Low lying counties along the lower Mississippi River have battled excess water for over two centuries. While today these lands are among the most fertile and heavily agricultural in the country, widespread development required protection from, and removal of, excess water. The Great Flood of 1927 is often characterized as the turning point toward federal control and success in these efforts. We argue using data from the US Agricultural Census that the reality is more complex. Local levee and drainage districts, formed mainly between 1870 and 1910, had by 1920 effectively coordinated landowners in lowland counties, resulting in improved farmland, rising land values, and increases in cotton production. It was the economic success of levee and drainage districts that created the tragic circumstances of the 1927 flood. In the reclaimed floodplains, large cotton plantations developed after the Civil War induced migration of the black workers most catastrophically affected. The economic productivity of these farms led to increases in landowners’ wealth and increases in their political power post flood in shaping state and federal policy. The destruction of these farms created a sympathetic narrative related to the broad public benefits of flood protection. While other authors have focused on 1927 as a failure of localism, we argue that it was also a demonstration of its success.
Introduction
The Mississippi River drainage basin, or watershed, is continentally large, covering parts of 32 states and two Canadian provinces. It collects rain and snowfall from as far west and north as arid north-central Montana and southeastern Alberta and as far east as the much better-watered New York state and Virginia. The main stem of the basin is the more or less north–south-oriented Mississippi River itself, but the drainage basin collects waters into the trunk line from a span of major rivers, including the Missouri, Ohio, and Arkansas, from a total of 740 million acres — 37% of the lower 48 states land area. The water so collected funnels in the main to a single outlet at New Orleans [1]. Figure 1 provides an overview of the entire Mississippi River Basin.
The map displays the Mississippi River drainage basin covering much of the central United States. The Mississippi River runs from the upper Midwest southward to the Gulf of Mexico. Major tributaries are labelled, including the Missouri, Ohio, Arkansas, Red, Tennessee, Platte, Yellowstone, and Kansas rivers. State boundaries and major cities such as Minneapolis, St Louis, Chicago, Memphis, and New Orleans are shown for reference. The river network illustrates how waterways from the Rocky Mountains, Great Plains, and Appalachian regions flow into the Mississippi system before reaching the Gulf of Mexico.Map of the Mississippi river basin
Source:https://commons.wikimedia.org/wiki/File:Mississippiriver-new-01.png
The map displays the Mississippi River drainage basin covering much of the central United States. The Mississippi River runs from the upper Midwest southward to the Gulf of Mexico. Major tributaries are labelled, including the Missouri, Ohio, Arkansas, Red, Tennessee, Platte, Yellowstone, and Kansas rivers. State boundaries and major cities such as Minneapolis, St Louis, Chicago, Memphis, and New Orleans are shown for reference. The river network illustrates how waterways from the Rocky Mountains, Great Plains, and Appalachian regions flow into the Mississippi system before reaching the Gulf of Mexico.Map of the Mississippi river basin
Source:https://commons.wikimedia.org/wiki/File:Mississippiriver-new-01.png
Of considerable hydrologic significance for the history of American settlement and agriculture is the stretch of the River south of St. Louis — the Mississippi Alluvial Plain — which periodically experiences flooding from the Mississippi and its tributaries as they connect with the Mississippi. Harrison (1961) terms this area “the Alluvial Valley of the Lower Mississippi River” and measures it at 30 million acres, about the size of Florida.
The lower Mississippi developed economically much later than did the Midwest, despite the enormous agricultural potential of the Alluvial Plain. In Edwards and Thurman (2025) we chronicle the critical importance of soil drainage to development in the Midwest and how technological innovation, importantly drain tile, spread from New England to the Midwest in the middle of the nineteenth century and transformed the swampy lands of the Upper Mississippi Valley into the Corn Belt. We discuss in that article how the transformation was the product of institutional and legal innovation in the form of typically sub-county size drainage districts. But the problems of saturated lands and periodic flooding along the lower Mississippi differed in degree, if not in kind, from those in the Midwest, and the period of institutional innovation to accomplish drainage was longer and came later. The Mississippi Alluvial Plain initially was mainly passed over by settlement, making it the final frontier of agricultural development in the United States.
Lowland counties are some of the most fertile in the country, but the inability of the government or landowners to unlock this value was reflected in depressed prices until the late 1800s. Today, land in these counties is almost entirely privately owned (97% of the “Mississippi Delta Cotton and Feed Grains Region,” as defined in United States Department of Agriculture, Natural Resources Conservation Service (2006)) and some of the most heavily used for agriculture in the country by percentage of total land area in production. Like in the Midwest, lowland counties of the lower Mississippi saw limited collective action to solve drainage and flood control problems until state legislation lowered the transaction costs of coordination. Local levee and drainage districts, formed mainly between 1870 and 1910, took considerably longer than their counterparts in the Midwest to effect drainage and flood control, largely due to the greater magnitude of projects required and the greater scale of coordination this entailed, typically the size of full counties or larger (Edwards and Thurman, 2025).
As in the Midwest, agricultural development in the Alluvial Plain required both technological and institutional innovations to enable the coordination across landowners needed to prevent flooding and drain soils. Coordination via districts required states to pass district-enabling laws, which occurred between 1859 and 1912. After much experimentation and failed efforts at coordination and financing, local markets began to reflect the successful removal of water as the feasibility of flood control and drainage was demonstrated. Land prices and development rose together in the early twentieth century as coordination mechanisms developed, often in a back-and-forth competition with the river’s floods. Today, a web of special districts owns levees, coordinates drainage, and collaborates with the US Army Corps of Engineers on design and investment.
In this paper, we develop an economic understanding of the emergence of drainage and levee institutions along the Lower Mississippi, using historical data and an array of contemporaneous and contemporary accounts to trace how state capacity and transaction costs shaped the evolution of water removal.
Levees on the Mississippi before the civil war
In this section, we trace the relevant history of the Lower Mississippi Valley prior to the formation and activity of levee and drainage districts, from early European exploration until the Civil War.
Economic development before the Louisiana purchase in 1801
The earliest European exploration of the Mississippi River occurred in the 1500s [2]. Prior and through this period the land was occupied by Native American tribes. In 1543, Hernando DeSoto became the first European to describe flooding along the River. In 1682, the French claimed for the king of France the “Louisiana Territory,” the entire area drained by the Mississippi. Levees were first built to protect the city of New Orleans from Mississippi floodwaters in 1717 and the city became the capital of the Territory in 1721. One of a number of Great Floods inundated New Orleans in 1735, erasing the previously built levees. Rebuilding after the 1735 flood resulted in there being new levees in place for 20 miles above and 30 miles below New Orleans by 1752 (Harrison, 1961, p. 55). During the next 50 years there was little new levee building, but plantations along the river were improved (Harrison, 1961, p. 56).
In 1762, as a result of the Seven Year War (also known as the French and Indian War), France ceded to Spain the western half of the Mississippi Basin and to England the eastern half. France regained the west bank of Louisiana territory in 1800 via land trades between Napoleon Bonaparte and Spain. This occurred just prior to the United States acquiring the territory through the Louisiana Purchase of 1801. The area was largely unexplored and mostly controlled by Native American tribes.
The newly acquired Louisiana Territory was subsequently subdivided and the much smaller state of Louisiana was created in 1812. At this time there were a few settlements on the West Bank of the Mississippi, now owned by the United States, between the Red and Arkansas Rivers. Thin settlement characterized the East Bank between the Red and Yazoo Rivers but by 1820 there was “rapid expansion of cotton plantations in lowlands of the Mississippi River north of the Red and Yazoo Rivers (Harrison, 1961, p. 38).”
Early levees and drainage: 1801–1848
As development along the banks of the Mississippi proceeded in the early and mid-nineteenth century, states along the lower Mississippi River — Missouri, Arkansas, Mississippi, and Louisiana — began to address the public finance challenges of building and maintaining levees. At initial settlement, riverfront landowners were the most strongly incentivized to build and maintain levees. Subsequently, these landowners were made legally responsible by state law for construction and maintenance. For example, Harrison describes 1838 legislation in Mississippi that was patterned after earlier law in Louisiana:
[L]egislation of early Mississippi followed the practices of Louisiana, in that it placed the burden of flood control on the local holders of front lands. Fines were imposed for inspectors and planters who failed in their duty toward flood control. When nonresident holders could not be reached, work on their lands was to be let to the lowest bidder and the cost of the work held against the land by the county boards of police. (Harrison, 1961, p. 61)
Responsibility for levees was extended to backland beneficiaries of levees when the Mississippi legislature passed in 1846 a levee tax on owners away from direct contact with the River.
The idea of general taxes for levee building soon entered Mississippi Valley flood-control law and regulations, even though they continued to hold front owners responsible. [Inspectors were instructed to] ‘lay out the line of said levee, and to estimate the probable cost thereof, and report said amount to the president of the board of police, who shall thereupon convene the board, which, when so convened, shall assess a tax on said land for the requisite amount, to be called a levee tax, together with a sufficient sum for defraying the expenses in collecting the same.’ This act provided that the back lands that benefited by the building of the levee should be taxed in proportion to the benefits the proprietors of said back lands would receive. (Harrison, 1961, p. 62)
The same Louisiana legislation in 1846 permitted the creation of levee districts to coordinate construction and maintenance of levees. The power of levee districts to tax possibly unwilling participants in the levee building was established in an 1848 law:
In Mississippi, reliance on front landholders began to break down. The flood of 1844 made it clear that the riparian holders could not do the job unaided. Threats of fine or sale of property for failure to build proper levees and drains began to be replaced in the law by a system of general levee-tax proposals that would apply to all land protected. An act of March 4, 1848 provided for “[…] a uniform tax on all land in said county (Tunica), to be called a levee tax, which tax shall not exceed three cents per acre on said land, for any one year, but the same may be kept up from year to year, until the amount raised thereby (together with what the State may appropriate) shall be sufficient for the erection of said levee.” (Harrison, 1961, p. 63)
Other downriver states adopted similar measures as the 1848 Mississippi legislation was mimicked in Arkansas and Louisiana.
While state law evolved to handle the challenges of flood control on private lands, substantial land remained in the federal estate as a result of the Louisiana Purchase. In 1829, Congress had appropriated $5,000 for the Army Corps of Engineers to survey and map the Ohio and Mississippi Rivers. Missouri became a state in 1821 and, in 1824, Congress appropriated $75,000 for snag removal in the Mississippi below the Missouri (i.e., south of Saint Louis) and in the Ohio River. These small investments were underwhelming relative to the more than 20 million acres of swamplands located along the Mississippi and its tributaries.
Swamp land acts to the civil war: 1849–1865
Congress began to transfer waterlogged lands back to the states with a series of Swamp Land Acts, which instructed states to sell the transferred swamp land to finance their efforts at flood control. The first such Act was passed in 1849 and transferred 9.4 million acres to the state of Louisiana. The Swamp Land Act of 1850 transferred federal land to other lower Mississippi states: Arkansas (7.7 million acres), Missouri (3.3 million acres), and Mississippi (3.3 million acres). The 1850 and later Acts also transferred land from the federal estate to state governments outside of the Lower Mississippi River Basin.
While Congress tried to induce states to fund drainage projects through the sale of the transferred land, these plans were eventually abandoned. Like the federal government before them, state governments at the time lacked the institutional capacity to coordinate or fund large public works projects. The undrained land was largely deemed worthless and sold to speculators. It was these speculators, or the successive landowners of the purchased lands, who eventually succeeded in developing the land. The disbursement of land to lower Mississippi states in 1849 and 1850 led to an extended period of legislative efforts to unlock the great wealth that the lowlands represented. For several decades, these efforts were stymied by the need to protect the lands from the periodic flooding of the Mississippi and its tributaries, the Missouri, Ohio, Arkansas, and Red Rivers [3]. Over time, state law evolved to enable local communities to organize investment and compel payment through levee districts patterned after Midwestern drainage districts. “Louisiana legislation that permitted the creation of levee districts also established the basis for taxation in the interest of flood control.” (Harrison, 1961, p. 62)
Flood protection and drainage efforts were brought to a halt by the American Civil War. When the war ended in 1865, local governing institutions were in disarray and the levees along the lower Mississippi and its tributaries had been damaged by strategic sabotage in battles, notably the dynamiting of levees by Union troops during the Vicksburg Campaign of 1863. Levees also suffered from the deferral of maintenance during the war.
The historical political economy of lowland levee construction
The end of the Civil War opened the possibility of continued development of the lowland counties, but obstacles remained. Although state governments had made legislative attempts to address the coordination problems inherent in building levees, a viable institutional structure to raise investment capital remained elusive. Over the decades following the war institutions arose that ultimately succeeded in building and maintaining levees and overseeing the drainage of protected lands. We contrast two explanations for the ultimately successful protection, drainage, and agricultural development of lowland areas and explore empirically their explanatory power.
One explanation is techno-centrist and emerged definitively after the Great Flood of 1927. It argues that flood control at the scale of the Lower Mississippi Basin is such a large and complex problem that it can only be planned and managed at the scale of the entire basin (Barry, 1997). An alternative explanation, suggested by the historical record, is one we attribute in part to Hayek (1945), who argued that decentralized information, specific to circumstance and place, becomes embodied in price signals through the coordinating actions of local entities. This second view holds that important reductions in the transaction costs of coordinating flood control and drainage in the Lower Mississippi were achieved between roughly 1870 and 1920, when states passed organic acts authorizing the creation of levee and drainage districts.
In this second view, local action is necessary to generate land wealth and price signals, which derive from the opportunities available to local actors. These market prices depend critically on the information available and the transaction costs of exchange. Both information and transaction costs, in turn, are determined in a context of property rights and “rules of the game.” Once the levee districts carried out effective (though not complete) flood control at the sub-basin level, land value and agricultural activity increased on the drained and fertile soils. While great damage was done to levees by the Great Flood of 1927 and lesser floods before then, it was the value created by levee district coordination that motivated and enabled the post-1927 coordination of levee districts by the federal government, specifically the Army Corps of Engineers.
Geographic definition of the study area of lowland counties
To develop this second interpretation conceptually and, in following sections, empirically, we define the study region to which our theory applies: the lowland counties adjacent to the Mississippi River stretching from its confluence with the Ohio River at Cairo, IL to its confluence with the Red River in Louisiana, below Natchez, MS. This is the lowland area stretching from the southern tip of Illinois to the point where the ankle meets the foot of the Louisiana boot. Figure 2 presents three similar collections of such counties taken from prior work: the Lower Mississippi River floodplain mapped in 1887 by the Mississippi River Commission and digitized by Allen (2025); “Bottomland Counties” from Otto (1999); and the five northern regions of the LMRB defined by Ladd and Travers (2019). The region is bounded on the east by the Mississippi River in Kentucky (between Cairo, IL and Memphis, TN) and again, moving downriver, in the middle of Louisiana (South of Vicksburg, MS). Between these two stretches the boundaries of the lowland area reflect hills that rise away from the River to the east and serve as natural boundaries to flooding [4].
The figure presents two side by side maps of the lower Mississippi River region. The left map shows the Mississippi floodplain circa 1899 and the Mississippi alluvial plain, overlaid on state and county boundaries. Distinct shaded areas trace the river corridor from the upper reaches through the Delta to the Atchafalaya and coastal deltaic plains near the Gulf of Mexico. The right map highlights bottomland counties along the Mississippi River, with these counties shaded and labelled, while surrounding lowland counties are outlined as the full sample. Rivers and major waterways are shown in both maps to indicate spatial alignment between floodplain geography and county boundaries.Regional definition of lowland counties
The figure presents two side by side maps of the lower Mississippi River region. The left map shows the Mississippi floodplain circa 1899 and the Mississippi alluvial plain, overlaid on state and county boundaries. Distinct shaded areas trace the river corridor from the upper reaches through the Delta to the Atchafalaya and coastal deltaic plains near the Gulf of Mexico. The right map highlights bottomland counties along the Mississippi River, with these counties shaded and labelled, while surrounding lowland counties are outlined as the full sample. Rivers and major waterways are shown in both maps to indicate spatial alignment between floodplain geography and county boundaries.Regional definition of lowland counties
We exclude from this study the Mississippi River Delta in Louisiana south of the Red River Confluence. This region — the Atchafalaya and Deltaic and Chenier Plains — is the newest and lowest portion of the alluvial plain [5]. By 1828, levees in this region were continuous from New Orleans to Red River Landing (Harrison, 1961, p. 57) to enable navigation and protect cities and agricultural land immediately adjacent to the river. It is lower and wetter than the upriver regions, with more saltwater interaction and less extensive agricultural development, in part because farmers were not able to develop the backwater areas away from the river as they did north of the Red River. As a result of these features, the southern two regions in Figure 2 have a distinct history of development and we do not include them in the relatively homogeneous set of alluvial counties above the Red River.
We will refer to our study area, the set of counties shown in the right panel of Figure 2, as lowland counties. These are flat counties with fertile soil, periodically flooded, requiring drainage for agriculture when not flooded, but not typically submerged for large fractions of the year. They are also the counties which we can identify from the historic record as having drainage or levee districts prior to 1920 (discussed in detail in the section “How Lowland Levee Districts Evolved” and Online Appendix Table A1).
Summary statistics
| Variable | Upland counties | Lowland counties | ||||
|---|---|---|---|---|---|---|
| 1860 | 1910 | 1959 | 1860 | 1910 | 1959 | |
| Observations | 239 | 244 | 242 | 43 | 47 | 47 |
| Total farm value (mil 2020$) | 56 (56) | 235 (251) | 203 (147) | 113 (103) | 172 (95) | 380 (242) |
| Land value per acre (2020$) | 299 (241) | 913 (2,017) | 865 (564) | 652 (392) | 909 (395) | 1,290 (419) |
| Prop. of county improved | 0.14 (0.12) | 0.40 (0.25) | 0.29 (0.18) | 0.12 (0.15) | 0.29 (0.16) | 0.49 (0.22) |
| Total number of farms | 630 (351) | 2,612 (1,084) | 1,438 (616) | 293 (128) | 3,520 (2,174) | 1,656 (711) |
| Total acres in farms | 173,287 (94,255) | 266,787 (96,062) | 235,501 (99,419) | 160,404 (99,695) | 190,083 (80,902) | 284,394 (112,917) |
| Bushels of corn | 470,187 (387,022) | 957,908 (969,989) | 912,586 (1,279,652) | 301,364 (215,568) | 647,755 (568,127) | 563,151 (821,110) |
| Bales of cotton | 5,235 (9,731) | 4,532 (6,188) | 2,800 (5,859) | 18,631 (23,295) | 19,141 (15,132) | 62,428 (48,278) |
| Median soil wetness index | 52.04 (10.22) | 52.09 (10.18) | 52.12 (10.21) | 69.12 (10.49) | 69.70 (10.30) | 69.70 (10.30) |
| Median productivity index | 8.03 (3.62) | 7.95 (3.63) | 7.97 (3.61) | 10.84 (1.00) | 10.81 (0.97) | 10.81 (0.97) |
| Median elevation (m) | 190.99 (104.20) | 188.22 (104.95) | 188.27 (105.35) | 53.53 (24.98) | 51.85 (24.54) | 51.85 (24.54) |
| Std. dev. elevation (m) | 32.21 (24.29) | 31.93 (24.12) | 31.92 (24.13) | 10.75 (9.09) | 10.05 (8.99) | 10.05 (8.99) |
Flood control requires coordination beyond that required for drainage
In what became the Corn Belt across Illinois, Indiana, Iowa, Minnesota, Wisconsin, and Ohio, coordination among landowners through drainage districts spurred investment over millions of acres (see Edwards and Thurman, 2025). In these areas, without large, low gradient river systems, the individual drainage district remains the primary governance structure today. Small drainage districts were less successful in the Lower Mississippi, where lowland areas are periodically flooded — on average twice a decade and sometimes for weeks and months at a time. Drainage cannot profitably be implemented if the drained land isn’t protected from flood.
In the Midwest, drainage investment on an individual scale was achieved with coordination across hundreds or thousands of acres (Prince, 2008; Wright, 1907). Water removal on the lower Mississippi required an entirely different scale of coordination. Burns (1954) reports that 92 drainage districts formed in the early twentieth century in Blue Earth County, MN. The average size of a Blue Earth County district in 1930 was 1,161 acres. In Story County, IA there were 95 districts by 1920 with an average size of 2,080 acres per district (Hewes and Frandson, 1952). Comparing these averages with the 285,000 acres comprising the Cypress Creek Drainage District and the 40,000 acres covered by the Ross Drainage District, both in Arkansas, suggests the difference in required scale (Deaton, 2016).
Further, the larger geographic scope of drainage institutions in the Lower Mississippi created the potential for conflict among organizations that was absent in the Midwest. One set of challenges faced by the newly created levee districts was the external effects imposed by, mainly upriver, districts resulting from their own efforts at fortifying and raising the levels of levees. In 1861 a federally funded and much celebrated report by civil engineers Humphreys and Abbott alluded to the possibility of one region being flooded by the efforts of another and, on the other hand, the possibility of one area being protected by the efforts of another:
[…] much of the want of success attending the efforts to secure the alluvial lands from overflow has arisen not from inherent difficulties in the construction of works of protection, but from the adoption of systems which have allowed one district to be submerged in consequence of the insufficient character or faulty execution of the laws of another, or left to be protected by taxes levied upon another. (Humphreys and Abbott, 1861, p. 152; as quoted in Harrison, 1961, p. 63)
It was this issue coupled with the large-scale flooding along the Mississippi River in the late nineteenth and early twentieth centuries that encouraged the gradual reentry of the federal government into the management of the swamplands for which it had abdicated responsibility a few decades earlier. This reentry accelerated after the Great Flood of 1927, spurring major investment by the federal government into existing districts.
By the 1930s, the federal government had substantially enhanced its institutional capacity to undertake natural resource and governance actions, capacity that it lacked at the time of the Swampland Acts in 1850. Evidence of this institutional capacity comes from the large-scale irrigation projects in Western states begun in 1908 and carried out by the Bureau of Reclamation, established in 1902. By 1927, the federal government could leverage large amounts of money and its engineering expertise (in particular, the Army Corps of Engineers) to coordinate management across levee districts and states, even when interests of districts and states diverged among themselves. Yet federal investment in the lower Mississippi lowlands did not occur in a vacuum. It was the success of levee districts, and the increases in land value their achievements induced, that motivated politicians to mediate the coordination challenges among districts.
How lowland levee districts evolved
The post-war period saw a succession of levee districts created. Early and notable was the Mississippi Levee District (or Board) in 1877, “the model for other bottomland levee districts.” (Otto, 1999, p. 26). The Mississippi Levee District contracted to rebuild levees in southern Delta counties after the Civil War, and the rebuilt levees held in the flood of 1884. Otto describes the activities of the District and the beginnings of federal involvement with districts:
Selling bonds and imposing taxes to fund levee work, the Mississippi Levee District (MLD) contracted to rebuild the levees in the southern delta counties […] Although these handbuilt levees resisted floodwaters in normal years, they failed during the great flood of 1882. Following this destructive flood, the Mississippi River Commission, which had been created to improve river navigation, began providing federal funds for levee construction in the Mississippi River Valley […] The successful levee-building of the MLD inspired the state of Mississippi to create the Yazoo-Mississippi Delta levee District (Y-MDLD) in 1884. Responsible for protecting the northern Delta counties, the Y-MDLD began building a coherent levee line in the northern Delta. (Otto, p. 26)
Inspired by success of the Mississippi Levee District, the Yazoo-Mississippi Delta Levee District (YMDLD) was created in the northern part of the Delta in 1884. To raise the capital needed to build levees, the district issued Levee Bonds for the first time in 1884 in the amount of $500,000. The bond paid an 8% coupon and matured in 25 years. Over the next 40 years the Board issued a total of almost $6M in bonds. Initially, the market viewed the bond proposition as a risky endeavor, with the spread over the average railroad bond rate exceeding 3.8 percentage points. Over time, and especially after 1900, the market began to signal much lower risk, with the spread decreasing dramatically relative to high-grade railroad bonds as seen in Figure 3.
The line chart plots values over time from about 1885 to 1922 on the horizontal axis, with the vertical axis ranging from 0 to 9. Two series are shown. The dashed line with dark markers represents the Yazoo Mississippi levee coupon rate. It starts near 8 around 1885, declines to about 6 by 1887, remains near 6 until the late 1890s, drops to around 4 in the early 1900s, and then rises gradually to just above 5 by the early 1920s. The solid line with light markers represents railroad bond yield. It begins near 4.2, declines to about 3.3 by the late 1890s, stays near that level in the early 1900s, and then increases steadily to around 4.3 by the early 1920s.Yazoo-Mississippi delta coupons relative to railroad bond yields
Note(s):YMDLD bond coupons from Harrison (1951). Railroad bond rates from Macaulay (1938)
The line chart plots values over time from about 1885 to 1922 on the horizontal axis, with the vertical axis ranging from 0 to 9. Two series are shown. The dashed line with dark markers represents the Yazoo Mississippi levee coupon rate. It starts near 8 around 1885, declines to about 6 by 1887, remains near 6 until the late 1890s, drops to around 4 in the early 1900s, and then rises gradually to just above 5 by the early 1920s. The solid line with light markers represents railroad bond yield. It begins near 4.2, declines to about 3.3 by the late 1890s, stays near that level in the early 1900s, and then increases steadily to around 4.3 by the early 1920s.Yazoo-Mississippi delta coupons relative to railroad bond yields
Note(s):YMDLD bond coupons from Harrison (1951). Railroad bond rates from Macaulay (1938)
The declining spread indicates that the tax revenues resulting from levee construction were viewed as less risky by investors over time. When issuing its first bonds, the district was reeling from three successive years of flooding in 1882, 1883, and 1884. Economic activity was depressed, and land for sale had few buyers (Harrison, 1951). Over time, more secure levees allowed for increased agricultural production and higher tax revenues, enhancing the financial standing of the district. At its first offering, the district faced strong skepticism over its ability to pay back bonds, in part due to the State of Mississippi’s bond default in the 1840s, as well as the high ongoing rates of land foreclosure in the Delta Region resulting from failed prior bids at raising capital for levee construction. The district initially relied on both ad valorem and commodity taxes to repay the bonds, with its authority to impose a cotton tax on all production in fields behind the levees a key condition of finding buyers (Harrison, 1951).
Following YMDLD, the Tensas Basin and Fifth Louisiana districts were created in northeast Louisiana in 1884 and 1886, respectively, the Clay and Greene districts in Eastern Arkansas in 1887, and the Laconia and Red Fork levee districts in Eastern Arkansas in 1891. In 1893, the St. Francis Levee District was created in eastern Arkansas on the sunken lands that were replumbed by the 1811–1812 New Madrid earthquakes. The year 1897 saw the greatest flood on record in the Mississippi to date. Subsequently, local levee districts began rebuilding a patchwork levee line (Otto, 1999, p. 27). “By 1897, state levee districts lined the lower Mississippi Valley.” (Otto, 1999, p. 35). Figure 4 shows the levee organization associated with each lowland county [6].
The map displays levee and drainage districts distributed along the Mississippi River from southeastern Missouri through Arkansas and Mississippi into northeastern Louisiana. Counties are outlined and labeled, with river channels shown as winding lines. Distinct shading and patterns identify individual levee districts, each labeled with a district name and year of establishment, including Mississippi Levee District, Yazoo Mississippi Delta Levee District, St Francis Levee District, Fifth Louisiana Levee District, and others. Counties such as Scott, New Madrid, Coahoma, Sunflower, Yazoo, Madison, Concordia, and Catahoula are visible within the levee areas. A legend lists all levee districts with corresponding patterns and founding years, indicating the historical spread of flood control governance along the river.Levee organizations on the lower Mississippi
The map displays levee and drainage districts distributed along the Mississippi River from southeastern Missouri through Arkansas and Mississippi into northeastern Louisiana. Counties are outlined and labeled, with river channels shown as winding lines. Distinct shading and patterns identify individual levee districts, each labeled with a district name and year of establishment, including Mississippi Levee District, Yazoo Mississippi Delta Levee District, St Francis Levee District, Fifth Louisiana Levee District, and others. Counties such as Scott, New Madrid, Coahoma, Sunflower, Yazoo, Madison, Concordia, and Catahoula are visible within the levee areas. A legend lists all levee districts with corresponding patterns and founding years, indicating the historical spread of flood control governance along the river.Levee organizations on the lower Mississippi
Drainage district formation was concurrent with levee protection. In 1905, the ambitious Little River Drainage District in the bootheel of Missouri was approved by the Missouri legislature and governor, combining drainage and flood protection functions in a single entity. A different approach was taken by the 1906 Alcorn Law in Mississippi, which created separate entities tasked solely with drainage, often in areas already protected from flooding by a levee district. Patterned after the Illinois Drainage Law of 1879, the Alcorn Law called for drainage district creation through local courts and county boards of supervisors. (Brandfon, 1967, p. 125) These drainage districts worked behind levees to remove standing water and move it away from farmlands. Thus, while the levee organizations were responsible for the main thrust of levee investment in the lowland counties, other, smaller organizations maintained other infrastructure and undertook drainage.
In all four states, average agricultural land values in the lowland counties fell or stayed constant following the Civil War and through 1880. As districts began to form and demonstrate their effectiveness, land values began to rise and continued to grow until the onset of the Great Depression. Figure 5 plots the average county total farmland value in each of 12 major levee districts in the study region. In the next section we build on these stylized observations to examine comparative agricultural development in the LMRB, and how it coincided with the growth of levee districts.
The figure presents four line charts labelled Missouri, Arkansas, Mississippi, and Louisiana, each showing county farmland value over time from 1860 to about 1965 measured in 2020 dollars in millions. In Missouri, Little River Drainage District and St Johns Levee and Drainage District rise slowly until 1900, peak near 1920 around 300, dip in the 1930s, then increase sharply after 1950 to about 1000 and 700. In Arkansas, six levee districts follow similar paths, with modest values before 1900, steady growth to around 150 to 300 by 1920, declines around 1930, and strong increases after 1950, reaching between 400 and 1000. In Mississippi, Mississippi Levee District and Yazoo Mississippi Delta Levee District increase to about 400 by 1920, fall in the 1930s, then rise sharply after 1950 to around 700 to 750. In Louisiana, Fifth Louisiana and Tensas Levee Districts start higher in 1860, drop by 1870, grow gradually to about 120 by 1940, then increase rapidly after 1950 to roughly 500 and 400.County farmland value by levee district
The figure presents four line charts labelled Missouri, Arkansas, Mississippi, and Louisiana, each showing county farmland value over time from 1860 to about 1965 measured in 2020 dollars in millions. In Missouri, Little River Drainage District and St Johns Levee and Drainage District rise slowly until 1900, peak near 1920 around 300, dip in the 1930s, then increase sharply after 1950 to about 1000 and 700. In Arkansas, six levee districts follow similar paths, with modest values before 1900, steady growth to around 150 to 300 by 1920, declines around 1930, and strong increases after 1950, reaching between 400 and 1000. In Mississippi, Mississippi Levee District and Yazoo Mississippi Delta Levee District increase to about 400 by 1920, fall in the 1930s, then rise sharply after 1950 to around 700 to 750. In Louisiana, Fifth Louisiana and Tensas Levee Districts start higher in 1860, drop by 1870, grow gradually to about 120 by 1940, then increase rapidly after 1950 to roughly 500 and 400.County farmland value by levee district
Empirical analysis of agricultural development
We construct a decadal panel of county-level data spanning 109 years, from 1860 to 1969, on Improved Acres, Total Farm Value, Farmland Value per Acre, Corn (in bushels), and Cotton (in bales) from United States Censuses of Agriculture digitized by Haines et al. (2015). To accommodate changes in counties over time, we scale data to 1910 boundaries using area-weight crosswalks constructed by Ferrara et al. (2024). We define lowland counties as in Figure 2 (these counties are identified in bold in Online Appendix Table A1). Table 1 provides summary statistics for lowland counties and upland counties in the four states in 50-year increments: 1860, 1910, and 1959. Upland counties generally include all counties in a state that are not lowland, except for those that we exclude from consideration because they lie within the Mississippi Alluvial Plain as defined by Ladd and Travers (2019). The excluded counties in the Alluvial Plain are either outside the study region (the Louisiana Parishes in the Atchafalaya and Deltaic and Chenier Plains) or those which we were not able to associate with a levee district (see Table A1). There are 43 lowland counties with agricultural census data in 1860, a number that increases to 47 in 1910 and 1959. There are 239 upland counties with census data in 1860, a number that increases slightly to 244 in 1910 then decreases to 242 counties by 1959.
The bottom four variables in Table 1 provide a lowland–upland comparison of four geophysical variables. Natural Soil Wetness Index (NSWI) represents the water content in the soil of a given county absent human modification (Schaetzl et al., 2009). The NSWI is an ordinal measure of long-term soil wetness ranging from 0 to 99. Soils with a NSWI of around 60 are generally termed “somewhat poorly drained,” while higher NSWI values represent more poorly drained up to 99, which is open water. The NSWI is derived from soil classification and slope and is not affected by drainage or irrigation. Land quality is measured using the Productivity Index (PI), an ordinal measure of the productivity of a soil (Schaetzl et al., 2012). The PI uses soil taxonomy information to rank features that tend to be associated with low or high soil productivity from 1 (least productive) to 19 (most productive). To understand topography, we construct a county-level measure of roughness: the standard deviation of 40-m grid elevation observations in a county.
Comparing lowland and upland counties, lowland counties are lower, flatter, and more productive than upland counties. They also have considerably wetter soils, which means farms in these counties benefit disproportionately from levee and drainage investments.
Absent extensive drainage and flood control, lowland counties in the four states had land values similar to upland in 1910 ($909/acre in lowland counties compared to $913/acre in upland counties). This changed by 1959, when per acre land values in lowland counties were 50% higher than those in upland counties. While upland counties saw large increases from 1860 to 1910 in proportion of land improved and total farmland value, increases in lowland counties occur later, from 1910 to 1959. Lowland counties also saw large increases in cotton production from 1910 to 1959, while upland counties did not.
The trends in means can be visualized more clearly, and separately by state, in Figures 6 and 7, which show total county farmland value and total county cotton produced, split into lowland and upland counties for comparison. Figure 6 shows that by 1920 lowland counties in Missouri, Mississippi, and Arkansas had seen multiple decades of increasing farmland value. Mississippi, whose levee districts were created first, saw total farmland value per county in lowlands exceed values on upland counties as early as 1890, earlier than Arkansas and Missouri. Because there was little expectation of significant federal involvement in the management of the river, these land value increases appear to be capitalizing the gains from successful local levees and drainage organizations.
The figure shows four line charts labelled Missouri, Arkansas, Mississippi, and Louisiana, plotting county farmland value from 1860 to about 1965 in 2020 dollars in millions for upland counties and lowland counties. In Missouri, upland values rise from about 60 in 1860 to around 420 by 1910, fall to about 180 in 1940, then increase to about 420 by the mid 1960s, while lowland values rise more strongly after 1900, reaching nearly 950 by the mid 1960s. In Arkansas, upland values increase gradually from about 40 to around 220, whereas lowland values grow faster after 1900 and reach about 820 by the mid 1960s. In Mississippi, upland values rise slowly from about 60 to around 250, while lowland values increase more sharply, peaking near 400 around 1920 and reaching about 750 by the mid 1960s. In Louisiana, both series start low after 1870, but lowland values rise faster after 1950, reaching about 400 compared with about 200 for upland counties.County means of total farmland value
The figure shows four line charts labelled Missouri, Arkansas, Mississippi, and Louisiana, plotting county farmland value from 1860 to about 1965 in 2020 dollars in millions for upland counties and lowland counties. In Missouri, upland values rise from about 60 in 1860 to around 420 by 1910, fall to about 180 in 1940, then increase to about 420 by the mid 1960s, while lowland values rise more strongly after 1900, reaching nearly 950 by the mid 1960s. In Arkansas, upland values increase gradually from about 40 to around 220, whereas lowland values grow faster after 1900 and reach about 820 by the mid 1960s. In Mississippi, upland values rise slowly from about 60 to around 250, while lowland values increase more sharply, peaking near 400 around 1920 and reaching about 750 by the mid 1960s. In Louisiana, both series start low after 1870, but lowland values rise faster after 1950, reaching about 400 compared with about 200 for upland counties.County means of total farmland value
The figure shows four line charts labelled Missouri, Arkansas, Mississippi, and Louisiana, with the vertical axis showing bales of cotton in thousands and the horizontal axis showing years from 1860 to about 1965. In Missouri, upland counties remain near 0 throughout, while lowland counties rise from about 5 in 1860 to a peak near 78 around 1960 before declining slightly. In Arkansas, upland counties increase from about 4 to around 11 by 1930 and then fall to about 2 by the mid 1960s, while lowland counties rise steadily from about 8 to nearly 75 by 1950 before easing to about 58. In Mississippi, upland counties fluctuate between about 6 and 15 and decline after 1930, while lowland counties increase from about 26 to around 75 by 1930 and remain above 70 thereafter. In Louisiana, upland counties vary between about 3 and 14 and decline after 1940, while lowland counties fall early, then rise from about 7 in 1910 to around 27 by the mid 1960s.County means of cotton production by type
The figure shows four line charts labelled Missouri, Arkansas, Mississippi, and Louisiana, with the vertical axis showing bales of cotton in thousands and the horizontal axis showing years from 1860 to about 1965. In Missouri, upland counties remain near 0 throughout, while lowland counties rise from about 5 in 1860 to a peak near 78 around 1960 before declining slightly. In Arkansas, upland counties increase from about 4 to around 11 by 1930 and then fall to about 2 by the mid 1960s, while lowland counties rise steadily from about 8 to nearly 75 by 1950 before easing to about 58. In Mississippi, upland counties fluctuate between about 6 and 15 and decline after 1930, while lowland counties increase from about 26 to around 75 by 1930 and remain above 70 thereafter. In Louisiana, upland counties vary between about 3 and 14 and decline after 1940, while lowland counties fall early, then rise from about 7 in 1910 to around 27 by the mid 1960s.County means of cotton production by type
In Figure 7, the path of development of cotton in lowland counties provides key insight into the role this crop played in the devastation of the 1927 flood. In all four states, the period from 1920 to 1930 saw rapid increases in lowland cotton production. This decade was characterized by volatile cotton prices, declining by half from a peak in the early 1920s. Even with declining prices and a large flood, production increased during the decade because more, high productivity land well-suited for cotton production became available due to the success of levee and drainage districts. The 1930 agricultural census was too early to see the impact from federal investment due to the 1928 Flood Control Act, and it is likely that cotton production had increased dramatically over 1920 levels by the time the flood hit in 1927. This increase put worker populations and higher field values at risk, increasing the human and economic toll of the flood relative to earlier floods.
We build on these insights using a statistical approach to better control for confounding variation. We regress outcome variables on a set of county and state-by-year fixed effects and then interact an indicator variable for lowland counties with each year:
where is the outcome for county i in state s in year t and Lowlandi is an indicator variable for a county designated as lowland. The model includes a county fixed effect, , and a state by year fixed effect, . The coefficient is the relative premium for the outcome variable in lowland counties, which varies by year.
The coefficients on the lowland–year interactions are plotted in Figure 8. Although there is no precise pre-district and post-district periods, all but one levee district in the lowland counties had formed prior to 1910 (that one district was St. John Levee District in Missouri formed in 1912). We therefore set the level of the lowland coefficients relative to 1900 to show the performance of lowland counties in an approximation of pre-district and post-district periods.
The figure contains four panels arranged in a two by two grid, each showing point estimates with vertical error bars by decade from 1860 to 1969, with a vertical dashed line at 1900 and a horizontal dashed line at 0. The top left panel shows proportion improved, near 0 before 1900 and rising steadily after, from about 0.03 in 1910 to about 0.45 by 1969. The top right panel shows farmland value per acre, negative or near 0 before 1900, then positive after, increasing from about 170 in 1910 to about 650 by 1969. The bottom left panel shows corn production in thousands of bushels, negative before 1900, then positive after, peaking near 650 around 1940 and declining to about 60 by 1969. The bottom right panel shows cotton production in thousands of bales, near 0 before 1900, then rising after, reaching about 50 by the late 1950s and easing slightly by 1969.Relative development of lowland counties before and after 1900
The figure contains four panels arranged in a two by two grid, each showing point estimates with vertical error bars by decade from 1860 to 1969, with a vertical dashed line at 1900 and a horizontal dashed line at 0. The top left panel shows proportion improved, near 0 before 1900 and rising steadily after, from about 0.03 in 1910 to about 0.45 by 1969. The top right panel shows farmland value per acre, negative or near 0 before 1900, then positive after, increasing from about 170 in 1910 to about 650 by 1969. The bottom left panel shows corn production in thousands of bushels, negative before 1900, then positive after, peaking near 650 around 1940 and declining to about 60 by 1969. The bottom right panel shows cotton production in thousands of bales, near 0 before 1900, then rising after, reaching about 50 by the late 1950s and easing slightly by 1969.Relative development of lowland counties before and after 1900
The top left panel of Figure 8 shows the relative percentage of lowland counties in improved acres. The rapid rise in improved land after 1900 results in lowland counties having almost a 50-percentage point higher share of land developed relative to upland counties in the same state. This is consistent with the intensity of agricultural production we see in lowland counties today, which account for around one-quarter of US cotton production and two-thirds of rice production. Pre-1900 the agricultural development of lowland counties was similar to upland, as indicated by coefficients statistically indistinguishable from zero at the 5% level.
By 1920, land values in lowland counties were over $500 per acre higher (in 2020 dollars) than upland counties (top right panel). These land value premiums did not exist in the 1870–1890 period. Although the 1880 and 1890 coefficients are significant at the 5% level, they are negative, indicating relative lowland values may have been rising even prior to 1900.
The bottom panels show corn and cotton production in lowland relative to upland counties. Relative corn production (left panel) increases from 1890 to 1900 and continues to increase through 1940. Relative cotton production (right panel) increases slightly through 1920 before increasing rapidly between 1920 and 1930.
Taken together, these results demonstrate that land values rose, and agricultural production increased dramatically, in the 50 years preceding the 1927 flood. We attribute these outcomes to the increasing effectiveness of levee and drainage districts over this period. In the following section, we explain why there is little reason to believe these increases were related to federal investment.
After the flood
The Great Flood of 1927 was a signal event of the twentieth century. Barry (1997) recounts the flood, its human and economic toll, and its legacy of extensive federal involvement. The prelude to federal involvement in the basin began in earnest in 1875, when Louisiana Congressman Randall Lee Gibson led a successful effort to create a House standing committee on Mississippi levees. Gibson and US Senator L.Q.C. Lamar of Mississippi used the committee to create the Mississippi River Commission (MRC) in 1881, which served as the main coordinator of federal River policy, under the Secretary of War, until 1928. Catastrophic floods on the Mississippi were common prior to 1927, with new federal intervention typically following each flood.
Initial appropriations for the Mississippi River Commission in 1891 totaled only $1 million. In keeping with traditional views of levees as protecting private landowners, initial appropriations to the MRC prohibited the use of funds in levee construction or repairs that primarily benefited private interests. Floods in 1890 led to new appropriations ($3.5 million) and allowed the MRC to fund levee construction and repairs directly. Floods in 1897 and 1903 led to calls from local interests for more federal funding. Consecutive floods in 1912 and 1913 led landowners to launch a public campaign urging additional federal intervention. These calls were primarily focused on obtaining additional funding, not the federal government’s expertise or coordinating ability, and we attribute the need for federal funding to intra-district coordination problems in raising sufficient funding to protect private land interests.
More focus on inter-district coordination occurred soon after the MRC aided the Southeast Arkansas Levee District (originally formed as the Chicot Levee District) in closing the Cypress Creek Gap, just south of the confluence with the Arkansas River, in 1921. In 1922, a record setting flood below the Gap was attributed by downstream landowners to the closure. Following this flooding, $60 million was appropriated directly for levee funding and after this work was complete, the MRC declared in its 1926 annual report that the levee system “is now in condition to prevent the destructive effects of floods.”
Following the 1927 flood, which Herbert Hoover declared “the greatest peace-time calamity in the history of the country,” Congress quickly acted. The 1928 Flood Control Act created the Mississippi River Valley Project (MRVP) and appropriated $325 million for its work improving the levee system from Cape Girardeau, Missouri to the Gulf of Mexico. This Act largely moved federal coordination from the MRC to the MRVP. As important as the dramatic increase in federal appropriations was Section “Levees on the Mississippi before the Civil War” of the Act, which laid out the public good case for federal involvement in protecting the property values of local interests:
That it is hereby declared to be the sense of Congress that the principle of local contribution toward the cost of flood-control work, which has been incorporated in all previous national legislation on the subject, is sound, as recognizing the special interest of the local population in its own protection, and as a means of preventing inordinate requests for unjustified items of work having no material national interest. As a full compliance with this principle in view of the great expenditure estimated at approximately $292,000,000, heretofore made by the local interests in the alluvial valley of the Mississippi River for protection against the floods of that river; in view of the extent of national concern in the control of these floods in the interests of national prosperity, the flow of interstate commerce, and the movement of the United States mails; and, in view of the gigantic scale of the project, involving flood waters of a volume and flowing from a drainage area largely outside the States most affected, and far exceeding those of any other river in the United States, no local contribution to the project herein adopted is required.
With the proverbial levee of federal funding breeched, local interests began securing additional federal funds. The 1938 Flood Control Act appropriated $375 million in new projects on major Mississippi tributaries including the Ohio, Tennessee, Missouri, Upper Missouri, Arkansas, White, Red, St. Francis, and Yazoo Rivers. The Act further removed local contribution requirements for reservoirs. The 1944 Flood Control Act authorized up to 150 additional flood control projects at an expense of $750 million. Complex economic, political, and hydrologic forces had coincided in 1927 to spur a generational shift in federal levee funding. It is telling that large changes in federal policy were not seen after earlier flooding events. The year 1927 was seminal not only because there was a great flood, but also because local levee and drainage districts had been successful enough to put considerable value at risk.
Supplementary material
The supplementary material for this article can be found online.
Notes
All waters of the Mississippi drain into the Gulf of Mexico, but the details become complicated at the confluence of the Red River with the Mississippi, roughly 80 road miles above Baton Rouge. At that point, the combined flow of the Mississippi and Red Rivers is split between the Mississippi and its natural distributary, the Atchafalaya, by the Old River Control Structure, operated by the Army Corps of Engineers. Thirty percent of the flow is diverted into the Atchafalaya, from where it drains into the Gulf of Mexico west of New Orleans. The rest passes to the Gulf through Baton Rouge and New Orleans.
For a systematic treatment of the “discovery” of North America through European exploration between 1492 and 1801, see Bernard DeVoto’s (1952)Course of Empire. The evolution of European understanding of the Mississippi River drainage is central to his telling of history.
See Harrison and Kollmorgen (1947) for the history of state inheritance of land and efforts to put it into private hands in Arkansas.
Generally, the three measures overlap in the Boeuf, Delta, Cache, and St Francis regions. While the Cache region is not part of the floodplain definition, it is included in Otto’s definition. The Grand Prairie region is only partially within the floodplain and not included in Otto’s bottomland county definition, but retains enough other similarities for our consideration. Despite being mostly excluded from the 1887 floodplain map, both the Grand Prairie and Cache regions partially flooded in the first but not second flood of 1927 (not pictured). The northern portion of the Delta region is notable for not flooding in 1927, potentially due to levee investment. Our lowland definition is also largely coincident with the “Southern Mississippi Alluvium,” or MLRA (Major Land Resource Area) 131A, as categorized by the Natural Resource Conservation Service of USDA. See NRCS-USDA (2022). MLRA 131A includes the deltaic plain south of the confluence of the Red River with the Mississippi, which we exclude from our study region. Online Appendix Table A1 shows the counties in each state included in our sample and their primary levee district.
Confusion is difficult to avoid, but note: the Mississippi River Delta referred to here is different from what is often called the Mississippi Delta — the home of Robert Johnson, the blues, and Highway 61. The Mississippi River Delta (the Atchafalaya and Deltaic and Chenier Plains in Figure 2), which we exclude from our study area, lies downriver and borders the Gulf of Mexico. It is evocatively referred to as a “place with no edge” by Mandelman (2020). The Mississippi Delta (Delta in Figure 2) is the farther upriver lens-shaped region of farmland lying entirely in the state of Mississippi and is included in our study area.
Each county is assigned to the primary organization within its boundaries. Many counties are marked as belonging to a levee organization that only covers a portion of the county’s area. District boundary and dates were compiled by the authors (see Table A1).
