The overall strategy proposed by “the money problem”

For Morgan Ricks, a financial crisis with serious real consequences is characterised by a generalised run on bank deposits or on any other short-term debt deemed equivalent to cash. Such runs are the manifestation of multiple equilibria in the financial sector. They have dire consequences for the growth trend of the economy under consideration. Consequently, any efficient financial regulation should, by design and as a priority, concentrate on preventing the occurrence of such runs.

For this, according to Ricks, it should first offer to the regulated institutions and against appropriate fees, public insurance on deposits and on all other short-term debt able to transform into cash in order to finance transactions. This would extend to the other relevant short term debts, but not necessarily for identical fees, the deposit insurance currently offered to chartered banks and which was successful in preventing generalised runs as long as the size of short term markets had remained limited. Second it should forbid institutions not submitted to such regulation to issue cash equivalents. Thus monetary financial institutions would be chartered under a common regime which would exclude other financial institutions and prevent them from operating like monetary ones.

The scope of the regulated financial sector would thus match the scope of the money issuing sector as defined traditionally by monetarist economists, encompassing money settling transactions rather than the financing of investment. This match and the related public insurance would, in turn, limit the need for a lender of last resort as well as the contagion effect of isolated runs. Generalised runs would therefore not occur. Indeed, once such consistency is imposed, capital requirements and portfolio constraints, which are the main supervisory tools, interfere much less with lender of last resort policies. The former can reach lower levels while the latter are comparatively less used. This ensures the stability of the banking sector and the optimality of monetary policy more naturally. This, in turn, leads to simpler and more efficient regulation. By contrast, without such consistency, the lender of last resort is often needed. Its action creates moral hazard, which interacts with and strengthens the need for capital requirements and portfolio constraints.^[1]

Ricks then fleshes out this overall strategy by developing what he considers as its natural consequences. For him, the economic purpose to be pursued by regulated institutions is not to intermediate savings and investment as this is the aim of the overall financial sector, which is wider than the regulated sector. By contrast, as the regulated institutions or banks are by assumption deemed to be the money-issuing sector, the economic purpose of the banking sector is to supply the amount of money necessary to clearing and settling transactions in this economy. This points to a distinction between commercial banking dealing with deposits and short-term debt and investment banking dealing with risky investment and trading.

Determining the adequate overall money supply may be done in practice by a monetary policy committee while the lending or purchase of assets may benefit from the skills and information of private agents. As a consequence, while the central bank or the state itself may also in principle directly perform the selection of corresponding assets and loans, it is rather natural for Ricks to hand it over to private but regulated banks and conceive of the banking sector as a public-private joint venture. If so, the fee charged for the public insurance of money-equivalent liabilities may well exceed the simple cost of bank failures usually charged in the context of deposit insurance: indeed, part of it would need to take the nature of seignoriage that should accrue to the state rather than to the private sector and would compensate for the exorbitant privilege conferred on banks by the public insurance of cash-equivalent claims. It would also show the “non-market” nature of such co-ordination and would reflect the role played by institutions in shaping the actual functioning of markets. This is another important feature of the regulatory design foreshadowed by Ricks.

This scheme is a priori consistent with several types of financial structures. With a hierarchical structure as currently, the central bank is the sole issuer of banknotes while banks can issue deposits and other short term liabilities. Its monopoly on banknotes then puts the central bank at the helm of the banking sector which in turn allows the central bank to become the location of the monetary policy committee. This can carry on.

However, while a central bank may be useful, it is not strictly necessary to the scheme: as noted by Ricks, the insurance offered by the state to regulated private banks against fees makes banknotes, deposits and other short term debts equivalent in the eyes of banks, economic agents, and the state itself. All types of money-claims equivalents can therefore be given a single name and called r-currency. This points to the possibility of different, less vertical or hierarchical, financial structures reflecting the “horizontal divide” between monetary transactions and the financing of risky investment more directly.

Hence, Ricks prefers and proposes to concentrate on the special case of a flat financial structure, where there is no central bank: all banks report to a state monetary agency which indicates to each bank their money supply cap, provides insurance and cashes the insurance fees. Moreover, he proposes an accounting scheme for transfers that avoids considering banknotes or deposits at other banks as the funds which allow the safe creation of loans and bonds. This safe creation, a third featuring aspect of the Ricks’ proposal, results from portfolio constraints and capital requirements combined to insurance fees ensuring that the private banks do not appropriate the whole value added.

Justifying the various steps of the strategy

On this basis, Ricks analyses several issues in more detail. First, he makes more precise the concept of money-equivalents. His view is that this category includes not only cash, but all other transaction media and instruments which are kept for the purpose of helping fulfil payments obligations. Such instruments must not only be liquid, that is be easily sold against cash, but also have little short term price volatility so as to avoid putting transactions in jeopardy whenever their price varies too much. He therefore identifies short term debt as the functional complement to cash and remarks that such category is called quasi-money by many monetarist authors. Also, the “money” markets are called this way precisely because they ensure the transformation of such short-term debt into cash with limited price volatility. Hence, if money markets extend up to a maturity of 12 months and capital markets cover longer maturities, only regulated institutions or “banks” should be able to fund themselves on money markets while other financial institutions should issue their liabilities on capital markets only. Thus the regulated banking sector is responsible for issuing a monetary aggregate very close to the definition of M3.

Secondly, generalised runs or panics, which arise when markets fail to allow the renewal of short term debt or cash-equivalent liabilities while depositors ask for redemption of their positions with banks, have for Ricks substantial, negative, and durable impacts on economic growth. Their occurrence is the main cause of the most severe recessions. This is in line with the views of Bagehot (1873) for the nineteenth century, those of Friedman and Schwarz (1963) for the Great Depression and those of Bernanke as expressed in his 2010 Jackson Hole speech for the Great Recession. This disagrees with the approaches of Irving Fisher (1933), Hyman Minsky (1992), Richard Koo (2003), or John Geanakoplos (2010) who all put more emphasis on a debt cycle of some sort, be it a debt deflation, balance sheet, financial accelerator or leverage cycle as the main cause. However, according to Ricks, panics explain actual developments better than debt dynamics. He rests his case both on the US and on Japan. In the case of Japan, the impact on GDP would have been limited until 1997, moment when a small default in the shadow banking sector led to a paralysis of the entire interbank market and accentuated this impact. This panic, often ignored by economists out of Japan, would coincide with the reopening of a protracted output gap in Japan and prove that panics are the major cause of durable growth slowdowns. As for the US, the shape of the unemployment series during the Great Recession resembles most closely the shape of the CDS-bonds basis as shown by Graph 4–13 of the book. Moreover, the panic was accompanied by a durable financing crunch. Rick therefore sees the panic as the major cause of the protracted US slowdown. Finally, to the question –why did the economy not bounce back once the panic was over?- Rick answers like Keynes that the economy may not be self-adjusting and that “the economy may find itself in stable equilibrium with [employment] at a level below full employment.^[2]” In other words, and as found by Fed economists in a recent paper on 23 advanced economies, growth is rarely faster after a recession than before anyway.

Thirdly, Ricks shows why banking, defined as the business of issuing money-equivalent claims against assets such as loans and bonds is an activity prone to runs, independently of fundamentals. In the absence of public insurance, this business depends on the expectations of customers concerning the actions of other costumers and on the degree of co-ordination of such expectations. If all expect that others will redeem their claims on banks, the runs will take place. If they do not have such expectations, no run will take place. This is a kind of co-ordination that markets cannot implement by themselves. This implies the need for a regulatory design consistent with its achievement.

Fourthly, the main tools available to regulators -that is portfolio and capital requirements- have their limits but these limits are less binding in the context of the proposed financial design. Portfolio requirements imposed to banks are one of these tools. They tend to limit profitability so that imposing them in view of limiting the occurrence of panics would imply comparing such impact on profitability with the social cost of panics. This would in turn tend to support extreme positions like 100 % reserves or narrow banking proposals. Capital requirements on heterogeneous assets need to be complemented by risk-weighting and contingent claims have to be taken into account if the requirements are not to be easily gamed. Thus Value at Risk techniques have to be used and parameters used may depend on the type of assets. This, like the specific kind of accounting suggested by Ricks in conjunction with the proposed financial design, also reminds us that accounting techniques may usefully differ depending on the needs generated by the institutional environment and their impact on markets.

Overall, portfolio constraints and portfolio requirements are easier to apply to a limited and well-defined set of assets but tend to become heavy and inefficient when applied to institutions that manage heterogeneous assets. Moreover, the more complicated, the more they limit the supply of money-equivalent claims in ways that might contradict the need to finance transactions. Consequently, being ready to supply large amounts of money-equivalent claims whenever needed from a monetary and macroeconomic viewpoint may be facilitated by ensuring that the assets of institutions to be regulated are of high quality and homogeneous.

As a result, the banking activity defined as proposed by Ricks is fundamentally different from the issuance of equity or long term bonds in order to finance credit. Beyond the need to be chartered, the payment services that money-equivalent claims offer make financing through the issuance of money-equivalent claims naturally cheaper than through equity or bond financing. Consequently, the Modigliani-Miller theorem does not apply to banks. Moreover, under the proposed financial design, systemic risk regulation would not be needed given that generalised runs would be avoided and that systemic risk as considered by Diamond and Dybvig (1983) would not apply to money-equivalent claims in the context of Ricks’ scheme.^[3] Monetary policy would therefore suffice to ensure financial stability, once the right financial design properly outlined.

Fifthly, the lender of last resort (LOLR) is an inefficient tool to control panics. The support of the LOLR is supposed to go to solvent banks and to be guaranteed by good collateral, as prescribed by Bagehot. However, how to identify solvent banks safely within a panic? How also to be sure of the value of collateral in a financial crisis?^[4] Moreover, its existence acts like a subsidy to the banking sector and leads to moral hazard, including occurrence of “too big to fail” behaviours. Finally, limiting the corresponding subsidy is equivalent to imposing a further constraint on the size of the money supply.^[5]

Also, when the need for a LOLR is avoided by the definition of the banking sector through the limitation of its liabilities other than equity to money equivalent claims and their insurance by the state, portfolio restrictions and capital requirements become the natural complements of the fees paid to the state for their insurance. Indeed portfolio restrictions restrict moral hazard connected to the insurance while the capital requirements impose a first-loss which corrects and aligns incentives. Moreover, this is also consistent with excluding banks from dealing with derivatives as derivatives tend to maximize rather than minimize the amount of risk taken per money equivalent created.^[6] This again makes banking an activity related to a monetary purpose rather than to a financial intermediation purpose.

Ricks is able to wonderfully combine arguments in support of his view that banking should be defined first and foremost as a monetary activity and that the simplest way to govern and manage this sector is to generalize deposit insurance to all money-equivalent claims. This public insurance would take care of the run-prone nature of the banking sector. As this would diminish the risks and eliminate the needs for a LOLR, this could make the implementation of portfolio constraints and of capital requirements more straightforward and more powerful. Lo and behold, this would diminish the need for complex regulations. The non-bank financial sector would ensure that savings and investment match. Monetary authorities would only need to fix the overall amount of money in line with needs.

Moreover, Ricks creates an expectation that the worst consequences of financial crises and in particular the protracted slowdown of growth which follow them, could be avoided. This implies even that the very occurrence of strong financial crises could be avoided.

But is that realistic?

Do generalised runs always constitute the main cause of severe recessions?

Ricks argues rightly that the most severe recessions in the 19th and twentieth centuries were accompanied with panics. This is consistent with an aggravating role of panics. Moreover, he rightly shows that such link can be found in US and Japanese data concerning recent financial crises. Moreover, it could be justified in the case of European countries too.

However, this aggravating character of panics is not necessarily indicative of a dominant impact of panic itself, especially if multiple factors have played a role and panic was just the last one to set in. Indeed, panics will in such case appear as associated with the worst part of the recession but may not constitute its main cause: such multiple factors may generate a substantial degree of slowdown in the economy already before the panic sets in, at which time the protracted slowdown of the economy turns into an even severer recession.

This reflects the possibility that, as economists know, causation is not equivalent to correlation. Ricks of course does not ignore such difficulty and takes care to explicitly consider the possibility of a specific alternative factor to panics, in particular in the case of the recent financial crisis in the US. He therefore asks: Is the severe recession the result of a specific factor, for instance the collapse of a debt-fuelled bubble, or is it the result of the panic that sets in at some point as a result of this collapse? Moreover, if it is the result of both factors, which one is most significant?

His answer in the case of the US is certainly acceptable. In the US recent financial crisis, the runs happened shortly after the burst of the real estate bubble which was not more severe than the real estate crisis associated to the Savings and Loans crisis at the beginning of the 80’s. Moreover, the burst of the real estate bubble did not create losses larger than those associated to the burst of the 2001 Internet bubble which had a temporary and limited impact. Moreover, the financial sector was still perceived as very profitable just before the onset of panic. Hence, the role of the panic in the US recent financial crisis must have been most significant. Moreover, the causality from runs to recession is all the more convincing because the severity of the recession and the runs happen close to each other and early in the slowdown process.

But this was not the case in the example of Japan. Indeed, the panic of 1997 mentioned by Ricks was real and does show that panic aggravate recessions. However, it took place more than seven years after the burst of the real estate and stock market bubbles and followed a very long period of negative output gap, as shown by Figure 4.15 in Ricks’ book. Hence, the cost of the bubbles was substantial already before the panic, even if clearly amplified after the panic of 1997. This is why in the case of Japan, other explanations than the panic itself are often explored to explain the slowdown in growth and investment such as the slow closure of loans to zombie-firms: this slow closure delays the recovery of banks’ profitability, contributes to the slowdown of investment in innovative start-ups and thereby to the slowdown in growth (see Ahearne & Shinada, 2005 for instance).^[7]

Figure 1 below lengthens the period under consideration and starts in 1979 instead of 1990. This leads to giving more importance than Ricks does to the debt bubble factor in the history of the financial crisis in Japan. Indeed, the disappearance around the end of the 80’s of the 80’s original upward growth trend appears more clearly associated to and initiated by the debt bubble from 1986 on and its subsequent collapse in 1990. In such context, the slowdown between 1990 and 1997 appears more significant when contrasted with the 1986–1990 acceleration of growth while the impact of the 1997 panic on the subsequent period, although visible, is less telling.

Figure 1:

Japan GDP: lost decades?.

In dollars 2010, source OECD

Second, beyond the question of the 1997 panic impact, the impression is that two decades rather than one were lost by the Japanese economy. This leads to wondering if some factors common to the two periods would not usefully complement the panic story. Combining the debt bubble factor to the zombie-story and to the 1997 panic and taking into account the nature and speed of the Japanese structural adjustment may be a way to achieve it.

Indeed, in the absence of a delayed structural adjustment to the debt bubble, the first “lost decade” would not have been associated with a weakening of the profitability of the banking sector which made it even more fragile and sensitive to a banking panic in 1997. The differences of productivity between frontier and zombie firms after 1997 would not have been that large. The financial health of banks financing the innovative investments after 1997 would have been better and would have allowed a stronger recovery, broadening the impact of efforts made to stabilize the banking sector.

My argument is therefore not one of relative emphasis on two successive periods of time. It proposes a more general interpretation of the same facts and shows that such interpretation allows covering a broader history. Admittedly, it does not use the milder nature of the early 1990’s to conclude that the panic was the main driver of the prolonged slump. It rather argues that the delayed adjustment in the early 1990’s to the debt bubble and its burst made a more dramatic but later adjustment unavoidable. That adjustment took place both in the follow-up of the 1997 liquidity crisis and of the 2008 one as the “zombie firms” problem had to be faced by banks and later by the economy at large. It is still being faced by some sectors. Moreover, no wonder that the 2008 impact of the global liquidity stress was clearer than the one of 1997 but did not modify a trend established since the beginning of the 1990’s.

A similar, but more complex set of remarks applies to the case of the euro area. The crisis erupted in August 2007 in the Euro Area at the same time as in the US and the UK. It went on in a very similar manner as shown by interbank market spreads during 2008–2009 (see Figure 2 below). Runs and original liquidity dry-ups took place in the US and the Euro Area simultaneously. However, the level of those spreads^[8] was more limited in the Euro Area up to 2010 at which time they started to increase due to the development of the Greek crisis. That crisis was originally a fiscal crisis which through its impact on the balance sheet of Greek banks gradually became a banking crisis and, by contagion, a European and even global crisis. Beyond Greek fiscal events, other factors of course include events associated to progress (or lack of) in the field of Euro Area public debt management. This created a situation peculiar to the EU, where the identity and policies of the lender of last resort were either unclear or assessed differently across time by market participants. However, even under such circumstances, full scale liquidity runs did not materialise.

Figure 2:

Interbank market spreads.

Note: Spreads are the difference between 12-month Euribor/Libor and Overnight Index Swap rates, in basis points. Source: Bloomberg. Last observation: 30th November 2012

Hence, in that case, the liquidity runs at the start of the crisis in 2008–2009 do not suffice to explain its rekindling under the pressure of Greek fiscal information in 2010 and later. They do not explain the deepening of the crisis after 2010 and the clear negative impact on growth and investment also visible from 2010 on. Other factors must therefore have played a role independently from liquidity runs that the monetary policy of the ECB did explicitly prevent in the second part of the crisis thanks to its collateral and non-standard measures.

It is true that some broad stresses or worries on liquidity have continuously existed in Europe since 2008. However, after the liquidity run of 2008–2009 which mostly harmed Irish and Icelandic banks, these stresses or worries never culminated in a panic such as seen in the US after Lehman. Even in 2011, spreads did not reach in the Euro Area their levels of 2008–2009. Hence, the injection of liquidity by the ECB, which had started as early as August 2007, was not and never had to be as massive as the expansion of the Fed following the collapse of Lehman: it was not responding to a panic in the sense of a generalised run although it was probably not far from it. Only after the resolution of the most stressful episodes of the crisis (see Figure 3), that is after the start of 2012, did the size of the ECB balance sheet gradually increase to levels similar to Fed levels. This was from 2014 on rather in an effort to avoid the risk of deflation, support financial stability and relaunch growth and investment.

Figure 3:

Central Bank balance sheets (% of GDP).

Consequently, the banks that had funding difficulties found sufficient resources with the ECB in time to avoid a panic that would have dried fully the market. This policy of easy but not unlimited liquidity gradually led to segmentation and sometimes segregation on money markets but did not lead to the co-ordination of pessimistic expectations typical of a panic or generalised run. This explains that the main events driving liquidity stress in peripheral countries were of fiscal nature rather than of banking nature. This does not imply that banks were not in need of liquidity and that banking news played no role. But their needs, although increasingly significant between 2010 and 2012, were covered even at end 2012 by the access to the ECB facilities and the extended definition of collateral accepted by the ECB: for this reason, their stresses were more decisively driven by fiscal and public debt management news.

As a consequence, I find it difficult to believe that generalised runs always constitute the main cause of severe recessions. Panic-proofing, although useful, may not be full-proof. So panic-proofing, although it will modify and at least attenuate or delay the expression in the economy of financial instability factors, is unlikely to prevent their manifestation totally.

Widening the deposit insurance to other money-equivalent claims is certainly a good idea. However, it certainly needs to be considered in view of its likely impacts rather than of its putative consequences in an ideal world. I am therefore less sure than Ricks that the improvement brought by his suggestions would be so large and therefore so easily acceptable to the general public. This leads me to examine another issue which is never explicitly dealt with by Ricks.

Can monetary policy effectively determine the amount of money needed by the economy?

It is clear that with less runs the economy may be less unstable. Thus, the link between monetary aggregates and prices may become more stable. However, the question of whether the stability of money demand is sufficient to inform monetary policy decisions may remain, at least while the functioning of the banking system is affected by its reorganisation along the lines suggested by Ricks. After all, the Federal Reserve does not use monetary aggregates to explain its decisions anymore. Moreover, even the ECB, which does use monetary aggregates for such purpose, does not do it mechanically and uses instead a two-pillar approach to explain its decisions (Papademos & Stark, 2010). Determining how much money the economy needs unavoidably mixes rules and discretion.^[9]

Furthermore, the major factor in a world with less panics and runs may well be the pressure for innovation: the pressure to develop new money-equivalent claims outside the balance sheets of the banking sector or to diminish the use of bank liabilities in the transactions of the non-bank sector may strongly affect the role of monetary aggregates or make them more difficult to define.

Ricks’ proposal to account for all money-equivalent claims as r-currency raises the issue of the treatment of digital private currencies which, like Bitcoin, would allow for transactions at low cost without a central counterparty: would Bitcoins be considered as currency because it allows transactions and therefore be taxed or alternatively considered as an asset with high price volatility and not be taxed despite its use in transactions? Indeed, financial innovation could put into question the very link between the use of money-equivalent claims in transactions and their coverage by public insurance that defines the banking sector in Ricks’ approach.

The need for an implementation strategy

The future of regulation will thus depend on financial innovation. But it will also depend on the political and intellectual process accompanying it. Banking is always part of a political process due to the nature of money, which is a public good. The current regulatory system, which owes to this political process, constitutes a starting point that needs to be considered even though the final point will certainly be different.

Ricks does not actually develop this point in any depth although he does sketch “Getting There from Here” in Chapter 9. He mentions that his proposal implies a “public-private partnership” concerning the “banking sector” and fees for the insurance of money-equivalent claims that would not be determined at marginal costs but would represent seignoriage and hence would be higher. But the precise manner to determine and justify such fees may not be so easy to agree upon. As well-known, many innovations which make sense from a collective viewpoint are unlikely to materialize (see Moutot (2014) for an example in a competitive framework), especially if for some citizens the fees still appear to be low and for some banks excessive. The weight of the institutional logic put forward by Ricks may therefore be insufficient.

Even more problematic is the treatment of the non-banks which today, 10 years after the financial crisis, finance on the short term money markets large credit positions. Implementing Ricks’ proposal would imply that such non-banks would have to replace their short term funding with long term funding or decrease their credit positions. This may take time and be controversial, in an environment where such shadow banking sector has reached a very significant size, but has not been (yet) the source of any new runs. This may be also viewed as untimely, as such credit may compensate for some decreased funding from the banking sector itself adjusting to the new regulatory environment.

In the meantime, the strategy which consists in monitoring systemic risk to avoid runs may be inefficient or exaggeratedly complex but is still helpful.

If a new financial crisis were to occur however, Ricks’ proposal would certainly be one of the possible building blocks for a response to such new situation.