A Margin Call Gone Wrong: Credit, Stock Prices, And Germany’s Black Friday 1927
Board of Governors of the Federal Reserve
2015-02-15
FEDS Working Paper No. FEDGFE2015-54
Abstract:
Leverage is often seen as villain in financial crises. Sudden deleveraging may lead to fire sales and price pressure when asset demand is downward-sloping. This paper looks at the effects of changes in leverage on asset prices. It provides a historical case study where a large, well-identified shock to margin credit disrupted the German stock market. In May 1927, the German central bank forced banks to cut margin lending to their clients. However, this shock affected banks differentially; the magnitude of credit change differed across banks. Using the strong connections between banks and firms in interwar Germany, I show in a difference-in-differences framework that stocks affiliated with affected banks decreased over 12 percent during 4 weeks. Volatility of these stocks doubled. Relating directly bank balance sheet information to asset prices, this paper finds that a one standard deviation decrease in lending to investors increased an affected stock’s volatility by 0.2 2 standard deviations. These results are robust to the problem that banks’ lending decisions may be influenced by asset prices. The Reichsbank threatened banks to cut their short-run funding. Using the differences in exposure towards this threat, an instrumental variable strategy provides further evidence that a sharp decrease in leverage may lead to stock price fluctuations.
A Margin Call Gone Wrong: Credit, Stock Prices, And Germany’s Black Friday 1927 – Introduction
Leverage is often seen as villain in financial crises. Leverage of traders and financial intermediaries is procyclical and changes in credit are correlated with asset price movements (Adrian and Shin 2010). Further, leverage growth predicts excess returns in several asset classes (Adrian, Moench, and Shin 2010). In the recent crisis, sudden decreases in lending by financial intermediaries are regarded as one of the main culprits for fire sales and sharp increases in stock market volatility during 2008-2009 (Brunnermeier 2009). In frictionless markets, changes in lending to investors have no asset pricing implications. However, recent theories establish a direct link between changes in margin credit and stock price movements (Gromb and Vayanos 2002, Brunnermeier and Pedersen 2009). If investors cannot satisfy margin calls, a decrease in leverage may lead to fire sales. With downward-sloping asset demand, this induces price dislocations, which may reinforce further deleveraging.
There is growing support for the hypothesis that traders’ and intermediaries’ balance sheet conditions matter for asset pricing. Several empirical studies show a relationship between changes in margin credit and asset price movements. Broker-dealer leverage is a significant pricing factor for stock returns (Adrian, Etula, and Muir 2011). Changes in leverage of financial intermediaries are strongly correlated with stock market risk (Adrian and Shin 2010). However, it has proven difficult to establish a direct link between changes in credit and asset price movements. Leverage rarely varies exogenously; changes in credit are mostly endogenous decisions by financial intermediaries, other financiers, or investors. Further, when balance sheets are marked-to-market, changes in asset prices directly affect leverage.
In this paper I examine the asset pricing consequences of a large shock to financial intermediaries’ margin lending. In May 1927, the German central bank forced banks to decrease credit for stock purchases given out to their clients. However, this shock affected only a subset of banks. Clients of the affected banks mostly had to unwind their positions to meet the margin calls. Using a particular bias in their portfolios, I am able to differentiate between firms mostly held by clients of affected banks or unaffected banks. I show in a difference-in-differences framework that deleveraging had large asset pricing implications. During the weeks following the shock on margin credit, stocks connected to affected banks declined 50 percent more than other stocks. They experienced negative cumulative returns of almost 400 percent (annualized). Further, return volatility of these stocks doubled. Connecting stocks directly with specific intermediaries and changes in their balance sheets, I find that mean daily returns were not affected by the cut in margin lending. However, a one standard deviation decrease in margin credit increased a stock’s volatility by 0.22 standard deviations.
The historical setting is interwar Germany. During the mid-1920’s, increasing stock market valuations went in lockstep with an increase in margin credit. Yet it were mainly the six large “Berlin banks” that enabled their clients to buy assets on credit. Margins could be as low as 10 percent. However, the rise of stock prices and margin credit drew the attention of the German central bank, the Reichsbank. Mostly for political reasons, its president, Hjalmar Schacht, campaigned against the banks’ practices to constantly increase credit supply and to allow highly leveraged positions. This campaign culminated in the threat of the Reichsbank to cut short term funding for the Berlin banks. The threat was effective. On 12 May 1927, the Berlin banks issued a joint statement. Over the course of the following weeks, each bank would decrease their stock of margin credit by 25 percent while issuing margin calls towards their clients. The consequences were immediate – 13 May 1927 became known as Black Friday. The stock market declined by 13 percent. The large shock initiated by the Reichsbank trickled down to investors. During the following weeks, banks increased the margins of their clients’ portfolios. As most investors could not satisfy the margin calls, fire sales occurred. However, at a given bank these fire sales were concentrated on a special set of firms – firms that had a close relationship to the bank. Using evidence from the German Federal Archives to establish these bank-firm connections, I show that stocks differed in their reaction to the deleveraging: A firm’s stock price declined stronger and fluctuated more if the firm was connected to a bank that experienced a larger credit crunch.
Figure 1 summarizes the main result. The left panel shows two stock price indices, one index composed of firms connected to the large Berlin banks, and another index of firms with no connection to these banks. Both indices are normalized to 12 May 1927. At this day, the Berlin banks issued their joint statement and both indices declined. However, over the course of the following month “large bank”-stocks declined more than 12 percent, while other stocks dropped less than 8 percent. Further, volatility almost doubled for firms connected to the Berlin banks. This is shown in the right panel, which plots volatility indices for both groups of firms. Stocks more exposed to fire sales had more negative returns and experienced larger fluctuations. This causal impact of deleveraging on return volatility is robust to several criticisms. The result still holds when controlling for attributes such as the number of underwriters or firm size. An instrumental variable strategy further shows that reverse causality does not drive the results.
See full PDF below.