Populist Kansas: A Human Capital Interpretation

Eugene P. Sigel

Introduction

Why did nearly 107,000 men turn out in the Fall of 1890 to vote for a previously unknown political party? That is the Populist Puzzle. In this session I address that question through a new quantitative sample of farm protest behavior in the Populist era, the 1890 precinct voting returns from 1,263 minor polling stations in eastern and central Kansas. Using a multinomial logit regression model of party selection, I demonstrate that economic influences played an important role in motivating farm voters. Populism took hold amongst farm communities that had been shut out from the indirect benefits of railway construction. It took hold amongst less wealthy, unfenced farming townships in the early stages of economic development. And perhaps most importantly, it took hold amongst farm families that were experiencing a migration-induced devaluation of their human capital. I argue that Kansas Populism was triggered in 1890 because many native-born farmers found, when confronted by a multi-year infestation of agricultural pests and severe drought, that they lacked the farm management skills necessary for profitable exploitation of the subhumid agricultural lands. Populism, I argue, is fruitfully interpreted as a movement rooted in the frustrating regional adjustment difficulties faced by individual farm families, rather than as simply a democratic, collective movement in American politics.

The Populist Trigger

The sudden emergence of Kansas Populism in 1890 remains a puzzle because few analysts have accounted for the timing of the movement's growth. Goodwyn argued that it was because farmers had been mobilized through cooperative ventures prior to 1890; when established merchants blocked the growth of cooperatives a structure was in place to channel farmers' anger into politics. But in Kansas there were few cooperatives—estimated at just over one in a thousand businesses—by 1890. Mayhew suggested that farmers were reacting to the exigencies of commercialized farming, which made them more price sensitive. But Populism began in 1890, well after the sharp drop in field crop terms of trade from 1881-1884 and well before the national depression of 1893-1896. Nor were any of these price movements unique to Kansas. And although I note that an index of building materials prices did move sharply against farmers developing a treeless region, the index shows no sudden shift in Kansas in 1890. Higgs's suggestion that farmers were reacting to a flat trend in real freight rates may have been too pessimistic: the freight rate terms of trade for Kansas that I construct show the sharpest growth of any series over 1870-1900.

McGuire and Stock proposed that the Populist impulse was strongest in Kansas and the central West because greater regional uncertainty coincided with a heightened fear of foreclosure. But the Kansas farmer has often faced greater fear of foreclosure and always greater economic uncertainty than farmers in Ohio, New York, and Pennsylvania; that gets us no closer to understanding why Kansas Populism emerged in 1890 and died out at the turn of the century. The shifting crop mix provides the missing link: agriculture in Kansas was more uncertain in 1889 and 1890, I argue, because farmers across the central wheat belt of the state—the backbone of Populist support—were rotating their fields away from wheat and towards Indian corn. Rarely again would so many Kansas farmers grow so much corn and so little wheat as they did in 1889 and 1890. And by McGuire's own measures corn was substantially more "uncertain" than wheat in both prices and yields.

Yet if Populism emerged in Kansas because a short term shift in the crop mix greatly enhanced the level of regional uncertainty, what explains the change in the crop-mix? For DeCanio and Halcoussis the answer was relative prices: farmers mistakenly shifted towards corn and away from wheat because they failed to anticipate the rising relative wheat prices of the last few years of the 1880s; the result was an 18.7% cut in counterfactual 1889 gross farm income, the largest estimated in any year. But this explanation of the crop-mix places a strong faith in relative price movements. In fact relative yield— rather than price—ratios were far more volatile in central Kansas.

And that is why farmers grew a more risky mix of corn and wheat on the eve of the Populist revolt: they had a fear of income losses from yield-reducing insect infestations. Small insects called chinch-bugs appeared in farmer's fields in 1886 and 1887 and forced farmers to chose either corn or wheat, but not both. These bugs sucked the life from the ripening winter wheat in the spring and then migrated to corn fields in early summer. The suggested pest-management technique was to eliminate either wheat or corn altogether for several years. The choice of which crop to abandon depended on which region a farmer felt he was part of: the corn belt, or "the golden belt?" Native-born migrant farmers trained in the corn-belt commonly chose to eliminate wheat, with unfortunate consequences. In 1889 corn reached historically low prices; and in 1890 Kansas was hit with drought and hot winds that lowered corn output by 80% (but winter wheat output by only 19%) of the previous year's bountiful harvest.

The Populist impulse did arise in a region of greater economic uncertainty. But part of that uncertainty was rooted in individual choices that reflected a farmer's endowment of human capital. When confronted by bugs attacking both crops, farmers made decisions about how to react based on their accumulated endowment of management experience. The shift to corn was an especially risky strategy because most of Kansas is not well suited to Indian corn. Hot winds and the heightened likelihood of persistent drought in July and August make Indian corn more susceptible to environmental stress than winter wheats and sorghums. Ultimately Indian corn would come to play but a minor role in farm rotations across most of Kansas; yet it required the test of actual experience for farmers to make that determination. By contrast the Russian and German farming communities, which specialized in wheats to a greater degree, failed to greet the Populists enthusiastically in 1890. Their human capital endowments, many formed in the subhumid regions of southern Russia, led them to specialize in winter grains rather than corn as the chinch bugs and drought appeared.

This theory of Populism explains the rise of Populism in 1890 across Kansas as a response to the frustrations of adapting crop selections and pest management routines to the subhumid environment. The solution to one problem—pests—became the source of another—crop failures. This had a strong economic impact as well. Between 1889 and 1890 I estimate that per capita operating farm income fell on average by 45 percent. Farmers raised hell in the fall of 1890, in part, because they had not yet formed the farm management skills that could have prevented such a collapse of operating income.

A Protest Model

In order to test the hypothesis that Populism was triggered by crop failures, I have assembled a new sample of farm protest behavior: the official 1890 voting returns, as reported in local newspapers of record, from 1,263 polling stations in 72 counties of central and eastern Kansas. In total, 1,181 townships and 92% of all Populist voters in Kansas are included. These data provide what is most likely the best cross-sectional measure of the patterns of Populism. Other scholars have relied on Alliance membership lists or county voting returns to test the distinctiveness of Populist farmers. But presence or absence on the three surviving Kansas membership lists does not provide an adequate measure of how an individual voted on election day. And the county voting returns used by Williams—aggregated over an average 800 square mile area—cannot adequately disentangle the issue of "railroad monopoly power" from simply the presence or absence of any railway station.

The precinct voting data form the dependent variable in a multinomial Logit model of party selection, estimated by weighted least squares:

(1) Loge Pv,t = ßo + ß1 Distance + ß2 Stations + ß3 Competition + ß4 Town-Aid +

Ppop,t

ß5 County-Aid + ß6 Wealth + ß7 Density + ß8 $ Stock + ß9 Farm Income +

ß10 Wheat/Corn + ß11 %∆ Farm Income + ß12 % Fenced + ß13 %∆Land Value +

ß14 %German + ß15 %Other Ethnic + ß16 %Urban + ß17 %Black + ß18 %Abstain + ev,t ,

[t = 1,2,...Townships; v = Democrat, Republican; pop = Populist].

The eighteen independent variables in the model are divided into three groups: Railroads, Agriculture, and Population. The railroad variables were constructed from contemporary railroad maps and State records. "Distance" is the beeline distance (in miles) measured from the geographic center of each township to the nearest railway station. "Stations" is the number of stations in each township standardized by the square miles of each township. "Competition" is a binary variable equal to "1" if A) a township's shortest beeline distance station was served by two or more independent lines or B) if the township had several stations that were served by two or more independent—but not necessarily intersecting—railway lines. "Town-Aid" and "County-Aid" are binary variables equal to "1," respectively, if a township either voted municipal aid to a railway construction company in order to obtain a railway station, or if it was located in a county which had voted aid to obtain railway lines. Agricultural variables were derived from the manuscript township abstracts of the annual Kansas agricultural censuses of 1889 and 1891 (the 1890 manuscripts are incomplete). "Stock" is the $ value of cattle, milch cows, sheep, and swine per farm acre. "Farm Income" is net per capita operating farm income for 1890 from marketed grain, livestock products, and garden, horticultural and timber products. "%∆ Farm Income" is the percentage change in "Farm Income" between 1889 and 1890 as a result of drought. "%∆ Land Values" is the change in the self-reported value of a farm acre between 1889 and 1891. "Wheat/Corn" is the ratio of planted acres in wheat to those in corn. "% Fenced" is the percentage of farm acres that were fenced. Population variables were taken from a linguistic atlas and State records. "Wealth" is the per-capita assessed value of a township's wealth. "Density" is the number of people per square mile. "Urban" and "Black" are the percentages, respectively, of urban and black residents in each township. "German" and "Other Ethnic" are the percentages of people in each township, respectively, residing in German, or French Canadian, Norwegian, and Swedish, families. "Abstain" is the percentage of each township's voters that abstained in the 1890 election.

The railroad variables were designed to test various theories about railroad resentment. If Populist voters were strangled by monopoly power, the party should have done significantly worse in those townships served by several competing lines. But if Populist voters were motivated by a frustrated hunt for capital gains on land through proximity to infrastructural improvements, Populist candidates should have prospered as they moved farthest away from a rail station. And if Populist voters were motivated simply by a resentment over the fact that their township did not have a railway station then Populist candidates should have done better in a township with no railway station. A township which had voted "Town-Aid" to the railroad lines had a binding service contract enforceable by the State's railway commission; it did not need to worry about losing its station as outgoing freight declined in drought years. Voters there should have been less interested in the Populist platform. The "County-Aid" effect is less certain, for all of a county's voters had to pay for a share of the bonds, but only certain townships actually received the rail line. But the Populists did threaten to outlaw this activity if elected, and this should have frightened those counties which had successfully employed it.

If farmers were the basis of Populist support, we should see a steep falling off in Populist support in townships with an urban presence. And if James Turner's "relative isolation" hypothesis is correct, after controlling for occupation—"Urban" is the occupational proxy for non-farmers—we should find the more densely settled townships rejecting the Populists in significantly greater numbers. The stage of development should also have been important to a voter. A township with a lower level of wealth and a greater degree of unfenced farming acreage may have been relatively more dependent on "Eastern capitalists" for farm development. They were certainly the townships which must have felt the pinch of rising building materials prices: with a partially fenced farm often went a sod hut.

Finally if the Populists thrived in Kansas because they were successful at mobilizing voters who might otherwise have stayed away on election day, the Populist candidate should have been substantially more successful in townships with the lowest rate of abstention.

If the sharp swing in farm income caused by the failure of the corn crop motivated voters, then a larger %∆ drop in farm income should have been an important aspect of the Populist's success. Higher levels of net income, after controlling for the swing from 1889-1890, might have reduced Populist support. But higher income levels were most often recorded among the eastern corn belt townships. They had avoided the worst of the effects of the 1890 drought relative to their central neighbors, but they also had been most specialized in corn in 1889, when corn prices hit bottom; the effect is therefore ambiguous a priori. A larger %∆ in land values meant farmers felt the drought also decreased the future expected earnings from farm land; they too should have been more Populist. And if a long run strategy of wheat specialization made farms more profitable over a series of years, increases in the wheat to corn acreage ratio should have spelled lost voters for the Populist candidate. Finally if over the long-term a Teutonic predilection towards specialized wheat farming gave the German and Russian migrants an agricultural edge on the subhumid prairies of Kansas, they too should have been less Populist after controlling for short run income effects. This should not have carried over to the "other ethnics" however; they faced many of the same agricultural handicaps as the native-born—such as previous residence in the American corn belt—and the Swedish vote in particular may have been driven by a youthful rebellion against church regulation of political life.

The Results

The model was estimated over two samples: the "All Townships" sample of 1,181 townships and a wholly rural sample of 853 townships. The use of a rural subsample was necessitated by an inability to completely isolate the rural vote; not all counties separately reported the urban and rural returns within a township. Full results are available in the complete paper. In Table 1, I report the effect of a one standard deviation change in each continuous variable on the probability of voting Populist, calculated as the product of the estimated partial derivative (constructed at the sample mean) and a one standard deviation change in the indicated variable. All of the variables included in Equation (1) were significant at conventional levels—except for the "% Abstain" and the railway "Competition" effect—in at least one of the two party selection equations for each sample.

Table 1 The Probability of Voting Populist
Variable/Co
nterfactual Estimated Probability
The Sample
Meana
The Rural
Meanb
Wholly
Urbanc
Wholly
Germand
No Germansa .481558
.542792
.215217
.130143
.512447
All = 1,181
Townships
Rural = 853
Townships ∆ in Estimated
Probability
Variable/Co
nterfactual All Rural
One Standard
Deviation ∆-
Railway
Distance
-Railway
Station
-Township
Wealth
-Settlement
Density
-$
Livestock/Ac
e
-Farm
Income
-Wheat/Corn
Ratio
-%∆ Farm
Income
-%∆ Land
Values
-% Fenced
Acres
-% Germanic
-% Other
Ethnic
-% Urban
-% Black
-% Abstain
A New
Railway
Station
in a 36-
Square Mile
Township
-Railway
Station Effect
-Distance
Effect
-Township
Bond Effect
No 1890
Drought
-Land Value
Effect
-Income
Variation
Effect
-Increased
Income
Effect
.02174
-.01586
-.01738
-.00583
-.01409
.01599
-.00463
-.01792
-.00789
-.02713
-.05670
.00765
-.08178
-.01988
-.00351
-.02203
-.03226
-.03314
-.00417
-.05457
.03372
.02007
-.01885
-.02433
-.01695
-.00208
.02335
.00030
-.03454
-.00782
-.01891
-.05456
.01468
-------
-------
-.00142
-.02756
-.02827
-.02941
-.00445
-.11025
.05358
a = Calculated at the 1,181 township mean.
b = Calculated at the 853 rural township mean.
c = 1,181 townships, % Urban =100.0.
d = 1,181 townships, % German = 100.0.

An additional set of railroad variables which were not included in the final specification was a group of binary variables that recorded the independent line(s) which served a particular station; the thought being that certain lines may have earned a greater resentment amongst the settlers they served. As Mrs. Lease stated in 1892, "Kansas suffers from two great robbers, the Santa Fe Railroad and the loan companies..." But it proved impossible to reject the hypothesis that Kansas voters felt an equal resentment against all of the State's railway lines.

It also proved impossible to reject the hypothesis that it did not matter at all to voters whether their township was served by one or several independent railway lines. This "Competition" effect was in all cases extremely small and not statistically significant in any regression. What was most important was 1) whether a township had a railroad station and 2) how far the average farmer had to travel to reach a railhead. Adding a station to a 36 square mile township without one lowered the probability of voting Populist by .0543-.0558. And if it was a station secured by "Town-Aid" the probability of voting Populist fell another .0294-.0331; these townships had been guaranteed by the Board of Railroad Commissioners that they would not lose their stations as drought and depressions slashed the value and volume of outgoing freight.

From the urban effect, clearly the Populist voter was most often a rural farmer. But lower density also significantly enhanced the probability of a rural voter turning Populist. And so too did the "stage of development" variables, most important where the settlement density was lower. Moving one standard deviation down from the mean wealth meant a boost of .0174-.0243 in the probability of voting Populist, while a similarly sized drop in farm fencing added .0271-.0189 to the probability of voting Populist. Those farmers still in the process of developing a farm were a fertile market for the Populist product.

But perhaps most important was the counterfactual effect of "No 1890 Drought." Without the drought farm income and farmland values might have risen along trend, resulting in a drop in the probability of voting Populist of .061 in the Rural model, or 11 percent. Surprisingly the effect of higher income—statistically significant from zero in every estimation—was to increase the Populist voting base, but not by enough to offset the losses from the reduced negative swing in farm income. And this most likely underestimates the counterfactual effect, for the counterfactual conditions of "No Drought" are not well represented in the data sets. Only 38 of 1,181 townships saw any income growth between 1889 and 1890; the average decline between 1889 and 1890 was 45 percent. This swing in income could have been vitiated, if only farmers had raised more wheat and less corn in 1890.

Conclusion

The Populist movement rapidly spread across Kansas in a year of near complete failure in the fields of the chief feed grain—corn. It came on the heels of a multi-year infestation of agricultural pests. And it arrived suddenly at the end of a sustained—but uneven—railway construction boom. The farmer's response to crop failures, insect depredations, and the realization that his township might never get a railway station varied. Some farm families simply gave up and returned East; others joined protest movements, such as the Alliance or the Populists; but almost all of "those who stayed behind" in Kansas also turned inward, toward the farm enterprise itself. Populism has become a movement "in search of context" precisely because the story of how the farm families of the central West adapted to the subhumid agricultural conditions has always been subservient to the story of how a band of colorful political entrepreneurs achieved short-term success in the marketing of a new political product.

Lawrence Goodwyn has argued that in Populism one finds a clear example of how "an intimidated people could create the psychological space to dare to aspire grandly." But the regressions results indicate a simpler explanation. The Populist political entrepreneurs presented a new political product which appealed to farmers because it legitimized their collective failure, as farmers, in terms of outside influences: railroads, moneylenders, or the loss of foreign markets. Certainly these factors were important to the success or failure of farmers as a class. But individual farmers also failed because their human capital endowments led them to make the wrong choices. The cultivator's choice—which crops to plant in order to obtain the greatest return on invested resources—forced farmers to draw upon their own individual decision-making capacity: management skills and techniques learned through the trials of past experiences and transferred to them by their parents, neighbors and kin, and the periodicals of the day. It was this human capital that drove a farmer to specialize in either wheat or corn—but not both—as the chinch-bugs appeared prior to 1890. I conclude, just as the Secretary of the Kansas State Board of Agriculture argued in 1889, "much of the failure in farming in the State is justly attributed to a want of knowledge in regard to proper methods, and the kinds of crops to be grown."

This research was funded in part by an Alfred M. Landon Historical Research Grant administered by the Kansas State Historical Society, Inc. A complete version of this paper is available from the author at: Eugene Sigel, Department of Economics, 818 Thompson Hall, the University of Massachusetts—Amherst, Amherst, Massachusetts, 01003, tel. (413)- 634-5334, or through e-mail ESIGEL@econs.umass.edu.