The author is indebted to Barry Eichengreen, Zvi Griliches, Barry Hirsch, John James, Terry Seaks, Insan Tunali, Jeffrey Williamson, and the participants of the Economic History Workshops at Harvard University and the Research Triangle for helpful comments concerning this research in its earlier stages. Claudia Goldin, George Boyer, and an anonymous referee provided valuable editorial criticism. The able research assistance of Caroline Gramley and Nancy Robbins is gratefully acknowledged.Google Scholar

¹ Two excellent treatments of the farmers complaints are: Hicks, John, The Populist Revolt (Minneapolis, 1931), chap. 3;Google Scholar and Shannon, Fred, The Farmer's Last Frontier (New York, 1961), chap. 13.Google Scholar

² U.S. Census Office, “The Report on Home Proprietorship and Indebtedness” (Washington, D.C., 1895), vol. 13, p. 3–4, 113.Google Scholar

³ Davis, Lance, “The Investment Market, 1870–1914: The Evolution of a National Market,” this JOURNAL, 25 (09 1965), pp. 355–93. For the mortgage market see pp. 373–87.Google Scholar

⁴ In an allocatively efficient national capital market the rates of return on the marginal investment projects for each region would have been equalized after adjustment for the lending risk intrinsic to the loan and the normal costs associated with making and administering the loan. Departures from this benchmark could have occurred for a variety of reasons: extraordinary costs and risks (beyond loan default) incurred when transferring funds between regions, regional differences in lending costs due to the uneven diffusion of cost effective financial intermediation, or variations in monopoly power. All these institutional and structural characteristics of the market were external to the underlying return on the financed project, and are considered here to be sources of inefficiency. The presence of any of these factors will be referred to throughout as market segmentation.Google Scholar

⁵ Eichengreen, Barry, “Mortgage Rates in the Populist Era,” American Economic Review, 74 (12 1984), pp. 995–1015.Google Scholar

⁶ Eichengreen's risk-adjusted rates showed substantial variation across states. The important point here is that the variation in the risk-adjusted rates showed no distinct regional pattern.Google Scholar

⁷ For the more modern and less sympathetic view of the farmers' complaints see: Higgs, Robert, The Transformation of the American Economy, 1865–1914 (New York, 1971);Google Scholar and Lee, Susan Previant and Passell, Peter, A New Economic View of American History (New York, 1979), pp. 292–301.Google Scholar

⁸ Davis, “The Investment Market,” pp. 355–56,Google Scholar discusses the importance of capital market imperfections to economic growth. Sylla, Richard, The American Capital Market, 1846–1914 (New York, 1975), pp. 219–32;Google Scholar and Williamson, Jeffrey G., Late Nineteenth Century American Development (Cambridge, Mass., 1974), pp. 119–45 also discuss the issue.Google Scholar

⁹ The economy-wide share of mortgage debt made on acres (relative to lots) decreased from 45 percent during the period 1880 to 1884 to 37 percent between 1885 and 1889. This pattern was pervasive throughout the national market except for the South Central region. See U.S. Census Office, “Report of Real Estate Mortgages in the United States” (Washington, D.C., 1895), vol. 12, p. 29.Google Scholar

¹⁰ U.S. Census Office, “Report on Home Proprietorship and Indebtedness” (Washington, D.C., 1895), p. 4.Google Scholar

¹¹ Ibid., pp. 5–11. The Census Office reported that the enumeration covered 82 percent of the farm and 74 percent of the home families in the country.

¹² U.S. Census Office, “Report of Real Estate Mortgages,” p. 309.Google Scholar

¹³ The description and summary of the special investigation appears in ibid., pp. 269–95. The basic data for each county appears in ibid., Table 110, pp. 836–95. For a complete enumeration of the sample counties see Appendix Table 6.

¹⁴ The counties were selected from 35 states representing the five census regions (the Northeast, South Atlantic, North Central, South Central, and West). The North Central region (which contained 49 of the counties) has been subdivided into the East and West North Central regions here.Google Scholar

¹⁵ A description of the sample counties can be found in ibid., pp. 273–77.

¹⁶ By restricting the analysis to owner-operated farms and owner-occupied homes, mortgages on rental and business property have been excluded. This homogeneity in the sample is a desirable feature because the hedonic interest rate equation used here requires mortgages to be identical in as many unobservable characteristics as possible.Google Scholar

¹⁷ U.S. Census Office, “Report on the Proprietorship,” table 103, pp. 382–420 and table 108, pp. 430–66. Two counties in Georgia reported no information for home mortgages. It was a common practice in the South and West for the borrower to pay an initial commission on the loan in addition to the stipulated interest payments. The Census Office included these commissions when calculating the effective interest rate.Google Scholar

¹⁸ The average life of the mortgages reported in Table I represent the duration of all mortgages in force on acres and lots rather than only for those farm and home mortgages reported in Volume 13.Google Scholar

¹⁹ For the modern market see: Schaff, A. H., “Regional Differences in Mortgage Financing Costs,” Journal of Finance, 21 (03 1966), pp. 85–94;CrossRefGoogle ScholarOstas, James R., “Regional Differences in Mortgage Financing Costs: A Reexamination,” Journal of Finance, 32 (12 1977), pp. 1774–78.CrossRefGoogle Scholar

²⁰ Eichengreen does not include explicit measures of market segmentation in his reduced form equation. He examines the estimated risk-adjusted rates for regional differences. Eichengreen, “Mortgage Rates,” p. 1012.Google Scholar

²¹ Bogue, Allan, Money at Interest (Ithaca, 1955) p. 120;Google Scholar and Williamson, Harold and Smalley, Orange, Northwestern Mutual Life (Evanston, 1957), p. 119, document this practice.Google Scholar

²² I will refer here to the costs of default rather than foreclosure. When a borrower defaulted on interest payments additional costs would be incurred even if foreclosure did not follow. These might have included foregone interest (unless the contract stipulated a penalty), or additional monitoring costs. Of course, expected foreclosure costs would have been included in default costs as well, and were likely to have been quite large.Google Scholar

²³ Bogue, Money at Interest, chap. 6 discusses the process of evaluating mortgage loans in detail for the Watkins Land Company.Google Scholar

²⁴ Ibid., p. 146, documents that J. B. Watkins complained of just this behavior by borrowers in western Kansas.

²⁵ The reduced-form equation for the county averages can be derived from a similar structural equation for individual mortgage contracts which treats the calamitous loan, borrower stability, and mortgage age effects as dummy variables. The specification is preserved for the averaged data under the assumptions that the home and farm mortgages in the sample were randomly selected from the total county population for these three characteristics. The (1/SIZE_i) variable is not strictly preserved by averaging. The error term in equation 1 is heteroskedastic because the effective loan rates (and hence the errors) were averaged across the number of borrowers in each market.Google Scholar

²⁶ The general level of mortgage rates were subject to both cyclical and long-term variations. Identical mortgages of different vintage would, therefore, have different effective rates simply because of differences in market conditions when they were made. The measure used here discriminates between mortgages that were made during the depression years before 1885 and those made during the lending boom later in the decade.Google Scholar

²⁷ For both population and farm growth figures data in 1880 are unavailable for three counties that had yet to be organized—Brown and Kimball in Nebraska, and Dickey in North Dakota. For these counties the measures were calculated for the smallest area that contained these counties for which data was available in both 1880 and 1890.Google Scholar

²⁸ Bogue, Money at Interest, documents the problems that both the Davenport family of Bath, New York, and the Watkins Land Company had in hiring and retaining reliable loan agents.Google Scholar

²⁹ Out-of-state funds would have lowered actual rates in the receiving market relative to their counterfactual levels had the mortgage market been perfectly segmented. However, once those funds were mobile, arbitrage would have ensured that the participation of out-of-state investors would have had either no effect on mortgage rates in the receiving market (if the transfers had been costless) or a positive effect (as compensation for the costs incurred).Google Scholar

³⁰ In Volume 12 all nonresident mortgagees of record, as well as mortgage and land banks, and railroads were considered to be out-of-state lenders.Google Scholar

³¹ The specification does not account for the potential effects of usury ceilings. The level of usury ceilings are available for the samples, but belong in the specification only if they are binding. Nonbinding ceilings would introduce endogeneity problems because the same forces that led to relatively high rates in the West would have led legislators to adopt a relatively high maximum customary rate in a usury law. An attempt to identify potentially binding usury ceilings for the sample was unsatisfactory, and I conclude that any such categorization would have been arbitrary. Eichengreen concludes that usury ceilings had a weak effect on rates in the Northeast, but also did not incorporate usury ceilings in his interest rate equation.Google Scholar

³² The error term is found to be heteroskedastic and the equations are estimated by Generalized Least Squares. The observations are weighted by the square root of the number of mortgages in each county.Google Scholar

³³ Eichengreen, “Mortgage Interest Rates,” p. 1009, argues that this is the reasonable range for the long-term risk-free rate in 1890, based on the yields of government and railroad bonds.Google Scholar

³⁴ Bogue, Money at Interest, p. 120, reported that The Watkins Land Company (in 1881) charged an effective rate of interest that was 0.5 percent lower for a $1500 mortgage than for a $500 mortgage. The coefficients reported here imply a reduction in rates of 0.44 percent and 0.55 percent for the farm and home markets respectively over the same range.Google Scholar

³⁵ The results including the interaction term are available upon request.Google Scholar

³⁶ Eichengreen, “Mortgage Interest,” pp. 1010–11. I have reported the ALT2 risk adjusted rates and the PREM2 risk premia.Google Scholar

³⁷ Besides the more complete geographic coverage, Eichengreen's sample includes all mortgages made on acres, some of which were not agricultural.Google Scholar

³⁸ Eichengreen examines the encumbrance ratios for his sample and finds that they are stable across regions. Because his sample includes all mortgages made on acres, and not only those made on farm land, the average encumbrance per acre and average price of a farm acre that he examined are not strictly comparable. The regional variations in encumbrance ratios for homes and farms in the sample counties are roughly the same as those reported for all counties in the country. U.S. Census Office, “Report on Proprietorship,” p. 82.Google Scholar

³⁹ An additional assumption of the hedonic specification is that all characteristics are priced linearly. In the case of the borrower risk measure (the encumbrance to property ratio) the true specification may well be nonlinear for individual loans. This potential mispecification should be mitigated to some extent by the use of county averaged data. In any case, the low risk premia in the North Central regions (relative to Eichengreen's) could not be explained by such a mispecification because borrowers in these regions had the lowest encumbrance ratios in the country.Google Scholar

⁴⁰ See for example: Ostas, “Regional Differences.”Google Scholar

⁴¹ In addition to the complaints of lenders noted by Bogue, , Keller, Morton, The Life Insurance Enterprise (Cambridge, Mass., 1963), P. 135 documents that some large New York insurance companies were reluctant to invest in the West where rates were much higher because of the costs and expenses involved.Google Scholar

⁴² Recall that the regional interaction term on the out-of-state variable is statistically insignificant and not reported.Google Scholar

⁴³ Bogue, Money at Interest, pp. 230–31, found that contractual rates paid to local investors in Nebraska were higher than those paid to nonresidents. He concludes that the agent's commission in the latter case would have brought to effective rates to near equality.Google Scholar

⁴⁴ Myers, Margaret, The New York Money Market (New York, 1931), vol. 1, p. 294.Google Scholar

⁴⁵ Frederiksen, D. M., “Mortgage Banking,” Journal of Political Economy, 2 (03 1894), pp.229–31. Frederiksen notes that both the French and German mortgage banking sectors controlled substantial portions of those country's banking capital, while in the United States the mortgage banks controlled only 2 percent of total bank capital at their peak in 1890. In addition, the bonds of American mortgage banks were rarely listed and traded on the formal securities market, again in contrast to their European counterparts.CrossRefGoogle Scholar

⁴⁶ Ibid., pp. 208–9. Goldsmith, Raymond, Financial Intermediaries in the American Economy Since 1900 (Princeton, 1958), reports that intermediaries supplied 50 percent of home mortgage and 14 percent of farm mortgage funds in 1900. These estimates appear to be consistent with Frederiksen's.Google Scholar

⁴⁷ Neither Eichengreen's specification nor the one employed here allows for the potential effects of risk diversification by mortgage lenders. Instead, it is assumed that the risk of the mortgage loan to the lender is a function of the characteristics of the loan itself and not its covariance with the returns of other loans or assets the investor might have held.Google Scholar

⁴⁸ Davis, “The Investment Market,” pp. 384–85 reaches the same conclusion using the mortgage earnings of life insurance companies.Google Scholar

⁴⁹ Keller, The Life Insurance Enterprise, p. 135,Google Scholar reports that the investments of the large New York companies were heavily weighted to the metropolitan area. The Connecticut companies, the most active interregional lenders in the market, concentrated on urban properties in the Midwest, see Zartman, Lester, The Investments of Life Insurance Companies (New York, 1906), pp. 29–30.Google ScholarThe Northwestern, a Milwaukee-based company, allocated 80 percent of its mortgage portfolio to urban loans (primarily in Chicago) between 1880 and 1900, Williamson and Smalley, Northwestern Mutual, pp. 57–58.Google Scholar

⁵⁰ Frederiksen, “Mortgage Banking,” p. 221.Google Scholar

⁵¹ The share of new construction in net capital formation increased from 45 percent in the 1870s to over 70 percent by the 1890s. These shares were calculated from data in Kuznets, Simon, Capital in the American Economy (Princeton, 1961), p. 524.Google Scholar

⁵² Goldsmith, Raymond, A Study of Savings in the United States (Princeton, 1956), vol. 1, p.619.Google Scholar