The UK’s Tripartite Agreement between HM Treasury, the Bank of England and the Financial Services Authority, including the division of labour set out in the Memorandum of Understanding did not work during the Northern Rock crisis. That is not surprising, as the design is flawed. The fundamental flaw became more obvious every time Sir John Gieve or Paul Tucker included in their answer to some question put to them by the Treasury Committee words like: The Bank of England does not collect/have information on individual banks/institutions.
Since the Bank lost banking supervision and regulation in 1997 to the FSA, when the Bank become operationally independent for monetary policy, only the FSA has had the information on individual banks necessary to perform an individual institution-specific Lender or Last Resort operation. However, the FSA does not have the resources to provide a credit line like the Liquidity Support Facility provided by the Bank of England to Northern Rock. The Bank has the resources, as it is the ultimately source of liquidity through its ability to create legal tender in any amount and at the drop of a hat, but it does not have the institution-specific information to allow it to determine in time which bank is solvent and liquid, which bank is solvent but illiquid and which bank is insolvent (only in Russia did we use to have banks like Sberbank that were insolvent but highly liquid…). So the FSA had (or should have had) the information on individual banks but did not have the resources and the Bank had the resources but not the information. The Bank ended up with the responsibility of providing a LOLR facility without having to information necessary to discharge that responsibility.
So the Tripartite Agreement and the MOU will have to be changed. There are a number of options.
(1) The back to the future model.
Transfer banking supervision and regulation back from the FSA to the Bank of England. The FSA should retain responsibility for the consumer protection and customer protection, as such retail issues are not of systemic significance. The Treasury, then as now, should be responsible for deposit protection/insurance/guarantees, although the Bank should be consulted if changes are made to those arrangements, as it can have systemic implications. This would put the information required for being an effective LOLR and the responsibility for performing the LOLR function in the same institution –the Bank of England.
The main argument against this is that bail-outs, including LOLR operations to solvent but illiquid banks, are always and inevitably deeply political, and can easily become party political (e.g. Northern Rock – a Northern institution brought down by the Southern gnomes of London). Property rights and the distribution of wealth and income are inextricably intertwined with the LOLR function. Should the institution that was granted LOLR assistance turn out to be insolvent after all, the Treasury will have to carry the can, by compensating the Bank for any losses made as part of the LOLR operations. If it failed to do so, the Bank might no longer have the financial resources to pursue its mandated inflation target. How can the Bank be independent in the domain of monetary policy, when it is engaged in deeply political LOLR operations and may need to call on the Treasury to recapitalise it if things go wrong?
(2) The minimalist monetary authority model.
This is the same as (1), but with the Monetary Policy Committee taken out of the Bank of England. The Chairman of the MPC would no longer be the Governor of the Bank of England. It might be interesting to have the Governor of the Bank of England and the head of the FSA as ex-officio external members of such a new-style MPC. The MPC would continue to have the same mandate price stability (with a numerical inflation target set by the Chancellor) and subject to that, growth, employment and all things bright and beautiful. The MPC would have but one instrument, Bank Rate, interpreted as the target for the overnight interbank rate. The Bank would act as agent for the MPC in using its money market and repurchase operations in the overnight market to keep the overnight rate as close to Bank rate as possible. Everything else, the Standing Lending and Deposit Facilities, market operations and repos at maturities longer than overnight and foreign exchange market intervention would be the province of the Bank. So the Bank would have both the LOLR responsibilities for individual banks and the responsibility for providing adequate liquidity to the key financial markets as a whole. Again, the information and the resources required for effective fulfilment of the LOLR role would be with the same institution – the Bank.
(3) The FSA as Lender of Last Resort model
A third model would be the one I proposed in my inaugural lecture at the LSE in 2006. This is to have the current arrangement and division of labour between the FSA and the Bank, that is, the FSA as bank supervisor and regulator and the MPC in the Bank, but with one key modification: the Bank’s role in the LOLR function would be entirely passive. The FSA would be given a credit line (overdraft facility), uncapped and open-ended, with the Bank of England, guaranteed by the Treasury, to make sure the Bank’s financial resources are not impaired. The FSA would have to responsibility to decide whether to make a LOLR facility available to an individual bank, and on what terms. The Bank would be able, through open market operations, to undo any undesirable systemic liquidity consequences of the FSA’s LOLR operations. The Bank’s role in the process would be entirely passive – as the provider of the credit line to the FSA, guaranteed by the Treasury. The FSA’s access to the credit line with the Bank would be unconditional, but it would of course be accountable for its decisions.
This proposal too would put the information and the resources required for effective fulfilment of the LOLR role with the same institution, but here that institution would be the FSA.
I have a slight preference for the third option. There may well be other ways of skinning the cat. One thing is clear, though: any arrangement that, like the existing one, puts the information required to perform the LOLR function properly in a different institution from the one that actually has to perform the LOLR function, is doomed to failure.