Sherman’s March

In December 1864, Union General William Sherman sent Abraham Lincoln a Christmas gift. I’m not sure what else Lincoln received in what would be his final Christmas, but Sherman’s gift was probably hard to top. It was the city of Savannah “with one hundred and fifty heavy guns and plenty of ammunition and about twenty-five thousand bales of cotton.” 

The taking of Savannah marked the mid-point of Sherman’s March; the Union Army cut a swath of destruction from Atlanta (captured September 2, 1864) to Savannah, then north to Columbia, SC before finally defeating General Hood’s Confederate Army in Goldsboro, NC in March 1865. Sherman’s expedition is noteworthy because it was one of the first and only instances of “total war”—as Sherman wrote fighting “not only… armies, but a hostile people”—during the American Civil War.

The general may have enjoyed some of our data-centric Broadstreet posts as Sherman was (allegedly) an avid reader of the decennial Federal Census. Sherman “had the [1860] census statistics showing the produce of every county through which [he] desired to pass.” He boasted that “[n]o military expedition was ever based on sounder or surer data” (Trudeau 2008, p. 538).

If I were more like Sherman and a better planner, I would have sent out this post in December (or next December) but my coauthors and I have just (finally) resubmitted our paper on the march and so early May it is. This is for the Broadstreet readers disappointed that Jeff didn’t become known as the “Civil War Guy.” I should note that our working paper isn’t accepted/published yet, but that didn’t stop Wikipedia from citing us, so you should trust us.

Capital Destruction and Economic Growth

In “Capital Destruction and Economic Growth: The Effects of Sherman’s March, 1850-1920,” coauthored with James Lee and Filippo Mezzanotti, we study the effects of the march. Fite (1984) quotes a Georgia planter who anticipates our econometric strategy quite nicely: “I had the misfortune to be in the line of Sherman’s march, and lost everything—Devon cows, Merino sheep, Chester hogs, Shanghai chickens, and in fact everything but my land, my wife and children and the clothing we had at that time.” Our goal is to compare the outcomes for the unlucky counties in the way of Sherman’s scorched earth campaign and their luckier neighbors.

Sherman’s March and 1860 County Boundaries. Based on a War Department Map. The vertex cities on the march are excluded from our analysis: Atlanta (captured September 2, 1864), Savannah, GA (December 10, 1864), Columbia, SC (February 17, 1865), and Goldsboro, NC (March 23, 1865).

Sherman’s record of destruction sounds impressive. The Union Army “destroy[ed] mills, houses, cotton-gins, & c.,” burned railroads and telegraph lines, and confiscated over 5,000 horses, 4,000 mules, 13,000 cattle, 10.5 million pounds of corn, and 9.5 million pounds of fodder (US War Department 1901; Lee 1995). 

When we look at the data, what do we find? We show that destruction did have significant medium-run effects on the southern economy. In 1870, the values of farms and of livestock in “Sherman counties” (those unlucky counties that were marched through) were significantly lower, as was the share of improved acreage (our proxy for agricultural investment). As we detail in the paper, the effects were economically very large. Manufacturing—whether we measure value added, employment, establishments, or capital—also grew less quickly (contracted more) in Sherman counties from 1860 to 1870. But by 1880, manufacturing seems to be back on track, though we detect echos of Sherman in agricultural investment (improved acre share) until at least 1920.

Our findings echo diaries from along Sherman’s path. According to Rubin (2014, p. 20), one Sandersville, GA resident lamented that people “struggled with deprivation, coping with the lack of livestock and supplies” for years. A Hillsboro, GA native lamented that only when Sherman marched had the “beloved [Georgia] country [been left] to desolation and ruin” (Trudeau 2008, p. 175).

The decrease in livestock value is also consistent with the historical record. In the immediate postbellum period, farming was done with “Sherman horses,” the old, “sore-backed” and “abused” animals the Union Army had swapped for fresh rides along the path of the march (Rubin 2014, p. 50-51).

But one of my favorite examples of the extent of Sherman’s destruction comes from Thompson (2004). Apparently, several bankruptcy claimants in postbellum South Carolina blamed Sherman for destroying their records and complicating the unwinding of debt obligations.

The Destruction of the Banks

The number of banks in Sherman’s march and non-march counties was comparable before the Civil War. Immediately after the war, disappeared in all counties. Banks may have recovered slightly faster in non-march counties afterwards.

We also find that the destruction was made worse by the war’s destruction of the southern financial system. In counties with antebellum banks—and so most “treated” by the wholesale disappearance of banks after the war—the Sherman effects were magnified on manufacturing. 

In agriculture, the effects are a bit more complicated. The Southern banking system in the antebellum “was entirely inadequate to meet rural needs” (Fite 1984, p. 27). “[B]anks chose not to provide service to rural areas” and there is no evidence that “rural banks provided (…) credit to small farmers” (Nier 2007, p. 152). Instead, small farmers mostly relied on the credit extended to them by country stores or wealthy local planters. The effects of Sherman’s destruction were blunted in counties with these sources of financing for small farmers.

Lincoln Gimlets and Sherman Neckties

The Library of Congress has quite the collection of images of Sherman’s men destroying railroads.

Sherman and his men were especially focused on destroying railroads, both as a signal of their thoroughness (train tracks are not easy to destroy) and because railroads were essential to a  mid-19th century military operation. Barrett (1956) describes in detail the Union army process of destroying a railroad, including the need to bend the wrap into a twisted doughnut shape known as a “Lincoln gimlet” (Barrett 1956, p. 51). The deformed tracks were also known as Sherman neckties, bowties, or hairpins.

Did the destroyed infrastructure matter? In short, no. We find no evidence that Sherman counties had fewer miles of railroad track in the years after the Civil War. This too is consistent with the historical record. Atack and Passell (1994, p. 378-379) report that although the lines were “destroyed with great vigor by the Union,” the “repairs were immediate.” 

The idea is simple: railroads are network goods and the economic incentives to fix a destroyed mile or two or even three hundred of track are very high because the entire network benefits. This is also, roughly, the logic behind some newer work in economic history estimating the effects of railroads on American economic development; Donaldson and Hornbeck (2016) measure market access. Conceptually, there are markets your products (agricultural and manufacturing) can reach via the transportation network. As the network expands, market access grows and everyone on the network benefits. But repairing tracks is an even cheaper way to increase market access for everyone along the network.

Long-Run Effects?

Sherman had another goal: to “avenge the national wrong [Southerners had committed by] dragging [the] country into civil war” (US War Department 1901). As Vicky wrote earlier this week, revenge (and avenging) can play an important role in society. And the war did end only a month after the march. 

But were there long-run consequences on the people who felt the “hard hand of war”? We trace one important institutional effect in our paper: the rise of sharecropping. 

Sharecropping and tenancy were quite rare before the Civil War. But landowners, faced with a general scarcity of capital and difficulty making regular wage payments to a newly emancipated workforce, could have turned to new land and labor institutions like sharecropping and tenancy as a solution. (Disclaimer: The long (long) literature on sharecropping and tenancy and their development is too long (long, very long) to recount here and the preferences of freed people, the cotton factors, and local merchants all likely played a role in institutional change in the postbellum.)

We find a large increase in smaller farms (under 10 acres, 10 to 19 acres, and 20 to 49 acres) and a consistent decrease in the number of large farms (those 100 acres and over) in Sherman counties after the war. In 1880, about three-quarters of farms under 49 acres were sharecropped or tenant farmed. This is consistent with financial constraints as one explanation for the rise of tenancy and sharecropping in this period. That isn’t to say sharecropping only developed in Sherman counties, but the macro shock of the war and to the credit markets combined with capital destruction in Sherman counties could have made sharecropping especially common.

And it seems superfluous to say, but it seems likely that the history of the postbellum South (and postbellum America) would have been very different if different agricultural institutions had developed after emancipation.


  • I am an Assistant Professor in the Boston University Department of Economics. I am also a Faculty Research Fellow at the NBER in the Development of the American Economy program. My primary research interests are in labor economics and economic history.

One thought on “Sherman’s March

Leave a Reply