They are grateful for the helpful comments of John Baldwin, Alan Green, Ieuan Morgan and two referees, as well as for the help of Claudia Goldin. They also thank participants at the Fourteenth Conference on the Use of Quantitaive Methods in Canadian Economic History (Montreal, 1985) and at the Harvard economic history seminar for many useful criticisms. This paper was supported in part by the Social Sciences and Humanities Research Council of Canada (SSHRCC 410–85–0159).Google Scholar

¹ Regehr, T. D., The Canadian Northern Railway: Pioneer Road of the Northern Prairies, 1895–1918 (Toronto, 1976), p. 477.Google Scholar

² Fournier, Leslie T., Railway Nationalization in Canada: The Problem of the Canadian National Railways (Toronto, 1935), p. 33.Google Scholar

³ The Canadian Northern did operate or control some railways in the East, but these were a small part of the total system. Their gross earnings were $1.45 million, which compares with $13.8 million for the Canadian Northern proper; moreover these railways operated only 836 miles of track. Canada, Department of Railways and Canals, Railway Statistics of the Dominion of Canada, 1910 (Ottawa, 1911), pp. 19, 46.Google Scholar

⁴ After 1919, mileage did increase slightly, and by 1922, 11,518 miles were operated. Railway Statistics: 1917 (Ottawa, 1918), p. 7;Google Scholar and Canada, Dominion Bureau of Statistics, Statistics of Steam Railways of Canada: 1922 (Ottawa, 1923), p. 43.Google Scholar

⁵ Mackenzie and Mann also were required to merge the Canadian Northern with its roughly thirty subsidiaries, write down substantially the par value of the Canadian Northern common stock, and release the Canadian Northern from all debts still owed to Mackenzie, Mann and Co. on the construction account.Google Scholar

⁶ Moody, John, Moody's Analyses of Investment: Steam Railroads, 1921 (New York, 1921), p. 1240.Google Scholar

⁷ Royal Commission to Inquire into Railways and Transportation, Report of the Royal Commission to Inquire into Railways and Transportation in Canada, Drayton-Acworth Report (Ottawa, 1917), p. lxii.Google Scholar

⁸ The Liberal party was in power when the decision was made to support the Canada Northern and the Grand Trunk Railway, In time, both railways operated transcontinental lines which competed with the Canadian Pacific system.Google Scholar

⁹ This railway, which ran from Winnipeg to Quebec City, was operated by the Grand Trunk Pacific Railway, a western subsidiary of the Grand Trunk Railway. Debates of the House of Commons of the Dominion of Canada, 3rd sess., 9th Parliament, vol. 61 (Ottawa, 1903), pp. 8433–34.Google Scholar

¹⁰ Stevens, G. R., Canadian National Railways, Volume 2: Towards the Inevitable, 1896–1922 (Toronto, 1962), p. 470. Although wartime inflation lowered the Canadian Northern's net earnings, there is a sense in which the railway benefited from the unexpected increases in the price level. Much of the Canadian Northern debt was in the form of long-term bonds paying fixed coupon rates. The World War I inflation sharply reduced the real value of these bonds.Google Scholar

¹¹ Ibid., p. 470; Dixon, Frank H., Railroads and Government: Their Relations in the United States, 1910–1921 (New York, 1922), pp. 119–32.Google Scholar

¹² See for example Easterbrook, W. T. and Aitken, Hugh G. J., Canadian Economic History (Toronto, 1956), p. 441.Google Scholar

¹³ This increase in income came despite other war-related problems emphasized by Regehr. These included the inability to purchase sufficient rolling stock and the requirement that the railway annually provide to the government, without pay, services with a value of up to 3 percent of the Canadian Northern subsidy, or about $1 million. Regehr, The Canadian Northern Railway, pp.. 398–99.Google Scholar

¹⁴ Railway Statistics: 1914, p. 44; and Railway Statistics: 1917, p. 22.Google Scholar

¹⁵ The annual rate of change in the wholesale price index, which averaged 2.3 percent from 1901 to 1914, jumped to 18 percent from 1914 to 1918. Leacy, F. H. et al. , Historical Statistics of Canada (2nd edn., Ottowa, 1983), K33.Google Scholar

¹⁶ Fournier, Railway Nationalization, pp. 340–41. Easterbrook and Aitken also take Fournier's position on the effect of loan guarantees: “As a method of financing… [the guarantee system] had serious disadvantages since, by burdening the railway with heavy fixed charges for interest, it weakened its ability to survive depression or temporary declines in earnings.… This system had been rejected as unreliable and open to abuse at the time of the chartering of the C.P.R.” Easterbrook and Aitken, Canadian Economic History, p. 437.Google Scholar

¹⁷ Although the return is realized after the project is completed, the model is one-period rather than two because no time is assumed to elapse between the completion of the project and the realization of the return.Google Scholar

¹⁸ The equivalence can be illustrated as follows. The outlay of equityholders is the difference between the cost of the project (C) and the outlay of bondholders. Since bondholders are assumed risk neutral, they provide an amount equal to the market value of their debt (V _d). The object of equityholders, therefore, is to maximize V _e − (C − V _d), where V _e is the market value of equity. Since C is exogenous, this is equivalent to maximizing V_e + V _d, which is the market value of the firm.Google Scholar

¹⁹ Modigliani, Franco and Miller, Merton H., “The Cost of Capital, Corporation Finance and the Theory of Investment,” American Economic Review, 48 (06 1958), pp. 261–97.Google Scholar

²⁰ The assumption that a bankrupt firm takes on its liquidation value may overstate bankruptcy costs given that failing North Ameircan railroads typically reorganized and continued to operate. Assuming smaller (but positive) bankruptcy costs, however, complicates the analysis without changing the theoretical or empirical results.Google Scholar

²¹ Canadian Northern debt issues that were not guaranteed by government generally had first claim on the firm's assets. Moody, Steam Railroads, 1921, pp. 1103–7.Google Scholar

²² Note that the density function of returns is truncated at L because the firm always has the option of liquidating its assets. The probability of realizing L is given by the area under the density function from –∞ to L.Google Scholar

²³ Note that Y = E −C and Y′ = E′ − C.Google Scholar

²⁴ We do not think it appropriate to base our analysis on an earlier year given that the system would have been largely unfinished. Selecting a later year is defensible, but it would not change our qualitative findings.Google Scholar

²⁵ Drayton-Acworth Report, p. xliv.Google Scholar

²⁶ Ibid., pp. xcviii, xcviii.

²⁷ The Canadian Northern's debt consisted of the following (in thousands of dollars):Google Scholar

Edward E. Loomis and John W. Platten, Report on the Canadian Northern Railway System, Report submitted to the Board of Directors of the Canadian Northern Railway (1917), p. 20. Canadian Northern Railway, Annual Report: 1916.

²⁸ Edward Altman, as part of his analysis of railroad bankruptcy in the United States between 1939 and 1970, computed an average debt-equity ratio of I for railroads that did not go bankrupt. In 1916, the average debt-equity ratio of the six western U.S. railroads analyzed by Lloyd Mercer was 1.1. Altman, Edward I., “Predicting Railroad Bankruptcies in America,” Bell Journal of Economics and Management Science, 4 (Spring 1973), p. 194.CrossRefGoogle ScholarMercer, Lloyd J., Railroads and Land Grant Policy: A Study in Government Intervention (New York, 1982).Google ScholarMoody, Railroad Investments, 1917, pp. 130, 362, 534, 671, 726, 747.Google Scholar

²⁹ Drayton-Acworth Report, p. xliii.

³⁰ Following is an exerpt from evidence provided to the commission by A. J. Mitchell, comptroller of the Canadian Northern: Mr. Mitchell: In addition to that, Sir Henry, in connection with construction, there are no contractor's commissions on any of the actual construction. The work was turned in to the railway at cost. For instance, there is actual construction representing something like $204,000,000, in which there is no percentage whatever to the contractors … instead of getting a 10 per cent commission they got the stock of the Canadian Northern Railway. That was the payment the contractors received instead of a cash consideration. The 10 percent figure first mentioned by Mitchell is probably biased upward as he continued: “For instance, the Canadian Pacific Railway, we will say lets a contract to Foley and Company to do certain work … Foley and Company would get their commission. On a large contract they would perhaps work on a 5 per cent or 8 per cent basis.” Drayton-Acworth Report, pp. lix–lx. Z. A. Lash, counsel for Mackenzie and Mann, claimed that their commitment was over $70 million; and Regehr argues that one should add to this figure the $21 million debt that Messrs. Mackenzie, Mann and Co. forgave the Canadian Northern as part of the 1914 agreement with the federal government to save the railway. Regehr, The Canadian Northern, p. 375. Despite these claims we regard $20 million as an upper-bound estimate of Mackenzie and Mann's contribution, given their personal resources at the start of the project, and given the testimony to the Royal Commission of Vice-President Hanna, and Comptroller Mitchell of the Canadian Northern. As part of the 1914 plan to save the Canadian Northern from insolvency, the dominion government received 40 percent of the company's common shares. Since this was clearly not part of Mackenzie and Mann's original investment strategy, we derive our ex ante results on the assumption that Mackenzie and Mann owned all of the common shares.Google Scholar

³¹ Between 1939 and 1970, the average debt-equity ratio of U.S. railroads two years prior to bankruptcy was 7.5. Altman, “Predicting Railroad Bankruptcies,” p. 194.Google Scholar

³² Because government-guaranteed loans were available, it follows from equation 2 that, given an optimal capital structure, the face value of debt was greater than the firm's liquidation value.Google Scholar

³³ PRIVATE RATES OF RETURN ON SELECTED NORTH AMERICAN RAILROADS

.

³⁴ The company reported the value of its road and equipment to be $495 million as of June 30, 1916. Included in this figure, however, is the $100 million in common stock which we have argued represents about $20 million in incurred cost. Canadian Northern Railway, Annual Report: 1916. The total return is the present value of the implied infinite income stream, therefore, our lower-bound estimate of the standard deviation of the total return ($254 million) is computed as. 0245 ($415 million)/.04. Four percent was the average coupon rate on the debt issued by the Canadian Northern. Loomis and Platten, Report, exhibit B.Google Scholar

³⁵ Drayton and Acworth concluded their review of these projections with the following assessment: “making all possible allowances, we think that those responsible for the estimates of 1914, upon which Parliament acted, have incurred very serious responsibility.” Drayton-Acworth Report, p. xxxix.Google Scholar

³⁶ Edward E. Loomis was president of the Lehigh Valley Railroad Company; John W. Platten was president of the United States Mortgage and Trust Company.Google Scholar

³⁷ Drayton-Acworth Report, p. xl.Google Scholar

³⁸ This value is derived under the assumption that net income would have remained constant at $24.8 million after 1923. This is reasonable given that net income per mile of track on both the Canadian National Railway and the Canadian Pacific Railway showed almost no upward trend from 1923 to 1929. The present value of the implied net income stream, discounted at 4 percent, is $572 million in 1916. From this figure we subtract the present value of additional capital costs through 1923, estimated to have been $18.8 million per year over 5 years. Loomis and Platten, Report, p. 26 (items 1–6).Google Scholar

³⁹ This decision in the Western Rates Case resulted in a 7.5 percent reduction in freight rates. We apply this adjustment to all freight earnings, which gives a small upward bias to our estimate of the mean of X.Google Scholar

⁴⁰ The Canadian Northern based its own projections on a ratio of operating expenses to gross earnings of 0.7 which reflected their past experience. We also have computed hypothetical net income using actual operating expenses, adjusted for increases in an index of operating costs. Fournier. Railway Nationalization, pp. 87–88. This procedure gives, on average, results very similar to those in Table 3. Another approach, which gives similar results, uses actual freight traffic, passenger traffic, and operating costs, but allows nominal freight and passenger rates to keep pace with inflation. Assuming no trend in net earnings after 1922, the present value of the implied net income stream is very close to our intermediate estimate. It should be noted that, using this approach, net earnings, are considerably higher during the war and immediate post-war periods.Google Scholar

⁴¹ Net income is averaged over the 6 1/2-year period, July 1, 1916 to December 31, 1922. Net income over the period July 1, 1919 to December 31, 1919 is assumed to be one-half net income for the year 1919 (see Table 3).Google Scholar

⁴² Following is a breakdown of dominion and provincial loans (in thousands of dollars):

.

⁴³ If we take $499 million to represent the book value of the Canadian Northern's assets (see above), our inferred liquidation value of $179 million is 36 percent of book value. The true liquidation value is, of course, difficult to determine because the Canadian Northern was eventually taken over by the federal government without a liquidation. It is, however, unlikely to have been less than $179 million. The Canadian Northern had land and current assets valued at $65 million and the depreciated value of its rolling stock was about $37 million (a depreciation rate of 10 percent is assumed). These assets would not have sold at much of a discount; therefore, if the road had a liquidation value of at least $77 million (179 – 65 –37), which is less than 20 percent of its book value, Mackenzie and Mann issued no more debt than was consistent with the strategy of maximizing their expected net return. Canadian Northern Railway, Annual Reports, 1907–1916.Google Scholar

⁴⁴ As of 1916 all Canadian Northern debenture stock guaranteed by the dominion government or provincial governments, was rated A by Moody's. They were regarded as “being quite secure, with a permanent and substantial future.” Of the unguaranteed bonds, about half were rated Baa being regarded as “generally good investments, but … not so stable in position as … the ‘A’ bonds.” The remaining bonds were rated Ba and so were “considered as in a half-way position between sound investments and speculations. Such bonds are often well secured so far as equity in property is concerned, but there is still some uncertainty as to their ultimate position as related to earning power.” Moody, Railroad Investments, 1917, pp. 20, 1205–9.Google Scholar

⁴⁵ The face value of debt in 1916 was $391 million and the cost of equity was $51 million, which included the estimated cost to Mackenzie and Mann ($20 million), the income-charge convertible debenture stock ($25 million), and the common stock in affiliated companies ($6 million). It should be noted that debenture stock normally sold at a small discount, and so $442 million overstates somewhat the actual outlay of investors, we, nevertheless, apply this value in part because the bias is small and in part because we discount future returns at the average coupon rate on these debentures. Canadian Northern Railway, Annual Report: 1916.Google Scholar

⁴⁶ Note that the loan guarantees, under this scenario, would have provided Mackenzie and Mann with an expected net return of between $106 million and $134 million.Google Scholar

⁴⁷ Regehr, The Canadian Northern Railway, p. 308.Google Scholar

⁴⁸ One example of this is building of the Grand Trunk Pacific Railway, which contributed to the process of rent dissipation. In addition Canadian Northern officials complained about government demands that they build lines prematurely. See Stevens, Canadian National Railways, p. 482; and Regehr, The Canadian Northern Railway, pp. 449–50.Google Scholar

⁴⁹ As recognized by Fournier, replacing loan guarantees with direct subsidies would have greatly reduced the probability of failure, however this does not mean that government policy was misguided. Instead, direct subsidization may simply have been politically unacceptable.Google Scholar

⁵⁰ This is the subtitle of Volume 2 of Canadian National Railways.Google Scholar

⁵¹ We make the standard assumption that investors are risk neutral. This is certainly reasonable for debtholders, because they were able to diversify their risk. Mackenzie and Mann were the only major equityholders, but given that they could have sold some of their equity, and chose not to, suggests that they too were risk neutral (or possibly risk-loving).Google Scholar

⁵² We assume that other debt issues have priority over the guaranteed loans (see fn. 21).Google Scholar

⁵³ Our analysis is based on the assumption that a firm must liquidate when it is bankrupt. If we assume instead a bankruptcy cost, B, such that 0 < B < G, equation 13a still applies but has a different interpretation.Google Scholar The ex ante cost of the loan guarantees becomes:

and the effciency loss is: