It is lovely to see attention to this critical episode in the Great Depression. Isabel Schnabel repeats the familiar argument that bank behavior was a primary cause of the German crisis. We have argued elsewhere that the German crisis clearly began as a currency crisis, not a banking crisis.¹

The purpose of this comment is to examine the data that Schnabel's article brings to bear on this question and evaluate the evidence it presents in support of its hypotheses.

It should be noted that although Schnabel indicates in her title that the crisis was a twin crisis, in which problems of the mark loomed as large as problems with the banks, her article concentrates almost entirely on the banks. The article states early on that, “[T]he German crisis had two independent causes.” A page or so later, however, it says, “The most important data sources used in this study are the balance sheets of German banks before and during the crisis of 1931.” By the end of the article, Schnabel's claim is unequivocal:

Schnabel continues, “it is quite possible … that the crisis of 1931 would not have occurred if the banks had acted with caution in the 1920s.” It may have been a twin crisis according to Schnabel, but it was caused by incautious bank behavior.

Schnabel makes two claims in support of this view. First, the large German banks indulged in behavior indicating moral hazard because they were “too big to fail.” Second, the Reichsbank did not act as a typical gold-standard central bank.

The article advances two lines of evidence to show that the great banks in Berlin were acting under the influence of moral hazard. First, they had a large share of bills of exchange in their liquid “means” in November 1929 (Schnabel's table 3). A neighboring footnote, however, adds the apparently innocent qualification, “that in 1931 the shares of bills of exchange in liquid assets were still well below the ones obtained in 1913.” The data on bills as a share of liquid means in November 1929 in Schnabel's table 3 are updated to November 1930 in her table 6. This mode of presentation makes comparison difficult, but it is hard to see any change. The share of bills for monthly reporting banks went from 56.7 percent to 54.6 percent. The share for the great branch banks went from 66.1 percent to 68.9 percent. There is no evidence of a deterioration of bank balance sheets in these data even though they come from the period when Nazi electoral success and German budget deficits were roiling financial markets.

Second, equity ratios for the Berlin great banks were below those of other banks (Schnabel's table 4). This is not a new observation: both at the time and later, many commentators drew attention to the decline in the equity ratios of the great banks since the stabilization of 1924. Contemporary opinion, as the following comment from The Economist suggests, did not regard lower equity as a smoking gun that banks were more prone to take on risks. Indeed, on the very eve of the crisis, it was even hailed as part of a package that would lower overall risk:

Finally, Schnabel argues that the Reichsbank was not a typical gold-standard central bank. The evidence? “Instead of letting insolvent banks such as Danat bank fail, the Reichsbank exhausted its reserves by supporting them.” Schnabel raises this point to a critical one in her discussion of whether the crisis could have been avoided. She notes the Reichsbank's limited options in the crisis with this caveat: “However, it is well possible that an early discontinuation of support to insolvent banks such as Danat bank would have prevented the outbreak of the crisis.”

There is no evidence that the Reichsbank “exhausted its reserves” supporting the Danat bank. This can be seen clearly in Schnabel's figure 3, where Reichsbank reserves are shown to be increasing in April and May. In fact, the Reichsbank did not lose reserves until the first week of June, as Brüning geared up for his famous denunciation of reparations.³

Neither Schnabel nor anyone else has produced any evidence suggesting that the Reichsbank was supporting the Danat bank before June or that it was putting itself in danger by so doing.

There also is no evidence that undisclosed problems at the Danat bank threatened the Reichsbank before the currency crisis of early June. As shown in Schnabel's table 7, the Danat bank did not lose many deposits before June. On 13 June the Reichsbank raised its discount rate by two percentage points, a very strong reaction to the pressure on the currency. On 17 June news of the Nordwolle failure leaked. It is possible that these two events were related, that the increase in the discount rate increased the cost of Nordwolle's inventories, leading it to collapse. If so, the Danat bank, although possibly unwise to have loaned so much to Nordwolle, might not have been in trouble, or as much trouble, if there had not been a currency problem.

By June, when the Danat bank's trouble became public, the decisive factor was the pressure the currency crisis placed on the Reichsbank. The Reichsbank attempted to help the Danat bank, but it was prevented from doing so by its responsibility to preserve the gold value of the mark. In Schnabel's words: “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 the Danatbank.” That is the hall-mark of a currency crisis, not a banking crisis. It reveals that the Reichsbank indeed was a typical gold-standard central bank.

Turning to narrative, Schnabel acknowledges the difficulty of disentangling currency and banking problems and argues for comparison with earlier episodes. She presents data on various currency crises of 1929–1931 in her figure 3, leading to the following conclusion: “The timing of the crises suggests that the deposit 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…. The wish to transfer money abroad translated into deposit withdrawals at German banks.” But only a few pages later, Schnabel presents data on foreign deposits in her table 5 and asks, “How can we explain the high withdrawals of foreign deposits of foreign deposits at the great branch banks?” She suggests that the answer may lie in bank behavior, apparently forgetting the explanation she gave for deposit withdrawal a few pages earlier (and just quoted here). If foreign investors were fearful about the mark, not the banks, they would reduce their exposure in marks by reducing their deposits in German banks. So of course they would withdraw from the great banks, because, to paraphrase a famous American bank robber's explanation for why he practiced his trade, that is where the foreign deposits were. In the context of the budget catastrophe, paramilitary stirrings, the change in the international climate that followed the hare-brained German scheme for a customs union with Austria, and the German government's denunciation of reparations, invoking banking problems is superfluous. The evidence cited in this note does not indicate that they were material.

The two crises shown in the data therefore were identified as currency crises. Schnabel adds a third “crisis,” the FAVAG crisis of August 1929. This was a business failure, and the article reveals that the failure resulted only in a few bank failures, not any general banking crisis: “[T]he crisis can hardly be detected in the aggregate deposit series (figures 3 and 4).” By using this case as an analogy, Schnabel implicitly compares the failure of the Nordwolle in 1931 with the failure of FAVAG in 1929. But the latter failure did not result in a generalized banking crisis; the analogy does not explain why the former should have done so.

Schnabel also points to the heterogeneity of deposits at German banks as evidence of the importance of bank behavior in the 1931 crisis. The primary evidence appears to be that “some banks such as Danatbank and Dresdner Bank already lost fairly high shares of their deposits in May.” The source for this observation is Schnabel's table 7, in which it is clear that deposits in both of these banks remained above 95 percent of their size in March 1931 at the reporting date of 31 May 1931. Schnabel's claim in the text is contradicted directly by the cited table in the article; the prose in the article at this point shows the triumph of ideology over data.

Deposits in the Dresdner Bank did not fall more than the average of the great banks even in June. Deposits in the Danat bank did fall, but the Nordwolle failed in that month. The paper provides no evidence of any banking runs in May. As Ferguson and Temin showed, banks were not calling in loans, there was no run on demand deposits, and only time deposits were decreasing.⁴

Acceptances, including many of foreigners, actually were rising. Table 1 demonstrates this observation with simplified data from the same monthly bank reports used by Schnabel. “Demand deposits” (DD) were accessible in a week or less, “time deposits” (TD) were accessible in a week to three months. Demand deposits did not fall at all in the crisis, indicating that there was no panic among depositors at the great banks. Time deposits fell only in June; depositors were repositioning their assets in anticipation of possible currency problems.

Schnabel's data confirm this conclusion. As shown in her table 8, the data do not provide evidence of banking problems or moral hazard. At the end of May foreign deposits at Berlin's great banks were at 95 percent of their level at the end of March. Domestic deposits were at 99 percent of their March level. Domestic depositors do not seem to have been frightened; only foreigners appear to have made a small adjustment. The differential decline indicates fear of the currency rather than fear of banks. The first, relatively faint signs of trouble with the currency appeared toward the end of May, as word began leaking out that the Brüning government was contemplating a move on reparations as part of a broader budgetary retrenchment. In the first week of June, as the rumors turned into fact, the currency's problems snowballed. By the end of June, foreign deposits had fallen to 75 percent, while domestic deposits stayed at 99 percent. Even with the developing problems of the Danat bank, domestic deposits in the great banks did not fall! There was no crisis of confidence in the great banks. Only foreigners, reducing their mark assets for fear of devaluation or currency controls, drew down their deposits. This of course is the story in Ferguson and Temin's work.⁵

Schnabel's article confirms that story with additional data; it does not add a believable banking crisis to it.

In the interests of furthering this debate, we present a small amount of new evidence in Figure 1. This graph shows the daily price of Young Plan bonds in Paris from 1 April to 30 June. Young Plan bonds were traded widely, but we found the most complete series for Paris. It provides a good index of investor sentiment in the spring and summer of 1931. After rallying early in the year, the bond price stayed remarkably constant from March to May, and then fell sharply during the week of 27 May. Why? There was no news about German banks in late May, but German newspapers began by 25 May to discuss the rumor that Brüning was likely to ask for some sort of relief in regard to reparations, as he did in June.⁶

This, not phantom withdrawals from banks, was the beginning of the fatal run on the currency that paralyzed the Reichsbank precisely at the moment it needed reserves to foster domestic stability.