Most of the time the central bank sets interest rates – the price at which banks can borrow reserves – and allows the financial system determine the rate of growth in deposits (Scott Fullwiler has a superb blog on this). This is not determined solely by banks – although, subject to regulation, they determine the supply of credit – but also by borrowers. In this sense, the phrase ‘banks create money (deposits!)’ could just as easily be ‘borrowers create deposits’.
Now there is an asymmetry in the role of central banks in the growth of deposits. Anyone who has witnessed a currency board in operation or, I suspect, anyone who worked closely with Volcker at the Fed, will know that it is very easy for a central bank to cause a collapse in deposit growth: aggressively shrink the stock or availability of reserves. Some will say, well that’s just raising the target interest rate. True. Which illustrates clearly that targeting interest rates or targeting reserves are two sides of the same coin. A central bank can either say, I’ll supply X quantity of reserves, and to hell with interest rates (in a strict currency board the supply of reserves is delegated by the CB to the balance of payments), or the central bank can target a rate of interest and supply reserves on demand – the situation Scott Fullwiler describes. Choosing to do the latter definitively does not mean that altering the supply of reserves has no effect.
There is an asymmetry because the central bank cannot engage in the opposite exercise. This is where a lot of weak analysis preceded QE – and I included Cochrane’s bizarre view that only the IOR averted hyperinflation! If banks don’t want to lend, or more relevantly, the private sector does not want to borrow, supplying all the reserves in the world will not create endogenous deposit growth. The only way the CB can create deposit growth in this situation is to buy assets (as with QE), which increases reserves and new deposits in equal measure, or to do helicopter drops. If this is done on large enough scale, the central bank could engineer whatever rate of deposit growth it wants.
The future of banking. Should we have 100% reserve backing of banks? Here’s my problem with this: We already do. Let me explain: deposit guarantees are worth nothing unless they are backed by the central bank’s ability to create reserves. So deposit insurance amounts to saying that the central bank will provide reserves on demand if needed to cover insured deposits (this, of course, does not preclude the likelihood that the government would bond-finance any claim).
Now in the UK during the financial crisis, the government pretty much said ‘no depositor will lose any money’. In other words, the government provides a 100% guarantee of deposits when there is a systemic risk. Well, that means that in a panic, the central bank has created reserves and contingent reserves equal to 100% of deposits. Of course, this is only true if there is systemic risk, so it’s a contingent backing.
A curious question is why does anyone’s hold deposits with banks in excess of those covered by deposit insurance. All it takes to get a 100% guarantee is to open an online broker account and put anything in excess of the insured level into T-bills. Perhaps we should tax anyone who doesn’t – for laziness.
The more interesting issue is will genuine technology-driven disintermediation kill off the banking sector. To address this we need to consider why banks exist. This fundamental question is often obscured. Tobin for example simply states, “Financial intermediaries typically assume liabilities of smaller default risk and greater predictability of value than their assets.” He goes on to say how they do this: through administrative efficiency, pooling of risk, and regulation – including capital ratios and deposit insurance.
This is weak analysis. The two fundamental issues that banks address are a) a coordination problem, and b) payment and settlement. On one side of the coordination problem are agents with excess financial capital, differentiated along three axes: amount, risk and maturity. On the other side are those in need of financial capital, again differentiated by size, risk and maturity. Separate to this is the payment function.
Now banking is in fact a highly cumbersome and inefficient way of addressing the coordination problem. Individuals should be free to choose how to save along each axis, and so should those in need of capital. Instead of matching the discreet needs of both sides, banks offer savings deposits or demand deposits at zero credit risk (if govt guaranteed), to those with excess liquidity – and on the other side of their balance sheets offer loans which are differentiated by size, maturity and risk, which the banks try to price accordingly. The bank takes all the risk associated with this mismatch, and is therefore forced to maintain expensive buffers, of capital and liquidity. Both sides of the intermediation are losing to bank profits. It is highly likely that people hold inefficiently high amounts of savings as deposits, and similarly the cost of capital to many borrowers is too high.
Intriguingly, technology and securitisation have solved this coordination problem. Today, an individual can choose to have any percentage of their financial assets in anything from T-bills, to long duration government bonds, to equities, to corporate bonds, to online loans. Securitisation means an individual can choose any combination of size, maturity and risk – in a diversified portfolio. Other than a minimal level of deposits to meet payments, it is hard to see why anyone should hold them.
The oddity is the shortage of one-stop shop providers. If individuals want 100% ‘deposit’ insurance for large deposits, why don’t they just open an account with an online broker and invest in T-bills? Would it take much for a broker to issue credit cards? Some already do.
Lending is also being dis-intermediated by online exchanges. Individuals can lend to diversified portfolios of borrowers directly. And banks do not appear to have more efficient pricing – quite the reverse. If it can happen with receivables, SME loans, credit card debt, why not mortgages?
So, banks are unique. They have played a central role in previous booms and busts. But there are early signs that their days may be numbered. Does that mean no more crises? I doubt it. It probably means that future crises will be very different.
I have ammended this paragraph following a very useful twitter exchange with Frances Coppola. In summary, a confusing semantic issue arises with insured deposits. I am saying that insured deposits are contingent claims on reserves. But “backing” deposits with reserves, does not, in my view, mean deposits = reserves. In fact, it starkly reveals their different properties.