Financial Institutions: Failures, Insolvency, and Moral Hazard

Cliometric Sessions at 1990 ASSA Meeting--December 29, 8:00 AM

Lawrence Kryzanowski*
Professor of Finance
Concordia University


Gordon S. Roberts*
Bank of Montreal Professor of Finance
Dalhousie University


Current research on the interaction between the monetary and real sectors discusses extensively the experience of the Great Depression. Although, economists have advanced different explanations for its severity, widespread banking failures play a key role in competing theories.2

To elicit evidence from the Great Depression, numerous researchers have focussed on an important contrast between the United States and Canada: the two economies faced similar declines in output, but, in contrast with the U.S., there were no bank runs and no banks failed in Canada. The common explanation, which has evolved into a "stylized fact", links Canada's more positive experience to its branching system which promoted the growth of a few large banks which (due to diversification) remained solvent throughout the Depression.

Based on a reexamination of this stylized fact, the present paper argues that the diversification benefits arising from national branching were not primarily responsible for the absence of bank failures in Canada in the 1930's. We use market-value accounting to restate Canadian bank balance sheets for the period 1922 - 1940. Our analysis reveals that nine of ten Canadian banks were technically insolvent during the Depression.

The superior performance of Canadian banks in avoiding explicit failure should be attributed to the Canadian government's policy of monitoring performance, standing ready to lend to banks and most importantly, forcing failing banks into mergers with healthy banks. This policy provided an implicit guarantee that, after a major bank failure in 1923, no other bank would be allowed to fail.3 The role of national branching was to make such a policy feasible by reducing the number of banks, and lessening the degree of competition.

The paper begins with a review of prior analysis of branching and failures. We next discuss key features of Canadian banking in the 1920s and 1930s and develop the argument that an implicit guarantee of all deposits was in place.4

The third part of the paper employs market value accounting to estimate the year-end market values of assets and liabilities for each of the ten Canadian banks in existence during the Depression. Our estimates take account of credit risk and interest rate risk by asset categories.5

Rejecting the "stylized fact", this research finds that Canadian banks were actually insolvent and remained in business only due to the forbearance of regulators coupled with an implicit guarantee of all deposits--much like the recent situation in the U.S. savings and loan industry.6


In a key passage which has made the Canadian experience during the Depression well known as an example in research, Friedman and Schwartz [(1963): 352-3] state that bank failures "...were the mechanism through which a drastic decline was produced in the stock of money." They argue that there was a larger increase in the currency ratio in the U.S. because "...the bank failures made deposits a much less satisfactory form in which to hold assets than they had been before in the United States or than they remained in Canada."

Friedman and Schwartz (1963) examine three reasons for the 1930 bank failures in the United States: the decline in asset values, inaction by the Federal Reserve, and runs which forced liquidation at firesale prices.

While they are silent on why no runs occurred in Canada, Friedman and Schwartz [(1963):�352] began a tradition of linking the survival of Canadian banks (and implicitly, the absence of runs) with the branching system implying that diversification through national branching protected asset values as follows:

Canada had no bank failures at all during the depression; its ten banks with 3,000-odd branches throughout the country did not even experience any runs, although, presumably as a preventative measure, an eleventh chartered bank with a small number of branches was merged with a larger bank in May 1931.7

Later writers, who extend or dispute the conclusions of Friedman and Schwartz, continue the tradition of attributing the lack of runs to the branching system. In a chapter on export-driven economies, Kindleberger (1973), for example, cites Friedman and Schwartz as if the link were a well-known fact needing little elaboration.

The Canadian experience is also important to the advocates of privatizing deposit insurance.8 Ely (1988) discusses the experience of the 1920's as a backdrop to the 1930's. He argues that Canadian banks did better in both decades due to the advantages of unrestricted branching; namely, geographical diversification and greater operating efficiency. Specifically, Ely observes that:

The Canadian experience [in the 1930's], in which some banks operated hundreds of branches nationwide, demonstrates that widespread branching is especially safe and desirable during an era of severe price deflation.

The Canadian experience is also used by O'Driscoll [(1988):�177] to support the argument for private deposit insurance: As the Canadian experience in the 1930's illustrates, a nationally branched banking system with diversified assets can withstand even severe shocks, both real and monetary.

Bernanke (1983) argues that the 1930's banking crisis caused U.S. bankers' fear of runs to be translated into a shift into safer loans which disrupted the intermediation process, and thus worsened the Depression. He subscribes to the theory that the system of a few large banks prevented runs in Canada and goes on to develop a detailed example showing that Canada had a debt crisis but not a banking crisis.9 The example is used to support his theory on the breakdown of intermediation.10

Haubrich (1989) extends Bernanke's conclusions to Canada. He finds that although Canadian banks closed branches in the Depression, closures are not a proxy for bank failures and that bank stocks did better than equities of other industries in this period. He concludes that ...Canada's superior organization of banking prevented a financial crisis...Other sectors did benefit from that superior structure...[ Haubrich (1989)]

Examining a series of financial crises in six different countries, Bordo (1988) shows that the accompanying monetary contractions were generally most severe in the United States. He identifies as a key factor branch banking which "represents a method of pooling risks which proved quite effective in guarding against the type of "`house of cards' effects common to the U.S....banking system" [(1988):�230]. The second factor is that, unlike the other countries, the United States lacked an effective lender of last resort. Although Canada did not have a central bank before 1936, Bordo [(1988):�230-1] argues that:

...the chartered banks had, by 1890, with the compliance of the Government, established an effective self-policing agency, the Canadian Bankers Association, which acting in loco parentis successfully helped insulate the Canadian banks from the deleterious effects of U.S. banking panics in 1893 and 1907. The existence of such a mechanism, whether provided by the Government or by the private market, once it proved effective would educate and instil a sense of confidence in the public sufficient to prevent incipient crises.

Bordo [(1988):�230, fn. 28] notes that the activities of the Canadian Bankers Association included:

...quickly arranging mergers between sound and failing banks, by encouraging cooperation between strong and weaker banks in times of stringency.

Bordo's work is the closest antecedent to our own because he emphasizes the role of the Canadian Bankers Association and the government as factors separate from the advantages of a branch system in explaining the absence of banking failures in Canada during the 1930's. In the following section, an implicit government guarantee of bank solvency is identified, and it is shown that this guarantee stood behind the policy of forbearance which allowed insolvent banks to continue to operate without runs in the 1930's.


Between Confederation in 1867 and 1940, twenty-seven banks failed in Canada and some depositors incurred losses. Of the thirty-six amalgamations, many involved troubled banks [Beckhart (1964)]. Neufeld [(1972):81] comments that "[t]he large number of failures is rather surprising in view of the Canadian banking system's reputation for solvency." This reputation is largely based on the timing of Canadian bank failures: none occurred between 1923 and 1985.

In 1923, a major failure occurred--- the Home Bank of Canada which had 70 (mainly urban) branches. Its directors were later charged with falsifying the accounting of the bank to cover up losses from bad loans. As a result of civil actions, the directors were required to pay damages for "misconduct, malfeasance and negligence..." [Jamieson (1953):�60-61]. During 1923, several other banks announced losses and reduced dividends or were forced to seek mergers.

The provision of government funds to avert a bank failure appears to have been initiated by the Government of Quebec. The Quebec Government financially assisted the merger of the Bank Nationale with the Banque d'Hochelaga in 1923 as follows [Globe (1924): 6]: ... The arrangement between the Quebec Provincial Government and the Banque d'Hochelaga is a unique one. Whether the Quebec Government felt that it had a moral obligation to advance aid, or whether its motives were purely philanthropic is a most interesting question. So far as Ontario is concerned, it opens up the possibility that Home Bank creditors may press for similar consideration. Although the two cases are admittedly not parallel ones, the fact that a Provincial Government has come to the rescue in one case may suggest a line of action for interested parties in the other.

Partly based on this precedent, the depositors in the Home Bank petitioned the Canadian Government for compensation and received payment up to 35% of the value of their deposits.

After 1923, the Canadian government provided an implicit guarantee to the public that no chartered bank would be allowed to fail and cause depositor losses. This guarantee was implicit because it was never formally embodied in law, and it was equivalent to one hundred percent deposit insurance.11 Beckhart (1964) documents that government policy was to arrange forced mergers for insolvent banks. He argues that the impetus for mergers came primarily from smaller banks near failure and from government.

Evidence exists that bank mergers were designed to avoid firesale insolvency for the merger of the Bank Nationale with the Banque d'Hochelaga in 1923 (discussed earlier) and the takeover of the Weyburn Bank by the Imperial Bank in 1931.

While the impetus for mergers may not have come primarily from larger banks seeking to expand, they were willing participants and there was considerable "behind-the-scenes" manoeuvering by the larger banks to absorb each new target bank. "I think it a pity," said another banker, "that the opportunity [Merchants Bank] was not offered to the other banks to participate in the business of the Merchants, and thus distribute the assets and the load, whatever its nature may be." [Globe (1921a):�1].

With regard to the role of regulators, primary evidence for the existence of an implicit guarantee comes from parliamentary documents and the popular press during the 1920's. A report in the Globe [(1921c):�1] described the rationale for the Federal Government's approval of the merger of the Merchants Bank with the Bank of Montreal: "The merger is the only way out." That is the considered opinion of Sir Henry Drayton, Minister of Finance, when asked if some other method could not have been found of meeting the crisis brought about by the troubles in the Merchants Bank .... Sir Henry Drayton said that a merger was only justified when the rest of a bank had been wiped out, its capital impaired and the affairs of the bank in such a position that the interests of the depositors themselves required to be guaranteed. It is assumed here that that must be the position of the Merchants Bank.

"What would happen if you had not given the preliminary consent to such a merger?" Sir Henry was asked.

"The only alternative is insolvency, with a consequent loss to depositors," was the reply. "That is my answer to criticisms of the Government's action in permitting the merger."

Although a proposal in 1914 to merge the Bank of Hamilton with the Royal was not approved by the then Minister of Finance, Sir Thomas White, a proposal to merge the then ailing Bank of Hamilton with the Bank of Commerce in 1923 was readily approved. [The Financial Post (1923a): 1, 16]

Similar sentiments were expressed during 1923 and 1924 when the failure of the Home Bank was scrutinized. A former Minister of Finance, Sir Thomas White, stated Government policy in favor of forced mergers to bank failures as follows:

Under no circumstances would I have allowed a bank to fail during the period in question�...�If it had appeared to me that the bank was not able to meet its public obligations, I should have taken steps to have it taken over by some other bank or banks, or failing that, would have given it necessary assistance under the Finance Act, 1914. [McKeown Commission (April�24, 1924, Vol.�5): 324].

If I had believed that the Home Bank at that time was in danger of failing, closing its doors, was insolvent, I should have gone to The Bankers' Association and told them to take over that bank. Either to one bank or more banks .... I would have made them do it. When I say that I had no legal power, but nevertheless I feel confident that I could have got them to do it..." [McKeown Commission (April�25, 1924, Vol.�6): 359].

The Federal Government's support for the merger of the Sterling Bank with the Standard Bank in 1924 was described in the press [FP (1924):�9] as follows:

... The government is determined that there shall be no more bank failures, if reasonable action on its part will obviate them. Accordingly, once it was demonstrated to the acting minister of finance that the proposed merger would strengthen the banks interested, there was no doubt about permission being granted.

Thus, the "forced" mergers of several small with large banks and the Home Bank failure led to a federal government policy placing a safety net under chartered banks. This view of government policy from the early 1920's forward is reflected in newspaper articles on the need to guarantee bank deposits. One editorial writer even assured the public that such a guarantee was in place in statutory form. [The Financial Post (Commons Debates, April�9, 1924):�1195]

This opinion was also reflected in the Home Bank depositors' petition as follows:

(10) Yon Petitioners therefore submit that whether rightly or wrongly the depositors of the Home Bank of Canada were largely of the opinion that the Finance Department of the Government of Canada exercised such supervision over chartered Banks that it was impossible for a depositor to lose their savings entrusted to such a Bank, the charter of which had been renewed from time to time by Parliament and it is further submitted that the confidence of the people as a whole would be greatly restored if adequate relief were granted to these depositors. [Commons Sessional Papers 100B (Thursday, March�24, 1924)].

In summary, historical evidence shows that beginning in 1923, an implicit guarantee from the Canadian government (amounting to 100% implicit insurance) stood behind all domestic bank deposits. The government actively promoted mergers to avoid firesale insolvency and successfuly created public confidence that no banks would be allowed to fail.12

Further historical evidence of the implicit guarantee of deposits is obtained by applying market-value accounting techniques to bank balance sheets of 1922-1940. Our analysis uses information available at the time to show that all major banks were insolvent at market values from 1930- 1935 at a minimum. Despite such widespread insolvency, no bank runs occurred providing strong support for the existence of an implicit guarantee of deposits.


The present analysis adjusts loans; made up of current loans and call and short-term loans, to market values.13 The analysis uses a set of bond indices and a stock index to value collateral securities held for short/call loans, and an activity index and a bad loan account to value current loans. Other assets and securities, along with all liabilities and capital are assumed at par. Banking practice called for valuing securities at the lower of book or market value yet the reported figures are likely optimistic for this period.14

There are two major categories of loans. Call and short term loans represent short-term lending (up to 30 days) to investment dealers secured by inventories of securities. Stated banking practice in the late 1930's was to lend up to 80% of the market value of securities (20% margin) and to update margins with shifts in market conditions [Patterson (1947:45)]. In earlier years, even more generous margins prevailed. In addition, some securities were illiquid and market prices were sometimes outdated due to thin trading. Further, some underwriters experienced difficulties [Neufeld (1972:509)]. Accordingly, the analysis that follows sets margins at 10%.

To obtain the market value of call and short term loans, we develop a securities index to adjust the collateral to market value.15 The bond indices are based on price quotes and new issue prices paid for Dominion, provincial, municipal, and industrial bonds on the Montreal Stock Exchange drawn from The Monetary Times. Close quotes are used if given, else High/Low or Bid/Ask are averaged. In general fifteen issues, if available, are used to arrive at an average price.

The result for each class of bond is an average year-end price. This price divided by the base year price yields the index for that class.16

The common stock index is drawn from Total Common Stock Price (series J494) in [Urquhart and Buckley (1983)] - an aggregate of Bank, Industrial, and Utility common stock indices.

Market value of call and short-term loans is the lesser of book value or the market value of the collateral. To obtain the market value of the collateral, the index is adjusted to reflect a 10% margin and this margin- adjusted index is multiplied by the book value of call and short-term loans to produce their market value.17

Current loans to be market valued are made up of "other current loans and discounts in Canada, other current loans and discounts elsewhere than in Canada after making full provision for bad and doubtful debts, loans to provincial governments, and loans to cities, towns, municipalities and school districts." Cashflows and market values of these loans varied with economic conditions.18 Loans to the federal government are assumed default free.

With the exception of unreserved bad loans, all loans are assumed to mature after one year.19 The interest rate on bank loans is assumed to equal the yield on Canadian corporate bonds at the beginning of the year. The yields are taken from [Neufeld (1972: 565)] and range from 4% to just under 10%.20

It is assumed that the probability of default on current loans is driven by the change in economic activity as measured by relative changes in a broadly based activity index.21 The model treats banks as not providing adequate reserves for "bad" or "doubtful" loans. A new account is created to cumulate this shortfall---"unreserved bad loans".

With a deterioration in economic activity, the unreserved bad loans account increases with new bad loans while an improvement in the economy triggers a decrease in cumulative bad loans as recoveries occur.

The activity index is a weighted average of indices of national manufacturing production, national retail sales, and national wheat gross value.22 The weights used in the activity index are based on an average aggregate breakdown of total banking loans in Canada from 1934 to 1940 in [Bank of Canada (1946:18-19)].

To implement the model, we first find the new current loans at the beginning of the year and outstanding at the end of the year by eliminating the cumulative bad loans outstanding from the book value of current loans. The next step is to find cumulative bad loans at the end of the year. If this year's activity index declined, some new loans default and bad loans increase. If activity increased, some bad loans are recovered. New current loans, unlike bad loans, pay principal and interest at year end. If economic activity decreased, payment is scaled down by the percentage change in the activity index. With an improvement in economic activity, the full promised payment is received.

We discount the market value of the payment to the beginning of the year to give the market value of new loans at the end of the prior year. By employing the corporate bond yield at the end of the year as the discount rate, interest-rate risk is incorporated into the model.

The market value of current loans is the market value of the new current loans plus the book value of loans to the federal government. The market value of the complete loan portfolio is the market value of current loans plus the market value of call and short loans minus the loan loss provision on the bank's books.

To measure solvency, the market value of the total loan portfolio is subtracted from the corresponding book value and the bank's capital reduced by any positive difference.

The four components of capital are dividends declared and unpaid, rest and reserve fund, capital paid up and the profit and loss surplus. These four items are totaled and reduced by the amount written down on loans. If the writedown is more than 100% of the value of the capital then the bank is insolvent. Another component of capital is inner or hidden reserves which did not apppear on the balance sheet. Although they cannot be quantified systematically, inner reserves were of insignificant magnitude compared to bad loans.23 5. RESULTS AND SENSITIVITY ANALYSIS Nine of ten Canadian chartered banks experienced asset writedowns in excess of their shareholder's capital (insolvency) at least from 1930 to 1935, and frequently for longer periods.24

Sensitivity analysis recasts critical assumptions employed in the valuation of call and short term and current loans. To test the robustness of our major finding, we vary selected assumptions in the direction which mitigates insolvency. In no case, is our major conclusion altered - nine of ten banks remain insolvent from 1930 through 1935.


Contrary to the "stylized fact", diversification benefits arising from national branching were not primarily responsible for the absence of bank runs and failures in Canada during the 1930's when no legal deposit insurance system was in place. Restating loan portfolios using market value accounting shows that nine of ten Canadian banks were insolvent at market values for each year from 1930 - 1935 inclusive. Sensitivity analysis demonstrates the robustness of these results.

The better failure performance of banks in Canada as compared to the United States should be attributed to the Canadian government policy of forcing failing banks into mergers with healthy banks. As documented above, this policy provided an implicit guarantee that, after a major bank failure in 1923 and several "forced" mergers of "failing" banks in 1921- 1923, no other bank would be allowed to fail. National branching made such a policy feasible by reducing the number of banks.

Our analysis suggests caution in extrapolating Canadian experience during the Great Depression to the current U.S. banking scene. In particular, it is an oversimplification of Canadian experience to argue that larger, more diversified banks are less likely to fail without recognizing the critical role of regulators in arranging mergers, closing troubled banks before they become insolvent as well as in providing an implicit guarantee of bank solvency.

Our reinterpretation of the Canadian experience reinforces the lesson of the savings and loan disaster --- it is dangerous for regulators to think that larger, faster growing financial institutions are necessarily more solvent [Kane (1989)]. Our results are evidence in favor of proposals by Benston (1986) and Benston and Kaufman (1986), among other for risk- adjusted capital and early closure of troubled institutions.

* We thank George Kaufman for enunciating our major hypothesis. Financial support for this research was provided by Fonds pour la formation de chercheurs et l'aide " la recherche (FCAR), the Social Sciences and Humanities Research Council of Canada (SSHRC), and the Center for International Business Studies, Dalhousie University. The authors acknowledge the capable efforts of Jonathan Dean along with those of Ken Bowen, Twila Mae Logan, Andrew Munn and Michael O'Grady in providing research assistance. They benefitted from comments by Michael Bordo and Kevin Huebner and from suggestions on earlier versions from audiences at the Bank Structure Conference, Federal Reserve Bank of Chicago and the Northern Finance Association, 1989 Meeting. Comments are welcomed.

1 This paper is based on a longer, more detailed version available on request from the authors.

2 Friedman and Schwartz (1963) argue that banking failures caused important contractions in liquidity and the money supply. Bernanke (1983) holds that banking failures forced a contraction in financial intermediation services which caused real contraction. Also see Gertler (1988).

3 Thus Canadian policy provided an early precendent for the "too large to fail" approach prevalent in the U.S. (and Canada) today according to Kaufman (1989).

4 Sections 2, 3 and 4 draw on Kryzanowski and Roberts (1989a and 1989b).

5 Our treatment of market value accounting for financial institutions draws on Bennett, Lundstrom and Simonson (1986), Kane (1985) and Kane and Foster (1986) and Benston et al (1986) among others.

6 In the framework of Kane, Unal and Demirguc-Kunt (1990), the implicit, off-balance sheet guarantee constituted a major asset of Canadian banks in the Depression.

7 Schwartz (1987) restates this comparison between the U.S. and Canada.

8 Government deposit insurance was not introduced in Canada until 1967.

9 Specifically, [Bernanke (1983: 259)] notes that: "The U.S. system, made up as it was primarily of small, independent, banks, had always been particularly vulnerable. Countries with only a few large banks, such as Britain, France and Canada, never had banking difficulties on the American scale.

10 Consistent with Bernanke's argument, [Safarian (1959:163, fn.180] states: "There appears to have been some pressure by the banks to reduce credit in the downswing."

11 This is similar to the current day implicit guarantee of FSLIC as discussed in Kane [(1987): 83-84].

12 Our interpretation of the historical evidence is also supported by Neufield[(1972): 98].

13 Logit estimates for Canadian banks were calculated on book values using the coefficients obtained by White (1984). The model was able to identify all except one of the weak banks that underwent mergers in the period 1922-1940. All the banks showed lower values hovering near the insolvency point for the early 1930s.

14 On Spetember 21, 1931, financial markets were unstable after Britain went off the gold standard, Canadian banks and stock exchanges set floor prices for equities in place through mid 1932. The next month, an Order in Council authorized banks to value securities, stock and bonds, at the lower of book value or market price on Aug. 31, 1931 effectively setting a floor under market values. [Joseph Schull and J. Douglas Givson, The Scotia Bank Story, A History of the Bank of Nova Scotia, 1823-1982, MacMillian of Canada, 1982:152] Sensitivity analysis sets securities to market values. 15 The mix of collateral securities is assumed to be equally distributed over Dominion, provincial, municipal and industrial bonds and common stock.

16 Since the bond indices were constructed from quotes available in each year they reflect a survivorship bias. While it is possible to construct more refined indices, this was not done as the survivorship bias works against our hypothesis of insolvency.

17 An upper bound of unity is placed on the margin adjusted index.

18 Default risk on provincial and municipal loans was significant. "From 1936 to 1945, Alberta defaulted on the principal of its maturing issues and paid interest at only one-half the coupon rate...Saskatchewan's credit rating also suffered in te 1930s" [Neufeld (1972: 5670] Municipal defaults were also common [Jamieson (1953: 78)].

19 Loans by the banks are consistently described as short term. Trade bills, a mojor component, averaged six weeks in maturity [Patterson (1947: 46)]. Commercial loans and loans in general are consistently described as short term and farm loans in 1933 were made for 3 to 4 month but were usually not repaid for 6 to 12 months [Royal Commission on Banking Currency (1933: 72)].

20 While bank loans likely had greater default risk than corporate bonds, they were shorter term. According to the Royal Commission (1933: 32-33)] the rate on good commercial loans was 6% ranging up to 10% for small loans and in small branches. Higher effective rates resulted from discounting and compounding cnventions. Agricultural loan rates ranged from 6% in the East to 7% in the West.

21 Haubrich (1989) provides a strong precedent for linking the strength of Canadian banks with economic indicators. The probability of default is take as the same for principal and interest.

22 The indices are, respectively, series Q8, series T53, and series M251 all from [Urquhart and Buckley (1983)].

23 In 1937 chartered banks were requested by the Government to bolster their inner reserves. The Royal Bank transferred $15 million from the published reserve account and reversed the transaction in 1946. [Ince (1969: 48)]. According to our analysis, cumulative bad loans for the Royal Bank were over $115 million in 1937.

24 The exception, Barclay's, was formed in 1929 under the control of the British parent institution [Jamieson (1953; 70)].