The Costs of Sovereignty:

Law and Finance on American Indian Reservations∗

Rachel Wellhausen

University of Texas at Austin

rwellhausen@utexas.edu

29 January 2015

Draft. Comments welcome.

Abstract

In 1953, Congress supplanted the tribal civil law on some American Indian reservations

with the civil law of the US state in which they are located. While this compromised some

tribes’ sovereignty, Congress’s action reduced external financial actors’ uncertainty over the

enfocement of contracts on some reservations. I argue that this change in jurisdiction has a

causal impact on the costs of capital today, and, specifically, that it manifests at the level of the

individual. Using data on 20,000 home loans to tribal members guaranteed by a US Housing

and Urban Development program (1995-2013), I show that loan recipients governed by US state

civil law pay consistently lower interest rates. But this advantage may not be insurmountable:

quantitative and qualitative extensions suggest that neither the presence nor the magnitude of

the effect offsets at least some tribes’ prioritization of sovereignty.

∗My sincere thanks go to the Indian leaders, government officials, and industry actors who contributed to theresearch, particularly Chairman Kevin Leecy and the Bois Forte Band of Chippewa; Chairman Roger Trudell andthe Santee Sioux Nation; Townsend Hyatt, David Johnson, Gavin Clarkson, and participants at the 33rd NativeAmerican Finance Officers Association conference (20-21 April 2015); the Smithsonian’s National Museum of theAmerican Indian; and the Harry S. Truman Library. I also thank David Singer, Pat McDonald, Bat Sparrow, MikeFindley, Laura Evans, Harrison Wagner, Amy Liu, Noel Johnston, Mike Tomz, Raymond Hicks, and participants atthe University of Chicago and the 2015 International Studies Association. Thanks to Gus Van Harten and DominicParker for sharing data. Cici McNamara, Amy Leung, Yefan Zhou, and Connor Smith provided excellent researchassistance. Funding was provided by the University of Texas at Austin (IRB 2015-02-0074).

1

1 Introduction

Consider the similarities between the 326 federal American Indian reservations and very small

developing countries.1 A tribal government’s sovereignty – over issues like membership criteria,

taxation, and regulation – reflects that of a typical country in many ways.2 Yet as in a developing

country, a tribal government must negotiate the constraints on sovereignty that emerge “in the face

of political, economic, and social disadvantage.”3 Of the approximately 2 million American Indians

living on or near a reservation in 2010, some 23 percent earned incomes below the poverty line,

and many tribal members encountered local unemployment rates of 50 percent or more.4 Given

such impoverishment, tribal governments prioritize bringing capital to where capital is scarce. In

so doing, tribal governments look to hands-tying institutions similar to those on which developing

countries rely. A partial waiver of sovereign immunity is de rigueur when tribal governments issue

debt or enter into contracts with non-tribal (“foreign”) firms. Put simply, tribal governments

regularly trade sovereignty for credibility. Yet as in developing countries, tribal governments are

not interested in foregoing sovereignty for credibility.

Tribes have built up domestic legal, economic, and social institutions over many genera-

tions.5 Such institutions can generate a cultural currency that encapsulates the beliefs and priorities

of a people. Unsurprisingly, tribes have been loath to wholly forego these tangible manifestations

of sovereignty. US policymakers fought against tribal sovereignty during the “termination era” of

1For ease of exposition, I refer to all federal Indian lands as “reservations,” whether they are technically calledpueblos (e.g., in New Mexico), rancherias (e.g., in California), missions, villages, communities, etc. The Bureau ofIndian Affairs counts 326 reservations, although they acknowledge this as an approximate number as definitions vary.Technically, the federal government calls this “Indian Country.” Not every of the 567 recognized US tribes have areservation, but some reservations are home to multiple tribes. The federal government holds approximately 56.2million acres in trust for tribes.

2Chief Justice Marshall noted that the “Indian nations had always been considered as distinct, independentpolitical communities, retaining their original natural rights, as the undisputed possessors of the soil, from timeimmemorial.” Worcester v Georgia (1832). I use the term sovereignty, or the authority to govern a particularterritory, because this is the terminology used by actors in the American Indian community. The sovereignty ofindigenous groups is, however, generally at the behest of the overarching state.

3Evans 2014: 279. See Evans (2014) in general for a call for political scientists to study American Indians andother indigenous groups.

42013 American Indian Population and Labor Force Report, US Department of the Interior (16 January 2014).In the 2010 Census, 5.2 million people living throughout the United States identified as American Indian/AlaskanNative (1.7 percent of the US population).

5Barsh 1986, Cornell and Kalt 2000, Wilkins and Stark 2011, Evans 2011, Cornell 2015. The 1934 IndianReorganization Act (Wheeler-Howard Act) allowed tribes to organize, adopt tribal constitutions, and incorporate andfunction as units of local government. Tribes organized under the Act also qualify for federal benefits. American Indianlaw establishes that tribes can develop court systems because of their retained sovereignty (and not, say, because ofan act of Congress) (Trentadue 1987/1988: 4-6). Cornell and Kalt (2000) link variation in tribal constitutions toreservation unemployment and income levels.

2

US-Indian relations in the 1950s. The top-down zeitgeist was “to look forward...to the time when

we will completely liberate the Indians and remove them from any suspicion of being wards of

the Government.”6 In particular, Congress saw a problem with “lawlessness” on reservations and

came up with a fix: Public Law 83-280 (PL280), which was imposed without tribal referenda and

in the face of significant Indian protest. PL280 replaced criminal jurisdiction on reservations with

the criminal jurisdiction of the US state in which the reservation is located. And, without explicit

justification, Congress replaced tribal civil law with state law as well.

In fact, PL280 set up a quasi-natural experiment in which some tribes were “treated” with

the civil law of their US state while others retained the preexisting tribal jurisdiction.7 This is

because Congress mandated state law only in those states with constitutions that did not require

amendment to do so. Some other states went on to adopt similar legislation (again without tribal

referenda), but I demonstrate below that this can be thought of as an as-if random “roll-out” phase

to the experiment.

When PL280 and roll-out laws replaced some tribes’ civil law with that of the state, po-

tential economic effects were not part of the discourse. Yet scholars have shown that externally

credible institutions, particularly concerning contracting and property rights, spur development.8

From this point of view, strong civil legal institutions are a means to an end – a mechanism for

government actors to establish the enforcement of rule of law, private property, and principles of

commercial exchange. Imposed civil law is deeper than a waiver of sovereign immunity, because

it keeps the soveriegn from intervening in either its own or in private transactions outside of an

externally interpretable framework of rule of law.

In this paper, I leverage the PL280 quasi-natural experiment to demonstrate that tribes

treated with the civil law of the state indeed enjoy economic benefits, because external lenders

expect easier contract enforcement. This strong research design moves beyond literature that has

6Statement of Hon. Walt Horan, State Legal Jurisdiction in Indian Country, Hearings before the Subcommitteeon Indian Affairs on HR 459, HR 3235, and HR 3624, 82nd Congress, 28 February 1952: 3. Horan was in factconcerned when laws were imposed on tribes without tribal referenda.

7“Untreated” reservations are actually subject to joint tribal and federal jurisdiction. However, civil legal mattersare effectively handled by tribal rather than federal courts on reservations, as federal prosecutors typically do notprioritize civil cases (LaFontaine 1974, Wilkins 1998). Thus, this paper uses simplifying language, comparing stateor tribal civil law.

8E.g., North and Weingast 1989; North 1989; Haber 1996; Keefer and Knack 1997; La Porta et al 1997; La Porta etal 1998; Rajan and Zingales 1998; La Porta et al 2000; Acemoglu, Johnson, and Robinson 2001; Johnson, McMillan,and Woodruff 2002; Easterly and Levine 2003; Rodrik, Subramanian, and Trebbi 2004; Levine 2004; Nunn 2007.

3

had to rely on clever identification strategies to break the endogeneity between institutions and

economic development.9 What is more, this paper uses contemporary, micro-level data to measure

the effect of variation in civil legal institutions on economic outcomes at the level of the individual.

I find that the benefits of state civil law accrue to individual tribal members who get cheaper

access to capital, in the form of better interest rates on home mortgages. Unambiguous access to

state (rather than tribal) courts lowers transaction costs for non-Indian lenders and provides them

heightened levels of certainty that their contracts will be enforced, resulting in lower interest rates.

Thus, externally understood civil legal institutions not only improve economic outcomes but do

so in a way that, by improving relatively impoverished borrowers’ access to capital, can mitigate

inequality.

Whether improved individual access to capital compensates for “termination era” policy

that has compromised tribal sovereignty is, of course, an important normative question. In line

with global histories of imperialism and colonialism, it took coercion to replace (some) tribal

legal institutions. This coerced replacement meant the exchange of some elements of sovereignty

for a more hierarchical relationship under another’s laws.10 PL280 reinforces the hierarchical

relationship between some tribes as “domestic dependent nations” in the United States.11 Other

US entities – like the Northern Marianas Islands or Puerto Rico – also navigate economic and

political hierarchies with the US while maintaining some sovereign rights. States around the world

mirror the US experience with indigenous and post-colonial territories. In short, the contemporary

tradeoff between sovereignty and development under study here, and the coercion behind it, is not

unique. In fact, “race to the bottom” arguments about relations between developed and developing

countries revolve around the same coercive tradeoff.12 What this paper does is help to quantify

the upside of coercion.

As it turns out, at least with regard to borrowers under the popular Section 184 home

loan program, the magnitude of the effect of state civil law may be thought of as small – on

the order of tens to hundreds of dollars per person per year. While even a small effect has real

economic consequences for impoverished borrowers, that individual-level effect may not be sufficient

9E.g., Acemoglu, Johnson, and Robinson 2001.10Lake 2009.11This is Chief Justice John Marshall’s famous characterization of tribes in Cherokee Nation v Georgia (1831).12E.g. Strange 1996.

4

to offset the tribal-level prioritization of sovereignty. I shed light on this tradeoff with qualitative

data collected from two tribes that petitioned to withdraw from PL280. I demonstrate that these

tribes in principle understood that there could be economic costs to retrocession. But, elected

officials in both tribes articulate a vision that rejects incremental sacrifices of particular aspects

of soveriegnty, in addition to pursuing practical steps intended to overcome the potential negative

individual economic consequences of tribal sovereignty. This evidence suggests that, despite the

robust finding of advantages to PL280, the coercive imposition of external law remains at minimum

a controversial tribal development strategy.

The paper proceeds as follows. After introducing the literature on law and finance, I

explain how PL280 and the as-if random treatment of tribes constitute a quasi-natural experiment

on civil legal institutions. I then describe the 20,000 Indian mortgages (1995-2013) under study

and present difference-in-means, regression, and matching results showing the advantages of state

civil law. Robustness analyses and extensions address default rates; home loan limits; and the

extent to which other federal programs mitigate PL280’s effects.13 The conclusion considers what

this quantification of the costs of sovereignty means for American Indian tribes and capital-poor

developing countries more generally.

2 Law and Finance

The study of political economy has taught us that the exercise of sovereignty comes with costs.

In international political economy, we know that expropriation deters foreign direct investment,14

sovereign default can deter sovereign lending,15 and, in general, governments’ domestic policies can

have negative consequences for external financial market actors under a variety of circumstances.16

Governments face the problem of credibly committing to protect external actors’ property rights

while still acting as sovereigns. Internationally, governments have turned to institutions like the

World Trade Organization or investment treaties to tie their hands, increasing market actors’ confi-

dence that the government will not unduly interfere in its own or in private transactions. Nonethe-

13In the Appendix, I consider the influence of other property rights issues on reservations (i.e. “checkerboarding”)on results around PL280.

14E.g. Li 2009, Allee and Peinhardt 2011, Jensen and Johnston 2011, Wellhausen 2015.15E.g. Tomz 2007, Gray 2013.16E.g. Milner 1997, Bernhard and Leblang 1999, Guisinger and Singer 2010, Bodea and Hicks 2015.

5

less, conducting economic transactions in a foreign country exposes an actor to that country’s

domestic law, such that investors are sensitive to domestic institutions. Even in an increasingly

globalized world, the credibility of domestic institutions matters for capital access that, in turn,

matters for development outcomes.17

A variety of scholars have measured the effects of domestic institutions on development

outcomes.18 In a review, Levine links broader considerations of institutional effects back to capital

access, concluding that institutions that have created deeper financial systems ease external financ-

ing constraints facing firms.19 Along these lines, in a survey of firms in post-communist countries,

Johnson, McMillan, and Woodruff find that weak property rights discourage firms from reinvesting

their profits.20 More broadly, Li and Resnick argue that institutions providing strong property

rights protections have a significant, positive effect on access to foreign investment.21 La Porta

and collaborators argue that institutions in the common law tradition, in particular, are associ-

ated with better economic outcomes.22 Jha further connects variation in historical institutions to

modern economic outcomes, inter-ethnic trust, and peace.23 Most famously, Acemoglu, Johnson,

and Robinson distinguish the effects of exploitative institutions from property rights-enhancing

institutions by using colonial settler mortality rates as an instrument.24 Easterly and Levine pit

the AJR explanation against endowment and policy explanations to find that “tropics, germs, and

crops” indeed have acted through institutions to influence development.25

Berkowitz, Pistor, and Richard argue that it is not institutions per se but the method of

institutional transplantation that has a significant effect on economic development.26 They argue

that for legal institutions to be effective, citizens must be invested in the law’s operation and

enforcement and that legal practitioners in society must be able to improve the law in response

17Ahlquist 2006, Farrell and Newman 2014.18For commentary on the development of new comparative economics around these issues, see Djankov et al (2003).

See Glaeser et al (2004) for criticism of statistical techniques in the literature.19Levine 2004.20Johnson, McMillan, and Woodruff 2002. Others have leveraged identification strategies focused on the effects

of specific domestic institutions in particular country contexts. For example, land rights in Ghana (Besley 1995),insurance in India (Townsend 1995), and securities market reform in Brazil (Haber 1996).

21Li and Resnick 2003.22La Porta et al 1997; La Porta et al 1998; Claeser and Shleifer 2002; Djankov et al 2006; Djankov et al 2007.23Jha 2013.24Acemoglu, Johnson, and Robinson 2001.25Easterly and Levine 2002. Greif (1994), in contrast, looks to culture to help explain “why societies fail to adopt

the institutional structure of more economically successful ones.”26Berkowitz, Pistor, and Richard 2003.

6

to citizen demand.27 Otherwise, countries are subject to the “transplant effect”: law unfamiliar

to the domestic population, not adapted to local conditions, and imposed via coercion will be

ineffective. Ineffective institutions, in turn, undermine economic development. This argument

reflects work by Rodrik that emphasizes the importance of local experimentation with institutions

for good development outcomes.28 However, a positive causal effect of institutions on development

and a negative “transplant effect” could co-exist. The effect of transplanted law on the endogenous

development of economic activity may be complicated, especially in the short run.29 But even if law

is imposed unilaterally, on a different nation, and without adaptation to the domestic population,

it can improve outsiders’ understanding of transacting in a society. In fact, these stark conditions

may be best suited to improving a society’s access to external sources of finance that can, in turn,

improve development outcomes. Compromising sovereignty in order to match the institutions of a

hierarchical power can bring increased access to the resources of that hierarchical power, even if it

has ambiguous effects on domestic actors.

A small literature has used American Indian tribes as a laboratory to consider the re-

lationship between property rights institutions, development outcomes, and possible “transplant

effects.”30 Some of this work focuses on complications around land tenure on American Indian

reservations. The Dawes Act of 1887 led to a “checkerboarding” problem, whereby tribal trust,

individual trust, and fee simple (private) land are extremely intermixed within reservations.31 An-

derson and Lueck demonstrate that reservation agricultural productivity decreases as more land is

held in trust relative to private land, because private land can uniquely be used for collateral.32

Broader than the “checkerboarding” problem is the issue of the credibility of tribal institu-

tions. In 1984, a Presidential Commission on Indian Reservation Economies reported that “busi-

nesses see uncertainty in situations where law is subordinate to the whims of tribal councils...There

is a fear that tribal courts will not protect the property rights of non-Indians by according them

27Berkowitz, Pistor, and Richard 2003: 166-167.28E.g. Rodrik 2000, 2007.29See de Soto (2000) for consideration of property rights protections and domestic entrepreneurship.30See Anderson 1992, 1995; Cornell and Kalt 2000; Miller 2012. Anderson, Benson, and Flanagan (2006) compare

institutional effects in US and Canadian reservations. For other studies of the political economy of indigenouspopulations, see work including Gitter and Reagan (2002) and Kuhn and Sweetman (2002).

31The process of allotment that caused checkerboarding ended with the 1934 Indian Reorganization Act. See Akee(2009) for a study of the effects of checkerboarding on housing investment on one California reservation. I considerthe relationship between “checkerboarding” and the argument presented here in the Appendix.

32Anderson and Lueck 1992. For more on the importance of land rights to tribal development, today and in history,see Anderson, Benson, and Flanagan, eds. (2006).

7

due process of law or protecting individual non-Indian civil rights. Uncertainty increases risk and

risk increases the cost of doing business on Indian reservations.”33 In 2012, a US senator posited,

“the patchwork of legal rules and regulations has discouraged business activity on reservations,

especially by financial institutions. The problem is only made worse by what some have described

as a casual approach to the rule of law.”34 These US officials speculate that external financial

actors are reluctant to use a given tribe’s “domestic” civil law to govern their relationships with

the tribe. They also raise a different but related issue: financial actors may not be confident that

the contracts they make under US law will be enforced by tribes. I focus on this issue of enforce-

ability and demonstrate that, in fact, external actors’ expectations of enforceability vary due to

“termination era” federal government policy and Public Law 280.

3 Public Law 280 as a Quasi-Natural Experiment

American Indian-US government relations took a new turn around the 1950s. This was the “ter-

mination era,” when many in the US government advocated eliminating the reservation system,

removing federal jurisdiction over reservations, and making tribal members subject to general state

law.35 In 1953, the sentiment in Congress was to assimilate tribes into the general population, so

that the Bureau of Indian Affairs could “begin going out of business in an orderly manner.”36

PL280 was a manifestation of this: it gave some states the unilateral right to supersede the stan-

dard combined tribal and federal legal jurisdiction and assert state jurisdiction on Indian lands.

Note that, when it comes to civil matters, what is otherwise joint federal and tribal jurisdiction on

Indian lands is effectively tribal jurisdiction, as federal prosecutors are known not to prioritize civil

matters.37 For simplicity, I will characterize civil law jurisdiction on Indian lands as either tribal

33Report and Recommendations to the President of the United States (30 November 1984): 36. See Haddock(1994) and Haddock and Miller (2006) for more on the external credibility problem facing Native American tribesand their leaders.

34Richard C. Shelby, Alabama. United States. Cong. Senate. Committee on Banking, Housing, and Urban Affairs.Opportunities and Challenges for Economic Development in Indian Country. 10 Nov. 2011. 112th Congress. 1stsess. Washington: U.S. Government Printing Office, 2012. 2-12. Print.

35This was the subject of House Concurrent Resolution 108, 83rd Congress, 1953. The termination era policyeroded in the 1960s and was decried by President Nixon. It was not formally revoked until 1988 (25 USC Sec.2501(f)): “Congress repudiates...any policy of unilateral termination of Federal relations with any Indian Nation.”See Anderson (2012). During the termination era, Congress terminated the special status of 1,400,000 acres of Indianland and 13,263 Indians. Cowger 1999: 100.

36Representative Harrison. 99 Cong. Rec. 9263 (1953). Quoted in Jimenez (1998): 1959-1661.37LaFontaine 1974, Wilkins 1998.

8

or state.38

Congress’s “foremost concern” in passing PL280 was “lawlessness on the reservations and

the accompanying threat to Anglos living nearby.”39 Thus, PL280 was primarily about states taking

over criminal jurisdiction on Indian lands. The National Congress of American Indians (NCAI),

the most prominent national Indian lobbying organization, protested that tribes had law-and-order

institutions already and, moreover, that tribal members would not understand state law and would

not be treated fairly in state courts.40 Most of all, tribes were angered that Congress imposed

PL280 on tribes without tribal referenda on the matter.41 Indeed, when he signed PL280 into

law, President Eisenhower said he had “grave doubts as to the wisdom of certain provisions,”

particularly the imposition without tribal referenda.42

In a study of PL280, Goldberg-Ambrose posits that, “most likely, civil jurisdiction was an

afterthought...added because it comported with the pro-assimilationist drift of federal policy and

because it was convenient and cheap.”43 Indeed, while the extension of criminal jurisdiction was

thoroughly discussed in hearings on PL280 and related bills from 1950 to 1953, the reason for civil

jurisdiction was never clearly articulated.44 A 1975 US circuit court ruling confirms this; the judge

held that there is little in the legislative history to convey a “congressional rationale” for extending

state jurisdiction over civil matters on reservations.45 Neither the NCAI nor individual tribes

articulated specific concerns about civil jurisdiction when opposing the legislation, but neither

did they articulate support.46 This paper focuses on the effects of the imposition of state civil

38It is possible that, if a creditor fails to win in tribal court, the creditor could obtain review of the tribal court’saction in federal court, under the 1968 Indian Civil Rights Act; but again, the US federal government has historicallybeen reluctant to interfere in tribal civil legal affairs (LaFontaine 1974: 31, Wilkins 1998). For a review of case lawon civil jurisdiction on Indian reservations, see Gover (1980).

39Ackerman, David. Background Report on Public Law 280, Congressional Research Service: 541. Quoted inAnderson (2012). See also Jimenez (1998): 1659 and Goldberg-Ambrose (1997): 48.

40Cowger 1999: 99-111. Additionally, PL280 did not come with requirements that states increase their budgets forlaw enforcement in tribal areas, nor did it come with any additional federal appropriations.

41E.g. Report to Accompany HR 1063, Senate Report No. 699, 83rd Congress (Senate), 29 July 1953: 7.42“The failure to include in these provisions a requirement of full consultation in order to ascertain the wishes and

desires of the Indians and of final Federal approval, was unfortunate.” Statement by the President upon Signing BillRelating to State Jurisdiction over Cases Arising on Indian Reservations, 165 Pub. Papers 564, 564-565 (15 August1953).

43Goldberg-Ambrose 1997: 50.44Interestingly, the presence of state criminal legal jurisdiction does not correlate perfectly with state civil juris-

diction. In the quantitative analyses below, several covariates (ruralness, local property values, etc.) control forvariation in the propensity for crime on Indian lands.

45Santa Rosa Band of Mission Indians v Kings County (9th Circuit 1975). Quoted in Jimenez (1998): 1660.46According to Congressional records and the Smithsonian National Museum of the American Indian archives of

the National Congress of American Indians, 1945-1960. Cowger claims that some California tribes were interestedin state civil jurisdiction (Cowger 1999: 103). There is a mention in the Congressional record that Hoopa Valley

9

law, an “afterthought” to the main purpose behind PL280, via the mechanism that state civil law

substantially increases external financial actors’ confidence in the enforceability of their contracts

on Indian lands.

“Termination era” policy was intended, ultimately, to reshape federal government relations

with all American Indians. However, PL280 focused on tribes in six “mandatory” states: Alaska,

California, Minnesota, Nebraska, Oregon, and Wisconsin (see Table 1A). PL280 was limited to

these states, because these states did not need amendments to their state constitutions in order

to change jurisdiction on tribal land. In 1881, a Supreme Court ruling limited state authority in

“Indian Country” (i.e., Indian lands) with regard to non-Indian crimes but maintained the status

quo of tribal and federal authority.47 Prior constitutions and Alaska’s constitution, written in the

midst of “termination era” policy, allow the federal government to cede jurisdiction to the state.

[Table 1A about here.]

Since the 1881 Supreme Court ruling, the US government had required as a perquisite

to statehood that a state constitution include a provision ensuring federal jurisdiction on Indian

lands.48 Such provisions cemented the 1881 ruling and effectively removed the possibility of new

states bringing similar cases against the US government. Ironically, this perquisite for statehood

stymied the US government’s ability to unilaterally cede its jurisdiction in pursuit of “termination”

in the 1950s. For post-1881 states that include Indian lands, an act of Congress cannot itself change

jurisdiction. Rather, state constitutions must be amended. Thus, Congress’s action in 1953 left

states that received statehood after 1881 “untreated.”49 To the extent that the timing of statehood

and thus the content of state constitutions is plausibly random, we can think of PL280 as an

experiment. (To the extent that this plausibility is compromised, as considered below, results are

consistent when treating the data as merely observational.)

A few additional factors lead me to identify PL280 as a “quasi-experiment” (at best). First,

Indians in California “have adopted resolutions favoring the proposal to confer civil and criminal jurisdiction on theState,” and that “Representatives of other groups have also indicated their approval.” (Testimony by Hon. JohnR. Murdcok, Chairman Committee on Interior and Insular Affairs, House of Representatives. 82nd Congress, 2ndSession, Report No. 2161, 11 June 1952: 4) My best attempts to use American Indian archival resources to identifythese other groups have failed. Suffice it to say, Indian support for transferring jurisdiction to the state appearslimited to some sentiment in California.

47United States v. McBratney, 104 US 621, 624 (1881). “Indian Country” is the accepted legal term for what Irefer to as Indian lands or reservations.

48Wilkins 2002, Anderson and Parker 2008.49PL280 permitted jurisdictional change in the future should “untreated” states make the necessary amendments

to their constitutions. Anderson (2012): 938.

10

Nevada – home to Indian lands and a state as of 1864 – refused to be a mandatory PL280 state.

This is because PL280 did not include any subsidies to help states extend legal institutions to

Indian lands, nor could states tax reservation land to raise revenues.50 Thus, Nevada (and not

its tribes) initially opted out of the treatment for reasons plausibly exogenous to external actors’

certainty over civil law enforcement on Indian lands.

Second, Congress had already passed a PL280-like law three years earlier in 1950 that

applied New York state civil law to New York tribes (and a law in 1948 that extended New York

state criminal law).51 Interestingly, President Truman allowed New York jurisdiction to be changed

but vetoed an analogous law in 1949 that applied to the Navajo and Hopi tribes in (post-1881 states)

Arizona, Utah, and New Mexico. Why this difference in treatment? In his veto, Truman offers the

explanation that New York laws applied to “comparatively small groups, the members of which

through long association with neighboring whites, had reached the stage where they were prepared

to and wished to be governed by State and local law. The Navajo and Hopi...are, indeed, the

Indians who are probably least prepared for such a major change.”52 Despite Truman’s words,

records show that New York Indian leaders protested against and were “disturbed by the step-by-

step intrusion of the State Government” forced upon them, consistent with later Indian protest

against PL280.53 But were the Navajo and Hopi able to endogenously determine their treatment –

that is, did Truman respond to their protests? Elites in Arizona, Utah, and New Mexico were likely

reluctant to have Indians serve on state juries, as would occur if state law held on reservations; to

spend state resources extending legal and policing institutions to the reservations; or, frankly, to

forego decades of local prejudice that would accompany equal treatment for Indians under state

law. Recall, for example, that American Indians could only just vote in Arizona and New Mexico

following court rulings in 1948; Utah did not repeal anti-Indian voting laws until 1957. In other

5083rd Congress, 1st Session, Senate, Calendar No. 691, Report No. 699 (to accompany HR1063), July 29(legislative day July 27), 1953: 6. Additionally, PL280 does not allow states to legislate concerning property held intrust by the United States or concerning federally guaranteed hunting and fishing rights.

51Additionally, in response to several heinous crimes and at California’s request, Congress had already extendedcivil and criminal jurisdiction to California over one reservation: the Agua Caliente Reservation (as of October 1949).

52Ibid. Truman also wrote that the change of jurisdiction “is in conflict with one of the fundamental principles ofIndian law accepted by our Nation, namely, the principle of respect for tribal self-determination in matters of localgovernment.” No such claim was made on behalf of the New York tribes.

53Streator, George. “State Laws Urged to Benefit Indians.” New York Times: 22 May 1949 (ProQuest HistoricalNewspapers). US counsel in favor of the bill said that it allowed for “court reform that will make the Indians equalto all other people in the enforcement of contracts. He said that he could not answer the chief contention of someIndians that the measure was intended to destroy the reservation.” Ibid.

11

words, there is reason to believe that the “non-treatment” of the Navajo and Hopi (as well as

the “non-treatment” of the many other tribes in Arizona, Utah, and New Mexico) is not due to

actions by the Navajo and Hopi themselves. In fact, because Congress added the topic of legal

jurisdiction to what was otherwise a development funding bill, the Bureau of the Budget and others

in the Cabinet advised Truman to veto.54 Clearly, the administration did not see a change in legal

jurisdiction as furthering the goals of funds for economic development, and thus not a corollary to

access to capital as is being tested in this paper.

Third, a number of states chose to amend their constitutions to accede to PL280 and impose

state civil law on Indian lands after 1953. I think of this as a “roll-out” phase after the initial PL280

treatment. Between 1953 and 1968, Florida, Iowa, and Nevada asserted civil jurisdiction over their

reservations, again without tribal referenda on the issue. In the 1968 Indian Civil Rights Act,

Congress approved a provision requiring a state to obtain tribal consent before adopting PL280.

Put differently, since 1968 the decision on acceding to the treatment has been endogenous to the

will of the potentially treated. No tribe has consented to be “treated” by state jurisdiction.55

Nonetheless, several states have since gotten around the 1968 provision for tribal consent to PL280

by passing state-equivalent laws: Colorado, Connecticut, Maine, Massachusetts, Rhode Island,

Texas, and Utah. Table 1B summarizes these “roll-out” extensions of the treatment.

[Table 1B about here.]

Summing up, from a research design standpoint, Nevada’s initial rejection of PL280; New

York’s treatment and Arizona, Utah, and New Mexico’s non-treatment; and the roll-out phase

would ideally be plausibly exogenous to the state-level treatment. To examine this quantitatively,

I construct a series of hazard models predicting a state’s entry into PL280 or its roll-out including a

variety of covariates that we may worry pollutes the exogeneity of entry with regard to the ultimate

dependent variable, tribal member well-being. These hazard models predict time until entry to

the treatment for the 33 US states that are host to Indian tribes.56 Altogether, seventeen states

54One advisor specified that the issue of jurisdiction was “unrelated, and unnecessary, to the accomplishment ofthe rehabilitation program.” Letter to Mr. William J. Hopkins, The White House (17 October 1949), from RogerW. Jones, Assistant Director, Legislative Reference, Executive Office of the President, Bureau of the Budget. HarryS. Truman Library, White House Records Office Files, Box 57, Folder “October 17, 1949.”

55Two Indian tribes have been forced to accede to PL280 since 1968. For the Mashantucket Pequot Nation inConnecticut, Congressional recognition legislation provided that PL280 would apply notwithstanding the 1968 law(PL98-134). The Texas Ysleta Del Sur Pueblo were treated as if they had consented to PL280 (PL100-89). In theanalysis below, results are robust to either dropping or coding as state civil law the very few observations affectedby these impositions.

56These are the states that are host to today’s 567 federally recognized Indian tribes. While federal recognition of

12

enter into the treatment, from 1950 to 1987. Table 1C summarizes the states that do not enter the

treatment.57

[Table 1C about here.]

Consistent state-level political and economic covariates are available beginning in 1948,

which allows an appropriate point of left-censoring for a treatment that begins in 1950.58 Compa-

rable data extend through 2008; results are robust to right-censoring at that point or in 1988, after

the last state entered the treatment. I test whether covariates that capture a state’s development

level and economic health are relevant, since PL280 was unfunded and thus made demands on state

resources. I also test whether a state’s law enforcement capacity influences entry to the treatment,

since PL280 was first and foremost about criminal law. Helpfully, in bivariate Cox proportional-

hazards models none of the following are significant predictors of entry to the treatment: total state

revenue per capita, total state debt outstanding per capita, unemployment compensation expenditure

per capita, public safety spending per capita, or total state prison population.59

Two relevant covariates are significant predictors in bivariate models. The first, Indian

population per capita (in state), scales American Indian and Alaskan Native population by state

population, as a measure of density and as a proxy for American Indian political presence in state

politics.60 The hazard ratio is significant and extremely close to zero, suggesting that states with

greater concentrations of American Indians are less likely to enter the treatment.61 This makes

intuitive sense, as states with many tribes are also those western states that achieved statehood

only after 1881. The second, (state income per capita (logged) has a sizeable, positive effect on

entry, with a large hazard ratio in the bivariate analysis.62 This suggests that states home to

some tribes was in fact terminated during the “termination era,” other tribes remain such that those states do notfall out of the sample. Indigenous Hawaiian persons have historically been counted under a different federal regimethan that for American Indians and Alaskan Natives. Hawaii is thus not home to Indian lands and drops out of theanalysis.

57Some US states, such as North Carolina, Wisconsin, and Louisiana, recognize tribes that the US government doesnot federally recognize. These tribes are not considered, as their states are not potential recipients of the treatmentand thus are not part of the analysis.

58Unfortunately, state-level covariates are unavailable for Alaska pre-treatment or at the point of treatment, as itwas simultaneous with statehood. Thus, it drops out of the analyses.

59Data from the US Census State Government Finances and the US Statistical Abstract, various years. Otherinsignificant predictors include: total federal intergovernmental revenue per capita, total expenditure per capita, totalinterest on debt per capita, total corrections expenditure per capita, total judicial expenditure per capita, and policeprotection expenditure per capita. Consistent state-level unemployment rates begin in 1976; this is not a significantpredictor (Bureau of Labor Statistics).

60Evans 2011. Data is available only from the decennial Census.61Hazard Ratio: 1.36*10∧-33, p=0.09, for 32 subjects, 16 failures, and 1249 at-risk observations.62Hazard Ratio: 60.92, p=0.01, for 32 subjects, 16 failures, and 1281 at-risk observations. Data from US Census

13

more wealth are more likely to enter the treatment. Given these variables’ significance in simple

bivariate analyses, I include them as necessary controls for PL280 and its roll-out to reach “quasi-

experimental” status.63

What about possible unobserved factors that might be non-randomly distributed across

treated and untreated states? Here, I bring qualitative evidence to bear.64 To the extent that roll-

out states have a systematic motivation for opting-in, I expect them to continue to be motivated

as Congress and states were initially – to address “lawlessness” on reservations. Those states

opting in to the treatment may have more “lawless” reservations, which in turn would plausibly be

associated with worse (rather than better) developmental outcomes. The bias introduced by the

“lawlessness” motive would make it more difficult to find positive effects of state civil law on tribal

development. Nonetheless, if PL280 clearly solved “lawlessness” problems, it could be that states

would opt in to the treatment even if their problems were not particularly severe. On this point,

we have evidence that PL280 did not solve “lawlessness” problems in mandatory states. Recall

that PL280 was an unfunded mandate. Goldberg-Ambrose documents that PL280 states did not

increase appropriations for law enforcement on reservations, nor did federal money supplement state

budgets.65 As a result, the withdrawal of federal Bureau of Indian Affairs police and the absence

of compensating regular and adequate state police patrols left legal vacuums on reservations. For

example, when federal law enforcement withdrew from the Winnebago reservation in Nebraska, the

reservation was left with no law enforcement personnel whatsoever, as the state did not allocate

any officers to it. Even by the 1970s, the reservation had only poor police coverage.66 In fact, the

Santee Sioux Nation and the Bois Forte Band of Chippewa retroceded from PL280 exactly because

state criminal law enforcement was ineffective.67

State Government Finances. There are no regional measures of GDP per capita prior to 1963 (Bureau of EconomicAnalysis). Thus, the more comprehensive income per capita is the most useful measure of relative wealth anddevelopment in a state. The correlation between state GDP per capita and income per capita from 1963 to 2008 is0.8.

63Both variables are significant when included in a multivariate model. However, both fail a proportional-hazardstest. When time-varying covariates are added to the model, no variable achieves statistical significance. Nonetheless,out of an abundance of caution I include both variables in analyses below to best account for possibly influentialobservables. See replication files.

64Dunning 2012: 326.65Only one Department of the Interior grant, in 1955, assisted law enforcement development on one Oregon

reservation. Goldberg-Ambrose 1997: 57.66Goldberg-Ambrose 1997: 57, 103. In 1957, the Nebraska governor demanded that the Bureau of Indian Affairs

take back jurisdiction. As that was not possible under PL280, Nebraska eventually set up special deputy sheriffs forreservation areas (Goldberg-Ambrose 1997: 103).

67Interview, Chairman Trudell, Santee Sioux Nation, June 2015. Under PL280, only a handful of state trooperswere responsible for an area larger than the state of Rhode Island, including the Bois Forte reservation. Interview,

14

One final complication is introduced by the fact that the Santee Sioux, Bois Forte, and

36 other tribes have petitioned at different times to get themselves excluded from PL280 and its

roll-out (see Table 1D).68 These tribes, unlike other tribes in their states, maintain the traditional

tribal civil jurisdiction on their reservations and thus must be considered in the research design.

Exemption or retrocession from PL280 and roll-out legislation is difficult, often taking decades of

effort. Before enactment, tribes had to petition Congress or their state legislature to be exempted;

after enactment, tribes have had to follow a complex process in order to show that they have strong

law-and-order institutions and, in particular, strong tribal police and criminal justice institutions.69

In other words, a few tribes with very strong institutions are the ones that have opted out from

the state civil law treatment. Given the hypothesis that the treatment affects development by

reducing external actors’ uncertainty over the enforcement of civil law, the demonstrated strength

of retroceeded tribes’ institutions is a challenging source of bias. Lastly, federal legislation does

allow tribes across the country to opt-in to state civil law. However, no tribe has chosen to do so.

In the framework of this paper, tribes have opted out of but not in to a treatment that is expected

to cause better development outcomes.

[Table 1D about here.]

4 Effects of Variation in Civil Law: American Indian Housing

A group of scholars argue that PL280 and its roll-out have hurt reservation Indians.70 In contrast,

Anderson and Parker expect “greater prosperity” on reservations with state civil law, because

external investors are “willing to invest more in physical capital on the reservation, to provide

credit to reservation Indians, and...to engage in more spot-market transactions on reservations.”71

In support of this contention, Anderson and Parker’s regression analyses using the 81 most populous

Chairman Leecy, Bois Forte Band, June 2015. Note that if PL280 in fact increases crime on Indian lands, we areunlikely to find that it has positive developmental effects.

68The Yamaka Nation in Washington state has been approved to retrocede from PL280 in 2016, but that tribe hasbeen subject only to state criminal and not state civil jurisdiction.

69To speed up the process, the Santee Sioux adopted Nebraska’s criminal code lest the state legislature find faultwith their own formulations. It still took them from the 1980s to 2006 to officially retrocede from PL280. Interview,June 2015. In another example, the Menominee petitioned but failed to be exempted from PL280 in 1953 and didnot retrocede until 1976. For a call to ease the constraints on retrocession, see Anderson (2012).

70E.g., Goldberg-Ambrose 1997, Tweeten 2000, Goldberg and Champagne 2005.71“Even if individual tribal members could credibly contract out adjudication on a case-by-case basis, it would be

costly for each member to go through this process prior to engaging in each contract. And this approach would failto provide security over implicit contracts such as torts.” Anderson and Parker 2008: 648.

15

Indian reservations show that state jurisdiction is associated with an increase in Indian income per

capita by at least 30 percent between 1969 and 1999.72 Anderson and Parker navigate the poor

historical data available at the reservation level and produce an analysis analogous to cross-country

work on law and finance. However, I established above that state income per capita and American

Indians per capita may be predictors of entry into PL280 and its roll-out; we might therefore have

concerns about whether income effects have been mistakenly attributed to PL280 and its roll-out.

In addition to highlighting these potential omitted variable issues, the analysis here ad-

dresses problems of ecological inference. Cross-country or cross-reservation level analyses of law

and finance do not necessarily shed light on individual well-being. It could be that the benefits of

credible legal institutions accrue to elites, via mechanisms like corporate growth and government

borrowing.73 Broadly, literature on law and finance has not yet been able to establish whether

credible civil legal institutions are equity-promoting. To do so, we need a micro-level outcome

measure at the level of the individual.

I look at the effects of the quasi-natural experiment in state civil law on individual tribal

members’ access to loans. Usefully, several qualitative studies of finance on Indian lands provide

existence proofs that lenders pay attention to the effects of law on individual tribal members’

credit-worthiness. Trosper argued that, for tribes with sovereign civil law, “uncertainty to banks

accustomed to state law may raise the cost of capital to Indians.”74 In a study of tribal courts

in North Dakota, an untreated state, loan officers admitted “great reluctance” in making loans to

Indians on reservations, with several saying they thought of these as unsecured “character loans”

since collateral on the reservation is difficult to recover.75 The president of the North Dakota

Collectors Association said, “Certainly many millions of dollars in debts are going uncollected,

and Indians are not getting the credit they might otherwise.”76 To forestall arguments that the

treatment of individual tribal borrowers may have roots in racism, I focus on variation among

72Anderson and Parker 2008: 653-654. Controls include reservation size, checkerboarding issues, natural resources,education levels, ruralness, regional economic growth, gaming, and measures of initial conditions. They examinetribes with population over 1000.

73For information on Indian corporations and tribal borrowing, see Safty (2012), Clarkson (2007, 2008), Bernardi-Boyle (2001/2002). This paper focuses on individual access to capital and not tribal access for or through vehicles likegaming operations. For a consideration of the mixed developmental effects of tribal gaming, see Evans and Topoleski(2002) as well as Hansen and Skopek (2011) and, for Canada’s First Nations, Belanger (2011).

74Trosper 1978: 503.75Trentdadue 1987/1988: 45.76Ibid: 46. Mudd (1972) makes analogous arguments for Montana.

16

tribal borrowers and posit that these qualitative accounts do not hold the same power in “treated”

states.77

Specifically, I test whether variation in state civil law systematically affects the cost of

capital for tribal members by focusing on the price of home loans. Access to capital to support

home ownership is a key means of capturing the abstract notion of economic development at the

individual level.78 Evidence of improved access to mortgage capital for tribal members governed

by state civil law would demonstrate the treatment’s equity-promoting effects, reducing at least in

some small way the tremendous disparaties between tribal members and the general US population.

Obtaining housing on reservations incurs particular challenges.79 Much reservation land

is held in trust by the US government, for the benefit of the tribes, but that land cannot be

mortgaged.80 Since privately held (or “fee simple”) land exists in a “checkerboard” patchwork

across many reservations, assessing whether land can or cannot be mortgaged is an onerous process.

Thus, home lending has historically been underprovided for Indians and reservation residents in

particular.81 The Section 184 Indian Home Loan Guarantee Program, established by Congress

in 1992, intends to help with these problems. Section 184 loans are available for enrolled tribal

members, and they offer very favorable terms.82 Loans can be used for the purchase or refinancing

of single family, primary residence homes. Anecdotally, lenders, borrowers, and tribal leaders have

been happy with the program.83

To external lenders, the most attractive part of the Section 184 program is that the Office

of Loan Guarantee within HUD’s Office of Native American Programs guarantees these loans 100

percent. This means, in case of non-payment, the lender must follow civil law to establish that the

77I also note that the Equal Credit Opportunity Act (1982) requires lenders to extend credit to all credit-worthycustomers “without regard to race, sex, religion, color, age, or national origin.”

78In the United States at large, mortgage debt increased from 15 percent of household assets in 1949 to 41 percentin 2001 (Green and Wachter 2005: 93). Home ownership is of prime importance in the context of American culture(e.g. Krivo and Kaufman 2004).

79Cyree, Harvey, and Melton (2004) find that Indian applicants are rejected from government housing guaranteeprograms at higher rates than non-Indian applicants, and that poor economic conditions on reservations lead lendersto hold on-reservation borrowers to even higher standards.

80If land is held for an individual Indian, the BIA must give approval before a lien can be placed on the property.81One consequence of this is that mobile homes are common on reservations, since they can physically be siezed

and thus better act as collateral.82These include, as of 2014: 2.25 percent down payment on loans over $50,000; 1.25 percent on loans under $50,000.

A one-time 1.5 percent up front guarantee fee paid at closing, which can be financed into the loan. (Loans with aloan to value of 78 percent or greater are subject to an annual 0.15 percent mortgage insurance premium.) Theycannot be used for Adjustable Rate Mortgages (ARMs).

83In 2012, the program was so popular that HUD was forced to abruptly suspend the program for five weeks, dueto a shortage of funding. Letter to tribal leaders, Office of Public and Indian Housing, HUD.

17

loan is in default and foreclose on the home. Then the lender will be repaid 100 percent by the

US government. These facts give us precision in interpreting the cause of variation in loan terms.

Suppose that other determinants of lending risks are controlled for, experimentally or otherwise.

If the treatment has an effect on interest rates, then we can extrapolate that enforcement costs

are at issue. I hypothesize that external lenders expect to more cheaply enforce contracts under

state law, because they understand state law, have experience with state law, and can therefore

undertake enforcement proceedings in a routine manner – whether or not the lender believes a

given tribe’s law is of high or low quality in itself. This expectation will be reflected in the interest

rates contracted with individual borrowers. The magnitude of any effect is limited by the presence

of the government guarantee; in the absence of the guarantee, varied lender expectations over the

costs of enforcement (and thus interest differentials) would be higher.84

Section 184 loans began in 1995 but grew significantly by the late 2000s. As of 2014, nearly

$4 billion in loans have been made to individuals in 37 states.85 I gained access to data on 23,567

loans from 1995 through the end of 2013.86 Some data has been withheld under FOIA Exemption

6: loan guarantee fee dates, fee amounts, and certain types of loans.87 The dataset includes a

maximum of 20,181 loans after redactions. What remain are loans made with regard to “fee simple”

land, or land for which the individual can and does hold title and control of the property. Interest

rates range from 0 (6 observations) to 9.63 percent, with a mean of 5.00 percent. Sixty percent of

loans were made to tribal members in untreated states and 30 percent to members in treated states.

For comparison, 62 percent (1.22 million) of American Indians living on or near reservations live in

untreated states. The most loans have been made to residents of Oklahoma (untreated), followed

by Alaska (treated), Arizona (untreated), and California (treated). See Appendix Figures A1, A2,

and A3 for the distribution of addresses on and off reservations in PL280 mandatory, roll-out, and

84Of course, such guarantees are rare (at minimum) in inter-state relations, implying that this analysis will seriouslyunderestimate the potential magnitude of externally imposed law on individual outcomes in a more traditional inter-state context.

85An individual may receive a loan if she is a tribal member, even if she lives outside the reservation (or in anotherstate).

86FOIA Control No.: 14-FI-HQ-01570.87Redacted types include: allocated/individual trust, tribal trust, assignment, leasehold, and restricted fee. Section

184 gets around the problem of trust land, because the borrower only leases the land from the tribe for 50 years,meaning that the home and the leasehold interest (and not the land) are mortgaged. With the assumption thatborrowers on trust lands (backed by the US government) and tribally owned lands (backed by the tribe as a whole)would receive better interest rates than individually owned land, these redactions make it more difficult to find arewarding relationship between civil law status and mortgage lending.

18

untreated states. The data has time-series but not panel properties.

4.1 Analysis

If PL280 and its roll-out were a pure experiment, a difference-in-means test across treated and

untreated states would be sufficient to show effects. Analyzing the data navely like this, borrowers

in treated states enjoy 0.03 percent lower interest rates, but the estimate is imprecise with a p-value

of 0.13.88 However, this analysis does not account for state income per capita or American Indian

population density, which are predictors (albeit not wholly robust) of entry into the treatment.

Nor does it account for the fact that interest rates vary over time and space in non-random ways

that could be misinterpreted as being a product of the treatment.

Table 2 presents regression analyses that consider the data as a quasi-experiment as well as

analyses that consider the data without relying on an underlying experimental design. Models 1

and 3 analyze the data as quasi-experimental and control for the two state-level predictors of entry

to the treatment, state income per capita (logged) and Indian population per capita (in state).

I add three additional covariates that, while unrelated to entry to the treatment, could vary in

nonsystematic ways across home loan interest rate observations in treated and untreated states.

The first is the distance to metro area, calculated as the number of miles between the borrower’s

home address and the closest border of one of the 381 US statistical metropolitain areas.89 It could

be that borrowers closer to urban areas share characteristics that allow them to access capital more

cheaply, such as access to a more competitive lending environment. The next are the appropriate

(15- or 30-year) locally prevailing mortgage interest rates in the week each loan was made, using

data from the Freddie Mac Primary Mortgage Market Survey (PMMS).90 The PMMS data varies

by region, which helps to control for the effects that non-Indian, local economic health has on rates

offered to tribal borrowers. It also controls for the general availability of credit in the region.91 I

use year dummies to account for national-level time trends. Finally, I exclude from analysis the 34

88If the data is analyzed year-by-year, interest rates are significantly lower (p less than 0.1) as hypothesized in 12of 18 years.

89Statistical metropolitain areas, defined by the OMB, have at least one urban area of at least 50,000 people andextends over land that has strong commuting ties to that core. In 15349 observations (76 percent) the distance is 0as the borrower’s address is within the statistical unit.

90The 30-year interest rate covariate is turned on for 30-year mortgages and any other loans in the dataset thatare made for more than 15 years.

91The country is divided into five regions: North Central, Northeast, Southeast, Southwest, and West.

19

borrowers in the data living on the reservations that have retroceded from or been exempted from

state civil law jurisdiction (see again Table 1D).92

Models 2 and 4 remove the assumption that the data has experimental qualities and include

a variety of individual-level covariates that could affect a borrower’s creditworthiness.93 The models

control for the loan amount (logged), loan term (in months), whether he or she has a coborrower,

and whether the loan is for refinance as opposed to home purchase. These covariates also capture

relevant information about tribal-level phenomena, such as the effects that any rents from gaming

or natural resources may have had on the value of the local housing stock. The data also include the

originating lender (260 banks) and servicing lender (141 banks) for each loan, for which I generate

dummies.94 These dummies help to capture the effects that variation in particular lenders’ practices

might have on interest rates.

[Table 2 about here.]

Results show that borrowers living in states with state civil law (as opposed to tribal

civil law) are benefitting from 0.04 percent (Model 1) or 0.02 percent (Model 2) improvements in

interest rates. Again, recall that all loans in the data are 100 percent guaranteed, meaning that

the effect of law on the price of capital is particularly notable. The effect suggests variation in

transaction costs around invoking the guarantee, which are higher when the lender needs to deal

with a “foreign” tribal legal system in the process. Now, while Section 184 borrowers need to be an

enrolled member of a tribe, the home they are mortgaging may be located on or off of a reservation.

It can be confusing to lenders as to whether and to what extent off-reservation Indian borrowers

have access to tribal courts; for example, at least some Section 184 lenders in Oklahoma use the

same paperwork for mortgages on- or off-reservation, although these borrowers may have access to

different legal systems.95 Thus, Models 1 and 2 are appropriate if conservative tests. However, if

the mechanism at play is really the added cost of enforcing contracts on reservations with tribal law,

then it should be borrowers located on reservations that ultimately account for differential interest

rates (so long as every Section 184 lender does not see them as interchangeable with off-reservation

92There are also 236 borrowers who are members of state-recognized but not federally recognized tribes, and thusare in states that are outside the relevant population of states that can be treated. These are excluded as well.

93By controlling for individual-level determinants of creditworthiness, we do not need to assume that good appli-cants are randomly distributed across the treatment groups.

94The most common servicing lender is Wells Fargo (17 percent of observations), which has a robust NativeAmerican practice.

95Personal communication, large Section 184 lender in Oklahoma, April 2015.

20

borrowers). Thus, Models 3 and 4 include a covariate indicating whether the home address is on

Indian lands (home on reservation) and an interaction term with state law.96 The direct effect

of state law is now unclear (Models 3 and 4). Unsurprisingly, home on reservation results in a

direct, significant increase in the interest rate, whether the data is considered as experimental

(Model 3) or observational (Model 4). But, as expected, the interaction term is negative and

significant: borrowers mortgaging homes on reservations receive significantly lower interest rates

under conditions of state civil law. The magnitude of the coefficient of interest is larger, at 0.06

(Model 3) or 0.04 (Model 4).

Other covariates in Table 2 behave predictably. State-level indicators for income and Amer-

ican Indian population have significant negative effects on interest rates, and borrowers closer to

metro areas also benefit from lower rates. Higher regional interest rates track with higher bor-

rower interest rates. For individual-level controls, borrowers taking out bigger loans and those that

are refinancing receive significantly lower rates. Coefficients on the loan term (positive) and the

presence of a coborrower (negative) are signed as expected but are not significant.

Table 3 provides a robustness check that treats the data as observational and uses nearest-

neighbor matching. Both Model 5 and Model 6 match on the full battery of covariates: loan

amount, loan term, co-borrower, refinancing, originating lender, servicing lender, state income per

capita, Indian population per capita, regional 15-year interest rate, regional 30-year interest rate,

year, and home on reservation. Model 5 matches exactly on year ; Model 6 matches exactly on

year and home on reservation. A minimum of 4 matches are required, and standard errors are

robust. In each case, results are consistent, and the magnitude of the average treatment effect

on the treated is large: treated borrowers are benefitting from 0.09 percent lower interest rates

(Models 5 and 6).

[Table 3 about here.]

In sum, the effect size in these analyses varies between a 0.02 percent and 0.09 percent

difference on a home loan interest rate. In the data, the mean home loan amount is $169,000.97

96Based on US Census TIGER/Line Shapefile, 2012, Current American Indian/Alaska Native/Native HawaiianAreas (AIANNH) National Shapefile.

97The mean is lowest in 1998 ($97,900) and highest in 2013 ($180,200). Year dummies account for any aggregatetrends at play. Because less worthy borrowers may be increasingly selected out over time, we cannot say for certainthat the Section 184 program is reducing inequality among tribal members even if it does reduce inequality betweenIndians and non-Indians.

21

For this amount, over the course of a 15-year mortgage, those governed by tribal civil law would

pay between $268 and $1207 more. Over the course of a 30-year mortgage, those governed by

tribal civil law would pay between $559 and $2525 more. Are these effect sizes large or small? As

suggested by the mean home loan amount, Section 184 loans are regularly used for modest homes.

That state civil law can cheapen these borrowers’ access to capital demonstrates a meaningful cost

to tribal civil law. And, given that these effect sizes exist even within a population of loans that

are all 100 percent guaranteed by the US government, we can imagine that disparities would be

larger in non-guaranteed loans. Nonetheless, these costs add up to (only) tens or hundreds of

dollars a year. Whether those costs are sufficient to motivate further restrictions in soveriegnty is

an open question. In the next section, I consider extensions to these results and draw on qualitative

evidence to provide contemporary tribal perspectives on the merits of trading off sovereignty for

capital.

4.2 Robustness and Extensions

Here, I exploit additional data on Section 184 loan defaults; Section 184 loan limits; and the extent

to which other federal programs mitigate PL280’s effects.98 I also draw on qualitative data to shed

light on how some contemporary tribes manage trade-offs between sovereignty and housing access.

The Section 184 program has dealt with 665 defaults (foreclosures) from 1996 through

2013.99 More of these took place in untreated states (409) than in treated states (256). This is

consistent with the expectation that borrowers default more often when the law makes lenders

more uncertain about (and thus increases costs of) obtaining restitution. However, the proportion

of defaults in untreated states is roughly equal to that in treated states (3 v. 4 percent). This

more compelling statistic suggests that Section 184 lenders are “correctly” pricing mortgages in

untreated states so as to compensate for increased risks of default.

That lenders prefer Indian borrowers under state civil law suggests that lenders should be

willing to lend more to Indian borrowers under state civil law and that the US government should

be willing to guarantee those larger loans. HUD issues upper limits for the size of Section 184

loans for counties and/or census areas where the program is available. I use the 2013 upper limits

98In the Appendix I consider the influence of other property rights issues on reservations (i.e. “checkerboarding”).99FOIA Control No.: 15-FI-HQ-1317. Unfortunately, the data do not show whether a foreclosure took place at an

on- or off-reservation address.

22

to test whether tribal borrowers in PL280 and roll-out states have access to systematically more

guaranteed loan funds. In Table 4, the dependent variables are guaranteed loan limits for a single

(Model 7), duplex (Model 8), triplex (Model 9), or fourplex (Model 10) home in 1881 areas in

29 states. Models control for state income per capita (logged) and Indian population per capita

(in state) as possible determinants of entry into the treatment. Consistent with the hypothesis,

guaranteed loan limits for each type of home are significantly higher in areas governed by state law.

[Table 4 about here.]

While this paper demonstrates that variation in civil law affects individual Indians’ access

to capital, the notion that Indians altogether have poor access to capital, whether borrowing indi-

vidually, as businesses, or as tribes, is not new.100 One major federal government program provides

special financial, technical, and training assistance to financial institutions that serve individual

Indians: the US Department of the Treasury Native American Community Development Financial

Institutions (CDFI) Assistance Program (NACA). From 2004 through 2014, NACA provided $92

million in funding to 135 tribal-related CDFIs in 26 states.101 While we might expect that NACA

funding differentially supports those borrowers with poorer access to capital (in untreated states),

there is not a robust, significant difference in NACA funding across the treatment.102 This sug-

gests there may be room for improvement in NACA funding allocations, if one goal is to reduce

institutionally generated disparities in interest rates.

While they may not be aware of systematic differences in capital access across Indian lands,

many Indian leaders have “heard it before” that property rights protections matter for develop-

ment.103 Two tribes that retroceded from PL280, the Santee Sioux Nation in Nebraska (2006

retrocession) and the Bois Forte Band of Chippewa in Minnesota (1975 retrocession), eschewed the

state civil law “treatment.” Extrapolating from the analysis here, retrocession suggests that their

members are likely to have exacerbated trouble accessing capital for purposes like home ownership.

While it is difficult to say whether these tribal members’ problems are in fact more severe than

100For example, the Senate Committee on Indian Affairs has held hearings to address capital in “Indian Country”(e.g., 17 June 2015).101http://www.cdfifund.gov/awardees/db/index.asp data downloaded 23 April 2015.102Untreated states do receive more NACA funding (logged) than treated states in a simple t-test (p=0.10). However,

that result does not hold up in a regression analysis where NACA funding (logged) is modeled as a function of statelaw as well as state income per capita (logged) and/or Indian population per capita (in state). See replication files.Jurisdictional status is not a part of CDFI’s explicit considerations in locating its NACA funds. Interview, USDepartment of the Treasury staff member, April 2015.103Interview, Chairman Kevin Leecy, Bois Forte Band of Chippewa, June 2015.

23

others’, both tribes recognize a pressing shortage of housing for tribal members. Interestingly, they

have taken steps to rectify this outside of traditional external financial markets. The Bois Forte

Band, which itself has borrowed from other tribes (as opposed to non-Indian financial markets),

opened a tribal credit union in 2013 with long-term plans to provide subsidized home financing

through it.104 The Santee Sioux government issues partial waivers of sovereign immunity to borrow

from non-Indian banks, uses those funds to build homes on tribal trust land, and then leases those

homes to tribal members at rates based on their income.105 In this way, the tribe absorbs any

systematic premium in market interest rates and offers its members what it considers fair rates.

These two examples demonstrate that tribes can find creative ways to lower the cost of borrowing

for tribal members, even under conditions of tribal civil law.

With the Bois Forte and Santee Sioux experiences in mind, this paper does not demonstrate

that the imposition of state, rather than tribal, civil law is the best or the only way to improve

individual Indians’ access to housing or to capital more broadly. Maybe tactics that work around

external lenders’ reluctance to engage with tribal law can provide stable, quasi-market solutions.

However, this paper does suggest that opting into state civil law is one clear means to lower tribal

members’ borrowing costs. What is more, it is relatively cheap, given that it requires adopting

existing institutions rather than developing new legal codes, social programs, or complex lending

structures. Just how cheap it is, however, depends on one’s valuation of sovereignty.106 Given

American Indian tribes’ troubled history in maintaining their status as sovereign nations, it is not

surprising that tribal leaders may be willing to forego potential individual-level economic benefits

of reduced sovereignty in favor of nation-building goals. Explaining that “we don’t have ordinary

economics on this reservation,” one elected leader in the Santee Sioux tribe put it clearly: “We

aren’t here for individual well-being; we are here for a large group of people.”107 That tribe’s

central worry is that by “slicing” sovereignty, a tribe opens itself up to challenges against its

sovereign rights to undertake other activities – like gaming. In this tribe’s evaluation, the social

returns to sovereignty at the tribal level – and, indeed, its tribal-level economic returns – outweigh

104Interview, June 2015.105Interview, June 2015. Like that of the Bois Forte, the Santee Sioux government has also occasionally borrowed

from other tribes.106It also depends on whether opting in to some institutions will weaken others, as has historically been the case

with PL280 and criminal law enforcement.107Interview, Tribal Council member, Santee Sioux Nation, June 2015.

24

any temptation to opt in to state civil law. At minimum, this paper sparks future research into

the extent to which “institutionally disadvantaged” sovereign nations recognize that disadvantage

and the strategies they use to (partially) overcome disadvantage, whether by surrendering (some)

sovereignty or otherwise.

5 Conclusion

Roughly equivalent to very small developing countries in the world, American Indian tribes are

looking to balance economic development and their own cultures and traditions. Tribes navigate the

space allowed to them, to both benefit from and push back against state and federal governments.108

As particularly impoverished nations, many tribes view the need for non-Indian capital as crucial

to development. For “standard” capital-poor countries, literature in law and finance demonstrates

that externally interpretable, stable property rights improve access to capital. This paper uses

a quasi-natural experiment to show that the same holds for Indian tribes: borrowers that are

governed by the law of the US state enjoy lower interest rates than borrowers governed by their

own tribe’s civil law. The effect sizes found may be small, but they manifest even in a dataset of

loans that are 100 percent guaranteed by the US government. Importantly, this paper shows effects

not at an aggregate level, but at the level of the individual borrower. By improving individuals’

access to capital, US state civil law may also improve equity outcomes for American Indians.

Adopting state civil law would be a straightforward way for a given tribe to avoid the costs of

maintaining its own civil legal sovereignty. But the quasi-natural experiment that brought state civil

law to some tribes was the “termination era” policy PL280, intending to stamp out “lawlessness”

but part of a bigger push to eliminate the reservation system. Clearly, if adopting state civil law

leads to “termination” and losing all sovereignty, capital access is far from a compelling reason to

give up legal jurisdiction. While federal Indian policy has since shifted away from “termination,”

maintaining and reinforcing sovereignty remains a much-prioritized goal for Indian tribes. Indeed,

state civil law has been imposed only through coercion and no tribe has voted to adopt it.

108Evans (2014) makes a powerful call to study tribes as important political entities. See also Bays and Fouberg(2002) and Deloria and Lytle (1984). For more on how Indian policy entrepreneurs get their tribes federally ac-knowledged, see Cramer (2005). For an argument about how tribes use incremental progress to get their preferencesacknowledged in local and federal legislation, see Evans (2011). For an argument that tribes became relatively en-ervated by treaties with the US government, whethr it was the president or Congress who negotiated, see Spirling(2012).

25

What this paper offers is a quantification of a possible benefit of giving up civil legal

jurisdiction. In so doing, the hope is not only to contribute to our understanding of law and

finance, but also to provide information on the costs of sovereignty for decision-makers tasked with

development.

26

References

Acemoglu, Daron, Simon Johnson, and James Robinson. 2001. “The Colonial Origins ofComparative Development: An Empirical Investigation.” American Economic Review 91:1369-1401.

Ahlquist, John S. 2006. “Economic Policy, Institutions, and Capital Flows: Portfolio and DirectInvestment Flows in Developing Countries.” International Studies Quarterly 50: 681704.

Akee, Randall. 2009. “Checkerboards and Coase: The Effect of Property Institutions onEfficiency in Housing Markets.” Journal of Law and Economics 52(2): 395-410.

Anderson, Robert T. 2012. “Negotiating Jurisdiction: Retroceding State Authority over IndianCountry Granted by Public Law 280.” Washington Law Review 87(4): 915-964.

Anderson, Terry L. 1995. Sovereign Nations or Reservations? An Economic History of AmericanIndians. San Francisco: Pacific Research Institute for Public Policy.

Anderson, Terry L., editor. 1992. Property Rights and Indian Economies. Lanham, MD:Rowman and Littlefield Publishers.

Anderson, Terry, Bruce Benson, and Thomas Flanagan, editors. 2006. Self-Determination.Stanford: Stanford University Press.

Anderson, Terry L. and Dominic P. Parker. 2014. “Lessons in Fiscal Federalism from IndianNations.” In the Handbook of Law and Economics.

Anderson, Terry L. and Dominic P. Parker. 2008. “Sovereignty, Credible Commitments, andEconomic Prosperity on American Indian Reservations.” Journal of Law and Economics 51(4):641-666.

Barsh, Russel L. 1986. “The Nature and Spirit of North American Political Systems.” AmericanIndian Quarterly 10(3): 181-198.

Bays, Brad A. and Erin H. Fouberg. 2002. The Tribes and the States: Geographies ofIntergovernmental Interaction. Lanham, Maryland: Rowman & Littlefield Publishers.

Belanger, Yale D. 2011. First Nations Gaming in Canada. University of Manitoba Press.

Berkowitz, D., K. Pistor and J.-F. Richard. 2003. “Economic Development, Legality and theTransplant Effect.” European Economic Review 47(1): 165-195.

Bernardi-Boyle, Dao Lee. 2001/2002. “State Corporations for Indian Reservations.” AmericanIndian Law Review 26(1): 41-65.

Besley, Timothy. 1995. “Property Rights and Investment Incentives: Theory and Evidence fromGhana.” Journal of Political Economy 103(5): 903-937.

Carpenter, Kristen A., Matthew L. M. Fletcher, and Angela R. Riley. 2012. The Indian CivilRights Act at Forty. Los Angeles: McNaughton and Gunn.

27

Clarkson, Gavin. 2008. “Wall Street Indians: Information Asymmetry and Barriers to TribalCapital Market Access.” Lewis & Clark Law Review 12(4): 943-970.

Clarkson, Gavin. 2007. “Tribal Bonds: Statutory Shackles and Regulatory Restraints on TribalEconomic Development.” North Carolina Law Review. 85: 1009-1086.

Cornell, Stephen and Joseph P. Kalt. 2000. “Where’s the glue? Institutional and culturalfoundations of American Indian economic development.” Journal of Socio-Economics. 29:443-470.

Cowger, Thomas W. 1999. The National Congress of American Indians: The Founding Years.Lincoln: University of Nebraska Press.

Cramer, Renee A. 2005. Cash, Color, and Colonialism: The Politics of Tribal Acknowledgement.Norman, Oklahoma: University of Oklahoma Press.

Cyree, Ken B., Keith D. Harvey, and Michael R. Melton. 2004. “Bank Lending to NativeAmerican Applicants: An Investigation of Mortgage Flows and Government Guarantee Programson Native American Lands.” Journal of Financial Services Research 26(1): 29-54.

Deloria, Jr., Vine and Clifford M. Lytle. 1984 The Nations Within: The Past and Future ofAmerican Indian Sovereignty. Austin, Texas: University of Texas Press.

de Soto, H. 2002. The Mystery of Capital: Why Capitalism Triumphs in the West and FailsEverywhere Else. New York: Basic Books.

Djankov, Simeon, Edward Glaeser, Rafael La Porta, Florencio Lopez-de-Silanes, and AndreiShleifer. 2003. “The New Comparative Economics.” Journal of Comparative Economics 31:595-619.

Djankov, Simeon, Oliver Hart, Caralee McLiesh, Andrei Shleifer. 2006. “Debt EnforcementAround the World.” NBER Working Paper 12807.

Djankov, Simeon, Caralee McLiesh, and Andrei Shleifer. 2007. “Private Credit in 129 Countries.”Journal of Financial Economics 84: 299-329.

Dunning, Thad. 2012. Natural Experiments in the Social Sciences: A Design-Based Approach.Cambridge University Press.

Easterly, William, and Ross Levine. 2003. “Tropics, Germs, and Crops: How EndowmentsInfluence Economic Development.” Journal of Monetary Economics 50(1): 3-39.

Evans, Laura. 2011. Power from Powerlessness: Tribal Governments, Institutional Niches, andAmerican Federalism. Oxford, UK: Oxford University Press.

Evans, Laura. 2014. “Tribal-State Relations in the Anglosphere.” Annual Political ScienceReview 17: 273-289.

Farrell, Henry and Abraham L. Newman. 2014. “Domestic Institutions beyond the Nation-State:Charting the New Interdependence Approach.” World Politics. 66(2): 331-363.

28

Glaeser, Edward, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer. 2004. “DoInstitutions Cause Growth?” Journal of Economic Growth 9: 271-303.

Glaeser, Edward and Andrei Shleifer. 2002. “Legal Origins.” Quarterly Journal of Economics117(4): 1193-1229.

Gitter, Robert J. and Patricia B. Reagan. 2002. “Reservation Wages: An Analysis of the Effectsof Reservations on Employment of American Indian Men.” American Economic Review. 92(4):1160-1168.

Goldberg-Ambrose, Carole. 1997. Planting Tail Feathers: Tribal Survival and Public Law 280.Los Angeles: McNaughton and Gunn.

Gover, Kevin. 1980. “Jurisdiction: Conflicts of Law and the Indian Reservation: Solutions toProblems in Indian Civil Jurisdiction.” American Indian Law Review 8(2): 361-385.

Greif, Avner. 1994. “Cultural Beliefs and the Organization of Society: A Historical andTheoretical Reflection on Collectivist and Individualist Societies.” Journal of Political Economy102(5): 912-950.

Haber, S. 1996. “The Efficiency Consequences of Institutional Change: Financial MarketRegulation and Industrial Productivity in Brazil, 18661943.” NBER Working Paper No. H94.

Haddock, David D. and Robert J. Miller. 2004. “Can a Sovereign Protect Investors from Itself?Tribal Institutions to Spur Reservation Investment.” The Journal of Small and EmergingBusiness Law. 8(2): 173-228.

Hansen, Kenneth N. and Tracy A. Skopek, editors. 2011. The New Politics of Indian Gaming:The Rise of Reservation Interest Groups. Reno: University of Nevada Press.

Jimenez, Vanessa J. 1998. “Concurrent Tribal and State Jurisdiction under Public Law 280.” TheAmerican University Law Review 47: 1627-1707.

Johnson, Simon, John McMillan, and Christopher Woodruff. 2002. “Property Rights andFinance.” American Economic Review 92(5): 1335-1356.

Krivo, L. and R. Kaufman. 2004. “Housing and Wealth Inequality: Racial-Ethnic Differences inHome Equity in the United States.” Demography 41(3): 585-605.

Kuhn, Peter and Arthur Sweetman. 2002. “Aboriginals as Unwilling Immigrants: Contact,Assimilation, and Labour Market Outcomes.” Journal of Population Economics 15: 331-355.

LaFontaine, Frank. 1974. “The Native American Credit Problem.” American Indian Law Review2(1): 29-40.

La Porta, R., F. Lopez-de-Silanes, A. Shleifer and R. Vishny. 1998. “Law and Finance.” Journalof Political Economy 106: 1113-1155.

La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny. 1997. “Legal Determinants ofExternal Finance.” The Journal of Finance 52(3): 1131-1150.

29

Levine, Ross. 2004. “Finance and Growth: Theory and Evidence.” NBER Working Paper No.10766.

Miller, Robert J. 2012. Reservation “‘Capitalism’: Economic Development in Indian Country.”Santa Barbara: Praeger.

Milner, Helen. 1997. Interests, Institutions, and Information: Domestic Politics and InternationalRelations. Princeton University Press.

Mudd, John O. 1972. “Jurisdiction and the Indian Credit Problem: Considerations for aSolution.” Montana Law Review 33(2): 307-316.

Safty, Graham. 2012. Federal Diversity Jurisdiction and American Indian Tribal Corporations.The University of Chicago Law Review 79(4): 1593-1628.

Spirling, Arthur. 2012. “US Treaty Making with American Indians: Institutional change andRelative Power, 1784-1911.” American Journal of Political Science 56(1): 84-97.

Townsend, Robert. 1995. “Financial Systems in Northern Thai Villages.” Quarterly Journal ofEconomics 110(4): 1011-1046.

Trentadue, Jesse C. 1987/1988. “Tribal Court Jurisdiction over Collection Suits by LocalMerchants and Lenders: An Obstacle to Credit for Reservation Indians?” American Indian LawReview 13(1): 1-58.

Trosper, Ronald L. 1978. “American Indian Relative Ranching Efficiency.” The AmericanEconomic Review 68(4): 503-516.

Wilkins, David E. 1998. “Tribal-State Affairs: American States as Disclaiming Sovereigns.”Publius: The Journal of Federalism 28(4): 55-82.

Wilkins, David E. and Heidi K. Stark. 2011. American Indian Politics and the American PoliticalSystem. Lanham, Maryland: Rowman & Littlefield Publishers.

30

Tables

Table 1a: PL280 Mandatory States

State No. of tribes∗

Alaska 1∧

California 107Minnesota 11Nebraska 6Oregon 10Wisconsin 11∗ Number of federally recognized tribes as of 2015.∧ Alaskan Native Villages have separate legal status.

Table 1b: Roll-out Adopters of PL280 or Analogous Statutes

State No. of tribes∗ Year enforced

Colorado 2 1984Connecticut 2 1983Florida 2 1961Iowa 1 1967Maine 4 1980Massachusetts 1 1987Nevada 19 1967New York 7 1950Rhode Island 1 1978Texas 3 1987Utah 7 1980∗ Number of federally recognized tribes as of 2015.

31

Table 1c: States with Tribal Civil Jurisdiction (in sample period)

State No. of tribes∗

Alabama 1Arizona 21Idaho 4Kansas 4Louisiana 4Michigan 12Mississippi 1Montana 7New Mexico 23North Carolina 1North Dakota 4Oklahoma 37South Carolina 1South Dakota 9Washington 29Wyoming 2∗ Number of federally recognized tribes as of 2015.

Table 1d: Tribes Excluded or Retroceeded from PL280 and Roll-out

State State Status Tribe (year of exclusion)

Minnesota PL280 mandatory Red Lake (1953), Bois Forte (1975)Nebraska PL280 mandatory Omaha (1970), Winnebago (1986), Santee Sioux (2006)Oregon PL280 mandatory Warm Springs (1953), Umatilla (1981), Burns Paiute (1979)Wisconsin PL280 mandatory Menominee (1976)Colorado Roll-out Ute Mountain (1984)Connecticut Roll-out Mohegan Nation (1983-1993)Maine Roll-out Aroostook Band of Micmancs (1980-1990)Nevada Roll-out Retrocession for 18 tribes (by 1975), Ely Colony (1988)Texas Roll-out Kickapoo Traditional Tribe (1987)Utah Roll-out Goshute, Navajo, Shoshoni, Skull Valley Band of Goshute,

Ute, Ute Mountain (all in 1980)

Total: 38 tribes.As of 2015.

32

Table 2: State law and Section 184 home mortgage interest rates (1995-2013)

Dependent variable: Interest rate (0 to 9.63)

Model 1 Model 2 Model 3 Model 4

State law (PL280) -0.043∗∗∗ -0.019∗∗ 0.005 0.022∗∗

(0.009) (0.009) (0.009) (0.010)Home on reservation 0.056∗∗∗ 0.038∗∗∗

(0.007) (0.007)State law * Home on reservation -0.136∗∗∗ -0.134∗∗∗

(0.016) (0.016)State income per capita (logged) -0.271∗∗∗ -0.137∗∗∗ -0.247∗∗∗ -0.122∗∗∗

(0.042) (0.043) (0.042) (0.043)Indian pop. per capita (in state) -0.736∗∗∗ -0.787∗∗∗ -0.806∗∗∗ -0.807∗∗∗

(0.049) (0.049) (0.053) (0.054)Distance to metro area (miles) -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗

(0.000) (0.000) (0.000) (0.000)Regional 15-year interest rate 0.802∗∗∗ 0.803∗∗∗ 0.802∗∗∗ 0.804∗∗∗

(0.011) (0.014) (0.011) (0.014)Regional 30-year interest rate 0.776∗∗∗ 0.768∗∗∗ 0.776∗∗∗ 0.768∗∗∗

(0.009) (0.009) (0.009) (0.009)Loan amount (logged) -0.090∗∗∗ -0.089∗∗∗

(0.007) (0.007)Loan term (months) 0.000 0.000

(0.000) (0.000)Coborrower -0.007 -0.008

(0.005) (0.005)Refinance -.029∗∗∗ -0.029∗∗∗

(0.006) (0.006)Servicing lender dummies Yes YesOriginating lender dummies Yes YesYear dummies Yes Yes Yes YesConstant 5.944∗∗∗ 5.583∗∗∗ 5.736∗∗∗ 5.460∗∗∗

(0.529) (0.532) (0.530) (0.533)

Observations 20174 20173 20174 20173R-squared 0.89 0.89 0.89 0.90

Robust standard errors, ∗ ∗ ∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1.

33

Table 3: Matching results (average treatment effect on the treated) (1995-2013)

Nearest-neighbor matching (4 draws)

Model 5 Model 6

State law (PL280) -0.085∗∗∗ -0.088∗∗∗

(0.014) (0.014)

Observations 19703 19682Maximum matches 7 8

Robust standard errors, ∗ ∗ ∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1.

Matching on loan amount, loan term, coborower, refinancing, originating lender, servicing lender,state income per capita, Indian pop. per capita (state), distance to metro area, regional 15-yearinterest rate, regional 30-year interest rate, home on reservation, year.(Model 5) Exact match on year.(Model 6) Exact match on home on reservation, year.

34

Table 4: Section 184 loan limits and state civil law (2013)

Model 7 Model 8 Model 9 Model 10

State law (PL280) 0.170∗∗∗ 0.170∗∗∗ 0.170∗∗∗ 0.170∗∗∗

(0.012) (0.012) (0.012) (0.012)State income per capita (logged) 1.210∗∗∗ 1.210∗∗∗ 1.210∗∗∗ 1.210∗∗∗

(0.064) (0.064) (0.064) (0.064)Indian pop. per capita (in state) 0.564∗∗∗ 0.564∗∗∗ 0.564∗∗∗ 0.564∗∗∗

(0.070) (0.070) (0.070) (0.070)Constant -0.397 -0.150 0.040 0.257

(0.671) (0.671) (0.671) (0.671)

Observations 1881 1881 1881 1881R-squared 0.73 0.73 0.73 0.73

Robust standard errors, ∗ ∗ ∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1.

35

Appendix

Figures

Figure A1. Addresses on and off reservations in PL280 mandatory states

36

Figure A2. Addresses on and off reservations in PL280 roll-out states

37

Figure A3. Addresses on and off reservations in tribal civil law states (untreated)

38

Appendix 1: “Checkerboarding” and civil law variation

Many Indian lands are “checkerboarded,” in that land that was “allotted” under the 1887

Dawes Act and can be privately owned (and used as collateral) is deeply intermixed with land

owned by the tribe, land owned by the US government and held in trust for the tribe, and land

with other ownership status. Effectively, this introduces variation in the amount of land available

for private home ownership on different reservations and the clarity that actors have about the

ownership status of a particular parcel of land.

Do borrowers on severely checkerboarded reservations derive the same net benefits from

state law as found in the main analysis? I limit the sample to only those borrowers with addresses

on reservations and explore whether variation in allotment further determines variation in their

interest rates. Note however that, without a model of the selection process that drives a borrower

to live on or off a reservation, the representativeness of these borrowers is circumspect.

I treat the data as observational and use the same covariates as in Table 2. I add covariates

about the reservation acreage: total acres (logged) (Model A1), total allotted acres (logged) (Model

A2), alloted acres as a pct of total (Model A3), and an interaction term, alloted per total * state

law (Model A4). These data are difficult to acquire. We took them from a 1974 report by the US

Department of Commerce, “Federal and State Indian Reservations and Indian Trust Areas,” Stock

Number 0311-00076.109

The results in Table 5 show that this approach washes out direct effects of state law. Thus,

the full population of Section 184 borrowers, both on and off reservations, is important to generating

the results in the body of the paper. The absolute size of a reservation does not have a significant

effect on an on-reservation borrower’s interest rate (Model A1). Borrowers on reservations with

more allotted land, that is ownable but is highly intermixed with other kinds of land, are facing

higher interest rates (Model A2). Allotted acres as a percentage of total reservation acres is my

measure of checkerboarding; the coefficient is positive in Model A3 and positive and significant in

Model A4, as expected. If borrowers on more checkerboarded reservations benefit more from state

109There are 236 observations for which borrowers live on a reservation recognized by the state but not by the federalgovernment; these are marked as missing. There are 205 observations for which borrowers live on a reservation butfederal reservation land amounts are not available in this source (28 reservations). As Alaskan Native Villages werenot subject to the Dawes Act and have special status with regard to the types of property ownership possible,addresses in Alaska are excluded from the analysis.

39

civil law, then the interaction term in Model A4 should be negative and significant. It is negative,

but it does not reach statistical significance. Thus, there is not conclusive evidence that state civil

law is especially helpful to borrowers on checkerboarded reservations. There is evidence, however,

that checkerboarding itself is correlated with higher interest rates.

[Table 5 about here.]

40

Table 5: Effects of “checkerboarding” on Section 184 home mortgage interest rates for borrowerson reservations (1995-2013)

Dependent variable: Interest rate (0 to 9.63)

Model A1 Model A2 Model A3 Model A4

State law (PL280) 0.037 0.043 0.038 0.060(0.042) (0.042) (0.042) (0.045)

Total reservation acres (logged) 0.001(0.003)

Allotted acres (logged) 0.001∗

(0.001)Allotted acres per total 0.016 0.018∗

(0.011) (0.011)Allotted acres per total * State law -0.092

(0.082)Constant 2.290 2.292 2.270 2.377∗

(1.434) (1.425) (1.424) (1.430)

Observations 9376 9376 9376 9376R-squared 0.92 0.92 0.92 0.92

Robust standard errors, ∗ ∗ ∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1.

Not reported: state income per capita (logged); Indian population per capita (in state); distanceto metro area; regional 15- and 30-year interest rates; loan amount (logged); loan term (months);coborrower; refinance; servicing lender dummies; originating lender dummies; year dummies.

41