What banks should do
In a pair of recent articles (Bank or MoMaMu Funds? (31 August) and Give banks a chance (7 September)), Ashok V. Desai pointed out flaws in the argument that I presented (For safe and sound banking, 11 August). There is much merit in his position, and in this article I try to clarify and summarise the issues.
The key issue that we face is that of trying to reconcile the features of banking as we know them today with a desired level of systemic integrity. What banks do today is a very attractive package from the viewpoint of users (firms or depositors). However, there are essential logical flaws inherent in commercial banking as we know it. This is why we have observed a steady stream of banking crises across countries (both advanced and backward) and across centuries (both past and present). This is why Nobel Laureate Merton Miller calls banking a "disaster-prone nineteenth-century industry".
I believe that it is difficult to have two things at the same time: (a) commercial banking as we know it today, and (b) systemic stability. We have to give up one. Since most of us would favour obtaining systemic stability, we need to ask basic questions about finding an appropriate match between what we ask banks to do and their ability to obtain these features without running into crises, bailouts, etc.
What do banks do? The product at stake - the bank deposit - can be summarised as follows:
- A: It offers assured returns (i.e. near-zero risk).
- B: It offers complete liquidity.
- C: It is backed by investments in illiquid, risky assets.
- D: The bank does this on a gigantic scale; i.e. it is extremely leveraged. The 8:1 leverage of banks exceeds the leverage of many `highly leveraged' exchange-traded derivarives.
Banks take deposits and give out loans (which are illiquid and risky), while promising depositors that there is no risk and complete liquidity. However, from time to time, the returns obtained by the bank do fall short - when loans go bad. Owing to feature D - high leverage - small errors of this nature make the bank technically insolvent. When this happens, a run on the bank can take place: the first who manage to take their money out benefit at the expense of those late in the queue.
The Basle Accord is a sophisticated strategy for addressing these problems. It summarises to three steps:
- Banks are required to preserve a buffer of equity capital (i.e. their leverage is capped),
- If this buffer falls short of the minimum, shareholders are first given a chance to top it off back to the required level,
- If shareholders do not do so, the bank is closed down by the regulators while it is still solvent; the assets are liquidated and the depositors are paid off.
It is important to emphasise that without the sound implementation of the Basle Accord, it is not possible for banks to honour their promises to depositors. If a bank is not closed down ahead of time - i.e. while it is still solvent - then it could lurch on to becoming insolvent, and then promise A would be violated.
In practice, the implementation of this regime is rendered impossible by feature C: investments in illiquid assets. When a bank invests in illiquid assets, marking to market is not possible, so even a honest and sincere regulator who seeks to implement the Accord cannot do so.
This gives us the conundrum. Features A,B,C,D are extremely attractive, however it is not possible to have them all at once. We can design several institutions by choosing to give up different sets of features.
Strategy 1. In my article, I offered one new institution, a quasi-MMMF, based on dropping feature C. If a bank avoids investing in any illiquid asset, it becomes like a money market mutual fund. If all the assets are highly liquid, then (and only then) can the regulator actually implement the Accord. The bank assets can be marked to market daily; when the buffer of equity capital proves to be inadequate, the regulator can move into the bank, actually sell off the assets on a liquid market, and thus, uphold promises A and B. All this can be done with very high levels of leverage (feature D). I should point out that money market mutual funds, as we know them, invest only in fixed income securities and do not offer assured returns. However, the institution described here allows assured returns and investment in any liquid securities.
Ashok Desai points out that investments in illiquid assets (e.g. loans to small businesses) are an essential feature of the economy and should not be given up. He is right. We can design an institution which would be well equipped to engage in such activities. But to do this, we would have to give up features A and B.
Strategy 2. It is reasonable to think that a finance company which invests in illiquid, risky assets cannot promise liquidity and fixed returns to investors. The correct institutional design here is the closed end mutual fund. A closed-end mutual fund could be explicitly designed to collect a billion rupees from investors, engage in such investments for a period of five years (promising no liquidity or fixed returns in this period), and gradually move out of illiquid assets by the redemption date. As Dr. Desai points out, one example of such a fund is a venture capital fund. However the bulk of illiquid lending into firms that a closed-end fund would engage in is not venture funding. The fund would work closely with individual borrowers, engage in information production, deal with asymmetric information, etc. The key point to note is that while all these are desirable features in the economy, they cannot be mixed with assured returns and complete liquidity.
In summary, it is hard to see how we can have all the four essential features of banking as we know them today in one institution. In order to obtain sytemic integrity, we need to instead look for institutions (such as quasi-MMMFs or closed-end funds) which jettison some subset of features A to D in the quest for internal consistency and inherent stability.
All this adds up to a major churning in the financial sector, and there are important questions of transitioning. Price signals would play an important role here. When banks (as we know them) start migrating into Strategy 1, the interest rates on illiquid loans would go up sharply, with two consequences:
- This would generate strong incentives for firms to shift their fund raising into liquid securities, i.e. bond issuance would go up at the expense of loan financing. Firms which can issue securities would be much more prone to do so.
- It would also open up possibilities to launch businesses in Strategy 2 at high rates of return, and that would attract depositors away from money market mutual funds (Strategy 1) into closed end mutual funds (Strategy 2).
Back up to Ajay Shah's BS column