Open Market Operations under the 1930s Gold Exchange Standard

Given that the United States was on the gold exchange standard, it is reasonable to ask whether open market operations could have any effect at all. After all, a central lesson of open-economy macroeconomics is that in a small open economy with fixed exchange rates and perfect capital mobility, a monetary expansion that puts downward pressure on domestic interest rates relative to world rates results in an offsetting loss of foreign exchange. As a result, it ultimately has no effect on the money supply or domestic interest rates.

There are two main reasons why open market operations, such as those undertaken in 1932, could have increased the American money supply even if the United States was on a credible fixed exchange rate system. First, capital mobility was limited in the interwar period because of factors such as transportation costs and inefficiency in the banking system. David Peel and Mark Taylor show that deviations from covered interest parity between the dollar and sterling would not be arbitraged unless the deviation exceeded 50 basis points.⁶

They also find that deviations from covered interest parity were quite persistent even outside this remarkably wide neutral band. Therefore, open market operations could result in an expansion of the money supply that reduced U.S. interest rates by 50 basis points or more without triggering immediate gold flows.⁷

Second, because the United States was a large country, U.S. open market operations could affect world interest rates. As a result, the U.S. interest rate could fall by more than 50 basis points without triggering gold flows. This suggests that the possible increase in the U.S. money supply was even larger than the arbitrage findings suggest. Michael Bordo, Ehsan Choudhri, and Anna Schwartz argue that the large-country assumption is indeed appropriate for the United States in the 1930s.⁸

They present simulations that indicate that substantial open market operations in the United States would not have caused large gold outflows even under the extreme assumption of perfect capital mobility.

The 1932 Open Market Purchase Program

In 1932 the Federal Reserve decided to use its powers to increase the money supply. A consensus in favor of expansion grew gradually in early 1932. In an address on 8 January, George L. Harrison, Governor of the Federal Reserve Bank of New York, bemoaned the decline of bank credit over the last few years. He went on to argue that to arrest this decline, “the Federal Reserve Banks are in a position at the present time and on the basis of their present gold supply to expand Federal Reserve credit to their members by some 3 1/2 billion dollars.”⁹

In February, Congress passed the Glass-Steagall Act, which effected a substantial change in collateral requirements for the Federal Reserve. Under the original Federal Reserve Act, the Federal Reserve had to hold gold as backing for 40 percent of notes, and eligible (private-sector) securities for the remaining 60 percent. Because the Federal Reserve typically did not hold enough such private bills, it ended up using gold to back substantially more than the statutory minimum. The Glass-Steagall Act allowed the Federal Reserve's large holdings of government securities to be used as collateral for Federal Reserve notes, thus freeing up large quantities of gold to back increases in the money supply.

Once passage of the Glass-Steagall Act was assured, the Federal Reserve began to expand the money supply gradually. At its meeting on 24 and 25 February, the Open Market Policy Conference (hereafter OMPC), the precursor of the modern Federal Open Market Committee, authorized the purchase of $250 million in government securities at a rate of approximately $25 million per week.¹⁰

These initial open market purchases are evident in Figure 1, which shows the weekly change in Federal Reserve holdings of U.S. government securities.¹¹ We have drawn a solid line in Figure 1 at the last observation before the start of open market purchases, which is 24 February.

At its next meeting, on 12 April, the OMPC decided to dramatically accelerate the program of monetary expansion. The OMPC authorized the Executive Committee to purchase an additional $500 million in government securities and recommended “that these purchases, at least in the initial weeks, should be at a rate as rapid as may be practicable.”¹²

Because the April meeting represented a dramatic acceleration of the open market purchase program, we have drawn in another line (this one dashed) at the last observation before the acceleration, which is 6 April.

Figure 1 shows that the open market purchases continued in May and June, but at the reduced rate of approximately $50 million per week. By mid-June, the consensus in favor of aggressive expansion had begun to wane.¹³

However, open market purchases were not curtailed dramatically until early July. At its meeting on 14 July, the OMPC indicated that “except in unusual or unforeseen circumstances purchases should not exceed 15 million dollars a week.”¹⁴ By the statement week of 30 July 1932, open market purchases were effectively zero. We have drawn a solid vertical line in Figure 1 at 13 July because this was the last observation under the 1932 open market purchase program.

Was the Public Aware of the Federal Reserve's Actions?

The rapid monetary expansion could have sparked devaluation fears only if the public was aware of it. Because the Federal Reserve released balance sheet data each week and these data were presented in a number of news sources, there can be little doubt that the information was available. However, to see if the Federal Reserve's actions were likely to have entered the public's consciousness, we examine the coverage of the open market purchase program in the financial press in the United States and abroad. In particular, we examine reports in the Commercial and Financial Chronicle (abbreviated as Chronicle in what follows), the premier business paper in the United States, and in the Economist, a key international financial publication. The Chronicle is a particularly useful source because it also excerpted stories on financial issues from general papers, such as the New York Times.

The proposal and passage of the Glass-Steagall Act was the subject of several long articles in the Chronicle in mid-February 1932. The paper clearly felt that the new legislation set the stage for monetary expansion. It stated on 13 February: “The purpose is perfectly plain, the country is to be flooded with credit and with currency on the idea that thus it will be possible to stop deflation.”¹⁵

The Chronicle reported the increase in Federal Reserve purchases of government securities starting in late February. However, because discounts were falling at the same time, the paper suggested that “[n]o distinct sign of any change [in policy due to the passage of the Glass-Steagall Act] is as yet discernible.”¹⁶ The Economist also gave extensive coverage to the Glass-Steagall Act and its possible effects on American monetary policy.¹⁷ Unlike the Chronicle, it also saw a distinct move toward monetary ease in the early spring. It said on 19 March:

There was extensive press coverage of the acceleration of open market purchases approved in mid-April. According to an article in the Chronicle on 16 April: “On Wednesday, April 13, Governor George L. Harrison … revealed that the Federal Reserve System had embarked upon a new ‘easy money’ policy through open market purchases of Government securities at an accelerated speed, both as to rate and volume.”¹⁹

The New York Times stated: “By entering upon a policy of controlled credit expansion, designed to turn the deflation in bank credit and to stimulate a rise in prices, the Federal Reserve System has undertaken the boldest of all central bank efforts to combat the depression.”²⁰

The Federal Reserve's monetary expansion was the subject of lead articles in the Chronicle over the next few weeks. The Chronicle commented frequently on the rapid pace and the extreme size of the monetary expansion. On 23 April it wrote of the rate of purchases: “This is certainly going it fast.”²¹

On 30 April it referred to the “Federal Reserve easy money policy, pursued with such diligence and on such an overwhelming scale.”²² The foreign press was also very aware of the Federal Reserve's actions. The Economist, for example, reported the Federal Reserve's purchases of government securities each week and stated on 30 April: “The Reserve Banks are thus continuing and intensifying their easy money policy.”²³

There is no evidence that the public was aware of the Federal Reserve's intention by mid-June to scale down its program. Indeed, on 18 June the Chronicle wrote: “there has been no change whatever in Reserve policy with regard to the purchase of United States securities.”²⁴

Several articles in the Chronicle in late June also reported that the Federal Reserve was continuing with its monetary expansion program.²⁵

In early July, the Chronicle noticed that open market purchases had declined dramatically. It did not, however, act as though an obvious change in policy had occurred or that the decline was necessarily permanent.²⁶

On 23 July the Chronicle wrote: “It is to be observed, however, that though the further acquisition of U.S. Government securities during the week has been relatively small, there has nevertheless been some increase …. In other words the process of expansion still continues.”²⁷ The Economist also saw no signs that the open market purchase program was ending. On 16 July it declared: “Federal Reserve cheap money policy is unshaken.”²⁸ Thus, it appears that as late as mid-July, the public was not aware that the open market purchase program was effectively ended.

Motivation

The well-publicized expansionary open market operations in 1932 could possibly have led people to doubt the U.S. commitment to the gold standard. If such doubts had been severe enough, the United States could have experienced a catastrophic loss of gold and been forced off the gold standard. This possibility is formalized in the “first generation” models of speculative attack due to Paul Krugman and expanded in a large subsequent literature.²⁹

In the Krugman model, the government attempts to maintain a fixed exchange rate while issuing domestic credit to finance an excessively expansionary fiscal policy. Eventually, when reserves fall below a crucial threshold level, a speculative attack occurs.

Obviously, no such attack occurred in 1932. But this does not rule out the possibility that the monetary expansion led to more moderate doubts about the U.S. commitment to the gold standard. If expansion led to a moderate loss of credibility, this would have shown up in a rise in market expectations of devaluation. This implication of monetary expansion is also contained in the first-generation speculative attack models. In Krugman's original deterministic formulation, instantaneous devaluation expectations remain zero until the time of the attack. However, devaluation expectations measured far enough into the future to capture the attack would show the deterioration. Furthermore, the assumption of the Krugman model that people have complete information and can perfectly forecast the time of the attack is obviously unrealistic. In the stochastic version of the model by Robert Flood and Peter Garber, expectations of devaluation gradually increase as fundamentals deteriorate due to monetary expansion.³⁰

Thus, a test of the hypothesis that the 1932 monetary expansion led to a loss of credibility involves looking for the predicted change in devaluation expectations. To examine such expectations of devaluation, we use the relationship between forward exchange rates and spot rates. If the open market purchase program led to expectations of devaluation, forward dollar exchange rates should have risen relative to spot rates during the program. The use of forward rates to measure expectations of devaluation is standard in studies of both modern and historical episodes.³¹

We also use interest rate differentials to measure devaluation expectations. A basic arbitrage condition implies that interest rates on identical assets should be the same across countries except for transactions costs and expectations of exchange rate changes. Therefore, if American interest rates rose relative to foreign rates, we can deduce that investors' expectations of dollar devaluation had risen.

Data

The data on spot and forward exchange rates that we use are from Paul Einzig.³²

Einzig's data show the spot exchange rate and the three-month forward rate between various currencies and the British pound. We infer spot and forward rates between the dollar and various currencies by dividing the rate between the dollar and the pound by the rate between the relevant alternative currency and the pound.³³

The particular currencies that we consider are the French franc, the Swiss franc, the Dutch guilder, and the Belgian belga. We choose these four currencies because we are interested in expectations of dollar devaluation caused by American open market operations. Therefore, we want to consider currencies thought to be firmly wedded to gold in this period. France, Switzerland, the Netherlands, and Belgium, four of the “gold-bloc” countries, satisfy this criterion. Eichengreen discusses the fact that countries that were home to international financial centers, specifically the Netherlands and Switzerland, felt it was crucial to maintain convertibility and stability.³⁴

France and Belgium had also made it clear they were determined to remain on gold. In July 1933 these four countries (and a few others) issued a joint declaration affirming their commitment to the gold standard, and Switzerland, France, and the Netherlands did not finally devalue until October 1936.³⁵ For this reason, most movements in the spot and forward rates of the dollar relative to these currencies should reflect expectations about the American commitment to gold.

We calculate expected dollar devaluation from these data by taking the difference in the logarithms of the forward rate and the spot rate. We multiply the log difference by 100 to express it as the expected percentage devaluation.³⁶

Similar data on spot and forward exchange rates are available from other sources for this period. Both the Economist and the Financial Times report the spot rate and the three-month forward premium or discount each week for the pound against a large number of currencies.³⁷

Figure 2 shows a comparison of the expected devaluation of the dollar relative to the French franc measured using exchange rate data from Einzig, the Economist, and the Financial Times. This comparison shows that the three sources are very similar. There are minor differences, most likely due to the particular brokers surveyed and differences in whether daily averages or end-of-day figures were reported. Another notable characteristic is that the Einzig series is virtually always confirmed by one or the other of the alternative series, whereas each of the alternatives has some idiosyncratic observations. The same is true of comparisons of the three sources for the other three currencies we examine. For this reason, we use the Einzig data in the rest of our analysis.

Weekly data on interest rates for the same set of countries are more difficult to find. Most sources, such as the Economist and the Financial Times, report the central bank discount rate for a number of countries, but not an open market rate. The Commercial and Financial Chronicle, however, reports the open market discount rate for Paris and Switzerland each week.³⁸

Such private discounts are equivalent to bankers' acceptances in the United States. Although the exact horizon of the data is never stated in the Chronicle, private discounts are inherently short term, and 90 days was the most common horizon. Therefore, we use as our American comparison series the open market rate on 90-day prime bankers' acceptances. For consistency with the foreign data, we collect this data from the Commercial and Financial Chronicle as well.³⁹

To measure devaluation expectations, we subtract the foreign interest rate from the American rate. This difference should show the expected devaluation of the dollar. We divide the simple interest differential by four because the 90-day interest rates are expressed at an annual rate. Therefore, to see how much the dollar is expected to depreciate over the 90-day period, one needs to convert the annual-rate data to quarterly rates.

Behavior of Devaluation Expectations

Figures 3 and 4 graph expected devaluation of the dollar relative to each of the four comparison currencies for July 1931 to March 1933. Both figures show expected devaluation derived from the relationship between forward and spot exchange rates. Figure 3, which shows expected devaluation of the dollar relative to the French franc and the Swiss franc, also includes expected devaluation derived using interest rate differentials. In each figure we have placed solid vertical lines to denote the beginning and end of the 1932 open market purchase program. We have also drawn a dashed line to denote the acceleration of the program in early April.⁴⁰

One fact that is evident from the figures is that expected dollar devaluation measured using forward rates behaved quite similarly across the four comparison currencies during this period. These four currencies were rarely expected to move relative to each other.⁴¹

Thus, the data appear to support our supposition that it was expectations about the American commitment to the gold standard that were driving movements in the series, not questions about the commitment of the foreign countries. Figure 3 shows that when measured by interest rate differentials, the behavior of expected dollar devaluation relative to the French franc is very similar to the behavior of expected dollar devaluation relative to the Swiss franc (though somewhat different from expected dollar devaluation measured using forward rates). This again suggests that developments in the United States were driving movements in these series.⁴²

The behavior of expected dollar devaluation measured using forward rates shows little evidence that the open market purchase program led to fears that the United States would abandon the gold standard. All four series show a small fall in expectations of devaluation following the passage of the Glass-Steagall Act and the first rounds of open market purchases in late February 1932. Even more tellingly, expectations of devaluation fell again with the dramatic announcement of the program's acceleration in mid-April. At the time when the program was front-page news and open market purchases were the largest (mid-April to mid-May), expectations of devaluation were flat or declining.

Figures 3 and 4 show that there was a slight rise in expectations of devaluation measured using forward rates about two-thirds of the way through the 1932 program. The three-month forward rate rose slightly relative to the spot rate for all four currencies in late May and early June. However, the expected dollar devaluation was small. The maximum expected devaluation during the 1932 experiment ranged from 1.38 percent, for the dollar relative to the belga, to 2.84 percent, for the dollar relative to the Swiss franc. (For comparison, in September 1931, Britain devalued by 26 percent relative to the dollar.) Not only was the rise in expectations of devaluation in May and June small, it was largely uncorrelated with developments in the open market purchase program. As Figure 1 shows, open market purchases slowed in mid-May and remained steady at the lower level over the next six weeks.

Finally, devaluation expectations based on forward rates declined precipitously in mid-June, at least three weeks before open market purchases were seriously curtailed. As Figure 1 shows, this decline corresponds to a time when the Federal Reserve was holding open market purchases constant at a moderately expansionary level. Perhaps more importantly, as discussed in the first section, there was no news about the program that can explain the drop in devaluation expectations in mid-June. It is most definitely not the case that people expected the program to end. The Federal Reserve was continuing to buy bonds and, as described above, we find no discussion that this was expected to change soon.

If anything, the behavior of expected devaluation measured using interest rates shows even less evidence that the open market purchase program caused investors to doubt the U.S. commitment to the gold standard. The interest differentials in Figure 3 show that U.S. short-term interest rates fell very slightly relative to those in France and Switzerland at the start of the 1932 monetary expansion in late February 1932. They fell more dramatically with the acceleration of the program in April 1932, and then remained low until early 1933.

That the interest rate differentials behave somewhat differently from expected devaluation measured using forward exchange rates is not surprising. As discussed in the first section, substantial capital immobility in the pre–World War II period allowed large deviations from interest parity. The fact that expected devaluation measured using interest rates is always below that measured using forward rates probably reflects slight differences in the riskiness of the assets in the United States relative to those in the comparison countries.

The fact that interest rates declined in the United States with the start of the open market purchase program suggests that the program had the intended salutary effect on liquidity and credit availability. That the interest differential did not rise at any point in the spring of 1932 is evidence that any effects of the open market purchase program on the perceived credibility of the U.S. commitment to the gold standard were very small.

Statistical Tests

Although the graphical analysis suggests little correlation between the open market purchase program and expectations of devaluation, it is useful to test this conclusion more formally. This is particularly true for the expectations of devaluation based on forward rates, because they rose slightly during the program, while those based on interest rates clearly fell.

Our goal is to determine whether the open market purchase program caused expectations of devaluation to differ from what they would otherwise have been. It is therefore natural to consider regressions of the form

where EXPDEV_t is a measure of expected devaluation of the dollar relative to some other currency, OM_t is an indicator of the open market purchase program, and e_t reflects all other influences on expected devaluation. Our interest is in b, the impact of the program on expected devaluation.

One obvious complication in estimating b is that e_t is not white noise. As Figures 3 and 4 show, movements in the measure of expected dollar devaluation were highly persistent. To deal with this, we model e_t as following an AR–1 process.⁴³

There are two natural ways to specify the indicator of the open market purchase program, depending on the mechanism through which it potentially affected expectations. One possibility is that the program affected expectations simply through its existence and news reports that it was occurring. In this case, the appropriate measure is a dummy variable for the weeks the program was in existence. Alternatively, there may have been great uncertainty about how expansionary the program was likely to be. As a result, people may have placed particular emphasis on the scale of open market operations in forming their expectations. In this case, the appropriate measure is a dummy variable for the weeks the program was in existence calibrated by the weekly level of open market purchases.⁴⁴

The sample used for estimation is the period when all five countries under consideration were on the gold standard. It begins with 30 June 1928, the first observation after France officially returned to gold. It ends with 4 March 1933, just before the U.S. suspension of convertibility. We run the regressions separately for expected dollar devaluation (based on forward rates) relative to each of the currencies considered.

Table 1 reports the results. The coefficient estimates confirm the impression given by the graphs: there is little correlation between expectations of devaluation and the open market purchase program. When we use the simple dummy variable, the estimates of b are all small and not significantly different from zero, and for two of the currencies they are of the opposite sign from that predicted by the gold standard hypothesis. When we use the calibrated dummy, all of the estimates of b are of the wrong sign.⁴⁵

This is surely a result of the fact that the slight rise in devaluation expectations that occurred during the program happened after open market operations had been scaled back somewhat. However, the estimates are insignificant for three of the four currencies, and mainly serve to dramatize the point that there is not the strong positive relationship between open market operations and devaluation expectations implied by the gold standard hypothesis. Thus, the statistical evidence suggests that the 1932 open market purchase program did not cause expectations that the dollar would be devalued relative to the currencies of the gold-bloc countries.

We have examined the robustness of the regression estimates along many dimensions. Starting the sample period with 10 October 1931, just after Britain's devaluation in September 1931, has no important effect on the results. Similarly, specifying the dummy variable to start with the dramatic acceleration of the program in April 1932 rather with than the gentle beginning in February has virtually no impact. Calibrating the dummy variable using the lagged level of open market purchases, the average level of open market purchases over the previous three weeks, or the average level of open market purchases since the start of the program also has no noticeable effects. Finally, to see if the existence of the program and the scale of open market purchases mattered jointly or one clearly dominated the other in its effects on expectations, we include the simple and calibrated dummy variables together in the regressions. The results are little different from when the two variables are entered separately. In short, there is no statistical evidence that the 1932 open market purchase program had a significant effect on expectations of devaluation.

Devaluation Expectations in Other Episodes

One concern with this analysis is the possibility that forward rates and interest differentials might not reveal expectations of devaluation that are nonetheless present. This could occur if the forward exchange market or the market for private discounts were thin, or imperfect in some other sense. One way to check for such imperfections is to look at devaluation expectations derived from forward rates and interest rates in episodes when such expectations are widely thought to have been present.

Two such episodes are provided by the French experiences of 1934 and 1935. Eichengreen describes how a political crisis in early February 1934 led to expectations of fiscal and monetary expansion.⁴⁶

These expectations led to fears of devaluation of the franc and a large loss of French gold in the first two weeks of February. On 31 March 1935 Belgium devalued. Eichengreen suggests that this, together with the French government's pursuit of “New Deal style” expansionary monetary and fiscal policies, again led to fears of devaluation of the franc in May 1935.⁴⁷

Figure 5 shows the expected devaluation of the dollar relative to the French franc for the extended period 1928–1935 measured using both forward rates and interest rates. Because the dollar was rebenchmarked to gold in January 1934, movements in the dollar/franc series starting in February 1934 were almost surely driven primarily by expectations about the franc. Because the graph shows expected dollar devaluation, a negative number is expected devaluation of the franc.

Figure 5 indicates that both French episodes show up much more strongly in the data than does the 1932 American open market purchase program. During the February 1934 episode, expectations measured using forward rates changed from an expected franc appreciation of around 1 percent to an expected franc devaluation of 1.5 percent. For comparison, in the 1932 U.S. expansion, expectations changed only from an expected dollar devaluation of about 0.5 percent to an expected dollar devaluation of 1 percent. Thus, the change in the 1934 French episode was roughly five times as large as the change in the 1932 American episode. This same pattern holds even more strongly in the May 1935 French episode. Expected franc devaluation measured using forward rates rose from 0.5 percent to 6.25 percent, and remained above 3 percent for several weeks. These episodes show that forward rates can provide evidence of devaluation expectations in the interwar period.

The May 1935 episode also shows up in interest rate differentials. The interest differential (U.S. rate minus the French rate) fell from around −0.5 percentage point in early May to −1.5 percentage points in mid-June. However, there was only a trivial fall in the interest differential in the February 1934 episode. Thus, while interest differentials can sometimes capture devaluation expectations in the interwar era, it appears that factors such as imperfect capital mobility and changes in risk can cause them to be an imperfect indicator.

An episode when fears of U.S. devaluation are commonly thought to have been present was late September and early October 1931. Britain's abandonment of the gold standard on 21 September 1931 could certainly have shaken confidence in the U.S. commitment to gold. And, indeed, a substantial quantity of gold flowed out of the United States in the six weeks following Britain's devaluation, suggesting increased concern about the stability of the dollar.

Figures 3 and 4 show that this episode led to the largest upward spike in our measures of expected dollar devaluation in the Depression era. Expectations of dollar devaluation relative to the French franc measured using forward exchange rates, for example, rose from approximately zero in the week before Britain's departure to +2 percent in mid-October. On average, the rise in devaluation expectations measured using forward rates is about three times as large in 1931 as during the 1932 open market purchase program. Devaluation expectations measured using interest rate differentials also spiked up in the fall of 1931, whereas they fell in the spring of 1932.

Although the 1931 episode clearly stands out in the time series for expected dollar devaluation, the expectations of devaluation were fairly modest. Friedman and Schwartz and Charles Kindleberger stress the Federal Reserve's rapid response to the gold drain.⁴⁸

That the Federal Reserve Bank of New York raised its discount rate 200 basis points between 8 October and 16 October 1931 may have quickly calmed any fears about the U.S. commitment to the gold standard. Another possibility discussed by both Kindleberger and Lester Chandler is that the gold drain was due to the Bank of France's desire to improve the appearance of its balance sheet following the losses caused by Britain's departure, rather than genuine fear that the United States was going to devalue.⁴⁹ Thus, expectations of American devaluation may genuinely have been fairly modest, as the data indicate.

The American experience in early 1933 provides a similar type of corroborating evidence. Because the United States suspended gold convertibility in March 1933, it is often presumed that expectations of devaluation must have been rampant during this period. And yet, the expected dollar devaluation series (relative to the French franc) given in Figure 5 shows no rise in the first two months of 1933. Expectations of devaluation measured using both forward rates and interest rates did spike up in early March, but even then the level of expected devaluation was relatively small. Our reading of the narrative evidence suggests that this picture of devaluation expectations derived from forward rates and interest rates is accurate, whereas the common presumption is not.⁵⁰

Perhaps the best evidence that devaluation fears were minimal in early 1933 is the fact that the gold price of the dollar remained constant for more than a month after the suspension of convertibility. Following emergency legislation surrounding the Bank Holiday, gold convertibility and gold exports were suspended by executive order on 10 March 1933. This left the dollar free to depreciate. However, according to Friedman and Schwartz, it was only after a further executive order on 20 April and a speech by Roosevelt that made it clear he favored depreciation that the price of the dollar fell relative to the remaining gold-bloc currencies.⁵¹

If the dollar had been widely expected to depreciate in early 1933, the fall in the price would have occurred as soon as convertibility was suspended.

A systematic reading of the Commercial and Financial Chronicle also suggests that devaluation fears were minimal. Throughout December and early January, uncertainty about the U.S. commitment to the gold standard and the possibility of devaluation were not discussed. Instead, the Chronicle contained numerous references to the fact that “gold is flowing into the country in a perfect stream.”⁵²

On 25 February 1933 the Chronicle expressed concern that “another scare seems to have developed abroad as a result of these inflation propositions and a new eruption of banking troubles in various parts of the United States and withdrawals of gold for export and for earmarking on foreign account are again proceeding on a large scale.”⁵³ But, it went on to say: “Of course the United States has enormous supplies of the metal.”⁵⁴ The Chronicle also reacted positively to Roosevelt's choice for Secretary of the Treasury, saying: “[fears] of a departure from sound economic principles … may now be dismissed.”⁵⁵ When the Bank Holiday and suspension of convertibility were declared in early March, the Chronicle saw some fears of devaluation, but suggested that they were short-lived “owing to the assurances against inflation given by the new Administration in Washington and the vigorous steps for control of the banking crisis.”⁵⁶ The absence of fears in early 1933, followed by only modest fears that were quickly resolved, is consistent with behavior of devaluation expectations shown in Figure 5.

Was the Federal Reserve Concerned about a Speculative Attack?

The beliefs and motivations of the Federal Reserve are well documented in the papers of George L. Harrison, Governor of the Federal Reserve Bank of New York. The Harrison Papers contain the minutes of the Open Market Policy Conference, of which Harrison was the chairman, and the memoranda of meetings of the Board of Directors of the Federal Reserve Bank of New York. They also contain a plethora of office memos and correspondence to and from Harrison.

The Harrison papers provide no evidence that the Federal Reserve was concerned that the open market purchase program would trigger fears of devaluation. When one member asked about possible international repercussions at the 12 April meeting of the Open Market Policy Conference where the program was accelerated, he was quickly reassured:

Indeed, the gold standard constraint was hardly mentioned in the initial or ongoing discussion of the program. The only exception to this pattern occurred at the meeting of the Board of Directors of the New York Federal Reserve Bank on 21 April. Eugene Meyer, Governor of the Federal Reserve Board, said:

Meyer's concerns about the gold standard were not echoed or discussed by anyone else at the meeting. Moreover, Meyer himself seemed to believe that these concerns mainly argued against lowering the discount rate, not in favor of ending the open market purchases.

A retrospective study of American monetary policy written by E. A. Goldenweiser, Director of the Division of Research and Statistics of the Board of Governors in the 1930s, confirms the Federal Reserve's lack of fear about gold losses and a speculative attack in 1932. In discussing the 1932 open market purchase program, Goldenweiser writes: “a Federal Reserve Bank could be forced to suspend by an excessive gold drain, but no such drain was in prospect and the possibility of suspension for such a reason was not contemplated.”⁵⁹

In May 1932 substantial quantities of gold began to flow out of the United States, primarily to France, but also to Switzerland, the Netherlands, and Belgium. The Federal Reserve attributed these gold flows to various factors, most of them unrelated to Federal Reserve actions. Governor Harrison stated:

Governor Meyer also believed that disturbances such as the passage by the House of the Goldsborough bill, which set as an explicit target of Federal Reserve policy reflation to 1926 levels, “had occasioned an unfavorable reaction in Europe and some gold withdrawals.”⁶¹

The Federal Reserve believed firmly that “only a very small part of this change in the System's gold position and the reserve ratio is attributable to purchases of government securities. Most of it has been due to the repatriation of foreign central bank dollar funds, which very probably would have occurred regardless of the policy of the System.”⁶²

The case of the French repatriation (which was by far the largest of the repatriations) provides clear evidence of both the motivation of the foreign central banks and the Federal Reserve's interpretation of the gold outflow. On 15 January 1932 Harrison thanked Governor Moret for informing him of the Bank of France's plan to withdraw its earmarked gold from the United States.⁶³

A subsequent letter on 9 March from Moret to Harrison explained that “the monetary Law of June 25, 1928, implicitly imposed upon the Bank of France the obligation to liquidate its foreign assets.”⁶⁴ Harrison responded to Moret on 3 April, saying: “[W]e have felt for some time that it would be desirable rather than hurtful from the point of view of our position to have the Bank of France gradually repatriate its dollar balances.”⁶⁵ Then in June, the Federal Reserve became concerned that the gradual repatriation was unsettling expectations. (And, indeed, the small hump in expected dollar devaluation shown in Figures 3 and 4 corresponds to this period of French repatriation of gold.) To dispel the uncertainty, Harrison sent Moret a telegram on 9 June asking him to “favorably consider the advisability of earmarking promptly and at one time all your remaining holdings of approximately $93,000,000.”⁶⁶ Moret readily agreed to Harrison's request.

In addition to having known about the intention of the European central banks to withdraw gold before the advent of the 1932 open market purchase program (and, indeed, having encouraged them to do so), the Federal Reserve expressed the view that the resulting gold outflow was a reason to undertake further monetary expansion, not a sign that the program should be stopped. The minutes of the February meeting of the OMPC record that:

The Chronicle reported that in a speech in late April, Treasury Secretary Ogden Mills declared that “the Federal Reserve program of buying Government securities could be fully justified on the ground of replacing exported gold and hoarded currency.”⁶⁸

During the height of the gold withdrawals in early June, Harrison argued to his directors “that he thought it would be a mistake to let up on our program at this point, in view of the country's heavy gold losses during the past week.”⁶⁹ Even James McDougal, Governor of the Federal Reserve Bank of Chicago, generally a staunch opponent of the open market operations, approved of their use to sterilize the gold outflows. In a letter to Harrison on 9 July, McDougal informed him that, “While purchases by the System for the purpose of offsetting gold exports were probably justified, we believe that the additional purchases made were much too large.”⁷⁰

Public Perceptions of the Credibility of the Gold Standard

Clearly, the Federal Reserve seems to have been remarkably unconcerned that expansionary open market operations would reduce the credibility of the U.S. commitment to the gold standard in 1932. To see if other informed observers were equally sanguine, we consider the reports in the Commercial and Financial Chronicle and the Economist.

The Economist thought that, if anything, the passage of the Glass-Steagall Act and the early monetary expansion increased the credibility of the U.S. commitment to the gold standard. It said on 27 February: “the loans of the Reconstruction Corporation combined with the Reserve Bank measure ought to put a brake on security deflation and remove some of the unfounded nervousness as to this country's retention of the gold standard.”⁷¹

On 12 March it reported that “Wall Street does not anticipate any stimulation to the gold outflow, except, perhaps, temporarily, and the Reserve feels itself well able to meet all demands for gold likely to be made upon it.”⁷²

The Chronicle railed against the Glass-Steagall bill because it felt it would be inflationary.⁷³

However, there was no discussion of possible repercussions for the gold standard. The Chronicle only emphasized that the new collateral provision “does not change the real gold position of the Reserve banks (which is exceptionally strong) in the slightest degree.”⁷⁴

Once large open market operations began in April, however, the Chronicle expressed concern that this policy had led to a loss of confidence in the dollar. For example, during the week that the OMPC accelerated the monetary expansion, the Chronicle said of the open market operations: “That there is menace in them, in the fears at least aroused abroad, appears in the resumption of gold exports.”⁷⁵

However, by the third week of the program, concerns over a potential devaluation had receded from the pages of the Chronicle. On 7 May the Chronicle once again discussed fears of a potential devaluation, but attributed them not to the current monetary expansion, but to the House's passage of the Goldsborough bill. Furthermore, the Chronicle reassured its readers that the “Federal Reserve banks, however, are so strongly fortified with holdings of gold … that the fear apprehended does not appear likely to come to pass unless Congress should become positively mad in its follies.”⁷⁶

The Economist was enthusiastic about the acceleration of open market purchases in April. It reported that the program “will help to stop deflation without bringing on dangerous inflation, it is felt by the sponsors of the policy, and at the same time there will be no danger of abandonment of the gold standard.”⁷⁷

The Economist was unconcerned about U.S. devaluation throughout the program. For example, it stated on 23 April: “The absurd rumours recently afloat in Paris over banking troubles here [in Washington] or the danger of departure of the country from gold were greatly disliked but were not taken seriously.”⁷⁸ In May, it stated: “The present rate of Federal Reserve Bank operations could continue a year or more without endangering the gold reserve …. No responsible person in the financial community seriously believes the country will be allowed to drop the gold standard.”⁷⁹

When gold exports to Europe accelerated in late May and early June, the Chronicle editorialized that “the large gold outflow reflects deep distrust of the performances in the United States—the Federal Reserve policy of large-scale purchases of United States Government securities and the various propositions finding favor or being urged in Congress involving expenditures of billions of dollars.”⁸⁰

The Economist also repeatedly mentioned the gold outflow in May and early June. It, however, attributed them primarily to Congress's failure to balance the budget and other expansionary legislative actions. For example, it discussed on 14 May “the strong movement of European currencies against the dollar, attributed mainly to anxieties over the Goldsborough Bill's passage.”⁸¹ In early June it stated: “continual gold losses are calculated to bring home to public opinion and also Congress at Washington the need for balancing the national finances.”⁸² The Economist at no time expressed fear that the open market operations were causing the gold flows.

Clear evidence that the gold flows were causing unsettled expectations, rather than reflecting fears of devaluation, is provided by the relief expressed at their end. On 10 June the New York Times reported that “the Bank of France seems to have ceased its sales of dollars on Thursday (June 9), and this sufficed to cause recovery in American exchange to a rate at which gold exports were no longer possible.” It went on to say that:

The Economist expressed the same view when it stated: “Equally keen is the feeling of relief that the major gold movements are probably at an end and that sentiment will not be so frequently disturbed by the publicity given to the gold outflow.”⁸⁴

Why Did the Federal Reserve End the Monetary Expansion?

If the Federal Reserve was not concerned about a speculative attack, why did it cease its open market purchases after only five months? Our reading of the Harrison papers suggests that the Federal Reserve decided to slow the monetary expansion in mid-June in part because its model of monetary policy led it to believe that monetary conditions were already loose and that further purchases would be of little use. As discussed by Elmus Wicker, David Wheelock, and Allan Meltzer, Federal Reserve officials in the 1930s focused on bank borrowing and excess bank reserves as their main indicators of monetary ease or tightness.⁸⁵

For example, on 12 May Harrison told the Board of the Federal Reserve Bank of New York that: “The best yardstick to use [for measuring the success of monetary policy] … would be the figures of member bank reserves.”⁸⁶ More importantly, policymakers at the Federal Reserve believed that once excess reserves were plentiful, further monetary easing could do little to stimulate recovery. At the same meeting, Harrison stated that:

A related view was that, because expansion would be ineffective when there were already large quantities of excess reserves, it did not make sense to expand when the demand for funds was low. Instead, Federal Reserve officials believed that the best time to expand was when confidence was high, or at least improving. For example, in reviewing the origin of the open market purchase program on 30 June, Harrison said: “It was thought best, however, not to use our ammunition until the chances of effective response from the banking and business community would favor the success of our undertaking.”⁸⁸

This model of the economy hastened the end of the open market purchase program in two ways. First, in late May and early June, some Federal Reserve officials believed that the program had already worked. The program had made excess reserves plentiful and so further expansion was not needed. On 26 May Harrison told his directors that “excess reserves of member banks are now at about the point where it had been thought they should be maintained.”⁸⁹

Goldenweiser recounts that “Federal Reserve authorities felt that their monetary policy had made bank credit expansion possible and that they were powerless to induce banks to lend more freely or even to arrest loan liquidation.”⁹⁰ Second, when a wave of banking panics in Chicago in late June led Federal Reserve officials to fear that widespread banking difficulties were about to begin again, some monetary policymakers concluded that further open market purchases would have no impact. Harrison, for example, said that: “There is no sense … in our purchasing Government securities merely as an offset to currency hoarding. That is an impossible task and an inversion of our program, which was based on a revival of confidence in the banking and credit structure.”⁹¹

As discussed by Gerald Epstein and Thomas Ferguson and by Meltzer, conflict among the twelve regional Federal Reserve banks also played a role in ending the program.⁹²

On 5 July Harrison informed his directors that the Federal Reserve Banks of Chicago, Philadelphia, and Boston were reluctant to continue the open market operations. He concluded that “if the other large Federal reserve banks are unwilling to proceed with the program, we cannot carry the burden for the entire System, while our reserve percentage is the lowest of any bank in the System.”⁹³ Owen D. Young, deputy chairman of the Board of Directors of the Federal Reserve Bank of New York, added that “if we cannot have the continuous participation of the Federal Reserve Banks of Boston and Chicago in the System program, I am for suspending the program.”⁹⁴

Why the Federal Reserve Bank of Chicago chose to drop out of the program when it did is a matter of some debate.⁹⁵

Harrison told his directors on 23 June “that Governor McDougal, always a reluctant follower of the System program, is now more reluctant than ever to participate heavily in purchases of Government securities because of the demands which may be made upon his bank as a result of the recent closing of a number of small banks in Chicago.”⁹⁶ There is some evidence that the Chicago Board of Directors subscribed to the view that banking panics made further increases in reserves useless. Young had talked to several of the Chicago directors and reported that:

Governor McDougal of Chicago cited as his reason for abandoning the program the notion that:

The conflict between the Federal Reserve Banks of New York and Chicago that ultimately ended the monetary expansion program reflected deeper structural problems of the early Federal Reserve System.⁹⁹

The Federal Reserve had been purposely set up as a decentralized system of 12 relatively autonomous central banks. Each bank had to back its note issues with eligible assets. Consequently, each Federal Reserve bank paid attention to its own reserves, and not to system-wide reserves. For example, on 23 June Harrison complained to his directors that the Federal Reserve Bank of New York had shouldered most of the burden of the open market operations and that “it is just as important for this bank to watch its reserve position as for any bank in the System.” Harrison also argued that with the reserve ratio of the Federal Reserve Bank of New York at 50 percent and that of the Federal Reserve Bank of Chicago at 75 percent, “it is difficult to see why we should pump funds into the market which will then be siphoned off to Chicago.”¹⁰⁰

In addition, until the Banking Act of 1933 was passed, each Federal Reserve bank could undertake its own open market operations. The OMPC had been established in 1930 to coordinate the activities of the twelve banks, but each bank could refuse to participate in the decision taken by the Conference. As a result, even during the few times when it was absolutely clear to the majority of the Federal Reserve's governors that expansionary monetary policy was urgently needed, such as the spring of 1932, the decentralization of policy resulted in paralysis. As Owen Young cogently summarized, under the institutional framework of 1932, “you may have two or three banks dictating the policy of the System at a critical time, just because of their ability to block a System program.”¹⁰¹

This is exactly what happened in the summer of 1932.