- Email: [email protected]
The parlous state of macroeconomics and the optimal financial structure
International Review of Financial Analysis 36 (2014) 78–83
Contents lists available at ScienceDirect
International Review of Financial Analysis
The parlous state of macroeconomics and the optimal financial structure C.A.E. Goodhart Financial Markets Group, London School of Economics, United Kingdom
a r t i c l e
i n f o
Available online 29 October 2014 JEL classifications: E40 E44 E50 E51 G20 G21 G28 G33
a b s t r a c t Macroeconomics remains in a parlous condition, largely because it has assumed away all financial frictions. Ultimately these latter depend on the possibility that borrowers might default on their repayments. Without default, there is no real role for most financial intermediations, collateral, liquidity or money. Yet default (especially of banks, the key ingredient of crises) is rarely modelled. In order to make banks safer, in the aftermath of the Great Financial Crisis, there are various proposals to restructure our banking systems, for example to dismantle universal banks into separate retail and investment parts. This partly derives from a mis-reading of the causes of the GFC, which was largely driven by an interaction between a housing boom and a bank credit expansion surfeit, thereby exaggerating leverage, mis-match and non-core bank finance. The need is for regulatory improvements that address these weaknesses. © 2014 Elsevier Inc. All rights reserved.
Keywords: Default Liquidity Money Stock Great Financial Crisis Bank Restructuring Liquidity (Mismatch) Ratio Ladder of Sanctions
1. Part I: the parlous state of macroeconomics 1.1. The recent history of macroeconomics One reason why I found the study of macroeconomics enjoyable is that it has been in such a mess. Consequently I thought (mistakenly) that I might be able to make a difference. As the old joke has it, ‘The questions in the Economics Tripos remain the same, but the required answers change.’ As Dennis Robertson once said, “Macroeconomics is rather like coursing for hares. If you stand in one place you will find that the hares will double back to you and you will see them as they return to the starting point.” When I was a callow undergraduate at Trinity College, Cambridge, in 1957, some 55 years ago, I had the opportunity to meet Dennis Robertson. At that time I was quite excited by studying those debates about macroeconomics which had raged in the 1930s, e.g. liquidity preference versus loanable funds. So I asked Dennis whether it had been exciting for him to have been a participant. I then saw the look of pain that went over his face, and I realised that I had put my foot in it. Anyhow, I can now, with reasonable confidence, proclaim that, having laboured in the field of money-macro for the last 50 years or more, I shall shortly leave it, with the subject probably being in a worse state than when I initially found it. Let me try to explain why E-mail address: [email protected].
http://dx.doi.org/10.1016/j.irfa.2014.10.014 1057-5219/© 2014 Elsevier Inc. All rights reserved.
and how this occurred. Let me start with a brief review of the main path of macroeconomics as a subject since the end of the 1950s and early 1960s, when I first found it. • We start with computerisation at the end of the 1950s and beginning of the 1960s. As an undergraduate in Cambridge, there was only one computer in the whole university, housed in a huge room, with no likelihood for any undergraduate, nor indeed a research student, being able to access it. This was rapidly changing, and when I went to Harvard, I found my first IBM computers, with punch cards as inputs, and a printer in the form of a typewriter sitting on top of the computer, with the keys going up and down without any human involvement, as the output got printed out. • Then we got the early Keynesian models based on the income and expenditure accounts, with consumption functions, investment functions, import and export functions, etc. All of these took the shape of reduced form fitted equations, trying to provide the best econometric fit to the dependent variable in each case, on the basis of whatever set of variables were supposed by theory, or empirical investigation, to give the best empirical explanation of the time series of the variable in question. • This, of course, led on to the Lucas Critique, that these reduced form equation were inconsistent, incoherent, and unrelated to any apparent pattern of micro-behaviour. This critique was, of course, correct.
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
• This led on to trying to rewrite the equations in the form of constrained utility maximisations (Euler equations) for every agent. But, if one was to have as many agents as there are in the economy, this would be impossibly complicated and virtually impossible to compute or to relate to aggregate empirical data. The result of this was to work on the basis of an assumption that everyone in each of the sectors involved was to be exactly the same, i.e. the Representative Agent assumption. • But, if there was to be a Representative Agent assumption, then either, should default occur, everyone would default and the sector would cease to exist, or no one would default ever. In effect, the latter assumption was made, though frequently this was implicit rather than explicit. But, with no default, there would be no risk premia, and no need for banks or money. Instead, there would just be one (official policy) interest rate, and expectations of its future path. • But, with no default, no banks and no money, and no risk premia, the resulting (DSGE) models were hopeless in a financial crisis. In effect, macroeconomics became totally divorced from finance, with finance depending critically on expectations of probability of default (PD) and loss-given default (LGD). In practice, in my view, the real advances in economics at the macro level in recent decades have occurred in the field of finance, rather than in macroeconomics. By abstracting from default, banking and money, macroeconomics has gone down a blind alley; the concept that all relevant action takes place in the real economy, with finance being nothing more than a veil, has not been helpful. 1.2. Default There is a need to model default explicitly, and that would bring with it the need to model liquidity, banks and other financial intermediaries, along with markets for financial assets. It is true that certain DSGE models now have financial add-ons, such as the Bernanke–Gertler and the Kiyotaki–Moore models, but these have default implicit in their models, even if not explicit. For example, some of these are based on the fact that borrowers need collateral in order to borrow, and the value of collateral varies with the state of the economy. But one only needs collateral in order to protect against default; so, the basis of such models effectively depends ultimately on the possibility of default. Moreover, default does occur; indeed, the default of banks has been at the heart of most of the worst downturns and depressions of the modern era; examples includes Lehman Bros in 2008, Credit Anstalt in 1931, Knickerbocker Trust in 1907, and several others that could be added. Default, however, is hard to model since it is not continuous. The response of Martin Shubik, and several colleagues, has been to reverse the problem and to focus attention on the repayment rate, in the event of default, rather than on the fact of default itself. When one is thinking about a sector, such as the productive sector, one can think of the percentage of non-performing loans or the percentage of write-offs that the bank has to apply. Even in the case of the failure of a single big bank, it is almost never the case that there is no repayment in the case of bankruptcy and liquidation. Even with Lehman Bros, the creditors (eventually) got a sizeable repayment. The repayment rate can vary continuously between 0 and 100% and both the actual, and expected, repayment rates are usually, almost always, somewhere in the interior. Thus, one can use repayment rates as a variable to include within the confines of the standard type of macro-model that is applied in our subject. 1.3. The determination of the money stock Even worse than the analytical treatment of the macro-economy is the recent, and indeed present state of the theory of the supply of money, and study and analysis of the operation of banks; and all this despite Richard Werner's good work (see, for instance, Werner, 2005) and recent book, with colleagues, entitled Where Does Money Come
79
From?: A Guide to the UK Monetary & Banking System (Ryan-Collins, Greenham, Werner, & Jackson, 2012). For example, the analysis of the supply-side determination of the money stock is still based on a money multiplier analysis, which is a purely mechanical relationship derived from identities. This relates the broader money stock, M to the monetary base, H, in a mechanical fashion dependent on two ratios, the currency deposit, C/D, and the reserve deposit, R/D, ratios. But, so long as the central bank sets the policy interest rate, the broad money stock is endogenous, since maintenance of market interest rates close to the policy interest rate requires the central bank to give the banks enough high-powered money to enable them to maintain their desired reserve deposit (R/D) ratio such as would be consistent with the policy-determined interest rate. If the central bank provided an insufficient (excessive) reserve base for the banks, they would push (short-term) market rates above (below) the policy rate, thereby forcing the central bank to inject (withdraw) reserves into the system to make their policy rate effective. But once the zero lower bound to interest rates kicks in, so that the policy interest rate is set effectively as close to zero as can be achieved, then the central bank can, and has with QE, attempted to vary H, the high-powered monetary base, exogenously, expanding H very sharply. So, policy has now changed, with the central bank setting the volume of base money, independently of the level of policy interest rates, which are stuck at zero. With policy changed, both the Lucas Critique and Goodhart's Law, then have kicked in. The increase in H, base money, in most of the central banks, Fed, BoE, BoJ and ECB, has increased by a factor of 3 or 4. Meanwhile, however, the overall money stock has grown, if at all, very grudgingly; and bank lending to the private sector has in most cases been stagnant, or even negative. The money multiplier has collapsed! If the money multiplier had remained stable, so that broad money had increased at roughly the same rate as the reserve base of the banking system, then the crisis would have been over, and, indeed, it might have been replaced by an upsurge in inflation. The bank lending channel has completely failed. Somewhat surprisingly, central banks have been too embarrassed to discuss this failure. What happened? Why did those who take banking decisions not want to use the increase in their reserves to expand their asset portfolios? It is bank CEOs who take the decisions on bank portfolio management. Why did nobody ask what would be such CEO micro-behavioural, constrained utility maximisation under such circumstances, rather than using the mechanical multiplier analysis? The CEOs' answer to their shareholders, and are concerned about the potential for takeovers in the equity market. Moreover, they themselves are almost always large shareholders; this is because they receive bonus payments quite largely in the form of shares in their own bank. So, naturally, they focus on the return on equity (RoE). Also, equity holders, including banks' CEOs, have limited liability. If there should be some disaster, they can depart, crying all the way to retirement with their amassed pot of gold. Equity involves an option call on the assets of a bank. The structure of pay-offs means that equity holders are necessarily riskloving; Northern Rock was the darling of the London Stock Exchange only a few months before it collapsed. Return on equity is generally maximised by expanding leverage, having a very small proportion of equity in relation to a much larger proportion of fixed debt. Consequently the equity holders get the benefit of all the upturn above the required payment to the fixed interest holders, while their downside is limited. So, equity will be minimised, and games will be played with risk-weighting, with banks subject to a riskweighted capital adequacy requirement (CAR) holding a massive portfolio of assets which have a supposedly minimal risk weight. In contrast, banks that are subject to an overall leverage ratio, will try to maximise RoE by holding riskier assets in their constrained volume of debt. Under these circumstances, it is only too likely that tail risk will occur. As a result, people will argue that equity ratios were, have been, and remain too low. And indeed this is so. Admati and Hellwig in their book on The Bankers' New Clothes, David Miles and colleagues in the Economic
80
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
Journal, and the Independent (Vickers) Commission on Banking, have all argued, correctly, that the equity ratio has been far too low. So, the cry has been to raise the equity ratio. But just requiring a higher equity ratio is actually the wrong way to get to a safer equilibrium situation. So long as the bankers continue to focus on RoE, they will try to get there by deleveraging, which will just worsen the downturn. The monetary authorities are, in each country, trying to protect financial intermediation in their own country by seeking and, almost blackmailing, their domestic banks into expanding, or at least not reducing, their lending within their own country; but at the same time arguing that they must enhance their equity ratio by cutting their lending abroad, and by selling assets situated in foreign countries. What should we have done instead? In this respect the response in the USA has been better than that in Europe; in the US banks were put under considerable pressure to raise the absolute volume of equity, rather than the equity ratio. But let us go back to the money multiplier; why are banks content with holding such a huge increase in their reserve/deposit ratio? The answer to this is that central bankers have, mistakenly, been willing to pay a low rate of interest on such safe assets; with zero risk weight, and requiring no capital, the holding of interest-bearing safe assets, which puts no pressure on either capital requirements, and obviously none on liquidity requirements, has seemed to CEOs to be highly desirable. So what should happen? One measure that the authorities could do, would be to reduce the marginal interest rate on excess reserves held at the central bank. Another would be to force the recapitalisation of banks with insufficient equity, if necessary through the use of taxpayer funds, until the capital constraint on uses of bank reserves to add to private sector lending became less unattractive. 1.4. Where should macroeconomics go from here? I would suggest that there are two main lessons that need to be applied, if macroeconomics is going to recover. • The first is that default and bankruptcy matter for macroeconomic analysis and for understanding the relationship between macro and finance. In particular, we need to end the separation between the analysis of finance and the analysis of macroeconomics. • Let us also do a Bob Lucas type micro-behavioural foundation model for the decision-making process of bank CEOs. In the process, we need to put the money multiplier where it belongs, in the deletebutton trash-file. 2. Part II: the optimal financial structure Having made the case that economics must consider banking and finance explicitly, and reflect the institutional reality of our present banking system, let me now turn to the question of how our financial structure should be designed or reformed in order to allow it to fulfil its intended function. 2.1. History is important Banking developed rather differently in Anglo-Saxon countries than on the European Continent and in Japan. In Anglo-Saxon countries, notably the UK and the USA, banks started up before the emergence of large scale industry. Such banks were usually small, unlimited liability partnerships in the UK, and financially fragile. An important determinant of successful continuing business was to avoid getting too involved in concentrated lending to (associated) private firms; lending was to be at arm's length and diversified. Where, even then, there was a need for large-scale finance, e.g. for canals and then railroads, this could and should be provided by, relatively efficient, capital markets, both for bonds and equities. But the entrepreneurs of such large firms and governments, at various levels, did not generally have the necessary information and skills to access
financial (and foreign) markets, so there sprung up another tier of financial intermediaries who used their market skills and information to provide such large entities with access to capital markets. These were the merchant banks (or Accepting Houses) in London, or the broker/dealers (at a somewhat later date, especially after Glass– Steagall) in the USA. Thus in these countries banks were, originally, primarily retail in character, serving the (well-to-do in the) local community, with a separate tier of investment banks acting as keepers of the gateway to efficient capital markets for those large institutions needing access to such markets. The history and experience on the continent of Europe and in Japan were different. Industrialisation occurred later, by which time industrial economies of scale had become more prevalent, e.g. in iron and steel, chemicals, and utilities, but the capital markets there were less efficient, and financial markets were less trusted in these countries. Meanwhile, the minimal needs of households for financial services, e.g. money transmission, were being met by other means, such as Post Office giro, savings and co-operative banks. So the purpose of the large, for-profit, commercial banks that were founded on the Continent and in Japan, from around 1850 onwards, was to finance and foster a closely associated group of large corporate entities, in the guise of the Haus Bank in Germany and the Zaibatsu in Japan. Such banks were, ab initio, established as universal banks, providing a full range of financial services to the large entities with which they were closely associated. Perhaps in part because such services were provided through financial intermediaries, i.e. by universal banks, rather than through markets, the capital markets in such countries remained less developed, thinner and possibly more open to speculation than in the Anglo-Saxon countries. Hence the tendency in such countries has been to value the services of universal banks much more, and to be much warier of the unfettered, free workings of capital markets – perhaps especially so in France – than in the Anglo-Saxon countries. This historical division continues. The British propose, in the Vickers' Independent Commission on Banking (The Vickers' Report) (2011), a reversion to a two-tiered banking system, ring-fencing (separating) retail from investment banks, while the French, German and Japanese defend the universal role of their national champion banks. In contrast, the French and Germans are moving to introduce taxes and constraints on their financial markets, in the shape of a Financial Transaction (Tobin) Tax and bans, or limitations, on computerised High Frequency Trading, while the Anglo-Saxons seek to defend the free and flexible workings of their capital markets. Given this historical and traditional split, it is unlikely, in my view, that there will be any European, let alone world-wide Basel endorsed, agreement to proceed far, if at all, in the direction of imposing a two-tier (retail/investment) banking division. Thus countries that want to move in this direction will have to do so on their own. As discussed later, this is likely to represent a particular handicap on the investment banking, and also perhaps the extraEuropean banking, operations of banks most prone to such constraints. The City of London will become subject to the Wimbledon effect, whereby the British provide the venue, but none of the champions. Whether this will be a good, or a bad, outcome is debatable. There is widespread myth that all investment banks do is to act as a casino, taking large proprietary bets with ‘other people's money’. In fact their main role is to act as gate-keepers to financial markets for those who need to access them. In most cases they are so keen to avoid taking a directional position (i.e. a bet on the markets' movement) that the initial response to filling a client's order, and thereby unbalancing the firm's own book, involves a whole series of rebalancing transactions amongst the banks involved. Thus the fact that the volume of financial transactions is a multiple, often 8 or 9 times, of non-bank client transactions is evidence of the desire of such (investment) banks not to take up a speculative position, rather than the reverse. Again many of the documented sources of large losses have come from failed hedges, rather than from outright speculation.
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
2.2. Structural separation will be problematical Be that as it may, there are a number of other reasons why the move back towards a two-tier banking system will be problematical. The first is that there is, in practice, no clear dividing line between wholesale/investment banking and retail banking, or between acting on behalf of clients and in the bank's own interests, as the difficult attempts to apply the Volcker rules in the USA have shown, (also see Scott, 2010, 2011). Even the largest clients need to undertake some functionally simple retail-type transactions, such as making and receiving payments, and many of the smaller clients could benefit from various kinds of hedging and derivative transactions. In view of the demise of ‘caveat emptor’ and the emergence of the doctrine that (supposed) ignorance of detail gives a buyer of a financial product grounds for a lawsuit if the hedge goes wrong, it is quite possible that financial innovations and products that would be of general benefit will not now be forthcoming. Of course, ring-fencing is not (quite) as disruptive as total separation, so a large client can continue to deal with several (ring-fenced) parts of the same bank, but it will add to complexity and inefficiency, perhaps particularly for clients with cross-border businesses extending beyond the European Union. The purpose of such ring-fencing is to restrict the ‘safety net’, and the contingent liability of taxpayers, to any further bank bail-out to those parts of the wider bank that no government can (politically) allow to close, i.e. the retail deposit and lending base and the (retail) payments system. But such divisions and restrictions will not necessarily make retail banking any safer, nor will the contingent liability of taxpayers be much less. Over the last 40 years, or so, there have been four main banking/financial crises in the UK. Three of these, i.e. the 1973–75 fringe bank crisis, the 1991/92 ERM crisis, and the 2007/8 blow up, have primarily involved boom/bust cycles in property, both residential and commercial.1 Whereas most people imagine banks as lending household deposit money to private non-bank (manufacturing) businesses, in reality nowadays banks intermediate mostly between net saving and net borrowing households, (see, for example, Adair Turner, 2010). Such mortgage lending, including to construction companies and on commercial property, forms the basis of most retail bank lending. Ring-fencing will cause the separate retail banking arms to focus even more on what has, historically, been the most cyclical and dangerous element of banking, as in Northern Rock, Anglo Irish and the Spanish Cajas. Lehman Bros went under not because of positions in derivative markets, which were profitable, but because it ventured into mortgage backed securities (MBS). Whether a safer, more diversified, profitable ring-fenced investment bank would (be legally able to) allow its own, separate equity to be used in support of a failing associated retail bank will remain to be seen. In several, though not all, respects retail banking business is relatively easy to restructure and/or resolve. Problems arise in the IT field and, for one of the SIFIs, the sheer scale of the exercise. Apart from such problems, insured deposits can be rapidly moved to a ‘good’ bank, or put in some other bank, with the remaining ‘good’ (mortgage) assets. The nonperforming assets can be put in a ‘bad’ bank funded by the bonds (other than covered bonds) and equity of the failing bank. The problem that the Resolution Agency and the authorities will have to face is what to do with the non-insured, i.e. over €100,000 and foreign deposits. In practice political self-preservation will usually probably mean that priority will be given to rescuing all domestic depositors (as in Iceland), with foreign depositors and bond holders getting shorter shrift. That, of course, is contrary to pari passu covenants. If these prove to be subsequently binding (the Icelandic case continues), it is possible that this will encourage a shift to explicit, legally-imposed, (domestic) depositor preference laws, which would have potential (unintended) consequences for the form and nature of (retail) bank financing, e.g. only by 1 The fourth, the LDC (or Mexico, Argentina and Brazil MAB) crisis in 1981/82, arose from (syndicated) loans to less developed countries which went bad when Paul Volcker moved to a tight money regime in October 1979.
81
deposit and secured, or covered, bonds, unless the regulators forcibly require some holdings of bonds that can be ‘bailed in’. The main reason why resolving a retail bank is relatively simple is that it has quite few connections with the rest of the financial system. In contrast an investment bank, primarily by virtue of its role as gatekeeper to financial markets for its (large) clients, has myriad, multiple interconnections with a large number of other financial institutions, including most other such investment banks, and most other financial markets. As noted earlier, the unwillingness of an investment bank to take a speculative position means that an initial client order will often trigger a far longer paper-chase of resultant (‘hot potato’) adjustment trades. Consequently the potential economic spill-overs and externalities that could result from a liquidation of an investment bank are far greater than those attendant on the liquidation of a retail bank. If one was to apply an economic, rather than a political, calculus it should be the investment, rather than the retail, bank that should be normally recapitalised by the authorities. The provision of access to financial markets, which is what investment banks do, though primarily for large clients, is as much a utility as the provision of retail services to smaller customers. The analogy of consultants and GPs, and barristers and solicitors, comes to mind. In any case when utilities such as railroads, gas and electricity go bankrupt, the capital infrastructure is almost never ripped up and sold for scrap. Instead, the share-holders, and perhaps the bond-holders, are expropriated, and the management sacked. If the government cannot find an appropriate buyer, it runs the utility itself. No one argues that because the railway lines or electric pylons are not sold for scrap metal, that this involves ‘moral hazard’. The infrastructure of banking is mostly intangible, human capital, skills and know-how; breaking that up has no more sense than destroying gas mains when the gas company goes bust. Perhaps, there are too many gas companies, and/or that gas is losing out to electricity. There can be a case for dismantling any utility company, but the idea that the default policy for a failing investment bank should be liquidation is an extremely dangerous concept, driven on by populist comments about ‘casino’ banking. 2.3. The danger of mis-match Be that as it may, our banking systems have patently become more dangerous and fragile in recent decades. There are two main reasons for this. The first has been that, despite the efforts of the Basel Accords, I, II and now III, the ratio of equity to total assets remains far lower than socially optimal. Admati and Hellwig (2013) have written about this in their book, The Bankers' New Clothes. While it would be socially beneficial to encourage bankers to raise the equity ratio from slightly over 3%, as now required in Basel III, (to total assets) to over 15%, the problem is that, in view of the incentive of shareholders/management to focus on the Return on Equity (RoE), rather than the Return on Assets (RoA), and of the existing debt overhang, any requirement for a higher ratio will provoke deleveraging rather than equity re-build. The need instead is to require each bank to hold a higher absolute level of equity, related to its initial (risk-weighted) assets, and prevent pay-outs to shareholders and/or management until that level is attained. But that issue, of the need for a much higher equity requirement, has already been fully aired, not only by Admati and Hellwig, but also by Miles, Yang, and Marcheggiano (2013). The second reason for the enhanced fragility has been the recent faster trend growth of credit, relative to deposits, (Schularick & Taylor, 2009). The growing gap between loans and deposits, both on and off balance sheet, has been primarily filled by recourse to funding from wholesale markets. Such funding, unlike deposits, is uninsured, and coming primarily from other large financial institutions is, again in contrast to retail depositors, quite well informed. That makes it peculiarly liable to runs (Gorton, 2012 and Gorton & Metrick, 2012). Initially lender concern about the possible solvency threats to (bank) borrowers tends to be exhibited in ever shorter funding maturities, so the borrower
82
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
has to roll-over its life-line on a day-by-day basis. Then, if anything comes to a head, the Central Bank/Treasury/regulatory authorities have almost no time to work out a resolution plan. One reason why banking systems remained calm between the mid 1930s and the 1970s was that the maturity mismatch between bank liabilities (deposits) and bank assets was calmed by a mixture of deposit insurance and quasi-automatic Lender of Last Resort (LoLR) actions by the Central Bank. When bank liabilities increasingly came from uninsured wholesale funding, when those using such funds had no access to the Central Bank (i.e. in the case of the US broker/dealer investment banks), and when the Central Bank felt constrained in its ability to offset such wholesale outflows (either on grounds of ‘moral hazard’, or because the banks in trouble might, indeed, be insolvent), the mismatch could lead to a (contagious) banking crisis. As in the 1930s, something has to be done about the source of fragility in mismatch. But what? At one extreme there are those who might argue that, so long as banks start with enough equity to avoid losses to other creditors, then the Central Bank ought always to offset net drains of liquidity, in part offsetting wholesale markets that become dysfunctional by becoming the market-maker of last resort. And indeed that is what Central Banks have, to a large extent, now done. The basic problem is always solvency. So long as that can be guaranteed, then each Central Bank should be active and innovative enough to prevent any liquidity problems from arising. One problem with this position is that equity ratios remain far too low, so that liquidity problems usually remain symptoms of underlying concerns about solvency. At the other extreme are those who would deal with the problem of mismatch by banning it by regulation. This is the essence of Kumhof's ‘Chicago Plan’ (Benes & Kumhof, 2012), involving a combination of ‘narrow’ banks for transactions purposes and financial trusts wherein loans are backed either by equity or long-dated debt. Kotlikoff's proposals are somewhat similar (see Kotlikoff, 2011). A problem with proposals of this kind is that they run counter to the revealed preferences of savers for financial products that are both liquid and safe, and of borrowers for loans that do not have to be repaid until some known future distant date. It is one of the main functions of financial institutions to intermediate between the desires of savers and borrowers, i.e. to create financial mismatch. To make such a function illegal seems draconian. But the greater the mismatch, the worse the fragility, for any given equity ratio. What should one then do? One obvious point is that the required net stable financial ratio of a financial institution should be inversely related to its equity ratio. The higher the equity ratio the stronger will be its solvency, the less the likelihood will be for runs, and the greater the confidence with which the Central Bank can extend LoLR. Of course, the equity ratio will need to be measured using market, rather than accounting values. Next, a liquidity ratio that has to be maintained at all times is an oxymoron; if it cannot be used, it is not liquid by definition. What is required instead is an increasing ladder of sanctions as the mismatch worsens. Probably the best form of sanction would be pecuniary, (Perotti & Suarez, 2011), with the costs of additional mismatch more than offsetting the usual rate spread between shorter and longerdated wholesale funding. Such pecuniary charges could be increased when there was a felt need to restrain excessive loan expansion (relative to the growth of deposits and/or GDP), and lowered during periods of recession, either across the board, or on a bank by bank basis, providing a further contra-cyclical instrument. Such charges could also go to finance an ex ante fund for meeting costs of resolution. The suggestion here that controls over liquidity and mismatch should be: • Related to the bank's (market value) equity ratio; • Based on a pecuniary ladder of sanctions; and • Capable of being cyclically adjusted.
This is, however, a long way from present Basel proposals either for the Liquidity Coverage Ratio (LCR) or the Net Stable Financial Ratio (NSFR). As presently outlined, they are unrelated to equity ratios, absolute rather than having a ladder of sanctions, and not cyclically adjustable. A particular weakness of the pre-2008 US investment banks was that they were primarily financed by non-deposit wholesale funding, and were left outside the official safety net. The effect of ring-fencing investment banks, as proposed by Vickers and Liikanen (2012), is to force them to be funded by non-deposit wholesale funds, and consciously to exclude them from the official safety net. Such extra fragility can be offset by imposing a tougher (mismatch) requirement on them via an NSFR (and LCR). But that would make their funding considerably more expensive, notably in comparison to continuing universal banks which could base their investment banking activities on a retail deposit base. Either ring-fencing will make the separated investment banks riskier, or less competitive, (or both). Whatever balance is struck, ring-fenced investment banks are likely to be heading for extinction. 2.4. Further thoughts Our banking systems did become more fragile in recent decades. This was primarily because we did not prevent bankers from keeping equity ratios too low in self-interested pursuit of RoE. But it was also because mismatch dangers were allowed to increase, as the expansion of loans relative to deposits encouraged greater recourse to (shorter-dated, uninsured) wholesale funding. These issues need to be tackled, far more aggressively than hitherto, perhaps in conjunction with measures to change the characteristics of bank management remuneration (and/or liability to loss) to make them dependent on RoA, rather than on RoE, possibly by making bonuses payable in bail-inable bonds, rather than equity. In contrast, the thesis here is that the structural changes so widely advocated, e.g. Vickers, Liikanen and the more extreme proposals of Kumhof and Kotlikoff, will be counter-productive, possibly making our financial systems even riskier and probably less efficient than at present. I have also argued that neither utilities nor banks should be scrapped. Decisions are made by people, not by abstract institutions. If the decisionmaking managers are sacked and their accumulated bonuses are decimated by the event of failure, then that should deal with the moral hazard, whether of a utility or of a bank. If that is not considered sufficiently draconian, try reverting to unlimited liability for a subset of senior managers! But if (universal/investment) banks are not to be liquidated, then might not the burden of recapitalisation, of meeting the loss, be too big for the government to meet? Such banks, as in Iceland, Ireland and Cyprus, may be too big for the government to save. Is there not then a case for limiting the size of a bank headquartered in Country X to some fraction of that country's GDP? There is also the argument that some banks may be too big, complex and geographically dispersed to be effectively managed, (as with non-financial conglomerates a few decades ago). Perhaps, but this would mean that only large countries, like the USA or China, could have large banks, and that small countries could only support small banks. If a bank headquartered in a smallish country found itself able, and willing, to expand, it might only be able to do so if it was to transfer its headquarters and country of residence to a larger country. Would that be acceptable to its prospective new home country, or to its previous home country? Why should all the large international cross-border banks reside in the USA, China or Japan? Would the establishment of a Banking Union in the Eurozone, with Eurozone-wide resolution financing mechanisms, be sufficient to allow Eurozone banks to become much larger than before? Perhaps. One reason why the doom-loop between failing banks and governments in the Eurozone has proved so toxic is that the trigger for state intervention to take over from current management has been too late, by which time severe losses have already been incurred. This is partly because the wrong metric continues to be used, accounting capital rather than the market value of capital, and partly because the authorities
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
are reluctant to abrogate the ownership rights of shareholders in a capitalist society. While the latter is understandable, it is not clear why it is any more consistent with free markets to impose structural controls over the allowable operations of a bank than to require official intervention when the market value of a bank's equity falls below X% of its total assets; moreover warrants could be given to such shareholders to allow them to share in any subsequent strong revival. To conclude, what I would advocate for my optimal banking structure would be: • A much higher equity ratio, supported by a ladder of sanctions, with a much higher minimum intervention point, measured via market, rather than accounting, values; • A liquidity (mis-match) ratio, inversely related to each bank's equity ratio, supported by a ladder of pecuniary sanctions, capable of contra-cyclical adjustment; • An acceptance that failure should imply the sacking of management and expropriation of shareholders, but not necessarily the liquidation of any bank. This could be reinforced by required changes to remuneration practices so that failure would be more painful to management. Intervention should come earlier, so as to restrict losses. Acknowledgements This policy paper consists of two keynote addresses, delivered by the author on 6 March 2013 at the 2nd European Conference on Banking and the Economy (ECOBATE 2013, ‘The Optimal Financial Structure’), in Winchester Guildhall, and on 25 April 2013 at the International Conference on the Global Financial Crisis (‘The Parlous State of Macroeconomics’), at the University of Southampton. References Admati, A., & Hellwig, M. (2013). The bankers' new clothes: What's wrong with banking and what to do about it. Princeton, N.J.: Princeton University Press.
83
Benes, J., & Kumhof, M. (2012). The Chicago plan revisited. International Monetary Fund working paper, WP/12/202 (August). Gorton, G. (2012). Some reflections on the recent financial crisis. National Bureau of Economic Research working paper no. 18397 (September). Gorton, G., & Metrick, A. (2012). Securitized banking and the run on repo. Journal of Financial Economics, 104, 425–451. Independent commission on banking (The Vickers report) (2011). Independent commission on banking: Final report. Written evidence, House of Commons Treasury Committee, 12 September. Kotlikoff, L. (2011). Jimmy Stewart is dead: Ending the world's ongoing financial plague with limited purpose banking. John Wiley & Sons. Liikanen, E., (Chairman), (2012), ‘High-level Expert Group on reforming the structure of the EU banking sector’, Final Report, European Commission, 2 October, see http:// ec.europa.eu/internal_market/bank/docs/high-level_expert_group/liikanen-report/ final_report_en.pdf Miles, D., Yang, J., & Marcheggiano, G. (2013). Optimal bank capital. The Economic Journal, 123, 1–37. Perotti, E., & Suarez, J. (2011). A Pigovian approach to liquidity regulation. International Journal of Central Banking (December). Ryan-Collins, J., Greenham, T., Werner, R., & Jackson, A. (2012). Where does money come from?: A guide to the UK monetary & banking system (2nd ed.). New Economics Foundation. Schularick, M., & Taylor, A. M. (2009). Credit booms gone bust: Monetary policy, leverage cycles and financial crises, 1870–2008. National Bureau of Economic Research working paper no. 15512 (November). Scott, H. A. (2010). Prepared written testimony of Hal S. Scott, Nomura professor of international financial systems at Harvard Law School and director of the committee on capital markets regulation, before the committee on banking. Housing and Urban Affairs, United States Senate (February 4. See http://www.banking.senate.gov/public/index.cfm? FuseAction=Files.View&FileStore_id=c372f56f-819f-4d93-bb5a-6c1802aeb18a). Scott, H. A. (2011). Written testimony of Hal S. Scott, director of the Committee on Capital Markets Regulation and Nomura professor and director of the program on international financial systems at Harvard Law School, before the Committee on Financial Services. United States House of Representatives (June 16. See http://www.capmktsreg.org/ pdfs/2011.06.16_House_Testimony.pdf). Turner, A. (2010). What do banks do? What should they do? Chairman of the Financial Services Authority speech slides. London: Cass Business School (17 March, see http:// www.fsa.gov.uk/pubs/speeches/at_cass_slides_170310.pdf). Werner, R. (2005). New paradigm in macroeconomics. Basingstoke: Palgrave Macmillan.
Contents lists available at ScienceDirect
International Review of Financial Analysis
The parlous state of macroeconomics and the optimal financial structure C.A.E. Goodhart Financial Markets Group, London School of Economics, United Kingdom
a r t i c l e
i n f o
Available online 29 October 2014 JEL classifications: E40 E44 E50 E51 G20 G21 G28 G33
a b s t r a c t Macroeconomics remains in a parlous condition, largely because it has assumed away all financial frictions. Ultimately these latter depend on the possibility that borrowers might default on their repayments. Without default, there is no real role for most financial intermediations, collateral, liquidity or money. Yet default (especially of banks, the key ingredient of crises) is rarely modelled. In order to make banks safer, in the aftermath of the Great Financial Crisis, there are various proposals to restructure our banking systems, for example to dismantle universal banks into separate retail and investment parts. This partly derives from a mis-reading of the causes of the GFC, which was largely driven by an interaction between a housing boom and a bank credit expansion surfeit, thereby exaggerating leverage, mis-match and non-core bank finance. The need is for regulatory improvements that address these weaknesses. © 2014 Elsevier Inc. All rights reserved.
Keywords: Default Liquidity Money Stock Great Financial Crisis Bank Restructuring Liquidity (Mismatch) Ratio Ladder of Sanctions
1. Part I: the parlous state of macroeconomics 1.1. The recent history of macroeconomics One reason why I found the study of macroeconomics enjoyable is that it has been in such a mess. Consequently I thought (mistakenly) that I might be able to make a difference. As the old joke has it, ‘The questions in the Economics Tripos remain the same, but the required answers change.’ As Dennis Robertson once said, “Macroeconomics is rather like coursing for hares. If you stand in one place you will find that the hares will double back to you and you will see them as they return to the starting point.” When I was a callow undergraduate at Trinity College, Cambridge, in 1957, some 55 years ago, I had the opportunity to meet Dennis Robertson. At that time I was quite excited by studying those debates about macroeconomics which had raged in the 1930s, e.g. liquidity preference versus loanable funds. So I asked Dennis whether it had been exciting for him to have been a participant. I then saw the look of pain that went over his face, and I realised that I had put my foot in it. Anyhow, I can now, with reasonable confidence, proclaim that, having laboured in the field of money-macro for the last 50 years or more, I shall shortly leave it, with the subject probably being in a worse state than when I initially found it. Let me try to explain why E-mail address: [email protected].
http://dx.doi.org/10.1016/j.irfa.2014.10.014 1057-5219/© 2014 Elsevier Inc. All rights reserved.
and how this occurred. Let me start with a brief review of the main path of macroeconomics as a subject since the end of the 1950s and early 1960s, when I first found it. • We start with computerisation at the end of the 1950s and beginning of the 1960s. As an undergraduate in Cambridge, there was only one computer in the whole university, housed in a huge room, with no likelihood for any undergraduate, nor indeed a research student, being able to access it. This was rapidly changing, and when I went to Harvard, I found my first IBM computers, with punch cards as inputs, and a printer in the form of a typewriter sitting on top of the computer, with the keys going up and down without any human involvement, as the output got printed out. • Then we got the early Keynesian models based on the income and expenditure accounts, with consumption functions, investment functions, import and export functions, etc. All of these took the shape of reduced form fitted equations, trying to provide the best econometric fit to the dependent variable in each case, on the basis of whatever set of variables were supposed by theory, or empirical investigation, to give the best empirical explanation of the time series of the variable in question. • This, of course, led on to the Lucas Critique, that these reduced form equation were inconsistent, incoherent, and unrelated to any apparent pattern of micro-behaviour. This critique was, of course, correct.
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
• This led on to trying to rewrite the equations in the form of constrained utility maximisations (Euler equations) for every agent. But, if one was to have as many agents as there are in the economy, this would be impossibly complicated and virtually impossible to compute or to relate to aggregate empirical data. The result of this was to work on the basis of an assumption that everyone in each of the sectors involved was to be exactly the same, i.e. the Representative Agent assumption. • But, if there was to be a Representative Agent assumption, then either, should default occur, everyone would default and the sector would cease to exist, or no one would default ever. In effect, the latter assumption was made, though frequently this was implicit rather than explicit. But, with no default, there would be no risk premia, and no need for banks or money. Instead, there would just be one (official policy) interest rate, and expectations of its future path. • But, with no default, no banks and no money, and no risk premia, the resulting (DSGE) models were hopeless in a financial crisis. In effect, macroeconomics became totally divorced from finance, with finance depending critically on expectations of probability of default (PD) and loss-given default (LGD). In practice, in my view, the real advances in economics at the macro level in recent decades have occurred in the field of finance, rather than in macroeconomics. By abstracting from default, banking and money, macroeconomics has gone down a blind alley; the concept that all relevant action takes place in the real economy, with finance being nothing more than a veil, has not been helpful. 1.2. Default There is a need to model default explicitly, and that would bring with it the need to model liquidity, banks and other financial intermediaries, along with markets for financial assets. It is true that certain DSGE models now have financial add-ons, such as the Bernanke–Gertler and the Kiyotaki–Moore models, but these have default implicit in their models, even if not explicit. For example, some of these are based on the fact that borrowers need collateral in order to borrow, and the value of collateral varies with the state of the economy. But one only needs collateral in order to protect against default; so, the basis of such models effectively depends ultimately on the possibility of default. Moreover, default does occur; indeed, the default of banks has been at the heart of most of the worst downturns and depressions of the modern era; examples includes Lehman Bros in 2008, Credit Anstalt in 1931, Knickerbocker Trust in 1907, and several others that could be added. Default, however, is hard to model since it is not continuous. The response of Martin Shubik, and several colleagues, has been to reverse the problem and to focus attention on the repayment rate, in the event of default, rather than on the fact of default itself. When one is thinking about a sector, such as the productive sector, one can think of the percentage of non-performing loans or the percentage of write-offs that the bank has to apply. Even in the case of the failure of a single big bank, it is almost never the case that there is no repayment in the case of bankruptcy and liquidation. Even with Lehman Bros, the creditors (eventually) got a sizeable repayment. The repayment rate can vary continuously between 0 and 100% and both the actual, and expected, repayment rates are usually, almost always, somewhere in the interior. Thus, one can use repayment rates as a variable to include within the confines of the standard type of macro-model that is applied in our subject. 1.3. The determination of the money stock Even worse than the analytical treatment of the macro-economy is the recent, and indeed present state of the theory of the supply of money, and study and analysis of the operation of banks; and all this despite Richard Werner's good work (see, for instance, Werner, 2005) and recent book, with colleagues, entitled Where Does Money Come
79
From?: A Guide to the UK Monetary & Banking System (Ryan-Collins, Greenham, Werner, & Jackson, 2012). For example, the analysis of the supply-side determination of the money stock is still based on a money multiplier analysis, which is a purely mechanical relationship derived from identities. This relates the broader money stock, M to the monetary base, H, in a mechanical fashion dependent on two ratios, the currency deposit, C/D, and the reserve deposit, R/D, ratios. But, so long as the central bank sets the policy interest rate, the broad money stock is endogenous, since maintenance of market interest rates close to the policy interest rate requires the central bank to give the banks enough high-powered money to enable them to maintain their desired reserve deposit (R/D) ratio such as would be consistent with the policy-determined interest rate. If the central bank provided an insufficient (excessive) reserve base for the banks, they would push (short-term) market rates above (below) the policy rate, thereby forcing the central bank to inject (withdraw) reserves into the system to make their policy rate effective. But once the zero lower bound to interest rates kicks in, so that the policy interest rate is set effectively as close to zero as can be achieved, then the central bank can, and has with QE, attempted to vary H, the high-powered monetary base, exogenously, expanding H very sharply. So, policy has now changed, with the central bank setting the volume of base money, independently of the level of policy interest rates, which are stuck at zero. With policy changed, both the Lucas Critique and Goodhart's Law, then have kicked in. The increase in H, base money, in most of the central banks, Fed, BoE, BoJ and ECB, has increased by a factor of 3 or 4. Meanwhile, however, the overall money stock has grown, if at all, very grudgingly; and bank lending to the private sector has in most cases been stagnant, or even negative. The money multiplier has collapsed! If the money multiplier had remained stable, so that broad money had increased at roughly the same rate as the reserve base of the banking system, then the crisis would have been over, and, indeed, it might have been replaced by an upsurge in inflation. The bank lending channel has completely failed. Somewhat surprisingly, central banks have been too embarrassed to discuss this failure. What happened? Why did those who take banking decisions not want to use the increase in their reserves to expand their asset portfolios? It is bank CEOs who take the decisions on bank portfolio management. Why did nobody ask what would be such CEO micro-behavioural, constrained utility maximisation under such circumstances, rather than using the mechanical multiplier analysis? The CEOs' answer to their shareholders, and are concerned about the potential for takeovers in the equity market. Moreover, they themselves are almost always large shareholders; this is because they receive bonus payments quite largely in the form of shares in their own bank. So, naturally, they focus on the return on equity (RoE). Also, equity holders, including banks' CEOs, have limited liability. If there should be some disaster, they can depart, crying all the way to retirement with their amassed pot of gold. Equity involves an option call on the assets of a bank. The structure of pay-offs means that equity holders are necessarily riskloving; Northern Rock was the darling of the London Stock Exchange only a few months before it collapsed. Return on equity is generally maximised by expanding leverage, having a very small proportion of equity in relation to a much larger proportion of fixed debt. Consequently the equity holders get the benefit of all the upturn above the required payment to the fixed interest holders, while their downside is limited. So, equity will be minimised, and games will be played with risk-weighting, with banks subject to a riskweighted capital adequacy requirement (CAR) holding a massive portfolio of assets which have a supposedly minimal risk weight. In contrast, banks that are subject to an overall leverage ratio, will try to maximise RoE by holding riskier assets in their constrained volume of debt. Under these circumstances, it is only too likely that tail risk will occur. As a result, people will argue that equity ratios were, have been, and remain too low. And indeed this is so. Admati and Hellwig in their book on The Bankers' New Clothes, David Miles and colleagues in the Economic
80
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
Journal, and the Independent (Vickers) Commission on Banking, have all argued, correctly, that the equity ratio has been far too low. So, the cry has been to raise the equity ratio. But just requiring a higher equity ratio is actually the wrong way to get to a safer equilibrium situation. So long as the bankers continue to focus on RoE, they will try to get there by deleveraging, which will just worsen the downturn. The monetary authorities are, in each country, trying to protect financial intermediation in their own country by seeking and, almost blackmailing, their domestic banks into expanding, or at least not reducing, their lending within their own country; but at the same time arguing that they must enhance their equity ratio by cutting their lending abroad, and by selling assets situated in foreign countries. What should we have done instead? In this respect the response in the USA has been better than that in Europe; in the US banks were put under considerable pressure to raise the absolute volume of equity, rather than the equity ratio. But let us go back to the money multiplier; why are banks content with holding such a huge increase in their reserve/deposit ratio? The answer to this is that central bankers have, mistakenly, been willing to pay a low rate of interest on such safe assets; with zero risk weight, and requiring no capital, the holding of interest-bearing safe assets, which puts no pressure on either capital requirements, and obviously none on liquidity requirements, has seemed to CEOs to be highly desirable. So what should happen? One measure that the authorities could do, would be to reduce the marginal interest rate on excess reserves held at the central bank. Another would be to force the recapitalisation of banks with insufficient equity, if necessary through the use of taxpayer funds, until the capital constraint on uses of bank reserves to add to private sector lending became less unattractive. 1.4. Where should macroeconomics go from here? I would suggest that there are two main lessons that need to be applied, if macroeconomics is going to recover. • The first is that default and bankruptcy matter for macroeconomic analysis and for understanding the relationship between macro and finance. In particular, we need to end the separation between the analysis of finance and the analysis of macroeconomics. • Let us also do a Bob Lucas type micro-behavioural foundation model for the decision-making process of bank CEOs. In the process, we need to put the money multiplier where it belongs, in the deletebutton trash-file. 2. Part II: the optimal financial structure Having made the case that economics must consider banking and finance explicitly, and reflect the institutional reality of our present banking system, let me now turn to the question of how our financial structure should be designed or reformed in order to allow it to fulfil its intended function. 2.1. History is important Banking developed rather differently in Anglo-Saxon countries than on the European Continent and in Japan. In Anglo-Saxon countries, notably the UK and the USA, banks started up before the emergence of large scale industry. Such banks were usually small, unlimited liability partnerships in the UK, and financially fragile. An important determinant of successful continuing business was to avoid getting too involved in concentrated lending to (associated) private firms; lending was to be at arm's length and diversified. Where, even then, there was a need for large-scale finance, e.g. for canals and then railroads, this could and should be provided by, relatively efficient, capital markets, both for bonds and equities. But the entrepreneurs of such large firms and governments, at various levels, did not generally have the necessary information and skills to access
financial (and foreign) markets, so there sprung up another tier of financial intermediaries who used their market skills and information to provide such large entities with access to capital markets. These were the merchant banks (or Accepting Houses) in London, or the broker/dealers (at a somewhat later date, especially after Glass– Steagall) in the USA. Thus in these countries banks were, originally, primarily retail in character, serving the (well-to-do in the) local community, with a separate tier of investment banks acting as keepers of the gateway to efficient capital markets for those large institutions needing access to such markets. The history and experience on the continent of Europe and in Japan were different. Industrialisation occurred later, by which time industrial economies of scale had become more prevalent, e.g. in iron and steel, chemicals, and utilities, but the capital markets there were less efficient, and financial markets were less trusted in these countries. Meanwhile, the minimal needs of households for financial services, e.g. money transmission, were being met by other means, such as Post Office giro, savings and co-operative banks. So the purpose of the large, for-profit, commercial banks that were founded on the Continent and in Japan, from around 1850 onwards, was to finance and foster a closely associated group of large corporate entities, in the guise of the Haus Bank in Germany and the Zaibatsu in Japan. Such banks were, ab initio, established as universal banks, providing a full range of financial services to the large entities with which they were closely associated. Perhaps in part because such services were provided through financial intermediaries, i.e. by universal banks, rather than through markets, the capital markets in such countries remained less developed, thinner and possibly more open to speculation than in the Anglo-Saxon countries. Hence the tendency in such countries has been to value the services of universal banks much more, and to be much warier of the unfettered, free workings of capital markets – perhaps especially so in France – than in the Anglo-Saxon countries. This historical division continues. The British propose, in the Vickers' Independent Commission on Banking (The Vickers' Report) (2011), a reversion to a two-tiered banking system, ring-fencing (separating) retail from investment banks, while the French, German and Japanese defend the universal role of their national champion banks. In contrast, the French and Germans are moving to introduce taxes and constraints on their financial markets, in the shape of a Financial Transaction (Tobin) Tax and bans, or limitations, on computerised High Frequency Trading, while the Anglo-Saxons seek to defend the free and flexible workings of their capital markets. Given this historical and traditional split, it is unlikely, in my view, that there will be any European, let alone world-wide Basel endorsed, agreement to proceed far, if at all, in the direction of imposing a two-tier (retail/investment) banking division. Thus countries that want to move in this direction will have to do so on their own. As discussed later, this is likely to represent a particular handicap on the investment banking, and also perhaps the extraEuropean banking, operations of banks most prone to such constraints. The City of London will become subject to the Wimbledon effect, whereby the British provide the venue, but none of the champions. Whether this will be a good, or a bad, outcome is debatable. There is widespread myth that all investment banks do is to act as a casino, taking large proprietary bets with ‘other people's money’. In fact their main role is to act as gate-keepers to financial markets for those who need to access them. In most cases they are so keen to avoid taking a directional position (i.e. a bet on the markets' movement) that the initial response to filling a client's order, and thereby unbalancing the firm's own book, involves a whole series of rebalancing transactions amongst the banks involved. Thus the fact that the volume of financial transactions is a multiple, often 8 or 9 times, of non-bank client transactions is evidence of the desire of such (investment) banks not to take up a speculative position, rather than the reverse. Again many of the documented sources of large losses have come from failed hedges, rather than from outright speculation.
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
2.2. Structural separation will be problematical Be that as it may, there are a number of other reasons why the move back towards a two-tier banking system will be problematical. The first is that there is, in practice, no clear dividing line between wholesale/investment banking and retail banking, or between acting on behalf of clients and in the bank's own interests, as the difficult attempts to apply the Volcker rules in the USA have shown, (also see Scott, 2010, 2011). Even the largest clients need to undertake some functionally simple retail-type transactions, such as making and receiving payments, and many of the smaller clients could benefit from various kinds of hedging and derivative transactions. In view of the demise of ‘caveat emptor’ and the emergence of the doctrine that (supposed) ignorance of detail gives a buyer of a financial product grounds for a lawsuit if the hedge goes wrong, it is quite possible that financial innovations and products that would be of general benefit will not now be forthcoming. Of course, ring-fencing is not (quite) as disruptive as total separation, so a large client can continue to deal with several (ring-fenced) parts of the same bank, but it will add to complexity and inefficiency, perhaps particularly for clients with cross-border businesses extending beyond the European Union. The purpose of such ring-fencing is to restrict the ‘safety net’, and the contingent liability of taxpayers, to any further bank bail-out to those parts of the wider bank that no government can (politically) allow to close, i.e. the retail deposit and lending base and the (retail) payments system. But such divisions and restrictions will not necessarily make retail banking any safer, nor will the contingent liability of taxpayers be much less. Over the last 40 years, or so, there have been four main banking/financial crises in the UK. Three of these, i.e. the 1973–75 fringe bank crisis, the 1991/92 ERM crisis, and the 2007/8 blow up, have primarily involved boom/bust cycles in property, both residential and commercial.1 Whereas most people imagine banks as lending household deposit money to private non-bank (manufacturing) businesses, in reality nowadays banks intermediate mostly between net saving and net borrowing households, (see, for example, Adair Turner, 2010). Such mortgage lending, including to construction companies and on commercial property, forms the basis of most retail bank lending. Ring-fencing will cause the separate retail banking arms to focus even more on what has, historically, been the most cyclical and dangerous element of banking, as in Northern Rock, Anglo Irish and the Spanish Cajas. Lehman Bros went under not because of positions in derivative markets, which were profitable, but because it ventured into mortgage backed securities (MBS). Whether a safer, more diversified, profitable ring-fenced investment bank would (be legally able to) allow its own, separate equity to be used in support of a failing associated retail bank will remain to be seen. In several, though not all, respects retail banking business is relatively easy to restructure and/or resolve. Problems arise in the IT field and, for one of the SIFIs, the sheer scale of the exercise. Apart from such problems, insured deposits can be rapidly moved to a ‘good’ bank, or put in some other bank, with the remaining ‘good’ (mortgage) assets. The nonperforming assets can be put in a ‘bad’ bank funded by the bonds (other than covered bonds) and equity of the failing bank. The problem that the Resolution Agency and the authorities will have to face is what to do with the non-insured, i.e. over €100,000 and foreign deposits. In practice political self-preservation will usually probably mean that priority will be given to rescuing all domestic depositors (as in Iceland), with foreign depositors and bond holders getting shorter shrift. That, of course, is contrary to pari passu covenants. If these prove to be subsequently binding (the Icelandic case continues), it is possible that this will encourage a shift to explicit, legally-imposed, (domestic) depositor preference laws, which would have potential (unintended) consequences for the form and nature of (retail) bank financing, e.g. only by 1 The fourth, the LDC (or Mexico, Argentina and Brazil MAB) crisis in 1981/82, arose from (syndicated) loans to less developed countries which went bad when Paul Volcker moved to a tight money regime in October 1979.
81
deposit and secured, or covered, bonds, unless the regulators forcibly require some holdings of bonds that can be ‘bailed in’. The main reason why resolving a retail bank is relatively simple is that it has quite few connections with the rest of the financial system. In contrast an investment bank, primarily by virtue of its role as gatekeeper to financial markets for its (large) clients, has myriad, multiple interconnections with a large number of other financial institutions, including most other such investment banks, and most other financial markets. As noted earlier, the unwillingness of an investment bank to take a speculative position means that an initial client order will often trigger a far longer paper-chase of resultant (‘hot potato’) adjustment trades. Consequently the potential economic spill-overs and externalities that could result from a liquidation of an investment bank are far greater than those attendant on the liquidation of a retail bank. If one was to apply an economic, rather than a political, calculus it should be the investment, rather than the retail, bank that should be normally recapitalised by the authorities. The provision of access to financial markets, which is what investment banks do, though primarily for large clients, is as much a utility as the provision of retail services to smaller customers. The analogy of consultants and GPs, and barristers and solicitors, comes to mind. In any case when utilities such as railroads, gas and electricity go bankrupt, the capital infrastructure is almost never ripped up and sold for scrap. Instead, the share-holders, and perhaps the bond-holders, are expropriated, and the management sacked. If the government cannot find an appropriate buyer, it runs the utility itself. No one argues that because the railway lines or electric pylons are not sold for scrap metal, that this involves ‘moral hazard’. The infrastructure of banking is mostly intangible, human capital, skills and know-how; breaking that up has no more sense than destroying gas mains when the gas company goes bust. Perhaps, there are too many gas companies, and/or that gas is losing out to electricity. There can be a case for dismantling any utility company, but the idea that the default policy for a failing investment bank should be liquidation is an extremely dangerous concept, driven on by populist comments about ‘casino’ banking. 2.3. The danger of mis-match Be that as it may, our banking systems have patently become more dangerous and fragile in recent decades. There are two main reasons for this. The first has been that, despite the efforts of the Basel Accords, I, II and now III, the ratio of equity to total assets remains far lower than socially optimal. Admati and Hellwig (2013) have written about this in their book, The Bankers' New Clothes. While it would be socially beneficial to encourage bankers to raise the equity ratio from slightly over 3%, as now required in Basel III, (to total assets) to over 15%, the problem is that, in view of the incentive of shareholders/management to focus on the Return on Equity (RoE), rather than the Return on Assets (RoA), and of the existing debt overhang, any requirement for a higher ratio will provoke deleveraging rather than equity re-build. The need instead is to require each bank to hold a higher absolute level of equity, related to its initial (risk-weighted) assets, and prevent pay-outs to shareholders and/or management until that level is attained. But that issue, of the need for a much higher equity requirement, has already been fully aired, not only by Admati and Hellwig, but also by Miles, Yang, and Marcheggiano (2013). The second reason for the enhanced fragility has been the recent faster trend growth of credit, relative to deposits, (Schularick & Taylor, 2009). The growing gap between loans and deposits, both on and off balance sheet, has been primarily filled by recourse to funding from wholesale markets. Such funding, unlike deposits, is uninsured, and coming primarily from other large financial institutions is, again in contrast to retail depositors, quite well informed. That makes it peculiarly liable to runs (Gorton, 2012 and Gorton & Metrick, 2012). Initially lender concern about the possible solvency threats to (bank) borrowers tends to be exhibited in ever shorter funding maturities, so the borrower
82
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
has to roll-over its life-line on a day-by-day basis. Then, if anything comes to a head, the Central Bank/Treasury/regulatory authorities have almost no time to work out a resolution plan. One reason why banking systems remained calm between the mid 1930s and the 1970s was that the maturity mismatch between bank liabilities (deposits) and bank assets was calmed by a mixture of deposit insurance and quasi-automatic Lender of Last Resort (LoLR) actions by the Central Bank. When bank liabilities increasingly came from uninsured wholesale funding, when those using such funds had no access to the Central Bank (i.e. in the case of the US broker/dealer investment banks), and when the Central Bank felt constrained in its ability to offset such wholesale outflows (either on grounds of ‘moral hazard’, or because the banks in trouble might, indeed, be insolvent), the mismatch could lead to a (contagious) banking crisis. As in the 1930s, something has to be done about the source of fragility in mismatch. But what? At one extreme there are those who might argue that, so long as banks start with enough equity to avoid losses to other creditors, then the Central Bank ought always to offset net drains of liquidity, in part offsetting wholesale markets that become dysfunctional by becoming the market-maker of last resort. And indeed that is what Central Banks have, to a large extent, now done. The basic problem is always solvency. So long as that can be guaranteed, then each Central Bank should be active and innovative enough to prevent any liquidity problems from arising. One problem with this position is that equity ratios remain far too low, so that liquidity problems usually remain symptoms of underlying concerns about solvency. At the other extreme are those who would deal with the problem of mismatch by banning it by regulation. This is the essence of Kumhof's ‘Chicago Plan’ (Benes & Kumhof, 2012), involving a combination of ‘narrow’ banks for transactions purposes and financial trusts wherein loans are backed either by equity or long-dated debt. Kotlikoff's proposals are somewhat similar (see Kotlikoff, 2011). A problem with proposals of this kind is that they run counter to the revealed preferences of savers for financial products that are both liquid and safe, and of borrowers for loans that do not have to be repaid until some known future distant date. It is one of the main functions of financial institutions to intermediate between the desires of savers and borrowers, i.e. to create financial mismatch. To make such a function illegal seems draconian. But the greater the mismatch, the worse the fragility, for any given equity ratio. What should one then do? One obvious point is that the required net stable financial ratio of a financial institution should be inversely related to its equity ratio. The higher the equity ratio the stronger will be its solvency, the less the likelihood will be for runs, and the greater the confidence with which the Central Bank can extend LoLR. Of course, the equity ratio will need to be measured using market, rather than accounting values. Next, a liquidity ratio that has to be maintained at all times is an oxymoron; if it cannot be used, it is not liquid by definition. What is required instead is an increasing ladder of sanctions as the mismatch worsens. Probably the best form of sanction would be pecuniary, (Perotti & Suarez, 2011), with the costs of additional mismatch more than offsetting the usual rate spread between shorter and longerdated wholesale funding. Such pecuniary charges could be increased when there was a felt need to restrain excessive loan expansion (relative to the growth of deposits and/or GDP), and lowered during periods of recession, either across the board, or on a bank by bank basis, providing a further contra-cyclical instrument. Such charges could also go to finance an ex ante fund for meeting costs of resolution. The suggestion here that controls over liquidity and mismatch should be: • Related to the bank's (market value) equity ratio; • Based on a pecuniary ladder of sanctions; and • Capable of being cyclically adjusted.
This is, however, a long way from present Basel proposals either for the Liquidity Coverage Ratio (LCR) or the Net Stable Financial Ratio (NSFR). As presently outlined, they are unrelated to equity ratios, absolute rather than having a ladder of sanctions, and not cyclically adjustable. A particular weakness of the pre-2008 US investment banks was that they were primarily financed by non-deposit wholesale funding, and were left outside the official safety net. The effect of ring-fencing investment banks, as proposed by Vickers and Liikanen (2012), is to force them to be funded by non-deposit wholesale funds, and consciously to exclude them from the official safety net. Such extra fragility can be offset by imposing a tougher (mismatch) requirement on them via an NSFR (and LCR). But that would make their funding considerably more expensive, notably in comparison to continuing universal banks which could base their investment banking activities on a retail deposit base. Either ring-fencing will make the separated investment banks riskier, or less competitive, (or both). Whatever balance is struck, ring-fenced investment banks are likely to be heading for extinction. 2.4. Further thoughts Our banking systems did become more fragile in recent decades. This was primarily because we did not prevent bankers from keeping equity ratios too low in self-interested pursuit of RoE. But it was also because mismatch dangers were allowed to increase, as the expansion of loans relative to deposits encouraged greater recourse to (shorter-dated, uninsured) wholesale funding. These issues need to be tackled, far more aggressively than hitherto, perhaps in conjunction with measures to change the characteristics of bank management remuneration (and/or liability to loss) to make them dependent on RoA, rather than on RoE, possibly by making bonuses payable in bail-inable bonds, rather than equity. In contrast, the thesis here is that the structural changes so widely advocated, e.g. Vickers, Liikanen and the more extreme proposals of Kumhof and Kotlikoff, will be counter-productive, possibly making our financial systems even riskier and probably less efficient than at present. I have also argued that neither utilities nor banks should be scrapped. Decisions are made by people, not by abstract institutions. If the decisionmaking managers are sacked and their accumulated bonuses are decimated by the event of failure, then that should deal with the moral hazard, whether of a utility or of a bank. If that is not considered sufficiently draconian, try reverting to unlimited liability for a subset of senior managers! But if (universal/investment) banks are not to be liquidated, then might not the burden of recapitalisation, of meeting the loss, be too big for the government to meet? Such banks, as in Iceland, Ireland and Cyprus, may be too big for the government to save. Is there not then a case for limiting the size of a bank headquartered in Country X to some fraction of that country's GDP? There is also the argument that some banks may be too big, complex and geographically dispersed to be effectively managed, (as with non-financial conglomerates a few decades ago). Perhaps, but this would mean that only large countries, like the USA or China, could have large banks, and that small countries could only support small banks. If a bank headquartered in a smallish country found itself able, and willing, to expand, it might only be able to do so if it was to transfer its headquarters and country of residence to a larger country. Would that be acceptable to its prospective new home country, or to its previous home country? Why should all the large international cross-border banks reside in the USA, China or Japan? Would the establishment of a Banking Union in the Eurozone, with Eurozone-wide resolution financing mechanisms, be sufficient to allow Eurozone banks to become much larger than before? Perhaps. One reason why the doom-loop between failing banks and governments in the Eurozone has proved so toxic is that the trigger for state intervention to take over from current management has been too late, by which time severe losses have already been incurred. This is partly because the wrong metric continues to be used, accounting capital rather than the market value of capital, and partly because the authorities
C.A.E. Goodhart / International Review of Financial Analysis 36 (2014) 78–83
are reluctant to abrogate the ownership rights of shareholders in a capitalist society. While the latter is understandable, it is not clear why it is any more consistent with free markets to impose structural controls over the allowable operations of a bank than to require official intervention when the market value of a bank's equity falls below X% of its total assets; moreover warrants could be given to such shareholders to allow them to share in any subsequent strong revival. To conclude, what I would advocate for my optimal banking structure would be: • A much higher equity ratio, supported by a ladder of sanctions, with a much higher minimum intervention point, measured via market, rather than accounting, values; • A liquidity (mis-match) ratio, inversely related to each bank's equity ratio, supported by a ladder of pecuniary sanctions, capable of contra-cyclical adjustment; • An acceptance that failure should imply the sacking of management and expropriation of shareholders, but not necessarily the liquidation of any bank. This could be reinforced by required changes to remuneration practices so that failure would be more painful to management. Intervention should come earlier, so as to restrict losses. Acknowledgements This policy paper consists of two keynote addresses, delivered by the author on 6 March 2013 at the 2nd European Conference on Banking and the Economy (ECOBATE 2013, ‘The Optimal Financial Structure’), in Winchester Guildhall, and on 25 April 2013 at the International Conference on the Global Financial Crisis (‘The Parlous State of Macroeconomics’), at the University of Southampton. References Admati, A., & Hellwig, M. (2013). The bankers' new clothes: What's wrong with banking and what to do about it. Princeton, N.J.: Princeton University Press.
83
Benes, J., & Kumhof, M. (2012). The Chicago plan revisited. International Monetary Fund working paper, WP/12/202 (August). Gorton, G. (2012). Some reflections on the recent financial crisis. National Bureau of Economic Research working paper no. 18397 (September). Gorton, G., & Metrick, A. (2012). Securitized banking and the run on repo. Journal of Financial Economics, 104, 425–451. Independent commission on banking (The Vickers report) (2011). Independent commission on banking: Final report. Written evidence, House of Commons Treasury Committee, 12 September. Kotlikoff, L. (2011). Jimmy Stewart is dead: Ending the world's ongoing financial plague with limited purpose banking. John Wiley & Sons. Liikanen, E., (Chairman), (2012), ‘High-level Expert Group on reforming the structure of the EU banking sector’, Final Report, European Commission, 2 October, see http:// ec.europa.eu/internal_market/bank/docs/high-level_expert_group/liikanen-report/ final_report_en.pdf Miles, D., Yang, J., & Marcheggiano, G. (2013). Optimal bank capital. The Economic Journal, 123, 1–37. Perotti, E., & Suarez, J. (2011). A Pigovian approach to liquidity regulation. International Journal of Central Banking (December). Ryan-Collins, J., Greenham, T., Werner, R., & Jackson, A. (2012). Where does money come from?: A guide to the UK monetary & banking system (2nd ed.). New Economics Foundation. Schularick, M., & Taylor, A. M. (2009). Credit booms gone bust: Monetary policy, leverage cycles and financial crises, 1870–2008. National Bureau of Economic Research working paper no. 15512 (November). Scott, H. A. (2010). Prepared written testimony of Hal S. Scott, Nomura professor of international financial systems at Harvard Law School and director of the committee on capital markets regulation, before the committee on banking. Housing and Urban Affairs, United States Senate (February 4. See http://www.banking.senate.gov/public/index.cfm? FuseAction=Files.View&FileStore_id=c372f56f-819f-4d93-bb5a-6c1802aeb18a). Scott, H. A. (2011). Written testimony of Hal S. Scott, director of the Committee on Capital Markets Regulation and Nomura professor and director of the program on international financial systems at Harvard Law School, before the Committee on Financial Services. United States House of Representatives (June 16. See http://www.capmktsreg.org/ pdfs/2011.06.16_House_Testimony.pdf). Turner, A. (2010). What do banks do? What should they do? Chairman of the Financial Services Authority speech slides. London: Cass Business School (17 March, see http:// www.fsa.gov.uk/pubs/speeches/at_cass_slides_170310.pdf). Werner, R. (2005). New paradigm in macroeconomics. Basingstoke: Palgrave Macmillan.