Rethinking Economics: reappraisal of too-big-to-fail

by George Hatjoullis

In Rethinking Economics: secular stagnation I introduced the idea that the asset risk preference of savers, as well as the aggregate level of saving, is important in keeping growth of GDP at potential. Moreover, I suggested (as in many previous blogs) that this phenomenon dates to the 2008 financial crisis. The key development was the loss too-big-to-fail status of large banks. This is most evident within the EU as it is central to the bank reform and banking union. However, the automatic bail-out of systemically important banks by the taxpayer is under review globally. Why is this so important?

In simple and conventional macroeconomic models, ex ante saving equal to investment is an equilibrium condition. It is a condition for GDP to grow at potential. The fact that investment generates risky assets whilst savers might prefer risk free assets is never discussed (to my knowledge). Implicitly there is some financial intermediation going on that automatically satisfies savers risk preferences irrespective of what they may be at any time. Banks play an important role in this financial intermediation or at least they did until 2008. Savers place cash with banks and banks lend to enterprises. The bank loans are risky but the cash in the bank has always been treated as risk free. The insured amount is, of course, risk free. However, the insured amount is relatively small (per account holder) and large deposits (over £85k in the UK, over Euro 100k in the eurozone) are uninsured. So why were they treated as risk free prior to 2008? The answer is that too-big-to-fail was always assumed. Moreover until 2008 this proved a safe assumption. The banking system magically generated risky loans to industry and the required risk free assets for savers irrespective of risk preference.It did so through an implicit state guarantee to all uninsured depositors (and senior bond holders). When the UK government bailed out RBS it really bailed out uninsured depositors.

In this context, there was no need to overcomplicate economic models with saver risk preference. Aggregate saving relative to investment sufficed. However, this is no longer the case. Uninsured deposits are just this; uninsured. They are risky assets and if the saver wants a risk free asset simply placing money in the bank is not sufficient. One should also note that netting is not automatic in bank resolution. A corporation or individual that has been granted a loan by a bank and has had the funds placed in a current account could find themselves in a sticky situation in the event of bank resolution. They might still owe the bank but the cash in the account could be used to recapitalize the bank. The Bank of England is presently looking at proposals to insure temporary large transaction balances arising out of events such as house sales and purchase. It would not be necessary to do this if too-big-to-fail still applied.

It was always understood that banks created a maturity transformation, converting short-term deposits into long-term loans. It was never fully appreciated that they also effected a credit transformation, creating risk free assets from risky assets through an implicit tax payer guarantee. This is no longer the case. For most individuals this is not an issue as savings are generally less than insured amounts. However, for corporations and high net worth individuals this is a big deal. Moreover, as most wealth and saving is concentrated in the hands of a few, this creates a real dilemma for much wealth and saving. The removal of the too-big-to-fail guarantee created an instant shortage of risk free assets, even if risk preferences had not changed. However, they did change after 2008. Savers became much more risk averse so the demand for risk free assets decline just as the demand increased. The result was an excess demand fro risk free assets.

The impact on investment rather depends upon the model of portfolio allocation one uses. In order to give a flavour of the consequences let us a consider a framework popular in finance; the capital asset pricing model. In this world investors achieve their desired portfolio risk by holding some combination of the risk free asset and the ‘market’ portfolio of risky assets. If investors wake up one day and discover what they thought was a risk free asset is in fact risky they will logically adjust their portfolios. They will reduce their holding of risky assets and try to increase their holding of risk free assets to get back to their desired portfolio preference. If the event in question also increased their collective risk aversion, this portfolio adjustment would be even more substantial. The market portfolio is the counterpart of real investment. It is the set of financial claims generated by real investment. Logically, it would reduce the desire for real investment. The other side of the coin is an increase demand for risk free assets. The yield on these assets would fall, at least relative to the yield on risky assets. Government bond yields would fall. They have fallen universally close to zero and in several cases have become negative. The removal of the too-big-to-fail status of banks has contributed to a suppression of investment intentions as well as historically unprecedented bond yields independently of other forces.

There are of course many other forces in play. The secular stagnation concept has validity in that aggregate saving has almost certainly risen relative to investment and has a secular tendency to do so. There have been other developments reducing the supply of risk free assets and increasing risk aversion. However, one significant new event that has reinforced these developments has been the removal of the too-big-to-fail backstop from large banks. The reason I have focused on this, and used the term ‘Great Deflation’, is because it has had dramatic implications for monetary policy. A large portion of the deposit base ceased to be ‘money’ and became a risky asset. Monetary theory would predict that a precipitous drop in the money stock would have deflationary consequences. We can argue about transmission mechanisms but the money stock was affected by removing too-big-to-fail is indisputable. Moreover, in the absence of some reforms, such as introducing Transaction Accounts suggested by the Iceland Reforms (discussed in earlier blogs), monetary authorities are unable to offset this drop in the money stock. The reason is that however much central bank money they pump into the system it cannot be held as central bank money. It is ultimately held with banks and a large amount is thus re-intermediated into a risky asset in the form of an uninsured deposit. Central banks have been unable to offset the excess demand for risk free assets by supplying risk free central bank money because there is no mechanism for holding economic agents to hold central bank money and bank money over the insured amount is no longer money.