The suggestions of Eugene Fama, Sidney Homer, Allan Meltzer, Norman Miller, Shyam Sunder, Joel Tarr and a referee are gratefully acknowledged. This research was partly supported by The National Science Foundation.

¹ U.S. Register's Office, Treasury Department, Statement of Public Debt of the United States, 1789–1885, (1886), p. 75.

² Bayley, Rafael A., National Loans of the United States (2nd ed.; Washington: U.S. Government Printing Office, 1882), p. 74.Google Scholar

³ The decrease in imports was apparently caused by high prices in Europe and Canada. An additional factor in the fall of customs revenues was a lowering of rates early in 1857. Dewey, Davis Rich, Financial History of the United States (New York: Longmans-Green and Co., 1934), p. 262Google Scholar. 1858 was the only year between 1851, when the record starts, and 1876 when the U.S. had an international trade surplus. U.S. Bureau of the Census, Historical Statistics of the United States, Colonial Times to 1957 (Washington: U.S. Government Printing Office, 1960), p. 565.Google Scholar

⁴ Texas Indemnity shares were really 5 percent, 14-year bonds conceded to creditors of the State of Texas in return for Texas relinquishing extensive territorial claims. In debate on the subject, it was said (presumably by a Texan) that “… her southeast corner was at the mouth of the Rio Grande, the region of perpetual flowers, her northwest corner near the South Pass in the Rocky Mountains, a region of everlasting snows; that she had a front on: the Gulf of Mexico of a thousand miles, a front on the Rio Grande of 2000 miles, and undisputed frontier on the Arkansas of 1000 miles, and within these boundaries an. area of 350,000 square miles, or more than equal to seven states as large as Pennsylvania,” Congressional Globe, 1st session, 31st Congress, p. 165. $7,750,000 in stock was actually paid to Texas for a few war materials she had had as a; republic and for these land claims. Not counting the armaments, this amounts to $.14 per acre, the current size of Texas being 267,000 square miles.

⁵ This may be contrasted with the 602 percent rise in federal debt from July, 1940 through July, 1945—from $43 billion to $259 billion.

⁶ Later, as Chief Justice of the Supreme Court, Salmon P. Chase wrote the majority opinion that ruled greenbacks unconstitutional. He had never liked them and accepted the duty of their issuance in 1862 only under the pressure of First Bull Run and McClellan's pending peninsula campaign. Cf. Hammond, Bray, “The North's Empty Purse,” American Historical Review, LXVII (Oct. 1961), 2Google Scholar. Also see Unger, Irwin, The Greenback Era, A Social and Political History of American Finance, 1865–1879 (Princeton: Princeton University Press, 1964), pp. 174–75Google Scholar. A. valuable source about the early Civil War years and the financing arrangements debated in the 37th Congress is Hammond, Bray, Sovereignty and an Empty Purse, Banks and Politics in the Civil War (Princeton: Princeton University Press, 1970Google Scholar). Homer, Sidney, A History of Interest Rates (New Brunswick: Rutgers University Press, 1963), pp. 306–12Google Scholar contains a concise history of Federal securities from 1861 to 1865 and draws attention to many unusual bond covenants.

⁷ Bayley, National Loans …, pp. 80–81.

⁸ New York Tribune, January 22, 1862 and February 1, 1862, italics added.

⁹ However, less than $1½ billion were outstanding at any given time as older issues were frequently redeemed.

¹⁰ Bond prices from 1862–65 were obtained from various issues of the Commercial Chronicle and Review; 1861 prices came from Banker's Magazine. Since prices were quoted including accrued interest, it was necessary to subtract the accrued amount from each quotation to eliminate discontinuities at coupon payment dates.

¹¹ The Southern victory at Chancellorsville was partly offset by Jackson's wounding and subsequent death.

¹² Grant became Commander-in-Chief on March 9.

¹³ Guaranteed greenback conversion to specie was suspended by the Government from January, 1862 to January, 1879.

¹⁴ Mitchell, Wesley C., A History of the Greenbacks (Chicago: University of Chicago Press, 1903) Chart III, p. 277.Google Scholar

¹⁵ Mitchell, p. 224.

¹⁶ Jellison, Charles A., Fessenden of Maine, Civil War Senator (Binghamton, New York: Syracuse University Press, 1962), p. 183.Google Scholar

¹⁷ Many authors have larnented the great profit gained by investors, especially English investors, who purchased government bonds when the gold premium was high, Cf. Martin, Joseph G., Martin's Boston Stock Market (Boston: published by the author, 1886), p. 123Google Scholar. When U.S. 6's of 1881 were trading for 42 in gold, their current gold yield was over 14 percent. In hindsight, this looks high because we know the bonds nearly doubled in price within a year. But the investor of 1864 could not have regarded 14 percent as better than a decent return after contemplating the condition of Federal finances.

¹⁸ This is the standard condition for prices in an “efficient market,” Cf. Fama, Eugene F., “Efficient Capital Markets: A Review of Theory and Empirical Work,” Journal of Finance, XXV (May 1970), 383–88CrossRefGoogle Scholar; Roll, Richard, The Behavior of Interest Rates (New York: Basic Books, Inc., 1970), ch. iiGoogle Scholar; and Samuelson, Paul A., “Proof that Properly Anticipated Prices Fluctuate Randomly,” Industrial Management Review, VI (Spring 1965), 41–49Google Scholar. The expectation is taken conditional on all current information including the past sequence of prices.

¹⁹ Obviously, bond prices are positively related to the probability of gold redemption as long as there is a premium on gold.

²⁰ See the references in fn.18.

²¹ For the distinction between strong-form and weak-form tests of competition, see Fama, “Efficient Capital Markets,” p. 388.

²² A price change is defined as p_t-p_t–j, where p_t is the market price at date t and j is the differencing interval.

²³ Turnovsky, S. J., “Empirical Evidence on the Formation of Price Expectations,” Journal of the American Statistical Association, LXV (Dec. 1970), 1441–54.CrossRefGoogle Scholar

²⁴ Bayley, National Loans…, p. 164; Commercial Chronicle and Review, January, 1865, p. 62.

²⁵ Bayley, National Loans…, p. 164.

²⁶ The 6's had a different defect—their principal was not definitely payable in gold.

²⁷ Only α's estimated with the 10–40 yield through 1879 are reported in Table 3.

²⁸ A negative anticipated gold price change persisted long after the end of the Civil War. Gold bonds yielded a higher (gold) return than paper bonds, until resumption of specie payment by the Federal government in 1879. After 1889, gold yields fell considerably below paper yields, an occurrence that Fisher attributes to the Free Silver movement. Fisher, Irving, The Theory of Interest (New York: The Macmillan Co., 1930), pp. 401–03.Google Scholar

²⁹ Assuming the absolute value of the gold price change,

is a measure of uncertainty, a regression between Z and the estimated gold price change, â was calculated as follows:

Durbin-Watson d = 1.89, R² = .77. Numbers in parentheses are standard errors. The negative, and significant coefficient of Z_t+1, the absolute value of the gold price change observed just after the predicted change â_t, supports the contention in the text. The lagged values Z_t and â_t-1 were added for econometric reasons—to reduce autocorrelation in the residuals. The coefficient of â_t-1 is an estimate of the firstorder serial correlation coefficient for the residuals from a two-variable regression of â_t on Z_t+1. The coefficient of Z_t should be approximately equal to minus the product of the coefficients of â_t-1 and Z_t+1. In fact, it is less than one standard deviation from this number.

³⁰ Friedman, Milton and Schwartz, Anna J., A Monetary History of the United States, 1867–1960 (New York: National Bureau of Economic Research, 1963), p. 73.Google Scholar

³¹ Ibid., pp. 70–73.

³² Ibid., p. 71. In a footnote on page 73, Friedman and Schwartz recognize their position is somewhat inconsistent because an expected fall in the greenback price of gold is sufficient alone to cause a positive spread between gold yields and greenback yields on bonds of similar risk. They remedy the inconsistency by offering the following “… reasons: (1) a fall could occur without taking the [gold] price [in greenbacks] back to prewar par,…” However, to generate the spread, investors need not have expected a fall all the way back to par. Any anticipated decline would have been sufficient. “… (2) even if it took it back to par, that might occur after repayment of the bond, and even if it occurred before, the coupon payments in the interim might be in greenbacks.…” Again, any decline in the interim would have brought real gains to holders of greenbacks. “… (3) the purchasers of government securities were a much more mixed and broader group than the speculators in foreign exchange were, so we are dealing with the expectations of two very different groups.” However, even if the two groups were very different, their expectations could have been the same. Their incentives were certainly the same and besides, a limited number of speculators active in both markets would have been sufficient to erase the arbitrage opportunities implied by Friedman and Schwartz. The active competition demonstrably existing in both markets implies that a greenback-gold yield spread is not just a fluke of one market's participants paying no attention to the other.

³³ One might think such a cost was illusory because if the greenback price of gold had increased, the government could have simply printed proportionately more greenbacks. Ultimately, of course, this could have resulted in hyper-inflation. It never did, and the price of gold was a real cost to the government, because Congress so effectively limited the total circulation of greenbacks.

³⁴ The bias will be downward only if greenback yields on the one-year bond are greater than greenback yields-to-maturity on the long-term bonds (assuming no gold payments at all). This relation is generally true for 1863 through 1865.

³⁵ The procedure follows: First, for an assumed value of β, yields-to-maturity were calculated according to the discounting formula

where, as before, P_{B, t} is the current bond price in greenbacks, d is the semi-annual coupon and is the probability of gold payment in period j (a function of β and of the current greenback price of gold, p_g.t). Defining the discounting formula by the.function a solution R* to

provides an estimate of α_t by

Of course, R_{L, t} was measured as before by the one-year greenback yield. Using the set of â 's thus estimated, the mean-squared prediction error is calculated as

where N is the sample size (in weeks).