In his book The End of Alchemy, former Bank of England Chairman Mervyn King has presented a riveting treatise on banking, monetary policy, financial crises and financial regulation.   The title of the book refers to what King calls the “alchemy” of banking, whereby liquid and safe deposits are transformed into illiquid, risky loans and securities.  This transformation includes a maturity transformation in which short-term or 1-day deposits fund long-term investments, and a risk transformation in which risk-free deposits fund investments with substantial risk of default.  The theoretical purpose of a bank is to manage these risks, the so-called “interest rate risk” and “credit risk.”

King’s choice of the word “alchemy” to describe this process suggests his view of its long-term viability.  Even with great risk management capabilities at the banks, it is not really possible to successfully manage these complex transformations (just as no one has figured out how to transform common materials into gold).   The problem shows up in periodic “runs” or financial crises where banks suffer losses and the depositors become scared and attempt to withdraw their funds.  With limited reserves and cash on hand, the banks may be unable to accommodate these withdrawals and may be forced to close their doors.  The standard “solution” to this problem, whereby deposits are insured by federal regulators, is a non-solution in that taxpayers end up on the hook.  In fact, in can be argued that deposit insurance makes the problem worse in that banks are able to acquire huge amounts of deposits because depositors have no reason to be concerned about the financial quality of their bank.

This problem has been identified before and has elicited potential solutions.  More than eighty years ago, economist Irving Fisher proposed the concept of “narrow banking” in which banks that took on deposits were restricted to hold only cash and risk-free investments.  Meanwhile, banks that offer risky loans must be 100% funded with long-term debt and equity.  Similar ideas have been proposed over the years by the great Milton Friedman and more recently by John Cochrane and Larry Kotlikoff.

While Chairman King supports the narrow banking concept, he believes he has an idea which is easier to implement, less disruptive, and likely to accomplish the objecting of ending the alchemy.  The basic idea arises from his experience as a central banker with two primary objectives.  The first objective is to strive for price stability (usually as expressed by 2% annual inflation) in normal times, and the second is to serve as the lender of last resort (LOLR) in times of crisis.  The historically accepted description of the LOLR function is from English writer Walter Bagehot:  “lend freely, against good collateral, at a penalty rate.”  King points out that this 1800’s vintage description does not apply today, because banks don’t hold sufficient amounts of good collateral!  The process today, as carried out by the Bank of England over the past decade, is to assign a suitable “haircut” (percentage between 0 and 100) to every piece of collateral and to lend against (100-haircut) % of the asset.  That is, suppose the central bank is lending against business loans.  The central bank will assign a haircut of, say, 40%.  That means the central bank will lend up to 60% of the book value of the business loans.  Apparently, that is what the Bank of England has been doing the past several years, and in doing so has acquired confidence in its ability to assign appropriate haircuts.

King’s proposal to end the alchemy of banking is to have the central bank come up with haircuts for every category of assets that a bank may want to use for collateral.  Adding up all the lendable amounts for a particular bank provides the concept of Liquid Assets (LA).  On the liability side, King proposes measuring Liquid Liabilities (LL) as the sum of deposits and short-term borrowings.  Then the very simple requirement for each bank is that LA must be equal or greater than LL.  This means that every bank is protected against a run – all “runnable” liabilities are covered by liquid assets.  (Of course, banks can still become insolvent – where the mark-to-market value of equity is negative.)  The beauty of this proposal is that it is very simple to implement, just one new requirement (LA>LL) that can be phased in over a number of years.

To take a simple example, consider a bank with $100 million assets comprised of $40 million business loans, $40M mortgage loans, $10 million consumer loans, and $10 million cash and government securities.  Haircuts for these asset types might be 40%, 20%, 50% and 0%, respectively.  This would suggest an LA total of $71 million (60%*40+80%*40+50%*10+100%*10).  Further, suppose the bank has $80 million of deposits, $10 million of short-term debt and $10 million of equity.  In this case LL is $90 million and the shortfall of LL over LA is $19 million.  Under the King proposal, this bank will have ten years to do a combination of increasing LA or reducing LL by $19 million.  The most straightforward, but maybe not the easiest, method would be to raise $19 of equity.  Alternatively, the bank could reduce its risky assets or increase its long-term debt.

What is interesting to me is the process by which the Central Bank will assign haircuts.   In the extreme case where the haircut is 100% for each risky asset, King’s proposal reduces to narrow banking.  In the case of less than 100% haircuts, the King proposal is an intermediate solution.   I would guess that the procedure would be to assess the worst case loss scenario and the percentage loss on an asset class in that scenario would be the haircut.  Consider the example of very well-underwritten mortgage loans.  In normal circumstances the credit losses are near zero, and in an extreme scenario the loss percentage (given a big drop in housing prices) might be 10% or so.  So, is 10% a reasonable haircut?  The answer is “no”, because this calculation only addresses credit risk.  Once you include interest rate risk, it is easily possible (think 1980) that a 30-year fixed rate mortgage loan could decline in market value by 50% if market mortgage rates increased a lot.  Thus, a reasonable haircut might be somewhere between 50% and 60%, depending on the correlation of credit risk and interest rate risk.  Yet, banking regulators have consistently assigned very low risk weights to well-underwritten mortgage loans.

Actually, the more I think about this idea, the more I want to wait and see how the Central Bank proposes to assess haircuts.  On the plus side, this approach is likely to provide significant employment opportunities for quantitative economists (like those graduating from CLU’s Masters of Science in Quantitative Economics program).  But, on the other hand, I suspect that the final negotiated haircuts may not fully reflect the risk of the asset.   In this light, the Cochrane narrow bank proposal makes a lot of sense.