Chart 5.3 shows,however, that total loan volume declined with GNP. Predicted loans are estimated from a simple regression in which loans depend only on nominal GNP. The decline in loans differed little from the decline in GNP. It seems fanciful to suggest that the decline in loans caused an immediate decline in GNP in each period. The more likely explanation is that households and businesses reduced borrowing as their incomes fell. Falling demand explains most of the decline in loans. Given the real return to lending, banks should have been eager to lend to solvent borrowers.
Haubrich (1990) tested Bernanke’s nonneutrality hypothesis for Canada. Canada had no bank failures, but banks closed many of their branches, probably disrupting lending arrangements. Canada also had a smaller share of interest payments on loans past due, but Canadian commercial failures were a larger fraction of GNP. Canada was effectively off the gold standard after January 1929. Haubrich finds no support for bank closings and commercial failures in Canada as a reason for decline.
The United States was not the only country experiencing bank runs and failures. The public in many countries expressed its fear of bank failures by withdrawing deposits and holding currency. Spain and the Netherlands experienced large increases in the ratio of currency to deposits. Austria, France, Norway, and Canada had smaller, but not negligible, increases. A central bank can offset the effect on the economy of an increased demand for currency by expanding the monetary base by more than the increase in currency demand. Under the gold standard, the government may have to temporarily suspend convertibility, as Britain did in such circumstances several times in the nineteenth century.
To study the effect of currency drains (or bank runs) on real income during the decline, 1929–32, I regressed percentage changes in the real value of the monetary base, the ratio of currency to deposits, and base velocity on the percentage change in real income for twenty countries in Europe, North America, and South America. The twenty countries for which data are available report very different experiences. Denmark, Greece, Hungary, Norway, and Spain show a rise in real income for the period as a whole. The United States, Canada, Brazil, Mexico, Austria, France, and Germany show double-digit declines in real income for the three-year period. Estimates and a list of the countries are given in appendix B.
The data attribute about half of the decline in real income in these countries to the combined effects of the change in the real value of the country’s monetary base and the change in base velocity. The change in base velocity includes changes in the demand for currency per unit of income. The larger the currency drain, other things equal, the larger the decline in base velocity. Currency drains and bank runs do not appear to have had any significant effects through a lending channel or other channel, independent of their effects on the real value of the monetary base and base velocity. The statistical results again suggest that monetary factors had an important role in the decline, but other factors affecting the demand for money were also significant for the twenty countries. I return to the role of gold in the concluding section.
Federal Reserve records suggest that the real bills or Riefler-Burgess doctrine is the main reason for the Federal Reserve’s response, or lack of response, to the depression. With few exceptions, the Federal Reserve governors accepted this framework as a guide to decisions. They believed that a low level of member bank borrowing and low nominal interest rates suggested there was no reason to make additional purchases. Additional purchases of government securities would expand credit based on speculative assets, which was inconsistent with the real bills doctrine and the gold standard.
Federal Reserve purchases at the start of the 1929 recession were much larger than at the start of previous recessions. One reason is that New York started to purchase at the time of the stock market break. Also, member bank borrowing was over $1 billion, far above the range that the Federal Reserve regarded as restrictive. At a March 1926 meeting of the Governors Conference, Strong restated the Riefler-Burgess doctrine and described how it would be applied at the start of a recession: “Should we go into a business recession while the member banks were continuing to borrow directly $500 or $600 million (if bills are included nearly $800 million) we should consider taking steps to relieve some of the pressure which the borrowing induces by purchasing government securities and thus enabling member banks to reduce their indebtedness” (quoted in Chandler 1958, 239).
Table 5.30 compares member bank borrowing and interest rates at the beginning and one year after the start of three recessions. The 1929–33 recession started with more member bank borrowing and higher interest rates than the others. By the end of the first year, the Federal Reserve had purchased $350 million to $400 million more than in the two previous recessions. In the 1923–24 recession, the Federal Reserve made seasonal securities purchases in December, seven months after the peak, but sold in January. Sustained purchases did not begin until February 1924, nine months after the cyclical peak. In 1926–27 the System made small-scale securities purchases at once, partly for seasonal reasons, since the recession started in October. Sustained purchases began in February, four months after the cyclical peak.
The earlier recessions reached a trough after thirteen or fourteen months. Eighteen months after the peak, the levels of borrowing or borrowing plus acceptances on the Federal Reserve balance sheet were very different after 1929 than after the earlier recessions. Interest rates on ninety-day acceptances are highest in the October 1926 cycle, lowest in the August 1929 cycle. Securities purchased show the same ordering. Judged by the measures that Riefler, Burgess, and Strong emphasized, Federal Reserve policy shifted from restraint to ease and back to restraint in the 1923–24 and 1926–27 cycles. In 1929–30 these measures indicated that credit conditions had eased substantially.
The monetary base shows a different pattern, largely unrelated to the Federal Reserve’s measures of credit conditions. In 1923–24 the base started to rise six months after the cyclical peak and continued to rise through the first six months of the recovery. Three months after the 1929 peak, the base was below the level reached at the August peak, and it continued to fall through the spring and summer of 1930, as shown in tables 5.5 to 5.8 above.150
150. In 1926–27, the base changed very little in the recession and in the early months of recovery.
CONCLUSION
People see most clearly what they are trained or disposed to see. The Riefler-Burgess version of the real bills doctrine was not a mechanical formula directing Federal Reserve policy, but it directed attention to member bank borrowing and market interest rates as measures of tightness and ease. In 1929–30, most members of the Federal Reserve Board and governors of the reserve banks accepted this framework. They believed they had acted decisively to ease credit conditions, and on their measures they had.
The real bills doctrine taught that central bank credit should finance self-liquidating commercial loans. Government paper, stock market loans, and real estate mortgages were “speculative” investments that had no place on a central bank’s balance sheet. Since speculative loans were not self-liquidating, they were considered inflationary finance.
To a modern reader, fear of inflation seems a strange concern after a year or more of falling prices. Yet there are surprisingly few proposals to restore the price level. The comments of W. Randolph Burgess, occasional comments by Governor Meyer or some New York directors, and the efforts of a few members of Congress are the only official comments to that effect that I have found.
Eichengreen (1992, 251–53) contrasts the Riefler-Burgess emphasis on borrowing and interest rates with the “liquidationist view,” so called because Treasury Secretary Mellon is said to have advised President Hoover to “liquidate labor, liquidate stocks, liquidate the farmers, liquidate real estate” (251). Mellon’s advice is entirely consistent with the real bills doctrine and the firm belief that Federal Reserve policy had financed speculative lending; its effects had to be purged (liquidated). An increased demand for borrowing to finance
real bills would, on this view, show that liquidation was complete and that recovery could occur without inflation. That is why the most extreme proponents of the real bills doctrine—Governors McDougal, Norris, and Young—typically opposed purchases. These men, and many others, repeatedly referred to the contraction as “inevitable”—the inevitable consequence of providing speculative credit.151 In 1929–33 their principles told them that deflation was both necessary and inevitable. For much the same reason, the Federal Reserve deflated in 1920–21 to eliminate the credit expansion based on war finance.
The volume of loans on securities by banks in New York and in the rest of the country did not increase disproportionately during the stock market boom. The main evidence of expanding stock market lending is the relatively large increase in loans to brokers and dealers, not to the public. At its peak, in the week ending October 2, 1929, total lending of this kind was $6.8 billion; the volume had nearly doubled since the end of 1927. Table 5.31 shows these data.
151. Senator Carter Glass held this view firmly. In hearings before his subcommittee, he attributed the financial collapse to neglect of the real bills principles (Senate Committee on Banking and Currency 1931).
Eichengreen and Bernanke correctly emphasize the transmission of deflationary impulses by the gold exchange standard. Falling gold stocks in many countries reduced the monetary base in countries that lost gold. Appendix B recognizes the importance of the decline in the real stock of base money as a factor reducing real income, but it also recognizes that the gold did not disappear. Some countries, including the United States and several members of the gold bloc, acquired gold reserves.
The United States experienced a gold inflow in the first year of the decline. Under gold standard rules, the increase in gold should have increased the monetary base. If the Federal Reserve had followed the rules, the money stock would have expanded by 14.6 percent from August 1929 to June 1930. This, of course, overstates the amount of gold inflow that would have occurred. However, an expansive monetary policy would have prevented at least some of the deflation and recession, so falling prices and fears of collapse would have been absent. The world would have been spared much of what followed.
The principles of the Federal Reserve Act called for passive policies. The founders intended the System mainly to respond to gold movements and offers of real bills. No one discussed what the System should do if the two signals gave conflicting commands, as in 1930 when gold flowed in and real bills declined. The Federal Reserve had abandoned strict adherence to the gold standard in World War I and in the 1920s. It followed the real bills guide. Policy was deflationary in 1930 when adherence to gold standard rules called for expansion.152
Eichengreen (1992, table 8.6) compared the behavior of surplus and deficit countries from 1929 to 1931. He showed that in 1929 and 1930 the twenty-six countries losing gold contracted reserves as they paid out gold. Surplus countries like the United States contracted also, so the well-known stabilizing process did not work. We can only speculate on why deficit countries followed deflationary policies instead of leaving the gold standard. One reason is that most policymakers, economists, and businessmen in these countries also believed that deflation was an inevitable consequence of the previous speculative boom in the United States. The world economic system could not return to stability until these “excesses” had been purged. Also, many countries attempted to protect their gold reserves by deflating, contrary to the advice of Bagehot and Thornton.
The data suggest that the United States economy and its monetary system experienced not one but a series of monetary and nonmonetary shocks in the forty-two months following the August 1929 peak in economic activity. Seasonally adjusted industrial production declined more than 50 percent, and the money supply declined by over 25 percent. Of the banks operating at the time of the peak, more than 25 percent—6,704 banks—failed or were merged into other banks.153
One long-popular belief is that the fall in output and the financial collapse were caused by a prior decline in stock prices. The decline in money and the waves of bank failures are attributed to the decline in loan demand. To be more than an example of the post hoc, ergo propter hoc fallacy, there must be some connection between the initial decline in stock prices and the series of shocks to the United States economy.
Prices of industrial shares had declined by percentages similar to the 1929 decline in the recessions of 1906–7 and 1919–20 without producing depressions of the same length and magnitude, although the earlier declines in the stock prices had been spread over a longer time. Following the October–November decline, stock prices rose in winter 1930 and summer 1932. By April 1930 the Standard and Poor’s index was only 3 percent below April 1929. It had recovered almost 40 percent of the decline from the September 1929 peak. Stock prices rose again in the summer of 1932 following the expansive monetary policy and despite an increase in tax rates during the spring.
152. Fremling (1985) notes that the United States was not the only country that did not follow gold standard rules. As noted above, France sterilized much of its inflow. Fremling’s conclusion that the rest of the world increased its holdings of foreign reserves and gold does not separate France from other countries, many of which were forced to deflate.
153. During the entire period 1864 to 1896, there were 1,562 bank failures—328 national and 1,234 state banks. In the worst year, 1893, 326 banks—approximately 4 percent of the total—failed. During the banking panic of 1907–8 there were 172 failures; fewer than 1 percent of the banks in existence on June 30, 1907, closed in the next two years. The number of active banks increased from 16,266 in 1906 to 17,891 in 1907 and 19,620 in 1908.
The number of suspensions per one hundred active banks during the early thirties was: 5.61 in 1930; 10.48 in 1931; 7.75 in 1932; and 12.86 in 1933. Banks that suspended operations between December 1929 and March 1933 had gross deposits of $5.5 billion, approximately one-third of the decline in total deposits for the period. All data are from Upham and Lamke 1934, 245, 247, and 250.
Banking data show little evidence of a prolonged effect of the October decline in stock prices. The data in table 5.32, and similar data on bank loans or loans and investments, show that in the first two years of the contraction, demand deposits in the larger New York banks rose approximately 10 percent, while deposits in all other banks rose about 1 percent. For the decline in money to result from the fall in stock prices, New York banks would have to have experienced a large loss of deposits. In fact, New York banks increased loans and deposits absolutely and relative to other banks for more than a year following the stock market break.
The data in the table support an alternative explanation. During the first two years of the contraction, gold flows and currency movements dominated the behavior of the monetary system. Deflation in the United States and risks abroad brought gold to the United States, as in 1920–21. The gold flows supplied relatively more reserves to the New York banks than to the small or regional banks in the interior. In part for this reason, the internal currency drains and rates of bank failures were much larger in the mid-western Federal Reserve districts than in others. Approximating the regional impact of the currency drains by the changes in the note issue of the various reserve banks shows that between December 1930 and December 1931 the total note issue of the reserve banks increased approximately 60 percent while the note issue of the Federal Reserve Bank of Chicago increased 275 percent and the note issue of the Federal Reserve Bank of Atlanta declined (Board of Governors of the Federal Reserve System 1943, 338).154
On the alternative explanation, the absolute and relative increase in deposits at large New York banks was mainly the result of the Federal Reserve’s contractive policy and the gold and currency movements of the period. When the gold flow reversed and the currency drains resumed after August 1931, deposits at New York banks declined, and the experience of the New York banks was more like the experience of banks in the interior.
Stock prices are one among
many measures of asset prices. Since shares are traded on open securities markets, share prices respond promptly to changes in anticipated future earnings and dividends. Suitably deflated by current output prices, share prices offer an approximate measure of the cost of available assets relative to the production cost of new assets.
Standard and Poor’s index of stock prices deflated by the GNP deflator reached a peak in third quarter 1929. The deflated stock price index was only 23 percent below its peak in second quarter 1930. If the decline had stopped there, deflated stock prices would have been more than twice their level four years earlier. Three quarters after the cyclical peak, stock prices showed no evidence that asset owners believed a further decline was inevitable, to use the term central bankers overworked so much at the time.
The decline in deflated stock prices is not a uniform or even a unidirectional movement that would characterize transmission of a single shock. Like the data on money and GNP, the movement of stock prices suggests a sequence of shocks. In addition to the initial 24.3 percent decline in fourth quarter 1929, three other quarters show declines of 20 percent or more—fourth quarter 1930, fourth quarter 1931, and a decline of nearly 38 percent in second quarter 1932, when bank failures reached a temporary peak. The last of these shocks brought the deflated index to its lowest point of the depression, three calendar quarters before the bank holiday. The earlier chronology identifies these quarters as periods of financial stress.
A History of the Federal Reserve, Volume 1 Page 58