Fri Jan 14th, 2011 at 08:58:33 AM EST
The financial system encompasses, roughly speaking, seven different kinds of activities, in declining order of importance:
- Retail banking - letters of credit, payment clearing, overdrafts, debit cards, checking accounts, etc.
- Maintaining orderly exchanges.
- Insurance (not including "insurance" against losses of financial assets - insurance schemes require losses to be mostly uncorrelated, and financial losses are almost always highly correlated).
- Investment banking - long-term credits to finance development of capital-intensive industrial plant.
- Long-term lending against existing capital - real estate being the prime example, but successful investment banking turns into this once the plant it funds is up and running.
- The rest - venture capital, corporate raiders, hedge funds, brokers' loans, etc.
We'll look at #1, 5 and 6: A lot of (but by no means all) crises originate in investment banking and mortgaging of existing assets, and retail banking usually gets taken along for the ride because of the joint ownership of retail and investment banking houses.
front-paged by afew
Traditionally, banking regulation has taken the form of solvency requirements and an overnight interest rate enforced by open market operations and liquidity requirements (plus the banks' internal risk management, which enforces lending standards). Under the traditional system a bank lends out money, which creates deposits. It then subsequently goes to the interbank market to obtain central bank reserves in order to meet its liquidity requirement. The central bank then buys sovereign bonds from the banks, thereby injecting reserves into the interbank market, in order to maintain the target overnight rate (repaying the loan reverses all these operations).
When you consolidate these transactions, the net result is that central bank has obtained a sovereign bond (thus extinguishing a government liability) and in exchange committed central bank reserves to the lending bank (thus creating a matching government liability); the lending bank has obtained a private bond or note (an asset for the bank), an amount of central bank reserves (an asset), an amount of interbank debt (a liability) and a deposit (a liability); and the rest of the interbank market has divested of a sovereign bond (an asset) and obtained an interbank loan (another asset).
This system has two crucial weaknesses and causes a moderate inconvenience on the side.
The first weakness is that there is no integral solvency check in the loan origination and funding process. It's all about liquidity. So bank solvency will have to be enforced at one remove from the daily operations of the bank. Solvency requirements thereby become an external constraint on banking activity, and the agency that enforces them is therefore likely to be perceived as a spielverderber, to be captured or worked around whenever possible. And working around or capturing solvency regulators is easy when they are not embedded in the core operations of a bank: Since the solvency enforcement is incidental to the daily operations, it depends on the solvency regulator asking questions. While you can penalise banks for not answering fully and honestly, there is no practical way to force the solvency regulator to ask questions about operations that it does not know is going on, or that it finds it impolitic to enquire into.
The second weakness is the interbank market. Recall the list of consolidated transactions above: The interbank market simply serves to swap assets that are, under all ordinary (that is to say non-crisis) conditions as close to being equally safe and remunerative as it is possible to be. The problem with this is that it causes bad banks to become "contagious" - a bad bank can take down the banks that lent it money on the interbank market. Now, you may argue that this is only fair - after all, those banks should know better than to lend to people who are crooked or incompetent.
But what about those banks who were two degrees of separation away from the bad bank? Three degrees of separation? At some point it becomes unreasonable to keep punishing them, but there is nothing in the interbank market's mechanics that prevents the transmission of a crisis beyond that cutoff. And, more importantly, interbank market panics carry such large macroeconomic costs that it is unfair to the non-financial sector to demand that banks several degrees of separation from the bad bank must be punished by the market before ordinary commercial activity can resume.
Under an exchangeable currency, such as the gold standard, there was an excellent reason for the central bank to insist on only taking on perfectly pristine assets: After all, under a convertible currency system, you can have a run on the central bank. But under a modern, non-convertible currency system there is no reason that the central bank must restrict itself to swapping reserves for sovereign bonds - it can now bypass the interbank market by using the discount window.
Another inconvenience arises from the separate enforcement of solvency and liquidity requirements: By grafting on solvency enforcement in a crude hack, rather than building it into the ordinary operations of the system, you necessarily split the solvency enforcement into enforcement of bank solvency and enforcement of borrower solvency. There is no compelling economic justification for this: What matters to the integrity of the financial system is the margin on the loans, not whether this margin is carried by bank equity or borrower equity. Moreover, it means that the financial regulators can only enforce the bank solvency part and must leave enforcement of borrower solvency to the banks' own risk management. In order to enforce a safe level of margin in the financial system overall, the financial regulators will therefore have to place the full margin requirement upon the banks. But the banks will naturally demand margin from their customers as well, so the financial regulator will be faced with the choice between accepting more margin than is necessary for the system as a whole or losing the ability to guarantee that sufficient margin is provided by bank and borrower taken together. This is obviously undesirable.
However, as hinted in the section on the interbank market, a non-redeemable currency raises certain interesting options for the central bank and other financial regulators. Specifically, since the central bank is now freed from having to worry about a run on the currency, it can take less liquid assets onto its balance sheet through the discount window. This would allow central banks to set margin requirements on the assets that are accepted as collateral at the discount window, and, by largely or wholly curtailing the interbank market, to discriminate between different sorts of banking when it comes to setting policy rates.
In order to use the discount window to set margin requirements, the liquidity requirement will have to be raised to 100 % of deposits and other important liabilities. This, in itself, will cause no trouble, since the central bank will always provide any amount of reserves that member banks desire anyway. The central bank can keep doing open market operations in sovereign bonds, or it can accept them as collateral against the discount window with a minimal haircut (though obviously doing both at the same time requires some care to ensure that traders can't arbitrage between the discount window and the sovereign bond market).
Under such a system, a bank lends to a borrower, and then goes to the discount window with the resulting note. The central bank then lends reserves against the note up to the limit of its value as collateral. The margin will have to be deposited by the bank, the borrower and/or other institutions or individuals who are prepared to extend unsecured loans to the bank. The bank and borrower don't have to possess the entire margin - they can borrow the balance from other banks - but those banks must have an excess of reserves, bonds or discount window collateral over what they need to secure their own deposits. In principle, the bank can demand that the borrower overcollateralises his loan to the tune of the entire margin, thus permitting the bank to operate with zero equity, or, at the other extreme, the bank can offer a no-money-down loan if and only if it is able to post the margin itself. In the real world, of course, even properly collateralised loans occasionally go tits-up, so it will be prudent to still enforce a solvency requirement for the banks. But tying margin requirements intimately into the lending process makes this a back-up rather than the main safety system.