Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Prepared for: World Economy and National Economies in the Interwar Slump, Theo Balderston, Editor (Basingstoke: The MacMillan Press). UNDERSTANDING THE GREAT DEPRESSION IN THE UNITED STATES VERSUS CANADA Pierre L. Siklos* Department of Economics Wilfrid Laurier University Waterloo, Ontario Canada, N2L 3C5 * Troy Elyea provided valuable research assistance. I am grateful to the Social Sciences and Humanities Research Council of Canada, and the Society of Management Accountants of Ontario (SMAO), for financial support. Part of the research was conducted while I was Wilfrid Laurier University University Research Professor for 2000-2001. Theo Balderston, the Editor, Angela Redish, Richard Burdekin, Lars Jonung, Lee Ohanian and LeRoy Laney provided helpful comments on earlier drafts. Previous versions of this paper were presented at the Western Economics Association Conference, San Diego, July 1999, and the Stockholm School of Economics, the Economic History Association, and the Canadian Network in Economic History. TABLE OF CONTENTS 1. Introduction 2. The Gold Standard as a Straightjacket for Monetary Policy 3. Impulse and Propagation of Shocks Leading to the Great Slump 4. The Great Depression as a Financial Crisis: Canada versus the US 5. The Consequences: Reform and Recovery 6. Concluding Remarks 1 1. Introduction It is now recognized that the Great Depression should be viewed as a global phenomenon with its roots in the Gold Standard so arduously restored in Great Britain in 1925 (Bernanke 1995). Temin (1989) and Eichengreen (1995) provide arguably the best recent accounts of the Great Depression. Not surprisingly, the literature on the interwar slump which afflicted much of the world has generated a vast literature which continues to this day to fascinate economists and social scientists more generally. Unlike Friedman and Schwartz (1963, 1982), whose masterful works documented the role of monetary policy to explain both the depth of the Depression and its propagation outside the United States, Temin (1989) stresses the influence of a large “economic” shock which was then magnified by an almost slavish adherence to the Gold Standard. Eichengreen (1995) would not disagree but also would emphasize the breakdown in international cooperation among government and central bank officials. Others, such as Kindleberger (1986), might argue instead that the collapse of the post World War I monetary order produced financial chaos which, like the now much discussed “Asian flu” of the late 1990s, caught on across much of the world. The interwar era has fascinated many due to the tremendous severity of the economic collapse, as is documented throughout this volume. Lately, however, interested researchers have been drawn not only to how the Depression was transmitted worldwide but whether the combination of economic ideology and politics conspired to generate such a spectacular slump. Since so much of the debate surrounding the Great Depression centers around financial issues, their real economic consequences, and the global nature of the event, comparisons across countries can provide useful insights about the role played by institutional factors. It is precisely for this reason that a review of the US and Canadian experiences is helpful, not only because of the obvious close ties the two countries share with each other but also because there were significant differences in the financial sphere which underscore the role played by the Gold Standard. Yet, despite the differences, there is one unmistakable similarity: both countries 2 experienced a Great Depression at roughly the same time and of comparable economic magnitude. As Bernanke (1993) points out, there have been relatively few references to Canada in various arguments about the causes and cures for the Great Depression. The rest of this chapter is organized as follows. The following section briefly sets the stage for a US-Canadian comparison by outlining what motivated the return to a Gold standard following World War I and why the policy, from a “modern” policy perspective, represents a “straightjacket”, otherwise referred to as the “cross of gold”. Section 3 explores the origins and transmission of the powerful and large shock that Temin (1989) argues lay behind the Great Depression and how this played out in the US and Canada. Some evidence about the relative predictability of deflation in the US and Canada is presented for the period of the Great Slump as well as for an earlier large deflation which took place in the early 1920s in both countries. Section 4 focuses on the financial aspects of the Great Depression. Arguably, this is a crucial ingredient of the global slump. Section 5 looks at the consequences of the Great Depression for economic reform and recovery. Section 6 offers a few concluding comments by asking how recent economic research sheds light on the overarching factors in the Depression. Needless to say, since the topics of interest are wide-ranging, the present chapter can only selectively survey the extant literature and, hopefully, whet the appetite of the interested student for a more serious study of this remarkable era in economic history. 2. The Gold Standard as a Straightjacket for Monetary Policy One cannot evaluate the magnitude of the Great Depression without comprehending the constraints imposed by the Gold Standard. For much of the 19th century commodity money regimes, such as the Gold Standard, shaped policymakers’ thinking. Not surprisingly then, some countries were keen to return to it following recovery from World War I. But the decision to return to Gold was not squarely based on economic considerations. As Winston Churchill who, as Chancellor of the Exchequer, led the United Kingdom back to gold noted: “But this [the return to the Gold Standard] isn’t entirely an economic matter; it is a political decision...” (Gilbert (1977, p. 92)). 3 One of the chief attractions of the Gold Standard among its supporters was that it operated automatically, leaving no room for political interference in the realms of international trade and investments. Combined with a commitment to central bank autonomy, the “rules of the game” insured that currencies could be converted into gold at the stipulated fixed exchange rate. Indeed, in its purest form, the Gold Standard comes closest to the fixed exchange rate system discussed in standard macroeconomic textbooks. Students of modern macroeconomics are taught that the demand for money is a function of income and the opportunity cost of holding money, usually measured in terms of some nominal interest rate. Moreover, the demand for money is assumed to be proportional to the price level, giving rise to the usual quantity theory relation. But in the purest gold standard system, i.e., one where the only money is gold coinage, the demand for gold is also a function of the price of gold, relative to those of other goods. This is because gold is a commodity with nonmonetary uses too. We can thus write M/PG = F(Y, P/PG , θ) (1) where M/PG is the money stock (M) deflated by the price of gold (PG ), Y is income, P/PG is the price of consumer goods (P) relative to gold, and θ is the “cost of holding money”.1 Notice that, unlike the Quantity theory, there is no reason why a change in P/PG should lead to a proportionate change in M/PG, holding Y and constant. By fixing the price of gold in terms of money, the price of goods in terms of money is also fixed, leaving no room for active monetary policy, making it “knave-proof”.2 The actual process which would guarantee this state of affairs was the “specieflow” mechanism.3 Hence, a country which consumed in excess of production, producing a balance of trade deficit, would have to export gold to cover its debt. Moreover, the resulting fall in gold reserves would normally lead to a contraction in the domestic money supply and, eventually, in the price of goods and services. To the extent that monetary policy can be used, at least in the short-run, to stimulate economic activity, the Gold Standard can be seen to act as a straightjacket. Moreover, as Eichengreen (1995) argues, the ideology of the Gold Standard 4 prevented due consideration of alternative policies that might have delivered better economic outcomes. The reality of the Gold Standard was, of course, different for a number of reasons. First, countries such as the UK and the US, which had central banks, could manage the monetary system by varying the terms under which they could extend credit to the banking system. Hence, a rise in domestic interest rates could offset the loss of gold reserves from a balance of trade deficit by attracting short-term capital flows. The higher interest rates would also serve to depress domestic aggregate demand thereby helping eliminate these pressures which produced the outflow of gold in the first place. Second, the actual Gold Standard was a fractional system since only a small portion of the money supply was held in the form of gold. Moreover, a central bank could resort to the practice of “sterilization”, that is, conduct open market operations. Hence, for example, a reduction in the money supply from gold outflows could conceivably be offset by buying government bonds held by banks or the public. Third, prices of goods and services do not adjust instantaneously to some imbalance between aggregate demand and supply. Fourth, the adjustment process is also dependent on how mobile capital is. Ostensibly then, the Gold standard could represent a straightjacket under the rules of the game, as sketched above. However, these rules were contingent on all of the above factors, including the ability of central banks and global financial markets to circumvent them as needed (e.g., see Bordo and Kydland 1995). What then of the US and Canadian experiences? First, it might surprise many that capital markets in industrial economies were highly integrated at the outset of the Great Depression making the current notion of “globalization” seem somewhat dated (Bordo, Eichengreen and Kim (1998)). Dick and Floyd (1992) describe in great detail how Canada operated under the Gold Standard until the outbreak of World War I and the degree to which capital markets in Canada and the US were integrated (also see Bordo, Redish and Shearer 1999). In particular, much of banks’ gold reserves were held in New York where a substantial amount of financial business was carried out. Indeed, these same authors present convincing empirical evidence to 5 suggest that capital market integration drove adjustment under the Gold Standard in Canada rather than adjustment in the price of goods and services.4 Table 1 shows the balance of trade, flows of short-term capital, and movements of gold for the period 1926-37. It is clear, for example, that the specie-flow mechanism worked differently in practice than in theory. Figure 1 plots the available data on short-term interest rates in Canada and the US. Whereas the traditional view of the specie-flow mechanism predicts that they should move in the opposite direction, to satisfy the condition of equilibrating gold flows from one country to the other, Figure 1 instead reveals a considerable amount of parallel movements in these rates before and after the Gold Standard period. Finally, Figure 2 plots the behaviour of the wholesale price index of the two countries from May 1925, when the UK returned to gold until the Gold Standard collapsed in 1931. This figure shows commonality of price movements in the two economies, not the inverse relation predicted by the specie-flow mechanism. (It also disproves the much-touted claim that the gold standard protected against inflation and deflation). Obstfeld and Taylor (1997) also point out that capital mobility was high not only between the US and Canada, but across all major economies at the time. The onset of the Great Depression led to a further contraction of the order of roughly 25% of cross-border flows of capital even between the US and Canada coming on top of an almost 50% drop in capital flows relative to the 1870-1913 period.5 3. Impulse and Propagation of Shocks Leading to the Great Slump The impulse to a business cycle expansion or contraction refers to an event or a “shock” of some kind which affects economic activity, output in particular, with a lag. This gives rise to changes in economic variables related to each other in time which is then called the propagation mechanism. Under this interpretation, the return of the Gold Standard sowed the seeds of a deflation but the straightjacket policy makers placed themselves in was largely motivated by the belief that the “credibility” of the Gold Standard rested on the maintenance of a regime that deemed crucial for delivering economic prosperity in the pre-World War I era. In Temin’s (1989) view, “major deflationary shocks” were the impulse for the Great Depression 6 and its roots therefore lay in the “ideology of the Gold Standard”. For Eichengreen (1995) the ideology of the Gold Standard, coupled with a breakdown of international cooperation among key central banks, produced a slump more severe than any previously recorded. That the shock, whatever its origins, was large is clear on inspection of Figure 3 which plots industrial production in the US and Canada as well as the dates of reference cycles constructed on the basis of various economic indicators and pioneered by economists at the National Bureau of Economic Research (Burns and Mitchell (1946)). In both Canada and the US, industrial production roughly doubled between 1921 and 1928 or 1929 followed by a contraction of approximately the same magnitude between 1929 and 1933, that is, in about half the time it took industrial production to rise to prior to the collapse.6 The peaks and troughs in the reference cycles – the former dated the beginning of a recession; the latter the end of a recession – tend to mirror the downturns in industrial production. Nevertheless, we do see more frequent and shortlived cycles in economic activity before the Great Depression while longer durations are apparent between recessions and recoveries since the Great Slump.7 Finally, note that, with the exception of the US recession of 1927, Canadian and US recessions are nearly coincident.8 This would suggest a sequence of recessions, including the Great Depression, that originated in the US but were quickly transmitted to Canada. The combination of the Gold Standard and the close economic ties between the two countries would reinforce the mechanism through which the original shock (i.e., the impulse) was transmitted internationally, as emphasized by both Temin (1989; see also 1993) and Eichengreen (1995). There is, however, disagreement on the point. Betts, Bordo and Redish (1996) have explained the parallelisms between the outset, depth and duration of the depression in both countries as a reaction to a common supply shock, namely a technical or resource shock originating in both economies simultaneously, which they did not manage to identify. Cole and Ohanian (1999, 2001) use a “neoclassical” model (of the kind popularized by Lucas 1987), which features a production function with constant returns to scale consisting of labour and capital as inputs. The model is then used to assess whether the downward rigidity of nominal wages or the role played 7 by the banking system were sufficient “shocks” to explain a significant portion of the Great Depression. They conclude in the negative. The strong parallels between the US and Canada detailed in this chapter may weaken the case for the Cole-Ohanian view as we shall see. More importantly, the basis for their contention of a weak recovery from the Depression is an unusual calculation of output performance (op. cit., n. 5) during the 1930s which is not justified by the authors and seems contradicted by Figure 3. Nevertheless, it is also striking, if not troubling, that the deflation of the early 1920s was more severe in both the US and Canada while the contractions experienced in both countries (shown in Figure 3) were considerably smaller in magnitude, again in both countries.9 These two explanations of the relationship between the US and Canadian depressions arise because of problems in the econometrics of distinguishing between them. Even a fairly straightforward model can admit several channels through which the variables of interest are affected. Hence, empirical testing requires that a priori judgments be made about what are, or are not, significant channels of influence, and scholars will differ in such judgments.10 Nevertheless, modern econometric tools at least permit the debate over the origins and magnitude of the Great Depression to be conducted in a manner that permits critical assessments, as well as progress, to be made over disgreements about the main outstanding issues. A number of examples follow to illustrate the point. Betts, Bordo, and Redish (1996) find against the view that Canada “imported” the Great Depression from the US (also see Fremling (1985)). Hence, relative price changes under the Gold Standard, whose role is highlighted in (1), implies that the Great Slump was an event which occurred simultaneously on a global scale. However, it is important to note that the authors are unable to identify short-run supply shocks which may have been common to both countries. Fackler and Parker (1994) and Burbridge and Harrison (1985) use similar techniques and find in favour of the impulse-propagation explanation of the Great Depression. Further, the shock to the US economy is thought to have been propagated worldwide (Romer 1993). Cecchetti and Karras (1994) identify the sources of the Great Depression in the US and conclude, in line with Temin’s 8 (1989) views, that a collapse in consumption, that is, an aggregate demand shock, was largely responsible for the drop in US output through late 1931. This drop was facilitated by monetary policy actions during this period, as emphasized by Friedman and Schwartz (1963) and Hamilton (1987). This was followed by an aggregate supply shock, via an unexpected decline in prices originating in the banking crises of 1930, 1932 and 1933, which contributed to further declines in output until 1933. It is important to note that these efforts do not formally allow a direct role for the impact arising from the high degree of financial integration between the two economies. As discussed in the next section, it is conceivable that while there was an idiosyncratic shock which led to deflation almost simultaneously in the US and Canada, the integration of financial systems, not explicitly accounted for in the foregoing research, may explain the global transmission of the Great Depression. Is there any single source which can be identified as setting in motion the Great Depression? While some popular histories (e.g., Garraty (1986)) point out the role of the stock market crash of 1929, more recent treatments of this event (e.g., Romer (1990)) suggest otherwise. First, the empirical connection between the stock market crash, lowered expectations for economic activity, and hence consumption, appears to be weak at best (e.g., Temin (1976), Dominguez, Fair and Shapiro (1988)). Moreover, unlike the present day situation, only a small portion of individuals’ wealth was invested in stocks so that this channel of influence on consumption is also likely not to have been operative. Yet, the combination of loss of confidence in financial institutions’ survival, and the fact that banks could underwrite share issues and thereby incur potential losses on their own account, both stemming from the stock market collapse, must also surely be an ingredient in the Slump. Romer (1990) argues that uncertainty about the future is what contributed to the drop in output. Although uncertainty can manifest itself in various forms, it is usually captured in the form of some measure of volatility. On this basis it can be shown that real income “uncertainty” appears to have risen before “uncertainty” in stock price movements.11 Even if uncertainty of some form is the proximate cause for the global collapse of the 1930s, was there something in 9 either the structure of the Canadian or US economies which can provide additional clues about the events which transpired during this period? Alternatively, perhaps, was the effect of the original shock amplified by its sheer unexpected size? Although both economies were dependent on the production and trading of commodities, Canada was perhaps relatively more vulnerable. For example, in 1919, the proportion of the labour force in non-agricultural pursuits stood at 73% in the US and 67% in Canada. By 1939, the size of the agricultural labour force shrank considerably in both countries (80.2% in the US; 73% in Canada) but the spread between the two actually grew. Hence, any underlying economic uncertainty might spread to uncertainty about future commodity prices. One candidate for the originating source of the Depression could be the pre-depression and depression-period movement of commodity prices. Figure 4 plots some available commodity price data for the US and Canada. The general downward trend in commodity prices until about 1932 is evident in both countries. Despite some stability in commodity prices between roughly 1925 and 1929, commodities fell by over two-thirds of their value in 1919 or 1920. Indeed, commodity prices fell by over 50% between 1919 and 1921 which alone constitutes a substantial shock. But modern macroeconomics would not treat even such large movements in prices as a “shock” if they were anticipated. The reason is that events which are fully anticipated should not upset individuals’ plans even if, as a result, their incomes fell, and if the downturn was expected to be temporary. Other than for economic historians, many would be surprised to know that, in the space of 40 years (from 1876 to 1922), the US experienced 12 years of deflation. During the same period Canada recorded a deflation on an annual basis 18 times.12 To what extent then were price changes during the 1920s, in particular, anticipated? Cecchetti (1992), Hamilton (1992), and Evans and Wachtel (1993) pose just such a question. Cecchetti (1992) points out that the historical behaviour of inflation in the US was such that the deflation of 1930-32 could have been anticipated. A straightforward explanation involves the notion that inflation and deflation are predictable variables because they are persistent. However, 10 there have been no comparisons of persistence between the US and Canada, nor have there been any comparisons between the deflations of the early 1920s and 1930s. Persistence means that inflation depends heavily on its immediate past history. A simple representation of this idea is to write π t = α + β π t −1 + εt (6) where π t is the inflation rate, α is a constant, β is the degree of persistence in inflation since it measures how last period’s inflation rate affects current inflation. Finally, ε t measures the error made in specifying that only past inflation affect current inflation. If we estimate (6) for the US and Canada we obtain the estimates provided in Table 2. We immediately notice that the degree of persistence did not change dramatically before and after 1928 in the two countries.13 Thus, for example, for both the US and Canada a 1% rise in last year’s inflation rate would raise this year’s inflation rate by roughly ½%, and vice-versa.14 To be sure, other variables might be helpful in explaining the current behaviour of inflation but we can go a long way by simply assuming that last year’s inflation rate carries with it considerable momentum. Put differently, asking whether the behaviour of prices in the 1930-32 period could be answered by using the estimates for the 1919-28 sample to forecast inflation for the years 1930-32. Figure 7 shows the results of this exercise. Clearly, (6) implies that the deflation was forecastable with essentially a one period lag. Of course, the term ε t in (6) implies that forecast errors are probable so that some portion at least of the deflation was a surprise. If so, can a more sophisticated model improve on forecast accuracy and, if economic agents are assumed to have, in effect, used such a model, would the deflation have been a surprise? Dominquez, Fair and Shapiro (1988) conclude in the affirmative. Burdekin and Siklos (1999) also rely on recent advances in econometrics to estimate whether there may have been a significant change in inflation persistence in several countries using over a century of data. They find significant breaks in US inflation in 1926 and 1933 while for Canada a break is found in 1930. Their findings suggest that if economic agents did not revise their model of inflation, it becomes even more likely that the deflation, to a significant degree, was unanticipated. 11 Hamilton (1992) also presents evidence that, despite the drop in commodity prices in the 1920s discussed earlier, markets in the US did not anticipate a deflation of the magnitude which actually occurred. Quite the contrary. Based on futures markets price data for key commodities, Hamilton’s analysis suggests that markets were betting at first on price increases. Futures markets are, of course, nowadays widespread in financial assets but their origins can be traced long ago to commodity markets. If these markets operate efficiently then a futures contract represents the market’s estimate of the price of a particular commodity expected to prevail in the future.15 In symbols, f j ,t − Et S j ,t +1 = εt (7) where fj,t is the futures price for commodity j at time t, Et Sj.t+1 is the market’s forecast, conditional on information at time t of commodity j’s price in the spot market next period (i.e., time t+1) and ε t is a “residual” which is defined to have a zero mean and a constant variance. A version of market efficiency requires that ε t fluctuate randomly. This hypothesis did not apparently apply to the markets examined by Hamilton, at least at first.16 Comparable data for Canada are not available but one can speculate that the US results would hold there as well. Evans and Wachtel (1993) point to uncertainty about the stance of monetary and fiscal policies to make the point that the deflation of the early 1930s, which was then followed by inflation, were both unexpected. Hence, if we view events after both economies reached the bottom, around 1933, as signaling a regime change then the process driving inflation would have changed. While proponents of the rational expectations hypothesis would argue that models such as (6) or (7) are at fault, because they presume that both the underlying relationship and the coefficients remain constant, Evans and Wachtel (1993) would argue that unexpected forecast errors remain possible so long as economic agents are unsure, in a probabilistic sense, about what drives future inflation. In particular, if individuals expect the deflation to end, and inflation to return, then it is conceivable that the severity of the deflation would be underestimated. While the argument seems persuasive, it ignores that deflation was a “normal” state of affairs since at least 1870. Therefore, disagreements persist over whether the deflation of the early 1930s and the 12 subsequent inflation need have been largely unanticipated.17 Nevertheless, the size of the forecast errors in this period suggest that they accumulate sufficiently over time to translate into a “large shock”. Friedman’s well-known theory of permanent income 18 provides additional perspective on the issue. It predicts that temporary changes in income should have little influence on permanent income. Hence, a shock to aggregate consumption would be signaled by a large and protracted fall in income relative to permanent income. Figure 5 shows an estimate of the ratio of per capita income to permanent income. There are some interesting features in the data. First, actual incomes are persistently below their long-run or permanent levels throughout the 1919-27 era in both countries. Second, the departures from permanent income are more prominent for the US than for Canada. Third, measured incomes begin to rise sharply relative to permanent income in both countries, beginning in 1933. Given the sharp drop in consumption, especially in the 1928-32 period, it is conceivable that the drop is measured to permanent income, which occurred simultaneously, is symptomatic of an unexpected shock. Since the ratio plotted in Figure 5 also shows a sizeable drop during the earlier deflation of the 1920s this still begs the question why a second deflation, temporally close to the first one, was such a surprise. Clearly, how one interprets individual behaviour, and the channels through which these lead to macroeconomic consequences of a particular variety, matter a great deal to the outcome of the debate over the degree to which the deflation was anticipated. Temin (1989), following Keynes (1931), places relatively more emphasis on the role of investment as the impulse for the chain of events which led to the Great Depression. An obvious candidate indicator of the real cost of borrowing money is the real interest rate. The higher real interest rates can be explained in part by the sheer size of the deflation and, not coincidentally, by central bank policies more concerned about preventing borrowing for “speculative” purposes but which, nevertheless, had the effect of choking off borrowing for more “productive” investments. Although the empirical connection between interest rates and investment has been found to be multifaceted and complex (e.g., see Romer (1996, chapter 8)) 13 the estimates presented in Figure 6, however imprecise, suggest that real interest rates were exceedingly high by historical standards. As shown in Figure 6 they averaged 5.37% in the US during the 1922-1930 period19 and 10.95% in Canada in the 1929-32 period. An additional indicator of future economic activity, much touted in recent financial economics literature, is the spread between long and short-term interest rates, also plotted in Figure 6.20 The potential link between the yield spread and future output operates along two dimensions. If short-term interest rates reflect the stance of monetary policy, then, for given long-term interest rates, a reduction in the spread would, in time, signal a contraction in economic activity as sectors of the economy sensitive to interest rates reduce spending. Alternatively, if firms expect future investment opportunities to improve, they will usually borrow in long-term market for funds. The resulting increase in the supply of bonds reduces bond prices thereby raising current long-term yields. As such, a rise in the spread would be associated with a future upturn in economic activity. As is clear from Figure 6, the spread in the US fell sharply in the years leading up to the Great Depression (until about mid-1929). In Canada, the only time the spread might give an indication of a fall in future economic activity is in 1931. Otherwise, the spread presages increases in future output. Either the message in the spread is misleading, or the differences between the US and Canadian experiences have their origins in the structure and response of the respective countries’ financial sectors. We explore this question in the following section. If the impulse for the Great Slump was the deflation in commodity prices (and the price level more generally) the role played by “ideology of the Gold Standard”, as Temin (1989) would perhaps emphasize, is not as clearcut as might appear to be the case at first glance. After all, Canada’s commitment to the Gold Standard in the interwar years was not as great as in the US. Moreover, the Act creating the Federal Reserve System prohibited it from inflating the economy by restricting central bank credit to be no larger than 2 ½ times the gold certificates held by the reserve banks. There was apparently less concern about the risks and consequences of deflation. With the price of gold fixed at $20.67/oz., the Fed felt mandated to use tight monetary policy to offset the potential inflationary effect from the resulting inflow of gold. In the 14 meantime Fed officials looked at the stock market’s performance during the mid-1920s, together with the expansion of credit, and concluded that monetary policy was becoming too easy.21 At this point the commitment to the Gold Standard did play a crucial role, especially given the attitudes of the French, who were fiercely committed to the monetary regime. The resulting tightening of monetary policy, evident from Figure 6, suggests that a form of international cooperation did work. Canada’s close financial and economic links meant that it would inherit the US experience. Nevertheless, if the impulse to the Depression was a combination of real factors built upon faulty institutional foundations, modern macroeconomic analysis also makes it clear that the propagation mechanism was financial in nature.22 Not to be forgotten in all this is the role of the exchange rate. As Eichengreen (1995, p. 240) points out the Canadian dollar ceased to be convertible after 1929. Floating exchange rates should, at least in theory, have insulated the Canadian economy for an “shock” emanating from the US. Yet, the $C/$US exchange rate was, in appearance, relatively stable between 1929 and 1931. Bordo and Redish (1990) argue that the Canadian government was intent on ensuring stability for at least two reasons. First, because of a long historical adherence to Gold Standard principles; second, because of large $US denominated debt. This state of affairs is somewhat puzzling since, some research has demonstrated that, during the Great Depression, countries with flexible exchange rates suffered milder contractions than countries with either fixed exchange rates or ones who adhered to the Gold Standard (Choudhri and Kochin 1980). While it is conceivable that credibility and political imperatives provided exchange rate stability it is less clear why there were deemed preferable over the maintenance of the Gold Standard orthodoxy. In the absence of credible commitment to exchange rate stability the behaviour of the exchange rate could reflect a “peso problem”. This refers to the phenomenon, apparently coined by Milton Friedman, to explain the performance of the Mexican peso during the 1970s which failed to be devalued despite widespread expectations that it would.23 After all, while average exchange rates were stable during the period in question, they were also considerably more volatile than in other years (figure not shown). 15 4. The Great Depression as a Financial Crisis: Canada versus the US As noted in the preceding section, deflation meant that real interest rates rose even in the face of relatively low nominal interest rates.24 But lower prices also raise the real cost to borrowers of repaying outstanding principal. Prospective borrowers, if they feel that prices are expected to fall further will also find credit to be expensive. In addition, default rates will be exacerbated if the deflation is unanticipated leading potential lenders to be scared away from making loans, thereby producing a severe “credit crunch”, to use the modern term. The foregoing ingredients are harbingers of a financial crisis and place the banking system at the centre of the story of the Depression. While Kindleberger (1986) was one who emphasized the role of financial crises in the present historical context, Bernanke (1983) is credited with working out the analytical details of how the Great Depression led to a massive failure of the credit allocation process (also see Bernanke (1995)) based on a hypothesis first put forward by Fisher (1933). Not to be forgotten is monetary policy. After all, few economists would nowadays disagree with Friedman and Schwartz’s (1963) view that the sharp contraction in the money supply was the main causal factor in the chain of events that led to the great contraction. In fact, Hamilton (1987) makes the persuasive case that the monetary contraction led to an unanticipated deflation, and hence the financial crisis which ensured, together with the attendant high real interest rates, the collapse of intermediation services. Figure 8 shows just how massive the monetary contraction was in both countries. However, a puzzling feature of the data is that Canada’s money supply peaks over a year earlier than in the US. So far as I am aware, there is no explanation for this result in the literature. If monetary policy and the banking system are important ingredients in the story, then do differences in institutional structure matter? There were two notable differences between the US and Canadian financial systems. First, until 1935, Canada had no central bank. Second, US banking laws aimed at preventing banks from forming oligopolies. The Canadian banking system permitted branch banking while the US system placed severe restrictions on this type of activity, 16 effectively encouraged such behaviour.25 As a result, Bordo, Rockoff and Redish (1994) report the relatively greater stability of the Canadian banking system over the US one in the sense of a reputation for soundness (also see Grossman (1994) in this connection). This translates into a lower probability of failure. Stability is then synonymous with the absence of bank failures. The importance of this feature lies in the concept of asymmetric information and the particular role it plays in the financial system. If insiders in the banking system are thought to possess a relatively better understanding of the state of the financial system than the outsiders (e.g., depositors), a shock to a particular bank or segment of the financial system may lead outsiders to paint all participants with the same brush, thereby prompting a run on the banks. This would lead to a further contraction of the money supply, through the well-known multiplier process,26 leading to a further diminution of the intermediation process. The consequences of asymmetric information for the severity of the Great Depression was noted by Mishkin (1978). While the number of bank offices or branches fell off substantially in both countries, and bank runs, whose social costs continue to be debated (e.g., Calomiris and Mason (1997)), figured prominently, the Canadian experience was marked by a contraction in the size and reach of existing banks (from 18 in 1920 to 10 in 1929), as only one bank, the Home Bank, failed in 1923. Thus, adjustment in Canada occurred mainly via mergers and acquisitions and not widespread bank failures as in the US.27 It is, among other factors considered, an important element in the story of the Great Slump because Haubrich (1989) argues that the contraction in intermediation services did not have a separate impact on Canadian output as did the US crisis in banking (also see Calomiris (1993)). Hence, whereas the costs of providing intermediation services should not have risen in Canada they did so in the US and so further exacerbated the output decline. Yet, banks in Canada (and the US) did engage in portfolio reallocation as loans declined by over 50% from 1929 until 1936 while holdings of government securities rose about four-fold over the same period.28 A more satisfactory explanation may be that Canadian depositors had more confidence than their American counterparts in the efficiency and productivity of the financial system. As a result, the US banking system was perhaps more 17 “fragile” than the Canadian one. While there is no evidence to support this hypothesis for Canada, Cooper and Corbae (1997) develop a model to suggest that the US banking system was fragile at the outset of the Great Depression. So we are still left with the perhaps startling conclusion that the much-touted stability of the Canadian banking did not make the country immune to the severity of the Great Contraction. Therefore, other factors must be at play. How important was the absence of a central bank (i.e., lender of last resort)? Under a gold standard rule, the automatic nature of the system would suggest the lender of last resort function is rather unimportant or unnecessary. As we have seen, however, the “rules of the game” were not strictly followed. In the Canadian context, monetary policy actions were influenced by government via the Finance Act of 1914, originally meant to assist the agricultural sector through crisis, but ended up extending loans more generally. The Government of Canada essentially acted as a lender of last resort, though it was a modest role at best.29 An additional source of funds for the Canadian banking system were reserves held in US dollars. Hence, while other countries (e.g., France) resisted treating the US dollar as a reserve currency, Canada did not.30 Moreover, Bordo and Redish (1987) argue that political considerations, not the desire to have a lender of last resort after the Gold Standard ended, explain the emergence of a Canadian central bank in 1935. Indeed, the creation of the Bank of Canada was a direct outcome of reforms is viewed as one of the ingredients in the recovery from the Great Depression necessitated by the need for instruments of monetary policy in a world without the automatic constraints provided by the Gold Standard (Shearer and Clark (1984)). Nevertheless, it is worth noting that in spite of institutional differences between the US and Canada in the realm of monetary policy, economic performance in the two countries is similar so it is hard to believe that the transmission of economic shocks from the US to Canada were insignificant (Siklos (1999)). 5. The Consequences: Reform and Recovery If economists are interested in the impulse and propagation of the Great Depression, they are equally interested in the timing of the recovery from that same event. As shown in 18 Figures 3 and 4 both economic activity in general, and commodity prices in particular, rebounded quickly and sharply from Depression era lows. As for explanations for this outcome, there is relatively less debate. Friedman and Schwartz (1963) argue that the deposit insurance in 1934 in the US marked the turning point. Wigmore (1987) suggests instead that the end of the Gold Standard in April 1933 marked the beginning of recovery. Since these two events are temporally close to each other it seems difficult to empirically identify the relative importance of one impulse over the other.31 Vernon (1994) points a finger at the enhanced role of fiscal policy in the aftermath of the slump. Eichengreen and Sachs (1985) and Choudhri and Kochin (1980), place a great deal of emphasis on the exchange rate regime, though neither article considers the Canadian experience. Cole and Ohanian (1999, 2001), however, are an exception. They argue that the recovery from the Depression was much milder than expected, and cannot be explained by the neoclassical approach. They “...conjecture that government policies toward monopoly and the distribution of income are a good candidate...” for the weaker than expected recovery, a theme also emphasized by Prescott (1999) who also cautions that international evidence is necessary. Beyond the initial impetus for change, following the earlier debacle in monetary policy, lay almost a decade’s worth of reforms that, as Timberlake (1993, p. 274) puts it: “... is almost too much for mortal mind to assimilate, let alone explain.” The abandonment of the Gold Standard and the centralization of Federal Reserve power at the Board of Governors, embodied in the passage of the Banking Act of 1935, were the highlights of monetary reforms of the 1930s. By contrast, the recovery in Canada was rather more placid with the introduction of the Bank of Canada being the major milestone. Unlike the US experience, deposit insurance was not introduced in Canada.32 It is interesting to note that studies referred to earlier which purport to show that the impulse for Canada’s Depression was not the US experience fail to come to grips with why Canada’s recovery so closely paralleled that of the US. Was it part of a natural secular renewal reflecting the depths of the downturn? Was international cooperation, in the form of the World Economic Conference in June 1933, a factor in the return to growth? In Eichengreen’s (1995) view, the Conference was an “utter failure”. Rather, the Tripartite agreement between the 19 US, the UK and France, combined with a large US devaluation undertaken by President Roosevelt, interpreted as a regime change in Temin’s (1989) view, were catalysts in freeing major economies from their “golden fetters”. Yet, there are aspects of the recovery that have not been fully investigated. For example, the apparent asymmetry in business cycles has not been fully explained nor the role played by institutional considerations or government involvement in the process. It seems clear, however, that recovery was not as complete in Canada as in the US (e.g., see Figure 3). The incompleteness of the Canadian recovery might suggest that an important element in its economic performance originates from domestic sources. Yet, both capital flows (Obstfeld and Taylor (1997, Table 2.1)), and the terms of trade (Safarian (1959, chapter 5)), recovered slowly so the international transmission of business cycles should not be underestimated. All studies of the recovery must contend with the “shock” of World War II which, in historical terms, came shortly after the bottom of the Depression had been reached. In part for this reason an alternative approach asks: What if policy makers had not been blinded by the ideology of the Gold Standard? In other words, what if the Great Depression had never happened? Speculations of this kind require constructing a “counterfactual” experiment. A simple example is Eichengreen’s (1995, chapter 10) argument that an open market operation (i.e., buying of government bonds) would have offset the fall in the money supply in 1931-32 in the US. A more elaborate approach (e.g., Bordo and Eichengreen (1997)) supposes that “stable” monetary policies in a (fractional) Gold Standard would have averted disaster including the unexpected deflation of the early 1930s. Unfortunately, the authors are unable to nail down the impact of the No Great Depression hypothesis on overall economic performance. Instead, they speculate that the Gold Standard in some form would have lasted beyond World War II. Bordo, Choudhri and Schwartz (1998) also use a counterfactual experiment to show that an expansionary monetary policy in the 1931-33 period would have averted not only the banking panics but the consequent loss of gold reserves would not have been large enough to reduce the gold ratio below the stationary minimum requirement. Interestingly, none of the counterfactuals consider how an 20 earlier introduction of a large fiscal stimulus may also have avoided the onset of the Great Slump. One of the most striking examples perhaps of a counterfactual experiment is the one suggested by Sims (1998). He asks: what would the behaviour of the Federal Reserve, as represented by movements in interest rates it can control or influence, have been during the inter-war era (1919-38), a sample that overlaps the period of the Great Depression, if central bank policy had been conducted along the same lines as in the post World War II era? Sims concludes that “better” monetary policy could not have prevented the Great Depression. The surprising result, however, is open to two strong objections. First, the structure of the economy during the inter-war years cannot possibly be compared with the one prevailing in the post 1945 era. Second, Sims’ approach does not explicitly take into account the fact that the 1919-38 period was marked by a string of financial crises while the post World War II era, for the most part, was not similarly afflicted. While interesting, counterfactuals assume that correct monetary decisions would have been possible. But, as the masterful works of Eichengreen (1995), Friedman and Schwartz (1963) and Temin (1989) show, one ignores the impact of the intellectual milieu of the era at some risk. Yohe (1990), Timberlake (1993), and Calomiris and Wheelock (1997) provide fascinating accounts of how ideology is a notion not easily embedded in any model. Notwithstanding the internal consistency of the Gold Standard system, simulations also presume that, if only a better policy option has been presented to decision-makers, the Great Depression could have been avoided. This type of optimism is not shared by those, such as adherents to the modern political economy literature, who argue that only clearly spelled out, accountable, and binding institutional constraints will lead to desirable and credible economic outcomes. There is, of course, another way out of the dilemma facing policy makers in how much self-imposed institutional constraints should be placed on them. Thus, for example, if an autonomous monetary authority – deemed desirable – implements “bad” policy, the fiscal authorities may simply decide not to “play the game” and introduce policies to offset the negative outcomes from the prevailing monetary policy. But then there still remains the potential 21 for economic activity to become more volatile than through institutional devices which ensure that “good” monetary policy is practiced. 6. Concluding Remarks Almost seven decades after the end of the Great Depression, economists have learned much from the Canadian and US experiences about how economic shocks are transmitted across countries and the role and importance of personalities who make and implement economic policies. Is there a consensus then amongst economists about why the Slump took place, why it was so severe, and the ingredients of the recovery? Eichengreen and Temin (1997) would have us believe this is so. Contrary to historians, the authors suggest that there “... now prevails a remarkable degree of consensus among specialists about the causes of the Great Depression”. They return to the themes listed in the opening section of this paper, namely the ideology of the Gold Standard and central bankers’ attitudes about the proper policies in the face of economic shocks. While this much is probably true there is still room for disagreement amongst economists. Whaples (1995) reports what is, in effect, a lack of consensus on the causes of the Great Depression. Based on a survey of 178 US members of the Economic History Association (EHA), only 31% of historians in the EHA agreed or partially agreed with the notion that monetary forces caused the Depression, while 52% of economists in the EHA agreed with this principle. By contrast, 61% and 51% of economists and historians, respectively, traced the impulse to the Great Slump to a real shock. Consensus, if any, appears greater about Temin’s great shock hypothesis than Friedman and Schwartz’s view. Interestingly, far more consensus was found concerning the proposition that the right mix of policies could have prevented the Depression (75% of economists and 78% of historians agreed with this idea). Indeed, this survey has also suggested that there is disagreement about the extent to which shocks are transmitted across countries. 22 The foregoing survey of views suggests that it is clearly unsatisfactory to suggest that a single cause is responsible for such a cataclysmic global event. Nevertheless, what is striking about the US-Canadian experiences is that, despite differences between the two countries’ financial structures, both countries suffered a tremendous slump. Either institutional structures are not sufficient to insulate a country from large foreign shocks or the role of institutional factors, and how economic policy can short-circuit their role in the Depression and its aftermath, is not yet sufficiently understood. One thing is certain, however. Regardless of how the Great Depression began and whatever the significance of US shocks on the Canadian economy, the principal lesson of the interwar years is that, in the area of monetary policy, ideology is no substitute for good judgement. One can only hope that the availability of better models and data that permit more objective assessments of the likely future course of the economy, an institutional by-product of the Great Slump as it happens, reduces the obstacles created by a slavish adherence to a particular economic ideology. 23 ENDNOTES 1 In its purest form, the cost of holding money is part opportunity cost, part cost of storing gold reserves, part cost of transacting in gold coins. 2 As argued by one of the supporters of the Gold Standard, Sir John Bradbury, at the time Churchill was debating whether or not the UK should return to gold. 3 David Hume, the famous 18th century Scottish economist, first described the process. “Specie” is gold, silver, or other precious metal used as money. 4 Trade and investment ties between the US and Canada were also strong, of course. During the period in question roughly 2/3 of imports were from the US while approximately 60% of foreign capital invested in Canada came from US sources. But ties with the UK also remained strong since over 1/3 of investment in Canada came from the UK while over 1/3 of Canadian exports went to the UK. Imports from the UK, by contrast, accounted for less than 20% of Canada’s total imports. Unfortunately, there are limitations in the available data which make it difficult to determine precisely how trading patterns were affected by the Great Depression. 5 Thus, financial dislocation, in the form of reduced, though still large, capital flows suggests one avenue through which the Great Slump may have been transmitted from the US to Canada. Yet, there is a suspicion that there is more to it than that since Figures 1 and 2 suggest that other forces may also have been at work. 6 While it is hazardous under the circumstance to compare the relative sizes of the two economies, the sheer size of the US economy is surely a factor by itself. US nominal GDP over the 1926-37 period was approximately 18 times Canada’s nominal GDP. 7 Indeed, Diebold and Rudesbusch (1990) find substantial differences in the duration of pre World War I versus post World War II business cycles. I do not discuss the comparative behaviour of unemployment rates in the two countries. Zagorsky (1998) shows that existing data, which purport to show significantly lower unemployment rates in Canada during the Great Depression, are misleading. Each country handled depression relief workers differently. The US treated these individuals as unemployed, Canada did not. 8 Peaks in the US business cycle are usually dated 1 to 5 months before those in Canada, while the dating of troughs is usually 1 month ahead in the US. 9 Indeed, comparisons between the 1920s and 1930s deflations, using a general equilibrium business cycle model, leads Cole and Ohanian (2001) to suggest that the Great Deepression remains a puzzle. In contrast, Bernanke (1995) argued that, while aggregate demand side related questions about the Slump were largely answered, there is as yet no convincing story about the aggregate supply response to the Depression. 11 Readers wishing a slightly more technical description of the problem are asked to consult the appendix. Sumner (1992) suggests that news contradicting the public’s expectations about the likely degree to central bank cooperation may have contributed to heighten uncertainty about the future monetary policy. A proxy for “uncertainty” would be the standard deviation in either a measure of real income (e.g., real GDP) or in a stock price index. Modern empirical macroeconomics tends to rely on more sophisticated measures of volatility derived from the variance of the residuals from some regression estimate 24 describing the relationship between economic series of interest. A fuller analysis of this procedure is, however, beyond the scope of this paper. 12 Based on data in column (2) of Table B.2 in Dick and Floyd (1992). 13 The choice of 1928 is arbitrary in a sense but it has the advantage that it omits the period most economists would describe as the Great Depression period. 14 Various other tests were conducted, including the pooling of US and Canadian data into a cross-section time series format. All such tests yielded the same result, namely that inflation persistence was roughly the same in the US and Canada and approximately the same in the 1920s as in the 1930s. 15 That is, at the time the contract expires at which time the relevant commodities are delivered at the agreed upon price. 16 For interested readers, an Appendix figure plots Hamilton’s data. t for the case of corn and oats, using 17 Indeed, using model (2), estimated during the period 1919-28, continues to perform quite well when forecasts for the 1935-37 period are made. 18 According to this theory, consumers look beyond their current level of income in determining consumption and savings levels. See Siklos (2001, Chapter 9). 19 The average real interest rate in the US rises to 10.21% if we begin calculations in 1920. 20 Bonser-Neal and Morley (1997) provide a thorough and introductory exposition on the subject. 21 For example, inflows of gold totaled 1432 millions between 1921-24. Meanwhile, loans by investment brokers to their clients rose 400% between 1919 and 1928 while installment credit rose almost 300% between 1919 and 1928. Based on figures in Historical Statistics of the United States: Bicentennial Edition (1997 CD-ROM), series X547 and X551 and Board of Governors of the Federal Reserve System (1976), Table 158. 22 An aspect of the debate not discussed here is the role played in the propagation mechanism by the flexibility of nominal wages. Unlike the apparent downward rigidity in nominal wages prevalent today, there was little such rigidity in the interwar era. Bernanke and Carey (1995) point out that Eichengreen (1995), for example, makes only a passing reference to the development of real wages during this era. See also note 4 above for an explanation of data limitations in the Canadian context. 23 A similar occurrence took place after Canada devalued the dollar in 1949. The Canadian dollar floated until 1961 over objections by the International Monetary Fund about violating the Bretton Woods monetary standard. At the time of devaluation the Canadian dollar was set at 90.9¢ US. In 1961, when the $C was pegged anew the rate was 92.5¢. 24 Obviously, nominal interest rates cannot go below 0% but real interest rates can rise with the severity of the deflation. 25 Through various regulatory measures meant to significantly raise the cost of entry into banking. See Siklos (2001, chapter 17). Wheelock (1992, 1995) also provides a thorough review of the US financial system during this period. 25 26 See Siklos (2001, chapter 16) for a basic introduction to the concept of a money multiplier. 27 If the adjustment was somehow assisted by a government bail-out then the stability of the Canadian banking system was a myth. This is the argument of Kryzanowski and Roberts (1989) but is persuasively refuted by Carr, Mathewson and Quigley (1995). Kryzanowski and Roberts (1999) reply that when one corrects for the difference between market value and book value on a bank’s balance sheet, some Canadian banks during the Depression were technically insolvent. 28 Based on data in Urquhart and Buckley (1983), series J191, J146 and J138. 29 Advances under the Finance Act never exceeded $82 million CDN (in 1929), less than 2% of bank assets at the time. Based on data in Urquhart and Buckley (1983), series J182 and J198. 30 For this reason, authors such as Rich (1988), and Dick and Floyd (1995), define the Canadian monetary base to include bank reserves of US dollars. 31 Coe (1998, chapter 3) uses some sophisticated econometric technique and concludes that the introduction of deposit insurance is the event which launched recovery in the US. 32 Siklos (2001, chapter 19) discusses the successful lobbying by Canadian chartered banks to prevent the introduction of deposit insurance in Canada (until 1967). 26 TABLE 1 TRADE, CAPITAL AND MONETARY MOVEMENTS: CANADA, 1927-1937 Year Current Account Balance Capital Account Balance Net Official Monetary Movements (Millions of dollars) Gold Standard Era 1927 1928 1929 1930 1931 Post-Gold Standard Era 1932 1933 1934 1935 1936 1937 -10 -32 -311 -337 -174 +3 +77 +62 -19 +54 +7 -49 -37 +36 -33 -96 -2 +68 +125 +244 +180 +60 +33 +27 +32 -2 -17 -3 -6 +4 +6 +5 +6 Source: Urquhart and Buckley (1983), series G83, G109, and G115. A positive sign implies a surplus in the current or capital account balances; a negative sign a deficit. A positive sign in Net Official Monetary Movements means a net inflow of gold; a net outflow occurs when the sign is negative. After 1935, the figure also includes foreign currency holdings held by the Bank of Canada. 27 TABLE 2 MODELLING THE BEHAVIOUR OF INFLATION Dependent Variable: Annual Inflation Rate US Independent Variables Constant Lagged inflation rate 1919-28 -3.05(5.26) .64(.07) 1919-37 -2.33(2.92) .54(.06) Canada 1919-28 -.17(2.22) .54(.08) 1919-37 -.96(1.20) .40(.06) Source: Inflation is the annual inflation rate calculated as (log P t - log P t-12 ) x 100. The Wholesale Price Index was used for the US; the Consumer Price Index for Canada. Lagged inflation is the previous year’s inflation rate. Standard errors (Newey-West corrected) are in parentheses. 28 FIGURE 1 SHORT-TERM INTEREST RATE MOVEMENTS IN CANADA AND THE UNITED STATES, 1930-37 Short-term Interest Rate Movements in Canada and the United States 6 Gold Standard 5 Percent 4 3 US Canada 2 1 0 30 31 32 33 34 35 36 Note: For the US, the yield on short-term government securities. For Canada, the yield on short-term Dominion of Canada bonds. Data on short-term interest rates for Canada not available prior to 1930. The Gold Standard is dated as ending in April 1933. Sources: For the US: Cecchetti (1988). For Canada, Nixon (1937). 29 FIGURE 2 PRICE LEVEL MOVEMENTS IN THE US AND CANADA, 1925-1931 US Wholesale Prices, 1926=100 140 130 100 120 90 110 80 100 90 70 26 Note: 27 28 29 30 Canada Wholesale Prices, 1935-39=100 110 US Canada 31 Monthly data on Wholesale Price Index Movements, April 1925-August 1931, inclusive. The Gold Standard was in effect throughout. Although the left hand and right hand side scales differ the plots are used to make inferences about the changes or the evolution of price changes, not their levels. Source: Brecher and Reisman (1957), Table 1 (Canada), Table 16 (US), Appendix A. The series are not seasonally adjusted. 30 FIGURE 3 BUSINESS CYCLES IN THE US AND CANADA US Business Cycles 70 Gold Standard > ends Index 1947-49=100 60 50 Ind. Ption 40 30 20 20 22 24 26 28 30 32 34 36 Business Cycles in Canada 140 Gold Standard > ends Index 1935-39=100 120 100 Ind. Ption 80 60 40 20 Sources: 22 24 26 28 30 32 34 36 Industrial Production: Brecher and Reisman (1957), Table 2 (Canada), Table 11 (US), Appendix A. Reference cycle dates: www.nber.org (US), and Hay (1967, Table 1) for Canada. The shaded areas refer to the peak to trough periods (recession). 31 FIGURE 4 SELECTED COMMODITY PRICES IN THE US AND CANADA Selected Commodity Prices: Canada, 1919-37 100 250 80 200 60 150 40 100 20 Wheat, cents per bushel Wood, WPI (1935-39=100) Oats, cents per bushel Gold Standard > ends Oats Wheat Wood 50 20 22 24 26 28 30 32 34 36 Selected US Commodity Prices: US, 1920-37 160 100 Gold Standard > ends 80 120 100 60 80 60 40 Oats, cents per bushel Corn, cents per bushel 140 Corn Oats 40 20 20 20 Sources: 22 24 26 28 30 32 34 36 For Canada: Urquhart and Buckley (1983), series K38 (wood products and paper), M228 (wheat), and M233 (oats). For the US: Hamilton (1992). The original data were trimestrial and annual averages were calculated and plotted above. Also, see notes to Figure 2. 32 FIGURE 5 THE EVOLUTION OF ESTIMATES OF MEASURED TO PERMANENT INCOME: US AND CANADA Ration of Measured to Permanent per capita Income The Evolution of Estimates of Measured to Permanent Income: US and Canada 0.95 0.90 0.85 0.80 Canada US 0.75 0.70 Gold Standard period 0.65 20 22 24 26 28 30 32 34 36 Source: Bordo and Jonung (1987). Appendix 1B explains how their measure of permanent income is constructed. 33 FIGURE 6 SELECTED FINANCIAL INDICATORS IN THE US AND CANADA Selected Financial Indicators for the US, 1922-37 40 3 20 2 0 1 -20 0 -40 -1 Great Depression -2 22 24 26 28 30 32 Short-term int. rate less inflation Long-term less short-term yield 4 Spread Real Int. rate -60 34 36 25 1.5 20 15 1.0 10 0.5 5 0.0 0 -0.5 -5 Great Depression -1.0 Spread Real Int. rate -10 29 Note: Short-term int. rate less inflation Long-term less short-term yield Selected Financial Indicators for Canada, 1929-37 2.0 30 31 32 33 34 35 36 37 The spread is the yield on long-term government bonds less the yield on short-term government bonds. The real interest rate is the yield on short-term government bonds less inflation in the Consumer Price Index (Canada), Wholesale Price Index (US). Inflation is the log difference of the chosen price level evaluated over a three month horizon. The real interest rate calculation assumes investors predicted exactly the inflation rate over the holding period (assumed here to be three months). The period of the Great Depression is dated 1929 October - 1933 December. Sources: CPI for Canada, series P10000 from Statistics Canada. WPI for the US from Brecher and Reisman (1957), Table 16, Appendix A. Yield on short-term government securities from Board of Governors of the Federal Reserve System (1976), 1919-28, and Cecchetti (1988), Table A1 (US); Nixon (1937), Table VI for Canada. 34 FIGURE 7 ACTUAL AND FORECASTED INFLATION: US AND CANADA, 1930-32 US Annual inflation (%) 0 -5 -10 Actual inflation Inflation forecasts -15 -20 30:01 30:07 31:01 31:07 32:01 32:07 Canada Annual inflation rate (%) 4 0 -4 -12 -16 30:01 Note: Actual inflation Inflation forecasts -8 30:07 31:01 31:07 32:01 32:07 Inflation forecasts based on the application of estimates of (6) for the 1919-28 sample for both countries. Estimates are given in Table 2. Source: Table 2. 35 FIGURE 8 MONEY SUPPLY IN CANADA AND THE US: M2, 1919-37 45000 2400 40000 2200 35000 2000 30000 Great Depression 1800 20 Note: 22 24 26 28 30 32 Millionsof $US Millions of $CDN Money Supply in Canada and the US: M2, 1919-37 2600 Canada US 25000 34 36 For the US, M2 is the sum of currency held by the public and total deposits (demand and time). For Canada, M2 is currency and total deposits. The Great Depression is dated 1929-1933. Sources: Bordo and Jonung (1987), Courchene (1969), and Metcalf, Redish and Shearer (1998). 36 FIGURE A1 THE BEHAVIOUR OF SPOT AND FUTURES COMMODITY PRICES: US, 1920-37 Futures price less spot price, one period ahead (4 months) The behaviour of Spot and Futures Commodity Prices: US, 1920-37 80 60 40 20 Fut. less spot(t+4)-OATS Fut. less spot(t+4)-CORN 0 -20 -40 -60 20 Note: 22 24 26 28 30 32 34 36 Data from Hamilton (1992). Also, see notes to Figure 4. 37 FIGURE A2 THE EVOLUTION OF BANK OFFICES IN THE US AND CANADA, 1919-37 The Evolution of Bank Offices in the US and Canada, 1919-37 35000 4800 US No. Bank Offices 4400 4200 25000 4000 20000 3800 3600 15000 3400 10000 3200 20 22 24 Source: Bordo and Jonung (1987). 26 28 30 32 34 36 Canada - No. Bank branches 4600 30000 No. Bank Off. - US No. Bank br. - Canada 38 FIGURE A3 EXCHANGE RATE BEHAVIOUR The Highs and Lows of the Canadian/U.S. dollar Exchange Rate Canadian dollars per 1 U.S. dollar 1.30 Convertibility > ends 1.25 1.20 1.15 1.10 1.05 1.00 0.95 20 22 24 26 Average 28 30 32 High Source: Urquhart and Buckley (1983), series J560, J561, J562. 34 Low 36 39 REFERENCES Bernanke, B.S. and K. Carrey, “Nominal Wage Stickiness and Aggregate Supply in the Great Depression”, Quarterly Journal of Economics 111 (August 1996), 853-83. Bernanke, B.S., “The World on a Cross of Gold: A Review of ‘Golden Fetters: The Gold Standard and the Great Depression, 1919-1939'”, Journal of Monetary Economics, 31 (April 1993), 25167. Bernanke, B.S., “The Macroeconomics of the Great Depression: A Comparative Approach”, Journal of Money, Credit, and Banking 27 (February 1995), 1-28. Bernanke, B.S., “Non-Monetary Effects of the Financial Crisis in the Propagation of the Great Depression”, American Economic Review 73 (March 1983), 257-76. Betts, C.M., M.D. Bordo, and A. Redish, “A Small Open Economy in Depression: Lessons from Canada in the 1930s”, Canadian Journal of Economics XXIX (February 1996), 1-36. Board of Governors of the Federal Reserve System, Banking and Monetary Statistics, 1914-1941 (Washington: Federal Reserve, 1976). Bonser-Neal, C. and T.R. Morley, “Does the Yield Spread Predict Real Economic Activity? A Multicountry Analysis”, Economic Review of the Federal Reserve Bank of Kansas City, (1997:3), 37-53. Bordo, M., A. Redish, and R. Shearer, “Canada’s Monetary Systematic Historical Perspective: Too Faces of the Exchange Rate Regime”, working paper, University of British Columbia, April. Bordo, M.D., E.U. Choudhri and A.J. Schwartz, “Was Expansionary Monetary Policy Feasible During the Great Contraction? An Examination of the Gold Standard Constraint”, working paper, Carleton University, July 1998. Bordo, M.D., B. Eichengreen, and J. Kim, “Was There Really an Earlier Period of International Financial Integration Comparable to Today?” NBER Working Paper 6738, September 1998. Bordo, M.D. and B. Eichengreen, “Implications of the Great Depression for the Development of the International Monetary System”, NBER working paper 5883, January 1997. Bordo, M.D. and F.E. Kydland, “The Gold Standard as a Rule: An Essay in Exploration”, Exploration in Economic History, 32 (October), 423-64. Bordo, M.D., H. Rockoff, and A. Redish, “The U.S. Banking System From a Northern Exposure: Stability versus Efficiency”, Journal of Economics History 54 (June 1994), 325-41. Bordo. M.D. and A. Redish, “Why Did the Bank of Canada Emerge in 1935?” Journal of Economic History XLVII (June 1987), 405-17. Bordo, M.D. and L. Jonung, The Long-Run Behavior of the Velocity of Circulation (Cambridge: Cambridge University Press, 1987). Brecher, I. and S.S. Reisman, Canada-United States Economic Relations (Ottawa, Ont.: Royal Commission on Canada’s Economic Prospects, 1957). 40 Burbridge, J. and A. Harrison, “A Historical Decomposition of the Great Depression to Determine the Role of Money”, Journal of Monetary Economics 16 (1985), 45-54. Burdekin, R.C.K. and P.L. Siklos, “Exchange Rate Regimes and Shifts in Inflation Persistence: Does Nothing Else Matter?” Journal of Money, Credit and Banking, 31 (May 1999). Burns, A.F. and W.C. Mitchell, Measuring Business Cycles (New York: National Bureau of Economic Research). Calomiris, C.W. and J.R. Mason, “Contagion and Bank Failures During the Great Depression: The June 1932 Chicago Banking Panic”, American Economic Review 87 (December 1997), 863-83. Calomiris, C.W. and D.C. Wheelock, “Was the Great Depression a Watershed for American Monetary Policy?” NBE working paper 5963, March 1997. Calomiris, C.W., “Financial Factors in the Great Depression”, Journal of Economic Perspectives 7 (Spring 1993), 61-85. Carr, J., F. Mathewson and N.C. Quigley, “Stability in the Absence of Deposit Insurance: The Canadian Banking System, Journal of Money, Credit and Banking 27 (1995), 1137-38. Cecchetti, S.G., “The Case of the Negative Nominal Interest Rates: New Estimates of the Term Structure of Interest Rates During the Great Depression”, Journal of Political Economy 96 (6, 1988), 1111-41. Cecchetti, S.G., “Prices During the Great Depression: Was the Deflation of 1930-32 Really Unanticipated?” American Economic Review 82 (March 1992), 141-56. Cecchetti, S.G. and G. Karras, “Sources of Output Fluctuations During the Interwar Period: Further Evidence on the Causes of the Great Depression”, The Review of Economics and Statistics 76 (1994), 80-102. Choudhri, E.U. and L.A. Kochin, “The Exchange Rate and the International Transmission of Business Cycle Disturbances: Some Evidence from the Great Depression”, Journal of Money, Credit and Banking, 12 (1980), 565-74. Coe, P., Money, Output and the United States Interwar Financial Crisis: An Empirical Analysis”, Ph.D. dissertation, University of British Columbia, 1998. Cole, H.L. and L.E. Ohanian, “Re-Examining the Contributions of Money and Banking Shocks to the U.S. Great Depression”, Macroeconomics Annual 2000 (Cambridge, Mass.: The MIT Press, 2001 forthcoming). Cole, H.L. and L.E. Ohanian , “The Great Depression in the United States from a Neoclassical Perspective”, Quarterly Review, Federal Reserve Bank of Minnesota (Winter 1999): 2-24. Cooper, R. and D. Corbae, “Financial Fragility and the Great Depression”, NBE working paper 6094, July 1997. Courchene, T.J., “An Analysis of the Canadian Money Supply: 1925-1934", Journal of Political Economy 77 (1969), 363-91. Dick, T.J.O. and J.E. Floyd, Canada and the Gold Standard (Cambridge: Cambridge University pres, 1992). 41 Dominguez, K.M. R.C. Fair, and M.D. Shapiro, “Forecasting the Depression: Harvard verus Yale”, American Economic Review, 78 (September 1988), 595-612. Eichengreen, B., Golden Fetters (New York: Oxford University Press, 1995). Eichengreen, B. and P. Temin, “The Gold Standard and the Great Depression”, NBE working paper 6060, June 1997. Eichengreen, B. and J. Sachs, “Exchange Rates and Recovery in the 1930s”, Journal of Economic History, (1985), 925-46. Enders, W., Applied Econometric Time Series (New York: Jonn Wiley and Sons, 1995). Evans, M. and P. Wachtel, “Were Price Changes During the Great Depression Anticipated? Evidence from Nominal Interest Rates”, Journal of Monetary Economics 32 (1993), 3-34. Fackler, J.S. and R.E. Parker, “Accounting for the Great Depression: A Historical Decomposition”, Journal of Macroeconomics 16 (Spring 1994), 193-220. Fisher, I., “The Debt-Deflation Theory of Great Depressions”, Econometrica, 1(October 1933), 33757. Fremling, G.M., “Did the United States Transmit the Great Depression to the Rest of the World?” American Economic Review, 75 (December 1985), 1181-85. Friedman, M. and A.J. Schwartz, Monetary Trends in the United States and the United Kingdom (Chicago and London: University of Chicago Press, 1982). Friedman, M. and A.J. Schwartz, A Monetary History of the United States (Princeton: Princeton University Press, 1963). Garraty, J., The Great Depression (New York: Harcourt, Brace and Jovanovich, 1986). Gilbert, M., Winston S. Churchill, Volume 5 (Boston: Houghton Mifflin, 1977). Grossman, A.S. “The Shoes that Didn’t Drop: Explaining Banking Stability During the Great Depression”, Journal of Economic History 54 (September 1994), 654-82. Hamilton, J.D., “Was the Deflation During the Great Depression Anticipated? Evidence from the Commodity Futures Market”, American Economic Review 82 (1992), 157-78. Hamilton, J.D., “Role of the International Gold Standard in Propagating the Great Depression”, Contemporary Policy Issues VI (April 1988), 67-89. Hamilton, J.D., “Monetary Factors in the Great Depression”, Journal of Monetary Economics 19 (1987), 145-69. Hay, K.A.J., “Money and Cycles in Post-Confederation Canada”, Journal of Political Economy 75 (June 1967), 262-73. Haubrich, J.G., “Nonmonetary Effects of Financial Crisis: Lessons from the Great Depression for Canada”, Journal of Monetary Economics 25 (1990), 223-52. 42 Keynes, J.M., “An Economic Analysis of Unemployment (Harris Lectures)”, in The Collected Writings of John Maynard Keynes, vol. 13, Donald Moggridge, Editor (London: MacMillan, 1973). Kindleberger, C.P., The World in Depression, 1929-1939, Second Edition (Berkeley: University of California Press, 1986). Kryzanowski, L., and G.S. Roberts, “Canadian Banking Solvency, 1922-40", Journal of Money, Credit and Banking 25 (1993), 361-76. Kryzanowski, L., and G.S. Roberts, “Perspectives on Canadian Bank Insolvency During the 1930s”, Journal of Money, Credit and Banking, 31 (February 1999), 130-34. Lucas, R.E., Jr., Models of Business Cycles (Oxford: Basil Blackwell, 1987). Metcalf, C., A. Redish, and R. Shearer, “New Estimates of the Canadian Money Stock, 1871-1967", Canadian Journal of Economics, 31 (February 1998): 104-24. Mishkin, F.S., “The Household Balance Sheet and the Great Depression”, Journal of Economic History 37 (1978), 918-37. Nichans, J., The Theory of Money (Baltimore: Johns Hopkins University Press, 1978). Nixon, S.E., “Interest Rates in Canada: The Course of Interest Rates, 1929-1927", Canadian Journal of Economics and Political Science 3 (1937), 421-34. Obstfeld, M. and A.M. Taylor, “The Great Depression as a Watershed: International Capital Mobility Over the Long-Run”, NBER working paper 5960, March 1997. Prescott, E.C., “Some Observations on the Great Depression”, Quarterly Review, Federal Reserve Bank of Minnesota (Winter 1999): 25-31. Rich, G., The Cross of Gold: Money and the Canadian Business Cycle, 1867-1913 (Ottawa, Ont.: Carleton University Press, 1988). Romer, C.D., “The Great Crash and the Onset of the Great Depression”, Quarterly Journal of Economics 105 (August 1990), 597-624. Romer, C.D., “The Nation in Depression”, Journal of Economic Perspectives, 7 (Spring 1993), 19-40. Romer, D.,.Advanced Macroeconomics (New York: McGraw-Hill, 1996). Safarian, A.E., The Canadian Economy in the Great Depression (Toronto: University of Toronto Pres, 1959). Shearer, R.A. and C. Clark, “Canada and the Interwar Gold Standard, 1920-35: Monetary Policy Without a Central Bank”, in M.D. Bordo and A.J. Schwartz (Eds.), A Retrospective on the Classical Gold Standard (Chicago and London: University of Chicago Press, 1984), 277-310. Siklos, P.L., “US and Canadian Central Banking: The Triumph of Personalities Over Politics?” in C.-L. Holtfrerich, G. Toniolo and J. Reis (Eds.), The Emergence of Modern Central Banking (Leicester, UK: Scolar Press for the European Association of Banking History, 1999), pp. 231-78. 43 Siklos, P.L., Money, Banking and Financial Institutions, Third Edition (Toronto: McGraw-Hill Ryerson, 2001). Sims, C.A., “The Role of Interest Rate Policy in the Generation and Propagation of Business Cycles: What Has Changed Since the 1930s?”, in Beyond Shocks: What Causes Business Cycles? (Boston: Federal Reserve of Boston), pp. 121-160. Sumner, S., “The Role of the International Gold Standard in Commodity Price Deflation: Evidence from the 1920 Stock Market Crash”, Explorations in Economic History 29 (1992), 290-317. Temin, P., “Transmission of the Great Depression”, Journal of Economic Perspectives 7 (Spring 1993), 87-102. Temin, P., Lessons from the Great Depression (Cambridge: The MIT Press, 1989). Temin, P., Did Monetary Forces Cause the Great Depression (New York: Norton, 1976). Timberlake, R.H., Monetary Policy in the United States (Chicago and London: University of Chicago Press, 1993). Urquhart, M.C. and K.A.H. Buckley, Historical Statistics of Canada (Ottawa: Statistics Canada, 1983) Vernon, J.R., “World War II Fiscal Policies and the End of the Great Depression”, Journal of Economic History 54 (December 1994), 850-68. Whaples, R., “Where Is There Consensus Among American Historians? The Results of a Survey on Forty Propositions”, Journal of Economic History 55 (March 1995), 139-54. Wheelock, D.C. “Regulation, Market Structure, and the Bank Failures of the Great Depression”, Review of the Federal Reserve Bank of St. Louis (March/April 1995), 27-38. Wheelock, D.C., “Monetary Policy in the Great Depression: What the Fed Did, and Why”, Review of the Federal Reserve Bank of St. Louis (March/April 1992), 3-28. Wigmore, B.A., “Was the Bank Holiday of 1933 Caused by a Run on the Dollar?” Journal of Economic History 97 (1987), 739-55. Yohe, W.P., “The Intellectual Milieu at the Federal Reserve Board in the 1920s”, History of Political Economy 22 (3, 1990), 465-88. Zagorsky, J.L., “Was Depression Era Unemployment Really Less in Canada than the US?”, Economics Letters 61 (1998), 125-131. 44 Appendix – Identifying Shocks in Econometric Models: Basics Suppose the economist believes that the correct model (called “structural”) linking, say, US and Canadian output can be represented by the following equations: C C ytC = β10 − β12 ytUS + δ11 y US t −1 + δ12 y t −1 + ε t (2) US ytUS = β20 − β21 y tC + δ 21 y tC−1 + δ 22 ytUS −1 + ε t (3) where yC is a proxy for output in Canada, and yUS is US output. The terms εt and εt , respectively, are C US residuals, that is, output movements not explained by the other right hand side variables. Hence, if economic agents are assumed to operate with a model such as (2) and (3) then the residuals capture the influence of unexpected policies on output. The model hypothesizes that Canada’s output is determined by its own current and past history and the current and past history of US output.33 Notice that the model presumes a contemporaneous link between US and Canadian output (i.e., yt C is affected by yt US, and viceversa). When economists attempt to empirically estimate a system of equations such as (2) and (3), they find it convenient to transform them (i.e., reparameterize) into a reduced-form which eliminates the contemporaneous effect of yt C on yt US and of yt US on yt C. Why? Simply because equations (2) and (3) are not reduced form equations where the right hand side variables are all lagged relative to the dependent variable. Since the details of such transformations are not essential,34 we simply state the result: ytC = α10 + α11 y tC−1 + α12 y US t −1 + ε1t (4) C ytUS = α20 + α21 ytUS −1 + α22 y t −1 + ε 2t (5) If the economist is interested in the coefficients in (2) and (3), but estimates (4) and (5), then 6 coefficients are estimated (α10, α11, α12, α20, α21, α22) when 8 coefficients are of interest (β 10, β 12, β 20, β 21, δ 11, δ 12, δ 21, δ 22). Clearly, the “structural” parameters cannot be estimated unless we impose some economically sensible restrictions. For example, it is sensible to suppose that US output affects Canadian output, but not vice-versa. This would mean β 21=0. The number of structural parameters is therefore reduced and this solves our problem.35 45 33 To keep the analysis simple I only allow a one period lag. Extensions allowing more lagged terms would not fundamentally change the arguments to be outlined below. 34 See, for example, Enders (1995, chapter 5) for a description. 35 An additional problem arises because estimation requires that the residuals in (4) and (5) be “well-behaved” (i.e., uncorrelated with a constant variance). Unfortunately, this requirement often does not hold since many macroeconomic time series contain trends (e.g., see Figure 2 for the case of wholesale prices in Canada and the US). A common solution is to take first differences in variables such as yt C and yt US. However, if the econometrician simply estimates (4) and (5) in first differences then an additional restriction is potentially ignored, namely one which might link output levels in the two countries (i.e., yt C and yt US), and presumably describes their long-run relationship.