How to stabilize the banking system: lessons from the pre-1914 London money market

Abstract

This article argues that the British financial system in the era prior to World War I provides modern policymakers with a successful model of how to stabilize the banking system. This model had two components: incentives were structured to ensure that all banks that originated or traded assets on the money market sought only to trade in high-quality assets; and macro-prudential regulation promoted the segregation of money markets from capital markets, monitored the growth of money market credit, and restricted trade on the money market in assets issued by entities and sectors whose money market liabilities were growing so fast that the most reasonable explanation was that the money market was being used to finance longer-term investment. These facts indicate that policymakers can successfully stabilize the banking system through a combination of structural reform and regulation of the growth of credit.

Recent empirical analysis of long-run cross-country banking data has found that there are two distinct eras of financial capitalism: the first includes the years from 1870 to 1914, when London was the center of the world financial system, and is characterized by credit growth that matched GDP growth and by crises with real effects that are relatively small; the second extends from 1945 and is characterized by credit growth that greatly exceeds GDP growth and by crises with real effects that are markedly greater than in the previous era (Schularick and Taylor 2012, pp. 1031, 1041).¹ I argue here that the relative stability of the earlier era was no accident, but instead the product of nascent central banking techniques and thus deliberate policy (Clapham 1945, p. 11; Sayers 1976, pp. 272–3; Duffy 1982, p. 76).² Indeed, a central banker from the era prior to World War I, had he been alive at the turn of the twenty-first century, would almost certainly have predicted the 2007–8 crisis and its severity. For this reason an understanding of pre-1914 British banking, which can be gained by studying the works of luminaries such as Henry Thornton, W. T. C. King and R. S. Sayers, provides us with very clear recommendations for what should be done to stabilize the banking system.

At the heart of pre-1914 British banking lay the London money market, which was structured to align incentives at both a micro- and a macro-economic level. Because incentives were well aligned, every money market participant sought to originate and to trade only assets that would not go into default. This, however, was not sufficient to promote financial stability. Equally important were the backstop of the central bank and the macro-prudential principles enforced by the central bank, which, first, treated unusual credit flows on the money market with skepticism, monitoring the market to ensure that such credit growth was repressed promptly, particularly if there was evidence that the credit flows were being used to finance long-term business activities, and, second, disapproved of the use of money market instruments to finance long-term assets more generally.³ Thus, in Britain financial stability, the origination of privately issued assets that had very low risk of default, constraints on credit growth and the segregation of money from capital markets were all closely related phenomena. This article argues that an understanding of how these elements interacted and sustained each other in pre-1914 Britain can help modern policymakers in their efforts to promote financial stability.

The relevance of the data cited above to this argument is premised on the view that credit policies pursued in the most important international money market can affect credit growth far beyond the borders of the dominant economy (Rey 2015). Thus, prior to 1914 the behavior of credit in many of Britain's trading partners could be affected by the rules of the London money market. By contrast, after 1945 as New York quickly became the dominant money market, London policy lost much of its influence abroad and became relevant mostly to the domestic market. In short, even though the London money market continued to operate on the basis of the same principles well beyond 1914, by the middle of the twentieth century London's policies played a much reduced role in conditioning the growth of bank credit worldwide. Because the United States developed its own approach to stabilizing credit markets, some of the fundamental principles of British banking were forgotten.

Notably, two concepts basic to the pre-1914 British understanding of financial markets have been rediscovered by modern theorists since the 2007–8 crisis: money market stability is founded on ‘safe’ assets,⁴ not the price discovery that characterizes capital markets; and the feedback loops that are created by money market finance of capital market assets foster asset price instability. Early twentieth-century financial markets in Britain were segregated precisely in order to combine stable money markets with price discovery – but not price instability – on capital markets. In fact, the stability of the London money market – and the British banking system – during this period is widely recognized (e.g. Calomiris 2010, p. 9; Dimsdale and Hotson 2014, p. 7; Turner 2014a, p. 8), as is the dominant role played by the London Stock Exchange in world capital markets at the time.

Note that the term stability as it is used in this article does not refer to a complete absence of panic or of bank failure, but instead to an environment where those crises that take place have relatively moderate effects both on the real economy and the banking system, as is consistent with the data cited above for the pre-1914 period. Thus the question of ‘how to stabilize the banking system’ is one of how to eliminate the danger of financial collapse or of crises that exhibit increasing severity over time, not a question of how to protect the economy from experiencing any adverse events at all.

This article analyzes the different components of the institutional structure of the pre-1914 London money market, and how they contributed to the success of the money market. Because the focus is on institutional structure and the goal is to answer the question of what elements comprise a ‘good’ financial structure, the underlying framework draws from game theory more than from competition theory. Thus, it is far from clear that there will be a tradeoff between stability and efficiency in the analysis. After all, if a country has a financial structure that is performing well below its potential because, for example, incentives are misaligned and fraud is common, it is reasonable to expect that improving this structure in a way that will greatly reduce fraud may promote both stability and efficiency at the same time. Even so, the macro-prudential, ‘real bills’-based components of British banking have been framed as promoting either efficiency (e.g. Sargent 2011) or stability, (e.g. Capie 1995), at the expense of the other, and it is important to acknowledge that there are arguments on both sides. In no small part because there is so little consensus on the relationships between real bills, efficiency and stability; however, this article takes no position on the tradeoff or indeed on the question of whether there is a tradeoff at all.

This article brings together a number of arguments made by other authors who have explored the foundations of the London money market's remarkable stability and argues that it was the combination of structural and prudential regulatory features that made it possible for privately issued assets to be safe and liquid. Benefits of the ‘real bills’ approach to banking have been explained by several authors. Charles Goodhart (2011, p. 137) discusses real bills as a form of macro-prudential regulation that promoted stability in part by constraining the growth of credit to match that of GDP. Allen (2014, pp. 8, 10) advocates a macro-prudential ‘eligibility policy’ that allows banks to treat certain forms of short-term commercial debt as high-quality liquid assets. Flandreau and Ugolini (2013, pp. 116, 118) discuss the micro-prudential regulatory role played by Bank of England in monitoring the bills that traded on the money market and argue that this helped establish the international status of sterling. Avner Offer (2014, pp. 159–60) and Goodhart and Perotti (2015, p. 1) argue that the real bills approach involved functional specialization of the different parts of the financial system and by separating the money supply from real estate finance promoted financial stability. This article acknowledges the importance of these aspects of ‘real bills’, but also explains the stability implications of the ‘real bills’-based segregation of capital markets – rather than just real estate – from money markets, details how ‘real bills’ functioned to constrain the growth of credit and argues that stability also required a market structure that aligned incentives on the money market.

Few modern authors have focused on the structural aspects of early twentieth-century British banking. Only Turner (2014a,b) argues that the extended liability faced by bank owners (and manager-owners) played an important role in stabilizing the banking system. This article goes beyond Turner's argument by emphasizing the important role in aligning incentives that was played by three additional structural factors: the market was structured as a discount market, with closely interconnected banks, and the banking system itself was required to bear the losses even of a ‘too-big-to-fail’ bank.

The first part of this article sets forth the basic characteristics of the London money market in the early twentieth century and the principles that governed the behavior both of the banks that traded on it and of the central bank that backstopped it. The second part focuses on the relationship between these principles and recent theory explaining the 2007–8 crisis. The third derives lessons that can guide policymakers in their efforts to stabilize the modern banking system.

I

How was it possible for the privately issued assets that traded on the London money market to be ‘safe’ assets? This section explains that it was possible due to a combination of carefully aligned incentives that ensured that it was in no bank's interest to originate or trade on the money market an asset that was likely to default and prudential regulation based on ‘real bills’ principles that deliberately discriminated against certain types of short-term financing.

Three important structural characteristics of the London money market differentiate it from modern money markets. First, even though many banks were structured as limited liability companies, in the event that a limited liability bank failed without enough assets to cover its obligations, shareholders were still liable for an additional capital contribution that amounted to 20 percent or more of the value of deposits in the bank (Turner 2014b, p. 145).⁵ In addition, since the directors of a bank were typically required by the articles forming the bank to hold shares in it, in Britain a portion of a bank manager's wealth was at stake in the event that the bank failed (Turner 2014b, pp. 145–6). This system of extended liability aligned the interests of bank owners with those of bank creditors, and the fact that managers were exposed to the losses of a failed bank aligned the interests of the managers with those of the owners.

Second, the London money market in the early years of the twentieth century was a discount market. In a discount market, every seller of an asset endorses it, guaranteeing that the asset will pay in full. Thus, for accounting purposes, once one has purchased an asset on a discount market, exposure to the credit risk of that asset cannot be eliminated by sale, but is only eliminated when the asset is paid off. This structure ensured that the interests of the banks that originated and traded money market assets were always aligned with those of the holder of the asset when it matured.

Third, the safe assets that traded on the London money market were privately issued, prime bank bills. Thus, even though then as now money market stability required that the assets traded on the market be safe, facing almost no price risk, in Britain privately issued debt dominated the market, and Treasury bills played only a trivial role in it (Roberts 1995, p. 155).

Given that in modern discourse the term ‘safe asset’ usually refers to government debt, it is important to understand what made it possible for privately issued prime bank bills to comprise the safe assets of the London money market. The most important characteristic of a prime bank bill is that it is the liability of more than one entity. Overlapping guarantees did not only ensure that the failure of a single firm could not result in money market losses, they also mitigated – or even eliminated – incentive problems. Indeed, the liquidity of the London money market was closely tied to its structure as a discount market and to the resulting alignment of incentives which ensured that it was in no market participant's interest to circulate debt of uncertain quality on the market.⁶ The fact that bank managers’ personal assets were at risk also mitigated incentive problems and fostered a liquid market.

Alignment of incentives alone, however, is not sufficient to stabilize a money market: it is a well-recognized problem that price competition between banks can erode underwriting standards – as banks that are poorly (or fraudulently) managed and undercharge for the risk that they are taking on capture market share from those banks whose business model is viable over the long run (Goodhart 1988, p. 48; Akerlof and Romer 1993, p. 41). In short, while the alignment of incentives does much to eliminate fraud, it is less effective at eliminating errors due to overconfidence as in the short run this type of error often results in an increase in market share. Thus, an important question for the structure of any banking system is how it mitigates the adverse effects of competition by either preventing the growth of banks with a short-run business model or ensuring that they fail quickly – before the banking system itself is destabilized.

In Britain central bank policy protected the money market from destabilizing competition. This was possible because an asset had to be eligible for discount at the Bank of England in order for it to trade on the money market. As a result, the Bank's discount policy was the key determinant of the liquidity of a money market instrument, and this gave the Bank effective control over the money market.⁷ By 1900 the Bank's control was so complete that even ‘when the Bank's grip was light, the banks … watched closely the trend of official policy, anxiously scanning … for any clue as to what the [next] action of Threadneedle Street would be’ (King 1972, p. 320).

Discount policy was designed to ‘maintain the standards of quality long associated with the London prime bank bill and hence its reputation as a liquid asset of undoubted security’ (Bank of England 1961, p. 28; cf. Sayers 1976, pp. 33–4). Two policy rules in particular supported this goal. First, only ‘two-name paper’, or debt that bore the guarantee of two ‘good’ names, was eligible for discount (Sayers 1976, p. 35). Thus, a London bank that originated a loan to a small tradesman could not discount that paper at the Bank. In order for the asset to be a money market asset the bank had to add its own guarantee, called an acceptance, to the loan, at which point any other bank that qualified as a ‘good name’ could discount the paper at the Bank. In short, interconnectedness was imposed on the banks by the structure of the market.

Second, the Bank of England had a policy of monitoring the list of ‘good names’ by carefully tracking its exposure to each name, not only directly at the discount window, but also indirectly in the form of acceptances or endorsements (Flandreau and Ugolini 2014, p. 86). In much the same way that a private banker would monitor a client's use of credit and begin to restrict that credit if there were signs that the client was rolling bills over excessively and that his business was not generating the funds needed to keep his debt from growing quickly, when the Bank of England's exposure to a particular name grew notably or if it had concerns about the changing quality of a name's liabilities, the Bank would take action (Sayers 1976, p. 273). The Bank had many tools at its disposal: it could simply communicate with the client; it could communicate with other market participants either directly or indirectly by charging higher rates on paper bearing the name; or it could refuse the paper at the discount office (Sayers 1976, pp. 273, 279). As noted above, the banks trading on the London money market were extremely responsive to any hints given by the Bank with respect to discount policy. For this reason the market was often anthropomorphized and described as ‘refusing to discount bills’ when, for example, an originating bank had placed so much paper on the market that it ‘aroused suspicion’ (e.g. Kerr 1908, p. 273; Cross 1932, p. 218). In any individual case it is often difficult to determine whether the money market itself – acting one assumes through the discount houses that functioned as gatekeeper banks – restricted credit or whether this took place at the instigation of the central bank, and only in a few instances do we have clear evidence of the Bank of England's role. On the other hand, King and Sayers, the scholars most familiar with the operations of the discount market, both concluded that the Bank exercised ‘control’ over the market (King 1972, p. 321; Sayers 1976, pp. 273, 275). The details of the role played by the central bank are, however, much less important than the fact that the Bank of England was universally understood to govern a money market that was widely recognized for its ability to act promptly to repress credit growth that went beyond the norms considered acceptable (e.g. Cross 1932, p. 218).

By most accounts, the Bank of England had only to hint its disapproval of a bank's activities in order for the bank to change its behavior. The force of such admonitions was surely strengthened by the consequences that had been suffered by bankers whom the Bank deemed to have abused the resources available on the London money market: all were aware that the Bank's decision not to support Overend, Gurney and Co. in 1866 was the proximate cause of that bank's failure and that the limited support provided to Baring Brothers and Co. in 1890 was carefully designed to leave the owners of Barings with their liabilities.⁸ Partners in both of these firms were reduced to poverty, losing their possessions and their status in society (Ziegler 1988, pp. 260–2; Ackrill and Hannah 2001, p. 46).

While the Bank of England's handling of the two failures was very different, in both cases the Bank established an incentive structure that would induce the banks to monitor each other's behavior. The failure of Overend implied that, if there was any significant risk that a bank would lose its status as a ‘good name’, it was not in the interests of the other banks to be exposed to the suspect bank's liabilities.⁹ In the case of the Baring Brothers, by contrast, certain of the partners were permitted to become directors of a new company, Baring Brothers & Co. Ltd, and to take over the sound mercantile business of the old company including its discounts and acceptances, protecting the value of these assets on the money market; this was only possible, however, because the Bank had persuaded the banks that were the most important of the old company's creditors to put up a guarantee fund that allowed the old company to liquidate its assets slowly.¹⁰ In short, the Baring Crisis put the banks on notice that even in the event that a ‘too-big-to-fail’ bank failed, they were the parties most exposed to the losses.¹¹

Overall, the Bank of England's discount market policies created a strong incentive for London banks to police each other's origination practices, and to refuse to buy the acceptances of a bank that was lowering the quality of money market assets.¹² At the same time these policies strengthened the leadership role of the Bank of England in the market, and by the early years of the twentieth century the Bank's ability to exercise control over the behavior of London banks was so great that there was a common saying that the Governor of the Bank of England had only to raise an eyebrow for a bank to fall in line (Flandreau and Ugolini 2014, p. 89; Sayers 1976, p. 273).¹³

The Bank's authority extended to the exclusion of suspect categories of assets from the money market (Sayers 1976, pp. 275–7). For example, in 1906 US finance bills were not paid off when the remittances from the fall harvest flowed in and this raised concerns in London that the London market was being used to finance longer-term US investment (Eichengreen 1992, pp. 50–1; King 1972, p. 316). The Bank believed that such debt was too risky for the money market, and indicated as much to the London banks by adjusting its discount policy (Sayers 1976, pp. 44–5, 55). The paper was allowed to run off, and the US had to weather the 1907 crisis without the ability to raise money in London using finance bills. Presumably, the reason that the London banks were so responsive to the Bank of England's concerns was that a bank's decision to support the market for US finance paper by accepting it would have raised questions in the market about the bank's own status as a ‘good name’.

Note in particular that when the Bank of England wished to discriminate against disfavored bills, it did not need to reject the bills outright at the Discount Office. It could instead impose terms on the transactions that were less favorable than those offered for ‘prime bills’ due to the quality of the bills presented. Indeed, the Bank insisted on its right – if not its duty – to evaluate every transaction presented at the Discount Office individually (Sayers 1976, pp. 36–7, 273).¹⁴ Thus, by the turn of the twentieth century the quality controls exercised through discount policy could be implemented incrementally and crisis-provoking shocks could be avoided.

While this explains the structure of the London money market and the central bank's role in it, what remains to discuss are the principles that governed the circulation of assets on the money market. I argue here that the Bank of England – and through it the London money market – was guided by a real bills principle, which favored real bills without imposing the stringent criteria of the ‘real bills doctrine’. A real bill is a short-term, negotiable debt that originates in a commercial transaction, whereas its counterpart, a finance bill, is a short-term negotiable debt that does not.¹⁵

According to the ‘real bills doctrine’, as long as the only bills that trade on the money market are real bills, the money supply and real output will necessarily grow together and both inflation and financial instability will be avoided – thus, in a ‘real bills’ world quantitative controls on the money supply are neither needed nor desirable (Smith 1776, p. 287; Sargent and Wallace 1982; Goodhart 2011, p. 137). This statement of the real bills doctrine is, however, also known as the real bills fallacy, because it is demonstrably false: in the absence of a price anchor nothing prevents the money supply and prices from growing together procyclically (Laidler 1984, p. 151).

This ‘critique’ was, however, understood from the time the doctrine was first expressed, and most proponents of real bills either explicitly or implicitly assumed that the gold standard would serve to determine prices (Smith 1776, pp. 307–8; Laidler 1981, pp. 197–8).¹⁶ Indeed, it was because proponents of real bills recognized the danger of procyclical movements in money and prices that they sought to exclude from the money market instruments that financed the purchase and carry of capital market assets or land (Smith 1776, pp. 290–301; Rockoff 2013, pp. 317–18). The destabilizing nature of such an asset-backed money supply had been illustrated by John Law's misadventures in eighteenth-century France (Smith 1763, pp. 211–18; cf. Law 1705, ch. 7),¹⁷ and every major British banking theorist recognized that an improperly backed money supply could foster asset price bubbles and offered trenchant criticism of Law's ‘debauch of paper money’ (Hawtrey 1919, p. 235; Smith 1763, pp. 211–18; Thornton 1802, pp. 239, 341; Bagehot 1873, p. 91).¹⁸ Thus, the traditional separation between money markets and capital markets in Britain was founded on an understanding of the dangers of using capital market assets as backing for the money supply.

In any case, the real bills doctrine, as stated above, quickly became irrelevant in England: by the 1830s the evolution of commercial practice and commercial law meant that finance bills had become so common that they were granted the same legal status as real bills.¹⁹ Only a few years passed, however, before two crises demonstrated the element of truth underlying the real bills doctrine: to avoid costly crises a financial system that relies on expansion of the money supply to fund business activity needs to guard against funding ‘zombie’ firms (King 1972, pp. 135–6) and those with few prospects of success (HC 1858, pp. xiii-xiv).

The real bills principle, then, defines two categories of finance bills that are considered speculative bills and are associated with banking crises: finance bills that the borrower has little prospect of paying off in the near future and that thus function as long-term or equity investment in the borrower's business, and finance bills that are used to fund the purchase and carry of long-term assets (Sayers 1976, p. 277). In short, the real bills principle is distinguished from the real bills doctrine because it takes a permissive approach to the circulation of finance bills on the money market – as long as the borrower's business is expected to generate the funds to pay off the finance bill within a relatively short time frame.

Because nothing in a finance bill's appearance distinguished it from a real bill, implementation of the real bills principle was necessarily imprecise. As has already been noted, the Bank of England monitored its exposure not only to discounters and acceptors, but also to the various sectors of the economy, and thus the real bills principle was implemented by the Bank's and the market's response to any exceptional growth in the flow of credit – the sort of growth that could be generated by a ‘zombie’ firm or industry gambling for redemption. Thus, the practice developed whereby bills that might be finance bills were carefully evaluated as they entered the market and the gatekeeper banks scrutinizing these bills would take into account recent trends in the quantity of finance paper on the market, and in the quantity of money market instruments financing both the issuer of the paper and the industry to which that issuer belonged (Gillett Bros. 1952, p. 47). Any bank or sector that drew too heavily on the market's resources would ‘arouse suspicion’ and find its ability to draw on the market curtailed (Cross 1932, p. 218; Sayers 1976, pp. 273–5).

From this perspective we see that the curtailment of credit to US markets in 1906 was a reflection of the real bills principle-based view that any large and unusual increase in the quantity of bills issued by a particular sector of the economy was evidence of destabilizing speculation which the central bank was expected to staunch, long before changes in interest rates or asset prices would cause the flow to reverse. To emphasize the relationship to the modern discourse on central bank practice, observe that a policy based on the real bills principle (i) has countercyclical effects because of the speed with which it can be implemented and (ii) is both micro- and macro-prudential because both individual entities and their aggregates, sectors of the economy, are monitored.

In summary, the London money market was able to originate privately issued safe assets through a combination of carefully aligned incentives and real bills-based prudential regulation. Incentives at the micro-economic level were aligned, because the banks that originated and traded the assets could not sell off the credit risk of the asset and the managers had ownership stakes in the banks and bore extended liability for the banks’ losses. Incentives at the macro-economic level were aligned, inducing the banks to monitor each other due both to structurally imposed interconnectedness and to the certainty that the banking system and not the government would bear the losses of even a ‘too-big-to-fail’ bank. Micro-prudential regulation took the form of central bank monitoring of its exposure to assets discounted and originated by each bank and in most cases it was sufficient for the central bank to speak softly – while carrying a very big stick – to the bank in question whenever the credit flows raised concerns. Macro-prudential regulation took two forms: first, the segregation of money markets from capital markets which actualized the generally accepted principle that money market assets should not finance capital market or other long-term assets, and, second, central bank monitoring of money market credit extended to the different sectors of the economy and again speaking softly whenever those credit flows raise concerns.

In short, the stability of the British banking system was founded on a carefully calibrated financial structure that was able to originate privately issued assets that were equally as safe as government debt, and two important components of this safety were real bills-based macro-level constraints on credit growth and the segregation of money markets from capital markets. Thus, when modern researchers find that credit during this period grew in concert with GDP, this empirical result should be viewed as an outcome of deliberate policy, not as an accident of history.

II

A few of the principles of pre-1914 British banking are being rediscovered by central bankers subsequent to the 2007–8 crisis. The fundamental distinctions between money markets and capital markets are discussed in a Bank of International Settlements paper, which explains that the money market and the capital market serve two very different purposes and are built on diametrically opposed logic: the money market can function well only in the presence of price stability – or an absence of price discovery – whereas capital markets function well when price discovery takes place (Holmstrom 2015).²⁰

An early twentieth-century central banker would agree with these distinctions, and explain further that the private sector's ability to create money market assets with stable prices must be carefully controlled so that money market expansion does not feed into capital markets and stimulate an asset price bubble. While the problem of asset price bubbles has become an important topic for modern researchers, the analysis of how the private sector can create safe money market assets has not.

Indeed at present the modern literature simply assumes that privately issued assets are intrinsically unsafe – and thus implicitly assumes that there is no way to structure markets so that privately issued assets are as safe as government debt (e.g. Dang et al. 2014). For example, a Federal Reserve study of ‘short-term safe instruments’ assumes that ‘public securities are preferable to private ones’ and goes on to compare commercial paper to Treasury bills (Carlson et al. 2014, pp. 5–6). Of course, from the point of view of pre-1914 British bankers, American commercial paper, on which only a single party is liable for payment, has never been structured to be safe. Indeed, the benefits that would accrue to the American financial system if it were to establish a ‘two-name’ paper market were discussed extensively in the early twentieth century (Warburg 1910; HR 1927, p. 431). From this point of view, the 2007–8 liquidity crisis that struck the shadow banking system in the US – which treated commercial paper as if it were a safe asset (Carlson et al. 2014, p. 3) – was entirely predictable.

Commercial paper is easily compared to the bills that traded on the London money market, because the majority of the bills were, like commercial paper, unsecured instruments. In modern money markets, however, collateralized instruments play a much more important role than they did in early twentieth-century Britain: central banks typically provide liquidity using collateralized loans rather than by discounting bills, and repurchase agreements have become an important money market instrument (Jobst and Ugolini 2014, pp. 13–17).

In this environment modern researchers have difficulty imagining how a banking system built on unsecured credit can function. Thus, the current literature on privately issued safe assets starts with the assumption that it is impossible for private entities to borrow unsecured, and develops a theory of collateralized lending. This literature emphasizes the value of opacity – that is, the usefulness of collateral that is so costly to value that no one does so, making it possible for money market instruments backed by low-quality and high-quality assets to have the same price. In this framework, a crisis takes place when the fraction of high-to-low quality assets has declined to such an extent that costly price discovery is worthwhile (Gorton and Ordoñez 2014, p. 368). The value of opacity in this literature, however, depends fundamentally on the inability of economic actors to borrow unsecured: if those with production opportunities can borrow unsecured, there is no economic benefit to offering alternative financing using opaque collateral (Gorton and Ordoñez 2014, p. 350).²¹ Thus, I suspect that an early twentieth-century British banker, on learning of the modern theory, would find that the theory is a good description of a low-quality banking system, and would therefore ask: why not put in place a high-quality banking system by aligning incentives so that safe, unsecured, privately issued assets are originated?

As was noted above, another branch of the modern literature that has drawn the attention of central bankers studies the relationship between asset price bubbles and money market instruments that are collateralized by capital market assets – that is, repurchase agreements. A much-cited paper co-authored by a Federal Reserve Bank of New York official shows that when the amount that banks are willing to lend on a short-term basis is determined by the value of collateral, feedback loops develop that foster asset price bubbles and crashes (Adrian and Shin 2010).²²

Once again a pre-1914 British central banker would find these results unexceptional. As R. S. Sayers (1976, p. 277) explained, the Bank of England when implementing its discount policy ‘regarded speculation – in commodities, or in the stock exchanges – as the most dangerous threats to the stability of the system’.²³ Indeed, our pre-1914 central banker would probably go on to explain that the best way to prevent the instability caused by asset price bubbles is to adopt a ‘real bills’ based approach by separating money market liquidity, which depends on price stability, from capital market liquidity, where the goal is price flexibility.

This section addresses only two important results that have been rediscovered pursuant to the post-crisis research agenda. While a more thorough comparison of recent theoretical developments with the preceding section's findings regarding the pre-1914 London money market would surely be of interest, that analysis is left for future work.

III

There are three lessons that modern policymakers should draw from this analysis of pre-1914 British banking theory: first, it is an error to assume that government debt is inherently safer than privately issued short-term unsecured debt, because the safety of the latter depends fundamentally on policy, including the incentives created by the structure of the money market, and the effectiveness of the central bank's macro-prudential regulation of money market credit and its growth. Second, a crucial component of financial stability is the central bank's macro-prudential exercise of control over the assets that circulate on the money market. Finally, the traditional separation of money markets from capital markets was an effective means of addressing one of the most important sources of financial instability, asset price bubbles.

Remarkably, over the course of the 1980s and 1990s when policies were adopted in the US that reversed the long-standing separation of money markets from capital markets, the concerns that were raised focused on the effects of the policy change on the banking system and there was virtually no consideration of the possibility that capital markets would be adversely affected by the change (e.g. Macey 1984; Calomiris 1995).²⁴ The idea that the integration of money and capital markets would be destabilizing, which seemed obvious to British bankers in the pre-1914 years, was beyond the scope of imagination a century later.

Many who have evaluated the consequences of this integration of money and capital markets have concluded that our current financial system is fundamentally unstable, and that only structural reform which eliminates the role played by banks in lending can address this instability (Kotlikoff 2010; Ricks 2012; Cochrane 2014; Wolf 2014; Levitin 2015). These ‘narrow banking’ proposals typically seek to separate banks’ role in the payments system from their lending activities, and to require that all bank deposits be backed by government debt or central bank reserves.

A few advocates of structural reform favor a ‘real bills’ approach (Offer 2014; Goodhart and Perotti 2015), and, interestingly, narrow banking reforms have much in common with such proposals. Both approaches prohibit banks from lending on a long-term basis and from financing long-term assets, reflecting the view that capital market prices should be determined mostly by the transactions of real money investors. The significant difference in the two approaches is that the real bills view treats safe short-term bank loans – and in particular loans that make liquid assets that have been earned, but not yet realized – as an appropriate banking activity, whereas the narrow banking view would prevent banks from playing any role in expanding the money supply.

Underlying this difference in approach is the skepticism that narrow banking proponents exhibit regarding the capacity of the banking system to originate any assets that are ‘safe’. The pre-1914 London money market demonstrates, however, that when the banking system is carefully structured to address incentive problems and when central bank macro-prudential policy is effective, the term safe, privately issued asset is not an oxymoron. Evidence from the crises that took place over the course of the half-century that spanned the turn of the twentieth century supports this view, because the real effects of these crises were comparatively modest (Schularick and Taylor 2012, p. 1041). Furthermore, most scholars have concluded that there were no crises of any significance in Britain during this period.

This article lays out in detail the four structural aspects of the London money market that supported the issue of private safe assets, few of which are present in modern money markets. First, the money market was structured as a discount market, so that every bank that originated or traded a money market asset was exposed to the credit risk of the asset until the asset was paid in full. This contrasts with the modern practice of originate-to-distribute, which involves the sale and securitization of bank-originated assets, which may then be funded on the money market. Second, both bank management and bank owners faced a measure of extended liability in the event that a bank failed and was unable to pay its debts. By contrast, in modern practice bank managers’ incentives are said to be properly aligned by the holding of shares in the bank when both they and the bank's shareholders have (unextended) limited liability. Third, the assets that could be sold to the central bank at the discount facility were limited to assets that bore the guarantee of at least three entities, an issuer, the originating bank, and the discounting bank. Modern discount window practices are typically much looser, dispensing with the preference for non-financial issuers and the requirement that the asset be guaranteed by a second bank in addition to the discounting bank.²⁵ Finally, the banking system had no expectation of access to government funds in the event of a bank failure, except that when the major banks were all willing to agree to a forbearance policy in order to prevent the forced liquidation of the failed bank's assets, the central bank could both broker the agreement and agree to abide by the forbearance policy on the same terms as the privately owned banks. Since the 2007–8 crisis – if not before (Stern and Feldman 2004) – the banking system has had reason to expect government support, although there is much debate over the effectiveness of post-crisis policies designed to reduce the likelihood of such support (e.g. Rehm 2013; Afonso and Santos 2015).

In short, based on the pre-1914 London money market we know how to structure a banking system that can originate safe, privately issued assets, and there is nothing remarkable about the formula for success. It is simply a matter of structuring the banking system so that incentives of the banks that originate and trade money market assets are all aligned together in favor of the circulation of assets with very low risk of default.

Such structural reforms address only the incentives created by banking, however, and are not sufficient to protect the banking system from the growth of new banks that underestimate risk, are initially very successful and yet sure to fail in the long-run – potentially after having decimated the income and weakened the balance sheets of their more staid competitors. These well-recognized adverse effects of competition in financial markets are the reason why macro-prudential regulation by the central bank is also needed. Thus, financial stability and the private sector's ability to originate safe assets also depends fundamentally on the central bank's assiduous monitoring of both the banks that originate money market assets and the assets that they originate in order to ensure that innovation is not having the effect of distorting the price of risk. Typically when such distortions occur, there is a sudden and unusual flow of money market credit into the new asset. Thus, such credit flows call for central bank attention and analysis. If the evidence indicates that these credit flows are not economically justified, then the central bank must exercise control over the money market and restrain these flows. This is the model of macro-prudential regulation that the pre-1914 London money market offers us.

Pre-1914 British macro-prudential regulation also addressed the problem of feedback loops between growth in the money supply and growth in asset prices. When the Bank of England suspected that an unusual credit flow on the money market was due to asset-backed lending, it made sure that the flow of this credit on the money market was restricted (Sayers 1976, p. 277). In addition the long-standing separation of money markets from capital markets was designed to discourage procyclical interactions that would result in asset price bubbles.

This pre-1914 framework stands in stark contrast to our modern financial structure, where the repurchase agreement (or repo) market institutionalizes the finance of capital market assets using money market instruments. This modern structure developed after Britain and the US eliminated the regulatory framework that supported the separation of money and capital markets. Britain acted first with the 1986 Big Bang and the US responded initially by reinterpreting the Glass-Steagall Act and finally in 1999 by repealing it. The repo market grew up in this deregulatory environment and played a key role in the 2007–8 financial crisis²⁶ – just as a British banker from 1900 would have predicted. The crisis, however, also saw the collapse of trade on the unsecured interbank markets that had been the fulcra of British and American banking through the twentieth century (Dudley 2014). Thus, one result of the crisis has been to make the financial system more dependent than it has ever been in the past on a collateralized money market that is, as Adrian and Shin's recent research has shown, intrinsically unstable. The advice of a 100-year-old central banker would therefore be to find a way to rebuild a money market that is based on unsecured debt, because we know that it is possible for a reasonably stable financial system to be founded on the basis of a properly regulated unsecured money market, but we have little reason to believe that it is possible for a collateralized money market to furnish a similar foundation.

IV

The one successful model of macro-prudential regulation that we have indicates that before such regulation can be effective in stabilizing the modern financial system structural reform is necessary, including the elimination of money market finance of long-term assets and the restructuring of money markets to align incentives. It is after all far from clear that financial stability is a reasonable goal in a world where the banks originating and trading assets are not incentivized to originate and trade high-quality assets and where it is common for money market instruments to be backed by capital market assets. After structural reform, promoting financial stability will require monitoring the use of credit on the money market, treating the rapid growth of any asset class as evidence (that is subject to rebuttal) of the use of money markets for long-term finance and restricting trade in such assets on the money market.

Critics will undoubtedly claim that embracing such a strict financial regime would result in a dearth of credit. They are almost certainly wrong. At the turn of the twentieth century Britain's carefully structured and monitored financial system was able to extend credit to markets around the world. Far from reducing the availability of credit, a well-structured financial system makes it possible for credit to be extended widely throughout not only the local, but even the world, economy.