Such transfer restrictions may have been well justified at one point in time. Many government programs that granted property rights to poor indigenous communities were passed by political coalitions of economist Bruce Yandle’s proverbial “bootleggers and Baptists,” and the “Baptists” may have correctly identified a need to protect newly created property owners from the “bootleggers,” who in this case were land‐hungry white settlers. In the long run, however, such transfer restrictions come at a heavy price: nontransferable property is noncollateralizable property, and a lack of collateralization is one of the biggest impediments to wealth creation for the global poor.
To investigate the long‐run consequences of limits on transfer rights, we leverage a natural experiment that resulted from the policy of land allotment on American Indian reservations in the early 20th century. This policy generated a patchwork of land titles on reservations, with some Native households owning their land in nontransferable “allotted trust” while other immediately adjacent Native households owned land under full fee‐simple property rights. To compare land use on plots with different land titles, we map the universe of historical land allotments from the Bureau of Land Management to the Public Land Survey System grid and from there to high‐resolution satellite data from the National Wall‐to‐Wall Anthropogenic Land Use Trends Database (NWALT).
Indian allotment began in 1887 and ended with the Indian Reorganization Act (IRA) of 1934. In the intervening half century, the federal government allotted millions of acres of previously tribe‐owned land to individual Native American households, starting with the 1887 Dawes Act and accelerating after the 1906 Burke Act. All land rights were first issued in nontransferable allotted trust and could then—after a period of trusteeship—be selectively converted into fee simple by a reservation’s local Bureau of Indian Affairs (BIA) agent. Had this policy run its full course, all reservations would have eventually been allotted and all allottees would have eventually obtained their land rights in fee simple. However, the 1934 IRA put an abrupt stop to it, ending all allotment for good and freezing all allotted‐trust plots into trusteeship in perpetuity. This created a patchwork of land tenures on reservations that has persisted to the present day.
Endogeneity problems in the comparison of allotted‐trust and fee‐simple lands on reservations arise from the fact that allotments were selectively converted into fee simple. There was the potential for both selection based on land characteristics (plots with certain characteristics getting converted at a higher rate) and selection based on the unobserved characteristics of the original allottees (allottees with certain characteristics having their plots converted at a higher rate). As a first step toward addressing this, we compare plots only inside neighborhoods that are small enough to make all observable differences in land characteristics disappear.
We then pursue a strategy that generates exogenous variation in whether an allotted plot was converted to fee‐simple title before the process ended in 1934 by utilizing an allotment’s issuance year, which—within a reservation—is fully explained by the birth years of the original allottees because all men and older children received their allotments simultaneously when a reservation was first allotted. Only young children and the unborn received their allotments in later waves and were thus less likely to see their allotments converted to fee simple before the program’s abrupt end in 1934. We also use the identities of the exogenously rotating BIA allotting agents on each reservation at different times. Our core finding from this strategy is that fee‐simple property rights increase land use by around 0.5 standard deviations.
The NWALT satellite data exist in five decadal waves (1974, 1982, 1992, 2002, and 2012), allowing us to pursue another approach to statistical identification; namely, we can include individual‐plot fixed effects to absorb all unobserved differences in invariant characteristics (of both the land and the original allottees). We find that the gap between fee‐simple and allotted‐trust land grew from 1974 to 2012. We also use the panel to separately explore land development and agricultural cultivation. Land development is of special interest because it is much less likely than agricultural cultivation to be endogenous to any 1887–1934 differences and because it best captures the process of structural transformation away from agriculture and into manufacturing, tourism, and services that has occurred on reservations since 1974. We find that there was no difference at all in land development in 1974, implying that the entire difference in 2012 is driven by subsequent divergence. This contrasts sharply with agricultural use, for which over 80 percent of the 2012 difference was already present in 1974.
While our focus is on comparing allotted‐trust land with fee‐simple land, we also extend the analysis to include tribally owned land, which still constitutes the majority of all reservation lands today. In the cross section, land utilization on tribally owned plots is a lot more similar to allotted‐trust plots than to fee‐simple plots. In the panel, however, development on tribally owned land increased over time relative to allotted‐trust land and at the same rate of divergence as fee‐simple land, suggesting a considerably more positive dynamic land utilization trajectory on tribally owned land than on allotted‐trust land.
There are two primary channels through which transfer limitations affect land use on reservations: one is obvious, while the other is a less obvious, indirect, and longer‐run channel. The obvious channel is the well‐known “de Soto effect”; this means noncollateralizable property does not give its owner the access to credit needed to make investments. This is a major problem on reservation trust land. The indirect and less obvious channel is that in the long run, transfer limitations create highly fractionated interests over the same plot, which create large transaction costs. We find evidence that both channels are driving the negative effect of allotted‐trust property rights on land utilization.
Finally, we develop a back‐of‐the‐envelope estimate of the negative impact of transfer restrictions on land values. To do so, we combine the estimated effect of fee‐simple title on land utilization with an estimate of the effect of land utilization on land values using county assessor data. This exercise suggests that fee‐simple title adds between $973 and $4,765 in value to an acre of land, or between $156,000 and $762,000 to a 160‐acre plot.
Our results indicate that conversion to fee simple would generate the biggest economic efficiency gains on allotted‐trust plots. But the alternative, which entails returning allotted trust to tribal control, may better safeguard the territorial integrity of tribes’ land base. This creates tradeoffs. Our view is that both the conversion to fee simple and the return to tribal ownership would be preferable to keeping land in allotted trust, but the choice of which (if either) path to pursue belongs to the individual tribes.
This research brief is based on Christian Dippel, Dustin Frye, and Bryan Leonard, “Property Rights without Transfer Rights: A Study of Indian Land Allotment,” NBER Working Paper no. 27479, July 2020, doi.org/10.3386/w27479.