Structure of the German Banking Sector

The German banking sector was dominated by the great banks in Berlin, which accounted for more than 20 percent of the banking sector's total balance sheet in the 1920s (Table 1).³⁰

Another important group were the public banks, namely the savings banks as well as the Staatsbanken, Landesbanken, and Girozentralen, which together held more than one-third of the banking sector's total assets. The mortgage and cooperative banks each had market shares of around 10 percent, the private bankers of 6 percent. The rest consisted mainly of provincial and other credit banks.

The Berlin great banks were universal banks: Besides the “traditional” banking activities, such as taking in deposits and extending loans, they were participating in investment banking and in international finance. While Deutsche Bank und Disconto-Gesellschaft (hereafter Deutsche Bank), Darmstädter und Nationalbank (Danatbank), Dresdner Bank, and Commerz- und Privat-Bank (Commerzbank) maintained widespread networks of branch offices, Berliner Handels-Gesellschaft (BHG) and the publicly owned Reichs-Kredit-Gesellschaft (RKG) did not maintain branch offices. The great branch and nonbranch banks also differed substantially in their deposit and loan structures: The clientele of the branch banks was mixed due to their presence in both large towns and rural areas, whereas the great nonbranch banks worked mostly with a relatively small number of wealthy and rather homogeneous clients on both the asset and the liability sides of their balance sheets. Thus, average deposits and average loans were much higher at the nonbranch banks than at the branch banks.³¹

The loan portfolios of the great branch banks were dominated by industry loans, whereas the great nonbranch banks also conducted a considerable part of their loan business with other banks. This was because provincial and other credit banks often preferred to conduct business with banks not standing in direct competition with them. The balance-sheet structure of the large provincial banks, who often maintained regional branch networks, was similar to the structure of the great branch banks, while the structure of large private bankers resembled that of the great nonbranch banks. In fact, some of the large provincial banks, such as Bayerische Hypotheken- und Wechselbank, Bayerische Vereinsbank, Barmer Bankverein, and Allgemeine Deutsche Credit-Anstalt (ADCA) and possibly also some large private bankers had total assets comparable in size to the great nonbranch banks.³² Therefore it is not justifiable by economic significance to neglect these banks in an analysis of the crisis of 1931.

The portfolios of small private bankers, cooperative banks, and savings banks were completely different: These banks were small and locally orientated and their average deposits and loans were small. The “little saver” was rather to be found at these banks than at the larger banks, even though the competition for these customers increased in the 1920s. The savings banks had their own central institutes, the Girozentralen, which managed the savings banks' mandatory liquidity reserves, served as central clearing organizations, and helped in the provision and investment of funds. Staatsbanken and Landesbanken belonged directly to the German states. Because their activities were similar to the ones of the Girozentralen, I subsume the three bank groups under one group called SLGs. The clientele of the SLGs consisted mainly of public nonbanks and public or private banks. Only at the public banks, the business with public entities played a significant role.

Major Changes in the 1920s

After the dramatic shrinkage of the banking system between 1913 and 1924, the second half of the 1920s saw a rapid expansion of the banking sector: Total assets increased by a factor of five between 1924 and 1930. This expansion had three noteworthy characteristics: First, it took place in an environment of intensifying competition; second, it was to a large extent financed by foreigners; and third, it was accompanied by increased risk-taking by banks, particularly by the great branch banks. I will describe each of these factors in turn before showing the links between them.

The intensification of competition in the banking sector was a consequence of the First World War and the following hyperinflation: By 1925, the total balance sheet of all banks had shrunk to one-fourth of the 1913 level (in real terms). Not all banks groups had been hit equally strongly: Assets at mortgage and savings banks had fallen by more than 90 percent, resulting in sharply reduced market shares.³³

In contrast, assets at the Berlin great banks had been reduced by “only” 50 percent, such that their market share had more than doubled. This was, however, mainly due to numerous acquisitions of smaller banks.³⁴ The market share of the SLGs also increased strongly because of the large number of new foundations. In the same period, the banking sector experienced an enormous physical expansion as the number of branches multiplied. In this manner, the banking system built up excess capacities that were out of all proportion to the shrunken business volumes.³⁵ As the competition through interest rates was limited due to existing cartel agreements, the regional expansion played an important role in the attraction of new customers. In particular, the great branch banks tried to expand into provincial areas, which until then had been dominated by the savings banks, cooperative banks, and small provincial banks. At the same time, the savings banks entered new fields of business that formerly had been reserved to credit banks. This was made possible by the loosening of the regulatory restrictions on the savings banks, which enabled them to compete with the universal banks.³⁶ Finally, there also was increasing competition through foreign banks who extended direct loans to German firms, foregoing the intermediation of German banks.³⁷

At the same time, concentration increased, especially because of the acquisition activities of the great branch banks: Between 1914 and 1928 these banks swallowed 210 smaller banks.³⁸

Thereby, the market share of the great banks as measured by total assets increased from 12.6 to 23.3 percent between 1913 and 1928. The concentration movement culminated in the merger of Deutsche Bank and Disconto-Gesellschaft in 1929. The merged bank had a market share of 9.2 percent, more than twice the size of the second largest bank Danatbank (Table 1).

The intensification of competition was a major determinant of banks' behavior in the 1920s. On the one hand, the battle for scarce deposits prompted the credit banks to raise part of their funds abroad; on the other hand, the erosion of rents in banking induced a general increase in risk-taking.

The most striking feature of the evolution of banks in the 1920s was the explosion of short-term foreign debt, which increased by a factor of 7.4 between 1925 and 1929. In 1929, 18 percent of all deposits in the banking sector were foreign.³⁹

Long-term foreign debt played only a minor role in the banking sector: At monthly reporting banks, long-term debt amounted to only 10.6 percent of total foreign debt at the end of 1928.⁴⁰ Long-term borrowing from abroad had been obstructed by a number of political measures, such as the installation of the “Beratungsstelle für Auslandskredite” for public and publicly guaranteed loans in 1924 and the abolishment of exemptions from the German capital yields tax for bonds issued abroad in 1926.⁴¹ Instead of reducing foreign debt, these measures led to a shift towards short-term debt and thus increased Germany's vulnerability to short-term capital reversals.⁴² In addition, foreigners were less and less willing to extend long-term loans when confidence in the economic and political stability of Germany was shattered towards the end of the 1920s. As long-term foreign debt at monthly reporting banks underwent virtually no change in the years preceding the crisis, the following analysis will focus on short-term debt.

Foreign debt was anything but equally distributed across banks: At the end of 1929, 82 percent of foreign deposits were held by monthly reporting banks and another 17 percent by private bankers. The remaining banks' foreign debt was negligible. The great Berlin banks alone accounted for 73 percent of total foreign deposits, 42 percent of their deposits were foreign (Table 2). For other reporting credit banks, the latter ratio was somewhat smaller (27 percent), but showed a strong dispersion, while foreign debt at the SLGs was close to zero. Hence, the expansion of foreign bank debt can entirely be attributed to two bank groups, credit banks and private bankers. Looking at the great banks separately, we find that the great banks without branches held higher shares of foreign deposits than the other great banks. In absolute terms, however, these banks' foreign debt was relatively small compared to the branch banks'; for example, the foreign debt of RKG and BHG each amounted to only 17 percent of the foreign debt held by Deutsche Bank.

Besides foreign debt, German banks also held considerable foreign claims, which amounted to 42 percent of foreign liabilities in 1928.⁴³

A large proportion of foreign debt was denominated in foreign currency. In addition, German banks held large foreign-currency claims towards domestic debtors as it had become common practice to re-lend foreign-currency debt in foreign currency in order to limit currency risk.⁴⁴ Hence, there probably was no currency mismatch in most banks' balance sheets.⁴⁵

The third remarkable factor was the increase in banks' risk-taking. In general, risk-taking can take two different forms: The first concerns the degree of maturity transformation and, thus, the risk of illiquidity, the second concerns the riskiness of the banks' assets and the levels of equity, and therefore the risk of insolvency. The banks who suffered most in the 1931 crisis, the great branch banks, were also the ones who were most vulnerable before the crisis with respect to both risk aspects.

After stabilization, first-order liquidity had been very high compared to the prewar period.⁴⁶

Over the second half of the 1920s liquidity ratios dropped substantially at all bank groups, but were still not much lower than they had been in the prewar period. However, these numbers conceal the underlying deterioration in banks' liquidity positions. First, the banks who had acquired high shares of “hot (foreign) money” were vulnerable to sudden reversals of international capital flows. Second, many banks—especially the credit banks—relied heavily on bills of exchange for their liquidity (Figure 1).

As has been forcefully argued by Knut Borchardt and Richard Tilly, one of the reasons for the extraordinary success of universal banking in Germany had been the “virtually unlimited access to the Reichsbank's discounting facilities,” which induced German banks to finance illiquid long-term projects that otherwise would have been considered too risky.⁴⁷

Because the Reichsbank had discounted all bills satisfying certain quality criteria without any quantitative restrictions before the war, bills of exchange could be considered close substitutes to cash with respect to their liquidity.⁴⁸ However, the banks received a serious blow in the turbulent year of 1924, when the Reichsbank's implicit liquidity guarantee clashed with its goal of stabilizing the currency. The Reichsbank declared a “Kreditstopp,” meaning that discount loans were frozen at their current level. Hence, bills could no longer be considered substitutes for cash. This could be the reason for the smaller reliance on bills in 1924 compared to the prewar period (Figure 1). However, there existed a number of exceptions from the discount restrictions, favoring banks who participated in the financing of agriculture, export industries, and large industrial firms.⁴⁹ Presumably, among those there were many large banks. In addition, the Reichsbank explicitly privileged the great branch banks by relaxing their quotas early on.⁵⁰ This may explain why the great banks reacted less strongly to the regime switch than the other credit banks (Figure 1): They expected to be spared from discount restrictions in the future due to their economic significance.⁵¹ Over the second half of the 1920s, the share of bills in liquid assets at the credit banks increased.⁵² Hence, banks' liquidity positions deteriorated even more than indicated by first-order liquidity.

Remarkably, the share of bills in liquid assets at the great branch banks was particularly large (Table 3). The highest share could be found at Deutsche Bank (72.7 percent). In contrast, the great nonbranch banks and the largest provincial banks held on average much smaller shares of their liquidity in bills. The high reliance on bills of exchange at the great branch banks was coupled with particularly low levels of expanded cash liquidity (Table 3).⁵³

Given their high levels of foreign debt and their low expanded cash liquidity, the great branch banks could quickly find themselves in a precarious situation if the Reichsbank refused to provide liquidity through the discount window.

The deterioration in liquidity positions was accompanied by a steady decline in equity ratios (Figure 2). The figure shows that the decline was most pronounced at the great banks where the equity ratio decreased from 19.5 percent in 1913 to 6.7 percent in 1929. In addition, equity ratios at the great banks were on average much smaller than at the other credit banks. A lower equity ratio indicates higher risk-taking and it also implies stronger incentives for taking risks.

Again the great branch banks stand out, as their equity ratios were particularly low (Table 4). In November 1929 Danatbank had the lowest (4.8 percent) and Dresdner Bank the third lowest (5.7 percent) equity ratio of all reporting credit banks, while the equity ratio of Barmer Bank-Verein was as high as 12 percent. The smaller credit banks had even higher equity ratios. Table 4 also shows that the decline in equity ratios was more pronounced at the great branch banks than at the great nonbranch banks and the large provincial banks.

Finally, there is some anecdotal evidence that risk-taking increased in the 1920s, and particularly so at the great branch banks. One striking feature was the common practice of extending loans large enough to lead a bank into distress in case of default. The most famous example is the 48 million Reichsmark loan of Danatbank to Nordwolle, which corresponded to 40 percent of Danatbank's equity. In addition, it seems that the great branch banks' monitoring was particularly poor. Deutsche Bank, for example, was involved in a series of financial scandals in the middle of the 1920s, which started a public discussion on whether Deutsche Bank was paying enough attention to monitoring its debtors.⁵⁴

Similarly, the chairman of Danatbank, Jacob Goldschmidt, had “gained a measure of negative notoriety” after the spectacular collapse of the Stinnes empire in 1925.⁵⁵ In contrast, the two nonbranch great banks were known for their prudential loan policies. At RKG, for example, loans were not extended beyond a certain maximum amount and every loan had to be approved by the directors' board.⁵⁶ Similarly, the two chairmen of BHG, Carl Fürstenberg and Otto Jeidels, were known to be conservative and prudent in their loan policies. One illuminating example is the refusal of BHG to give further loans to Nordwolle, which eventually led to the switching of Nordwolle to Danatbank, where it received abundant loans.

It, hence, seems that the great branch banks were particularly vulnerable with respect to their liquidity and their solvency positions even before the crisis had started. One may argue that there was no need for the great branch banks to have the same liquidity and capital buffers as the other banks, because they were more diversified on both the asset and the liabilities sides due to their branching networks. Although it is true that branching protected these banks from local asset or liquidity shocks, a branching network was of little help against an economy-wide crisis or against sudden withdrawals of foreign debt. The severity of the crisis to come could hardly have been expected, but capital reversals had happened in 1924 and 1927, and were not unlikely to happen again. Therefore, a cautious banker would have made provisions against this eventuality by holding higher levels of liquidity.⁵⁷

In fact, the large provincial banks, who also maintained branch networks, had relatively high equity and liquidity ratios, but much smaller shares of foreign debt.

Too Big to Fail?

The observed increase in banks' risk-taking has often been attributed to the intensification of competition in the banking sector.⁵⁸

This argument is certainly valid with respect to the general increase in risk-taking, but it cannot explain why the great branch banks exhibited a particularly risky behavior compared to other bank groups such as the great nonbranch banks and the large provincial banks. I therefore propose an alternative—and complementary—explanation, namely that the great branch banks were subject to a moral-hazard problem due to their enormous size. The great branch banks may have been expected to be saved by the Reichsbank or the government in case of a crisis because the breakdown of one great bank could threaten the stability of the whole banking system. Such a “too-big-to-fail” doctrine would, of course, lead to incentive problems showing up in excessive risk-taking at these banks.⁵⁹

With respect to liquidity, the moral-hazard problem arose directly from the implicit liquidity guarantee of the Reichsbank. In fact, the issue of moral hazard had already been fiercely debated before the war, when the generous liquidity provision of the Reichsbank had induced German banks to sharply reduce their liquidity reserves.⁶⁰

The “Kreditstopp” of 1924 presumably mitigated the moral-hazard problem for credit banks other than the great banks, but not at the great banks who had received preferential treatment during the discount restrictions. With respect to solvency, it is less obvious that the great banks were protected by an implicit guarantee as there was no precedent of the failure of one of the great banks. However, the great banks had become ever more influential, especially after the enforcement of a “corporatistic model” of banking regulation by the banking commission (“Enquête-Ausschuß”) in 1908, which provided for a close cooperation between the great banks and the Reichsbank in regulatory questions.⁶¹ Their influence was still increased by the concentration movement and the ensuing growth in the great banks' market shares. Therefore, a bail-out of the great banks could easily be justified by their “economic significance.”⁶² Hence, it was very likely that Reichsbank and Reich would try to prevent the failure of any of the great banks.

I would like to stress one special aspect in which the great branch banks were too big to fail, namely their high levels of foreign debt, particularly in comparison with the Reichsbank's reserves. Foreign deposits at the great branch banks exceeded the Reichsbank's reserves by 70 percent at the end of 1929 and were almost seven times as high as the “free” reserves above the statutory 40-percent gold cover. Due to the high levels of foreign debt, withdrawals of deposits at the great branch banks would tend to translate into pressure on the German currency because foreign creditors were likely to transfer the withdrawn funds abroad. This could easily lead to a demand for gold large enough to exhaust the Reichsbank's reserves. Therefore, the central bank had an incentive to prevent withdrawals from banks holding foreign debt if it wanted to protect the stability of the gold standard, in turn increasing the banks' incentives to attract short-term debt.⁶³

This is where the conflict of goals of the central bank comes into play. The provision of liquidity to the banking sector could at the same time endanger the gold standard, especially if the currency was also under pressure at that time. Furthermore, the Reichsbank's scope for interventions was limited if foreign debt was denominated in foreign currency, as it could not create foreign currency. Therefore, the credibility of the Reichsbank's implicit guarantee towards the banks hinged on its ability to obtain foreign currency from abroad.⁶⁴ International cooperation between central banks in times of crisis had been common in the times of the prewar gold standard, and was also present in the second half of the 1920s.⁶⁵ Political shocks that cast doubt on such foreign support could, however, easily decrease the credibility of both the gold standard and the implicit guarantee towards the great branch banks.

The intensification of competition probably exacerbated the moral-hazard problem. Before the war, high rents in banking may have restrained banks from taking excessive risks because they ran the risk of loosing these rents—the bank's “charter value” in economic parlance—in case of default.⁶⁶

The erosion of rents after the war increased the banks' incentives to take higher risks.

There remains one unresolved question with respect to the moral-hazard explanation. Why did the great nonbranch banks or the large provincial banks not take comparably high risks as the great branch banks? Looking at Tables 1 and 2, we find that these banks were not really large compared to the great branch banks. Therefore, these banks could be less sure that they would be “saved” in case of a crisis. In addition, the monitoring of the great nonbranch banks by their depositors presumably was much better than at other banks because there were relatively few large depositors.

We can conclude that the evidence is consistent with excessive risk-taking at the great branch banks due to an implicit Reichsbank guarantee and an increasingly competitive environment. This behavior was even reinforced during the crisis.

Currency Problems

Before 1931 there were two episodes in which the currency situation became acutely unstable, with gold reserves and the gold cover collapsing (Figure 3). The two crises have many similarities. Both episodes were triggered by political events, the first by the imminent breakdown of reparations negotiations in the spring of 1929, the second by the governmental crisis in July 1930 and particularly by the Reichstag elections in September 1930. As can be seen from Figure 3, the gold cover plummeted, almost touching the 40 percent boundary in April 1929. Increases in the discount rate could not stop the outflow of reserves, and the Reichsbank started to restrict lending through the discount window in April 1929. Interestingly, these restrictions were again applied selectively, favoring the Berlin banks, and in particular the great banks.⁶⁹

In both cases, reserve outflows ended only when the political crises had been settled.⁷⁰

The currency disturbances were accompanied by declines in deposits (Figure 3). The currency circulation decreased as well, which suggests that there was no substitution of deposits into Reichsmark. As there are no indications of a flight into stocks, bonds, or real assets either, it seems that the funds were transferred abroad. This is consistent with the observed gold outflows during the crises. The declines in deposits were common to almost all bank groups. Only at the savings banks, total deposits did not decrease during the two crisis episodes, but continued to grow until the end of May 1931. A decomposition of deposits shows, however, that the savings banks' giro deposits declined in both the spring of 1929 and the fall of 1930 and that the growth rates of savings deposits decreased in both episodes. Hence, the impact of the crisis episodes was discernible at all bank groups, while the magnitude of deposit losses differed across bank groups.

How did foreign creditors react during the two crisis episodes? Figure 4 shows that domestic and foreign deposits at monthly reporting banks generally moved in the same directions, but that foreign deposits tended to swing more strongly in both directions.⁷¹

Thus, banks holding foreign debt were hit more strongly than other banks. In addition, the withdrawals at the great banks were particularly strong. This is consistent with the view that foreign deposits were more mobile than domestic ones and that deposits at the great banks were on average more mobile than deposits at other bank groups.⁷² Thus, the intensity of deposit losses seems to have been linked to the mobility of depositors at different bank groups.⁷³

The timing of the crises suggests that the deposits withdrawals were a consequence of the currency problems and not vice versa. The crises started immediately after political shocks and ended when the political situation calmed down.⁷⁴

The political shocks shattered investors' confidence in the maintenance of the gold standard and raised fears of a sovereign default and capital controls. The wish to transfer money abroad translated into deposit withdrawals at German banks. In fact, doubts about the stability of the Reichsmark manifested themselves in disproportionate withdrawals of Reichsmark deposits: Commerzbank, who had the highest share of Reichsmark deposits in total foreign deposits, experienced the strongest foreign withdrawals in the spring of 1929, and the share of Reichsmark deposits at the great banks decreased substantially in both 1929 and 1930.⁷⁵ Moreover, the banks' attempts to conserve their foreign-currency claims during the crises suggest that they themselves distrusted the German currency.⁷⁶ There is no indication that deposit withdrawals were triggered by doubts about German banks, because conservative banks like BHG and RKG experienced withdrawals comparable to those of a more venturous bank such as Danatbank. Instead, the different intensities of deposit losses at different banks depended on the mobility of their creditors. Therefore, the savings banks survived the two crises largely unscathed because the “little savers” were relatively slow to react (Figure 4), while the great banks were the ones to suffer most.

Banking Problems

The third crisis episode originated from problems in the financial sector, namely the breakdown of “Frankfurter Allgemeine Versicherungs-AG” (FAVAG), the second largest German insurance company, in August 1929. The collapse was followed by a stock market crash and the failures of a number of banks connected with the FAVAG through affiliation or business relations. The number of failing banks in the fourth quarter of 1929 was so high that it even exceeded the number of failures in the stabilization crisis of 1924.⁷⁷

The high number of bank failures was connected to falling stock prices and the cyclical downturn, but in most cases these factors seem to have been the trigger rather than the cause. In fact, a majority of the bank failures have been attributed to high risk-taking and fraud.⁷⁸ The breakdown of FAVAG itself had been due to losses from dubious transactions outside the insurance business that had been veiled until 1929 by falsifications of balances.⁷⁹

Interestingly, the crisis can hardly be detected in the aggregate deposit series (Figures 3 and 4). Only domestic deposits of monthly reporting banks show a slight decrease, while deposits at savings banks and foreign deposits increased strongly.⁸⁰

However, the evolution of deposit flows was not homogeneous: Deposit losses were concentrated in certain regions, such as Frankfurt and Kassel, and in certain types of banks, such as civil servants' and grain banks.⁸¹ In my sample, there were 24 banks other than the failing banks that lost more than 10 percent of their deposits between September 1929 and February 1930, while aggregate deposits of monthly reporting banks increased by 3 percent in the same period. Deposits at the great banks were increasing as well, even though almost all of them—the exceptions being Dresdner Bank and BHG—were financially involved in the FAVAG scandal. A decomposition of deposits shows, however, that domestic interbank deposits declined sharply towards the end of the year, and particularly so at the great banks and at the large provincial banks, but these declines could be more than compensated by an increase in foreign deposits. As the failing banks were mostly small and their foreign deposits negligible, it is not surprising that the currency situation remained stable, and that the Reichsbank's reserves even increased slightly. This time, the Reichsbank did not inject liquidity into the system. Whereas bill holdings at the Reichsbank increased sharply during the two currency crises, they actually decreased in the end of 1929, apart from a seasonal spike in December (Figure 5).

Hence, this episode differs from the two other episodes in that deposit withdrawals were restricted to a relatively small part of the banking sector and that they were not accompanied by currency problems. It seems that depositors (correctly) perceived that the FAVAG scandal and the following bank failures were idiosyncratic events. They reacted by withdrawing deposits from “bad” banks and redepositing their funds at other banks that were believed to be solvent. If there was any contagion at all, it was restricted to certain localities and types of banks.⁸²

Hence, the crisis was primarily a solvency crisis bringing down insolvent banks, while confidence in the stability of the banking system as a whole and in the currency was not shattered in the fall of 1929.

Evolution of Balance Sheets

The crisis of 1930 left its marks on banks' balance sheets. Deposits at monthly reporting banks fell by 5.8 percent from their peak in June 1930 till the end of the year (Table 5). The time period between the fall 1930 crisis and the 1931 crisis is usually characterized as a phase of returning optimism and economic revival.⁸³

Yet, deposits at monthly reporting banks continued to fall after the crisis and did not recover as they had done after the spring 1929 crisis (Figure 4). The declines in foreign deposits again were particularly strong, and by March 1931, monthly reporting banks had lost 13.5 percent of their foreign deposits compared to June 1930 (Table 5). The strong inflow of foreign deposits that had been experienced in recent years had come to a halt in 1930.⁸⁴ A decomposition into different bank groups shows, however, that foreign withdrawals were directed mainly at the great branch banks. Foreign deposits at these banks declined by more than 20 percent between June 1930 and March 1931, while the decline at the great nonbranch banks amounted to a mere 2.2 percent. Foreign debt at other reporting banks—other reporting credit banks and SLGs—even increased by 18.3 percent.

With respect to domestic deposits, the strongest withdrawals were found at the great nonbranch banks (Table 5). This was mainly driven by a sharp decrease in interbank deposits, which dropped by more than 20 percent, while the respective drop at the great branch banks was below 7 percent. In addition, interbank deposits constituted a much larger share of domestic deposits at the nonbranch banks (33.2 percent in June 1930) than at the branch banks (14.5 percent). It is unlikely that the high withdrawals of domestic interbank deposits at the nonbranch banks were driven by doubts about their stability because the nonbranch banks were generally recognized to be sound.⁸⁵

Instead, the withdrawals may have been due to the scramble for liquidity at the depositing banks during the crisis of 1930. In addition, the losses of domestic deposits may have represented capital flight, which should have affected the nonbranch banks more than the branch banks due to the prevalence of large depositors at the nonbranch banks.

How can we explain the high withdrawals of foreign deposits at the great branch banks? The answer may lie in the combination of a deterioration in the great branch banks' liquidity and earnings positions and an increasingly instable political environment, which shattered investors' confidence in the Reichsbank's ability and willingness to support these banks in times of crisis.

As in earlier years, the great branch banks stood out with respect to their weak liquidity positions, but the difference had become much more pronounced (Table 6, compare Table 3). First-order liquidity had decreased at all bank groups apart from the SLGs. In contrast, the evolution of expanded cash liquidity and the shares of bills were very heterogeneous. Most dramatic was the situation at Danatbank and Dresdner Bank where expanded cash liquidity had dropped sharply, while the shares of bills had soared. Instead of increasing liquidity in light of the apparent volatility of their deposits, these banks had allowed their liquidity ratios to decline even more. In contrast, Deutsche Bank had been able to stabilize its expanded cash liquidity (but at a low level) and had somewhat reduced its share of bills. BHG experienced a reduction in its expanded cash liquidity and an increase in the share of bills, but both ratios were still relatively strong. The best performance could be found at RKG. Overall, the liquidity positions of the great nonbranch banks, and also at the other reporting credit banks, were much stronger than at the great branch banks. In November 1930 the great branch banks' cash plus deposits at other banks covered on average less than 10 percent of their short-term liabilities. Almost 70 percent of their liquidity consisted of bills, indicating that they depended more than before on the implicit liquidity guarantee by the Reichsbank.

The long downturn had also impaired banks' earnings, and again the great branch banks seem to have been hit more strongly than other bank groups. Even though all great banks showed profits in their annual accounts of 1930, it was widely believed that the great branch banks had incurred losses in 1930.⁸⁶

The reduction of dividend rates for 1930 has to be taken as strong evidence for the great banks' disastrous earnings situation, for they had so far zealously pursued a policy of constant dividends. The great branch banks had incurred large losses from the decline in the stock market, which had depressed their stock portfolios and their commissions from securities transactions. Moreover, they had accumulated losses from loan defaults, many of which involved enormous sums. The great nonbranch banks were largely spared from such spectacular losses, which certainly was a consequence of their more prudent loan policies.

The banks' desperate situations also showed up in a collapse of their stock prices. It became common practice at the great branch banks to buy up their own shares when the share prices came close to their par values. The extreme case was Danatbank, which had bought up more than 50 percent of its own shares by the middle of 1931.⁸⁷

This implies that actual equity ratios were much smaller than published ones. In contrast to the great branch banks, BHG avoided large purchases of own shares by arranging a deal with the Warburg group, who agreed to buy up large quantities of BHG shares.⁸⁸

We can conclude that the great branch banks' absolute and relative liquidity and earnings positions deteriorated substantially in the years before 1931. In addition, the great branch banks decreased their equity cover by buying up huge amounts of own shares. If the implicit guarantee of the Reichsbank had still been credible, there would have been no need to withdraw deposits. However, the fall 1930 crisis had made clear that the Reichsbank may not be able to fulfill this guarantee.⁸⁹

Increasing political tensions made international support in times of crisis less likely.⁹⁰ Without such support the Reichsbank would not be able to ensure the redemption of the banks' debt, especially their foreign-currency debt. Hence, political shocks not only decreased the perceived stability of the gold standard, but also the credibility of the implicit guarantee by the Reichsbank.

Currency Crisis

The events of 1931 show a number of similarities to the earlier episodes of currency turmoil. The currency problems started in June and were triggered by a political shock, namely the reparations memorandum, which raised new fears of a German moratorium or debt default. Shortly after, the gold cover plummeted, and so did deposits (Figure 3). The deposit withdrawals again affected all bank groups, but were stronger than in earlier years (Table 7).⁹⁵

High gold outflows indicate that a large share of the withdrawn funds was transferred abroad, but there also was an increase in currency circulation from June on, suggesting a flight into German cash. Part of the increased currency circulation probably reflected capital flight as there were numerous reports about people crossing German borders with suitcases stuffed with bank notes.⁹⁶

As before, the great banks experienced the highest withdrawals, but this time even the savings banks experienced deposit losses. Moreover, the decrease in foreign deposits was particularly strong: From March till June, total foreign deposits of monthly reporting banks fell by 21.2 percent, foreign cash loans even by 32.3 percent; domestic nonbank deposits showed only a mild decline in June, and collapsed no earlier than July (Table 8). According to archival material from the Reichsbank, German capital flight started no earlier than in the middle of June.⁹⁷

Thus, greater mobility of deposits again translated into higher withdrawals.

These observations suggest the following interpretation: As in earlier periods of political uncertainty, foreign and, to a lesser extent, domestic investors started to move their money out of Germany, which showed up in reserve losses at the Reichsbank. The run on the Reichsmark translated into a run on German banks, with the causality running from the currency crisis to the banking crisis. This, in fact, is the preferred interpretation of the 1931 crisis in basically all recent studies on the crisis.

However, there are a number of differences between the 1931 crisis and the earlier episodes, suggesting that this is not the whole story. One important difference is that there was a strong heterogeneity of deposit withdrawals across banks. Moreover, the crisis did not end when the political crisis was settled. Instead, the currency events were increasingly superimposed by problems in the banking sector. Another difference is that the German government apparently had exhausted its credit capacity, showing up in a refusal of further international loans. International loans had played a major role in calming down financial markets in the earlier episodes. These differences may explain why the outcome of the 1931 crisis was so much worse than in the other crises.

Banking Crisis

When two exceptional events happen in short succession, it is tempting to conclude that the event that happened first triggered the event that followed after. Therefore, it has long been uncontroversial that the German crisis was triggered by the Austrian banking crisis in May 1931.⁹⁸

However, one should distinguish carefully between the withdrawals of deposits at one bank (“bank run”) and the simultaneous withdrawals of deposits at many banks that together constitute a significant part of the banking sector (“banking panic”). The banking panic clearly did not start in May 1931 as the immediate repercussions of the breakdown of Österreichische Creditanstalt in Germany were small: Total deposits started to decrease in May, but the decline was modest and the deposit-to-currency ratio was still rising. The strongest effect was on domestic interbank deposits, which dropped sharply in May, indicating that the Austrian crisis led to an early scramble for liquidity at German banks (Figure 6).

The evolution of deposits at different banks and bank groups was very heterogeneous. Although some banks like Danatbank and Dresdner Bank had already lost fairly high shares of their deposits in May, BHG and RKG were able to increase their deposits (Table 7).⁹⁹

This suggests that contagion from Austria concentrated on those banks that were known to have similar kinds of weaknesses as the Creditanstalt, namely bad industry loans, low liquidity, depreciated stock portfolios, and low equity ratios.¹⁰⁰ The decrease in aggregate deposits at monthly reporting banks in May implies that the withdrawn funds were not fully re-deposited at German banks. There also was no flight into cash in May. Therefore, part of the withdrawn funds seems to have left the country, putting pressure on the Reichsbank's gold reserves.¹⁰¹

Deposit withdrawals accelerated sharply in June. This acceleration was driven almost entirely by foreign deposits at the great banks, which decreased by 21.9 percent in the month of June. As discussed, the acceleration of withdrawals presumably was related to the concurrent political and currency problems. The largest part of domestic withdrawals seems to have taken place no earlier than July, again indicating the slower mobility of these deposits. However, the heterogeneity of deposit withdrawals persisted. Table 8 shows that domestic nonbank deposits at BHG were rising sharply in June, while the total of domestic deposits at the great banks remained roughly constant in June. The steep decline in domestic deposits in July seems to have been common to all great banks. However, the level of domestic deposits at the great nonbranch banks at the end of July was almost 10 percent above the level of March 1931 and not much below the level obtained in June 1930. At the great branch banks, the level of domestic nonbank deposits at the end of July was 18.7 percent below the level of March 1931, and even 24.4 percent below the level of June 1930. Thus, the total amount of domestic withdrawals was much larger at the great branch banks than at the large nonbranch banks. The strongest withdrawals were again found at Danatbank and Dresdner Bank who lost 32.5 percent and 26.1 percent of their domestic nonbank deposits between March and July 1931, which has to be attributed to their involvement in the Nordwolle scandal. As in the fall of 1929, domestic depositors distinguished clearly between banks they considered to be risky, and banks they considered to be safe. The same is true for foreign deposits, even though the difference in withdrawals between the great branch and nonbranch banks amounts to only 6 percentage points as of 10 July 1931, relative to March 1931. However, we have seen that the discrimination against the great branch banks by foreign depositors had begun much earlier. Between June 1930 and 10 July 1931, foreign deposits declined by 43.5 percent at the great branch banks, and only by 24.4 percent at the great nonbranch banks. This indicates that both domestic and foreign depositors had lost their confidence in the safety of their deposits at the great branch banks. These observations underline the protracted nature of the banking problems: Instead of a sudden panic, there was a gradual shift away from the great branch banks, which started long before the final panic. Therefore, the crisis looks very differently from an American-style banking panic, as the ones observed regularly in the nineteenth century. However, this is not sufficient to dismiss the claim made in this article that weaknesses in the banking sector contributed significantly to the outbreak and the severity of the crisis.¹⁰²

The high deposit withdrawals put a lot of pressure on the banks' liquidity positions. As most advances were frozen, banks had to reduce their cash holdings and their interbank claims in order to meet deposit withdrawals.¹⁰³

This explains the strong reduction in domestic interbank deposits from May on (Figure 6). To ease the strain on banks' balance sheets, the Reichsbank discounted huge amounts of bills: Between May and July, bills at the Reichsbank increased by 1.5 billion Reichsmark (+82.7 percent) with bills at monthly reporting banks decreasing by approximately the same amount (Figure 5). Owing to the discount restriction, this way of obtaining liquidity was, however, limited to banks in Berlin, which is impressively illustrated by Figure 7. While discount loans to the Berlin banks almost tripled between 30 May and 11 July, the extension of loans to provincial banks was strictly limited by the discount restriction. The largest part of total discount loans in Berlin (73.2 percent) had been taken out by the great branch banks, another 5.7 percent by the great nonbranch banks. The largest individual amount had been extended to Danatbank, and in the last nine days before its failure discount loans to Danatbank still increased by 25.7 percent.

Remarkably, the Reichsbank also discounted bills that were generally not acceptable for discount¹⁰⁴

: Banks converted illiquid advances into acceptance loans by allowing their customers to draw a bill of exchange on them, discounting the bill, and exchanging the bill with another bank to obtain the required third signature. Then the bill, which clearly was a pure financial bill and, hence, not discountable at the Reichsbank, was discounted at the Reichsbank. This practice is reflected in the sharp increase in banks' acceptance liabilities from June on (Figure 6). The strongest increase could again be found at Danatbank where acceptance liabilities jumped up by 70 percent between March and July 1931. This makes clear that banks had run out of discountable bills long before the Reichsbank's denial to discount further bills in July. The proceeds from discount loans frequently were immediately exchanged into foreign currency. The largest part of the foreign-currency demand at the Reichsbank between 1 June and 10 July again stemmed from the great banks (77.7 percent).¹⁰⁵

These events show some similarities with the earlier episode of banking turbulence: Again there were certain banks which had run into trouble due to risky business policies, and again depositors seem to have distinguished between these banks and the remaining banking system. However, the protagonists in the 1931 crisis were very different from the ones in the FAVAG crisis. This time, the problem banks were no small, insignificant banks, but two of the major players in the banking system, Danatbank and Dresdner Bank, who moreover held huge amounts of foreign debt. Instead of letting these insolvent banks fail, the Reichsbank exhausted its reserves by supporting them. In fact, the Reichsbank explicitly violated Bagehot's rules by discounting poor quality bills on a large scale.¹⁰⁶

The Reichsbank even discriminated against other (potentially solvent) banks to the advantage of the great (partly insolvent) banks in applying discount restrictions selectively, neglecting that these restrictions became completely useless because the banks with the highest demand for foreign currency were excluded. When the currency problems came to a head, the Reichsbank could not keep up its liquidity support to the great banks, which triggered the breakdown of Danatbank.

The banking panic started only after the breakdown of Danatbank. Karl Erich Born reports that “as soon as the close-down of Danatbank became known, a run on banks and savings banks by the domestic public set in.”¹⁰⁷

High withdrawals forced banks to restrict, and later suspend payments. Foreign creditors also seem to have joined the run, but the withdrawals were small compared to the preceding weeks (Table 8). The question why the breakdown of Danatbank triggered the banking panic can only be answered in connection with the currency problems.

Analysis of the Twin Crisis

The preceding analysis has shown that there were two concurrent threads of actions, whose origins were largely independent. Political shocks provoked capital outflows and deposit withdrawals at many banks, and the Nordwolle scandal led to withdrawals at the main creditor banks of Nordwolle and to further capital outflows. However, the two strands became increasingly intertwined over the course of the crisis, and it is this interweavement, which makes the 1931 crisis so interesting for economic study.

At first glance, it seems that the banking panic of 13 July was triggered by the breakdown of Danatbank. The literature on contagion in banking distinguishes two channels of how individual bank failures can spread to the whole banking system, the first operating through information, the second through interbank liabilities. The first channel is based on the idea that the failure of one bank contains information on the value of assets at other banks.¹⁰⁸

This channel suggests that contagion should primarily affect banks that are considered to be similar to the failing banks. The second channel refers to the domino effect that can arise if banks are linked to each other through interbank liabilities.¹⁰⁹ In the FAVAG crisis, both channels seem to have been present. The extent of contagion was, however, limited because most banks were not considered to be similar to the failing banks and because the level of interbank liabilities was rather small. The same arguments are valid in the 1931 crisis: The breakdown of Danatbank had been caused by a bank-specific event—the Nordwolle default—and by an imprudent loan policy that was not representative for the banking sector as a whole, and the amount of domestic interbank liabilities was relatively small.¹¹⁰ Therefore, it is far from clear why the failure of Danatbank should induce a general banking panic.

Instead, it seems that the banking panic and the breakdown of Danatbank were triggered by the tightening of the Reichsbank's discount restrictions on 10 July, which had become necessary because the Reichsbank had exhausted its “free” reserves. The strict implementation of discount restrictions meant that the Reichsbank withdrew its support from the banking system and that it would not serve as a lender of last resort, even if a bank was able to present bills that were in principle eligible for discount. Thus, the liquidity risk of all banks—solvent or insolvent—jumped up, which triggered the general banking panic.

The proponents of the currency-type explanations of the German crisis blame the political disturbances in the beginning of June for the exhaustion of reserves at the Reichsbank. According to this view, the crisis was not caused by contagion from Danatbank, but by a macroeconomic shock that hit all banks at the same time. Indisputably, this explanation has some truth to it. However, this explanation is incomplete for two reasons. First, it neglects that part of the banking problems was unrelated to the currency problems. The comparison of different bank groups suggests that an important part of the withdrawals was triggered by doubts about the great branch banks' conditions. Hence, the great branch banks' risky business policies contributed to the exhaustion of the Reichsbank's reserves over and above the effect that the political shock had had. Second, the currency crisis explanation takes the vulnerability of German banks to a run on the currency as given instead of viewing it as the result of the banks', and especially of the great banks', decision to attract huge amounts of short-term foreign debt. Some banks, such as BHG, had taken precautions against sudden outflows of “hot” foreign money by holding comparatively high levels of liquidity and, in fact, managed to stay liquid in spite of heavy deposit withdrawals. At the great branch banks, however, high levels of short-term foreign debt went along with extremely low liquidity ratios, which made their situations particularly precarious. Therefore, it is not adequate to picture the great branch banks as mere victims of a macroeconomic shock.

Despite their particularly risky business policies, the great branch banks enjoyed preferential treatment by the Reichsbank. I have argued that the great banks were subject to a moral-hazard problem because they were “too big to fail.” Already in earlier crises, these banks had been granted privileged access to liquidity. But in the crisis of 1931, the Reichsbank went farther than that: It provided liquidity to insolvent banks, such as Danatbank, thereby endangering its ability to serve as a lender of last resort to solvent, but illiquid banks. Even when Danatbank was on the brink of collapse, the Reichsbank kept on discounting its bills.

By absorbing the Reichsbank's “liquidity” (foreign currency), the great branch banks exerted an external effect on all other banks. This effect was still exacerbated by the reluctance of the great branch banks to draw on their accounts held abroad to obtain foreign currency. The great banks preferred purchasing foreign exchange on credit at the Reichsbank to using up their own foreign-currency reserves. In the Reichsbank's records this shows up in the frequently made complaint that the great branch banks had used the Reichsbank not as a lender of “last,” but as a lender of “first” resort.¹¹¹

The preferential treatment of the great branch banks continued even after the outbreak of the banking panic, as each of them received large capital injections.

The Reichsbank's policy had its costs. It distorted competition to the advantage of the great branch banks, which furthered the crowding out of regional and local credit banks and of private bankers. In the crisis, it forced solvent, but illiquid banks for which the Reichsbank's discount window was closed to liquidate assets at high costs, and possibly drove some of them into insolvency. The Reichsbank itself wrote in October 1931 that “it can easily happen that healthy parts of the economy are demolished because its weakest parts … are protected by special governmental measures for the sake of their general economic significance; this is, for example, also the case with Danatbank and Dresdner Bank.”¹¹²

After the crisis, the relative positions of smaller banks were weakened even further by the capital injections into the great branch banks. In addition, the Reichsbank's policy provided incentives for the misallocation of funds at the great branch banks, which showed up in excessively risky loan policies and insufficient levels of liquidity. It probably even induced the great branch banks to assume particularly high levels of short-term foreign debt, thereby increasing the exposure of the German economy to the danger of short-term capital reversals. It, hence, seems that the policy of the Reichsbank came at a high cost, and that a system with implicit guarantees can be subject to much the same moral-hazard problems as a system with explicit deposit insurance.

The final question is why the German crisis ended in a general banking panic and the abandonment of the gold standard. This stands in stark contrast to the literature that predicts that there is either a currency or a banking crisis, depending on whether the central bank acts as a lender of last resort or not. However, the scope for lender-of-last-resort activities was limited because a large proportion of banks' deposits was denominated in foreign currency. Moreover, deposit withdrawals and reserve losses reinforced each other in a kind of vicious circle due to a coincidence of capital flows, which is depicted in Figure 8: Deposit withdrawals from weak banks translated into reserve losses and threatened the stability of the currency, while capital outflows induced by the currency disturbances implied deposit withdrawals and threatened the stability of banks. In turn, both resulted in further capital outflows and deposit withdrawals.

This vicious circle meant that in July 1931 the Reichsbank faced a dilemma: It could either continue to provide liquidity to the banking system risking a deviation from the gold cover prescribed by law or it could—as it actually did—curtail lending to banks, thereby risking the breakdown of the banking system. However, none of these measures were suited to putting an end to deposit withdrawals and reserves losses. In the absence of capital controls, further provision of liquidity would have led to an exhaustion of the Reichsbank's gold reserves that covered banks' short-term foreign debt by less than one third in the middle of July.¹¹³

This would have triggered a run on the Reichsbank's reserves and further deposit withdrawals at German banks. Discount restrictions, on the other hand, endangered the stability of the banking system directly, inducing further deposit withdrawals and, hence, reserve losses. Only an early abandonment of the gold standard coupled with capital controls could possibly have prevented the breakdown of the banking system. However, the Reichsbank seems to have hoped until the very end that foreign support would help to prevent both the abandonment of the gold standard and the banking crisis. The prevailing political tensions made such financial support unobtainable.

Hence, the German experience suggests that with high levels of foreign and foreign-currency debt in the banking system, a central bank cannot choose between the stability of the banking system and the currency because of its limited ability to serve as a lender of last resort and because of the mutual reinforcement of deposit withdrawals and reserve losses. In the absence of capital controls, only an “international lender of last resort” could have prevented the German collapse.