5. The Resilience of British Commercial Banks in the 1930s How accurate are assertions of ‘the vulnerability of the British financial system’? Grossman’s study of a number of countries argues that Britain did not experience a financial crisis, which he attributes to structurally robust banks, the effective use of the LLR function, and timely departure from the gold standard (Grossman, 1994). His analysis of the UK relies on published rather than ‘true’ data for bank profits and capital, but his conclusions are consistent with conventional wisdom about UK banks, which can be summarised in two quotes:
‘The British commercial banks were ... largely unscathed, finding strength in their branch structure and security in their traditionally cautious policy towards involvement in industry’ (Feinstein, Temin and Toniolo, 1997, p.110).
‘... in the UK, despite a number of traumas in domestic and overseas financial markets, there was no important bank failure, no widespread intervention by the authorities and no danger of systemic collapse’ (Collins, 1998, p.19).
Broadly, we do not disagree with these interpretations, and we believe them to be supported by a wider array of evidence than is currently in the public domain. In this section we bring together a variety of evidence relating to the clearing banks in support of the views set out earlier. Variations in operations and lack of consistency in the survival of data in archival records mean that it has not been possible to extract the same information for all institutions, but a reasonably clear picture does emerge.
5.1 Support Operations The large clearing banks neither needed nor received support from the Bank. A number of other bank support operations of various types were organised in the 1920s and 1930s, with the Bank usually playing a leading role. We summarise these in Table 4. Some of these operations were accompanied by significant losses. The bulk of the losses arising were borne by the Bank, with those from the support operations for Anglo International, Anglo-South American Bank (ASAB), Huths and Williams Deacons’ totalling around £8 million. Losses to the clearing banks were spread among various banks and over time. Virtually all these operations related to merchant banks or banks under British ownership operating overseas. The notable exception was Williams Deacon’s, the small Manchester-based clearing bank with a heavy concentration of lending to the cotton industry, whose takeover by the Royal Bank of Scotland was engineered by the Bank, which met those losses which arose from before the takeover. [INSERT TABLE 4 HERE]
In addition to participating in various of the support operations identified in Table 4, the clearing banks assisted the Bank’s efforts to support Sterling before the abandonment of the gold standard. Barclays (and presumably other clearing banks) entered into an exchange with the Bank, receiving Treasury bills for commercial bills, which the Bank then used as security for its loan of £25 million from the Bank of France in August 1931 (Tuke and Gillman, 1972, p. 31).15 Another form of support was participation at the request of the Bank in a scheme in the forward foreign exchange market organised by the Joint Committee of Clearing Banks and Accepting Houses. One of the smaller clearing banks noted that ‘... the General Manager reported that forward [US] dollars and [French] francs had been sold for account of the Bank of England to the extent of £1,700,000’ (Barclays, Martins Board Minutes, 22 September 1931). Sayers (1976, pp.408-9) notes that the total extent of this operation was about £20 million of forward sales, not insubstantial compared to total Bank gold and foreign exchange holdings of £141 million at 30 December 1931 (Sayers, 1976, Vol. 3, Appendix 37, p.355).16 5.2 Capital Flight from Britain A feature of the 1931 exchange-rate crisis commented on by several authors (for example: James, 1992, pp. 602-3; James, 2001, pp. 71-2; and Williams, 1963, pp. 525-8) is the withdrawal from London of short-term foreign capital. This capital flight developed in the face of balance of payments and government expenditure deficits and political crisis (Williamson, 1992). Some of this capital flight represented withdrawal of bank deposits, but it is also believed that some foreign holdings of treasury bills were liquidated. However, estimates of the size of the outflows involved are problematic. For example, Williams (1963, pp. 527-8) estimates total outflows of nearly £300 million in 1931. Sayers (1976, Vol. III, Appendix 32, pp. 307-34) discusses the deficiencies of UK balance of payments data in this period, and his data for capital movements in 1931 include a particularly large balancing item (Sayers, 1976, Vol. III, Appendix 32, p.313, Table C).
Commentaries on the 1931 capital flight are contradictory, both as to the sellers of sterling, and the amounts involved. In Kunz’s account (1987, p.94) ‘... the British themselves were rushing to get out of sterling. Another large portion of sterling sales were German-inspired ... French sales continued, though not by the Bank of France ...’, while US investors were not selling. Kunz goes on to state that at 29 July 1931 the French Treasury and Bank of France held £8 million and £75 million respectively in London (1987, p.86). Born (1983, p.269) suggests that: ‘from mid-July to 20 September 1931, £200 million of short-term foreign credits were withdrawn’ by American, Swiss and Dutch banks unable to withdraw funds from Germany and that the French government was also selling sterling. But Born also claims that at suspension the Bank of France held £65 million at the Bank of England, with a further £24 million held by the Belgian, Dutch and Swiss central banks (1983, p.283). Friedman and Schwartz (1963, p.315) refer to ‘runs on Sterling by France and the Netherlands’ immediately prior to abandonment of the gold standard. According to Sayers, at the time of gold standard suspension the Bank of France held £80 million in the London market, of which only £1 million was held at the Bank of England, which held £40-50 million of European central bank deposits (1976, p.414, footnote 12).
Ideally, archival records would help to reconcile these various accounts and reveal the extent to which foreign depositors withdrew funds from commercial and merchant banks, but few relevant records survive. Overall, it seems implausible that outflows of the magnitudes suggested can have come from the banking sector alone. The deposits of the clearing banks fell in 1931, but by only around £125 million in aggregate (Capie and Webber, 1985, p.437, Table III.(4) ). However, comparison of deposit totals at year-ends is potentially misleading, as monthly averages over the period 1930-32 varied little (Table 5). Consistent with the view that foreign depositors withdrew funds from London, deposits of the Big Five, with stronger London bias to their business, fell more sharply than those of the smaller clearing banks and the Scottish banks. Deposits of the merchant banks were small in relative terms, but would have included a higher proportion of foreign deposits. Truptil (1936, p.314) suggests that ‘the acceptance houses had £105 millions of deposits, nearly all of foreign origin’ at the end of 1930 and lost about 40 per cent of these, which Table 5 suggests is plausible. The British overseas banks, typically with distinct geographical biases to their activities, generally saw falls in their deposits of a similar magnitude to the clearing banks. The exception was ASAB, whose deposits contracted sharply, reflecting the fears for its future which forced the Bank of England to organise a major and long-running support operation (see above and Table 4). A review of data for deposit balances on a week-by-week basis throughout 1931 for one large clearing bank, Westminster, and one small clearing bank, District, suggests that variations during the year were not particularly wide (Table 6). Westminster’s deposit balances did touch their minimum immediately after suspension of the gold standard. [INSERT TABLES 5 AND 6 HERE]
Some direct evidence of deposits placed by foreign customers is available in relation to Westminster. This implies that only a small proportion of short-term capital outflows in 1931 would have been represented by withdrawals of deposits from the clearing banks: ‘When five years ago, England [sic] went off gold, we had approximately £20 millions of sterling balances from customers domiciled abroad ... by September 1934, such balances were reduced to £17¼ millions ... now the Foreign Sterling Balances are at the record figure of £31½ millions ... This Bank habitually holds more than one fifth of the Foreign Balances deposited with the eleven Clearing Banks’ (RBS, WES/1174/187, ‘Report of Chief General Manager’ to Board meeting of 29 September 1936). Foreign currency deposits at the Westminster were small. US dollar deposits and loans were roughly equal at approximately $11 million at 31 October 1927. At the time of German suspension there was a mismatch, with approximately $14 million deposits and $5 million loans at 13 July 1931, reduced to $10.5 million in deposits and $3 million in loans by 7 September 1931 (RBS, WES/1174/249). This mismatch, however, was very small in relation to the overall balance sheet.
5.3 Other Aspects of Balance Sheet Structures An examination of the clearing banks’ aggregate balance sheets for 1930 and 1931 (Table 7) suggests that their deployments of assets do not appear notably different from those in earlier decades, although higher proportions of the ‘bills discounted’ and ‘investments’ categories were now represented by government debt. The major change in balance sheets was the fall in advances in 1932 - in absolute and relative terms. This was offset by the rise in investments, almost all of which were British government securities.17 Thus these banks’ asset structures remained fairly constant, even through the most difficult period in 1931, though 1932 brought some changes. [INSERT TABLE 7 HERE]
Finally, in considering clearing banks’ balance sheets, we turn to their capital positions. As noted earlier, published financial statements consistently understated the strength of clearing banks’ capital positions due to significant ‘hidden reserves’ (Table 8). However, ‘true’ capital ratios generally showed some deterioration through the 1930s, although the pattern varies from bank to bank (Table 9). [INSERT TABLES 8 AND 9 HERE]
5.4 Profitability and Bad Debts Overall, the 1930s proved to be a less profitable decade than the 1920s and later decades with the exception of Midland (Table 10). Table 11 demonstrates, predictably, that advances were the clearing banks’ most profitable asset category. Investments were also profitable, and the policy of ‘cheap money’ from 1932 gave some banks a useful cushion of unrealized profits on investments (Table 12). [INSERT TABLES 10, 11 AND 12 HERE]
Clearing banks’ experiences of bad debts in the 1930s were mixed (Table 13). New provisions for bad debts were generally lower in the 1930s than in the 1920s, and actual write-offs no worse for the two banks for which records survive. Total provisions expressed as a percentage of advances appear significantly worse in the 1930s but the falls in advances exaggerate this deterioration. [INSERT TABLE 13 HERE]
5.5 Other Evidence and Variations in Experience The clearing banks, as with the merchant banks, cannot be considered as a uniform block. We have already seen evidence of different approaches to dealing with Standstill debt, with some banks accepting losses to extract themselves while others held out in the hope of eventual recovery. There were other notable differences between these banks, and their diverse experience can be considered a stabilising factor for the sector as a whole.
The archival records of individual banks yield useful information. In the ‘crisis year’ of 1931, the clearing banks remained sufficiently liquid to maintain normal business. For example, Barclays continued to make large advances to a wide range of customers before and after the German Standstill and abandonment of the gold standard (Barclays, 140/65, ‘Records of Advances Made, Mr. Murrell’s Section’), despite having been rumoured in May 1931 in the New York market to be in difficulty (James, 2001, p.71). A similar picture of ‘business as normal’ can be found, for example, in the approvals given by Lloyds’ Board of Directors (Lloyds, Main Board Minutes 1931).
Westminster, one of the smallest of the Big Five clearing banks, was banker to the merchant banks with the largest Standstill exposures. Westminster’s analysis of advances shows that its exposures at 17 September 1931 to the merchant banks Kleinworts and Sassoon were £1.76 million and £361,000 respectively (RBS, WES/1174/51). The corresponding analysis shows that in 1939 its advances to financial institutions increased by two-thirds over the previous year, with nearly half of this increase of £5 million accounted for by increased exposures to the merchant banks Kleinworts (+£1,696,000), Schroders (+£400,000), and Sassoons (+£160,000). A further 15 per cent of the increase is described as increased exposure to ‘Germans and Poles’ (RBS, WES/1174/50). This is consistent with the Bank’s attempts to ensure support for institutions with large Standstill exposures and the withdrawal of Standstill bills from the discount market during the later period.
The nature of the international activities of the clearing banks differed. Barclays had its French subsidiary, but its international activities were mainly in British colonies or dominions through it subsidiary, Barclays (Dominion, Colonial and Overseas) - Barclays DCO. Lloyds had relatively large indirect exposures to South America and Europe through subsidiaries or associates which it was forced to support. It owned 57 per cent of the Bank of London and South America (BOLSA) in the early 1930s, diluted to under 50 per cent after BOLSA’s 1936 merger with ASAB (Jones, 1993, p.140). Lloyds’ interest in BOLSA led it to decline to participate in the support operation for ASAB (Jones, 1993, p.241). Lloyds held a 50 per cent interest in Lloyds and National Provincial Foreign Bank (LNPFB - discussed below) and its 10 per cent, £100,000, shareholding in British Italian Corporation generated a loss of more than ten times its original investment (Jones, 1993, pp. 231-4; Table 4). The Midland also had fewer problems from international business, having pursued a different international strategy - its London-based overseas branch was consistently profitable but unlike the other Big Five banks it had no subsidiary or branch operations outside Britain (Holmes and Green, 1986, p.165).
The Midland had greater exposure to British trade and industry than other clearing banks and by 1930 already had ‘... a particularly heavy concentration [of doubtful or bad debts] in the textile industry, coal, iron and steel, and in stockbroking and the commodity trades’ (Holmes and Green, 1986, pp. 179 and 189). Furthermore, with its subsidiary Belfast Bank, the Midland had an exposure to the Royal Mail Shipping Group of £3.4 million, roughly equivalent to its initial German Standstill exposure (Holmes and Green, 1986, p.185). Lloyds’ exposure to heavy industry was less significant - for example, advances to collieries at 31 July 1931 amounted to only £1.43 million - less than 0.4 per cent of assets (Lloyds, Main Board Minutes, 28 July 1931). Its direct exposure to Standstill debt was proportionately lower than those of the other large clearing banks, and the bank showed greater willingness to accept losses in reducing this.
Of the Big Five banks, we know least about National Provincial, whose surviving archival records from the period are much sparser. This is unfortunate as it was banker to many of the accepting houses affected by the Standstill. When the Standstill appeared close to breakdown (24 October 1934) ‘... the chairman of National Provincial Bank ... was called to Bank of England and the Governor warned the [Bank’s] Committee of Treasury that the Bank might have to join in a rescue operation for four of the [accepting house] firms’ (Sayers, 1976, p.510).
Lloyds and National Provincial were equal partners in LNPFB, with its head office in Paris and most of its operations in France. This can only be judged a disastrous venture. The two parents provided a total of £1.5m in support up to the outbreak of World War Two compared to the approximately £15 million in assets shown in a pro-forma balance sheet at November 1935 (RBS, LBI/14/2078). The support was made up of £900,000 paid in March 1936 to cover guarantees previously made and a total of £600,000 in provisions made in the books of the parent banks to cover expected losses (RBS, LBI/14/2072).
6. The Bank of England as Lender of Last Resort Even if our arguments are not accepted and sufficient evidence could be brought forward to persuade that there was a risk of financial crisis, that the commercial banks were at risk and the payments system in danger, we would still argue that this was not a serious concern. The reason, as indicated above, is that there was a mature LLR in the Bank.
The story of the Bank’s emergence as LLR has been set out at length (Capie, 2002) and only a brief rehearsal of that is needed here. The Bank slowly learned that role over a long and necessarily complicated process in the course of the nineteenth century. Following the removal of the Usury Laws in the 1830s the Bank was able to use interest rates when supplying the market with funds. When there was a danger of a scramble for liquidity, which might arise for a number of reasons, the Bank, as the ultimate source of cash, could supply it endlessly. However, it was preferable that it did so at an increasing price, and those who held good quality assets could always get cash.
In the nineteenth century there was the complicating factor of the gold standard. Under such a regime the monetary base was determined via the balance of payments and the Bank had a limited role to play at best. The solution to this would be to suspend the gold standard when there was the risk of a financial crisis, and to print and supply all the cash needed to satisfy the demand. This could only be done where the supplying institution had fairly complete credibility, as was the case in the circumstances of the second quarter of the nineteenth century.
The LLR role developed in a favourable legislative period and near to ideal institutional setting. When financial institutions learned that they could always get liquidity they were less inclined to panic. Furthermore, the institutional arrangements of the time included the operation of the discount houses as a buffer between the banks and the Bank, which allowed anonymity to prevail and prevented special pleading, very much in line with the mood of that time in keeping special interests at a distance. At the same time, the banks were learning prudence and to work off appropriate operating ratios, which included a shift to more liquid asset structures (Baker and Collins, 1999; Collins and Baker, 2001a and 2001b).18 After about the 1870s these arrangements were in place and there was no financial crisis after that date. There may however have been LLR of activity on the part of the Bank i.e. providing liquidity when fear made an appearance. There was almost complete trust in the system and a complete appreciation of its workings. Thus even when other countries suffered financial crises, such as in 1873 or 1907, the British system remained stable. So in 1929-31 there was no need for panic and the banking system understood how the Bank would behave if any such panic began to appear - the Bank could secure the suspension of gold convertibility and supply the markets with all the liquidity they needed at an appropriate rate of interest. The markets had come to understand this and the result was the remarkable stability that characterised the British financial system from the 1870s to the late twentieth century.
7. Contrast to the German and US Experience Controversies continue over the causes of banking failure and financial crises in the 1930s. While there is insufficient space in this paper to give more than a cursory treatment to the positions in Germany and the United States, certain aspects of their experience can be contrasted to that in Britain. Many have argued that the policies of the German Reichsbank and the US Federal Reserve (‘the Fed’) provide notable examples of the inadequacy of macroeconomic policy. We have already commented on the Reichsbank’s failure or inability to act to stem the central European banking crisis.
The significant losses of deposits by German credit banks can be contrasted to the experience of British banks. By the end of June 1931 foreign deposits of German credit banks had shrunk to about 38 per cent of their level two years earlier, but still represented 18 per cent of their total deposits (Balderston, 1991, p. 564, Table 5). The total assets of the Berlin Great Banks had fallen by 30 per cent over the same period and their liquidity had declined sharply (Balderston, 1991, pp. 578-9, Tables 6 and 7). Balderston (1994) includes more detailed analyses of German bank deposit changes during the 1931 crisis, and also argues that the deposits of foreign banks in Germany had acted as a substitute for LLR activity by the Reichsbank (1994, p.61). As we have noted above, whilst it is not possible to produce comparable analyses of deposits for the British clearing banks, it is clear that major British banks did not rely to the same extent on international deposits, did not suffer comparable withdrawals and remained much more liquid. A dependence on short-term foreign bank deposits has been seen in more recent financial crises such as that in south-east Asia in the late 1990s.
Turning to the US, the debate on the precise causes and timing of the Great Depression continues. However, there is considerable agreement that the role of the Fed was crucial and no doubt that its behaviour was inept, its mistakes avoidable and not the result of powerlessness in the face of global forces. There were several waves of bank failures at the beginning of the 1930s and the public rushed into cash as fears of liquidity problems spread. In the 43 months from September 1929 to March 1933 that mark the depression years, commercial bank deposits fell by 42 per cent, while the ratio of cash to deposits rose from 9 per cent to 22.5 per cent, and the banks’ reserve ratio from 7.7 to 11.9 per cent (Friedman and Schwartz, 1963, pp.803-4, Table B-3). What was needed therefore was for the Fed to inject sufficient high-powered money to offset the reduction in the money stock. Although this was done, it was hopelessly inadequate. Bordo et al. (2002) have shown that at two critical points, between October 1930 and February 1931 and then again from September 1931 to January 1932, the Fed could have prevented the banking panics, and could have done so without endangering convertibility, such were the stocks of gold in the US. The Fed knew how to inject this cash and had done so before. The conclusion must be that there was no need for the normal US business cycle to turn into the Great Depression, which in turn would not have spread to many other countries for whom the American market was important.
Richardson and Van Horn (2007) support this interpretation, in demonstrating that failures of banks in New York City, the centre of the US money market, were not a result of contagion from European banking problems. Calomiris (2008), while also rejecting the view that contagion or panic contributed to significant bank failures in the US, is critical of the view that the Fed failed. He draws attention to sector-specific and regional shocks, magnified by the prevalence of unit banking in the US. However US experience is interpreted, it was not mirrored in Britain. The British amalgamation process had produced a small number of sound institutions - liquid, well-capitalised and well-diversified - and these were able and willing to support the actions of the central bank, which did not, and did not need to, extend to full LLR activity. Selective, rather than general, support was sufficient to preserve the institutional framework.