Debtors’ prisons in America: An economic analysis

Journal of Economic Behavior & Organization 84 (2012) 216–228

Contents lists available at SciVerse ScienceDirect

Journal of Economic Behavior & Organization journal homepage: www.elsevier.com/locate/jebo

Debtors’ prisons in America: An economic analysis Matthew J. Baker a,∗ , Metin Cosgel b,1 , Thomas J. Miceli b,2 a b

Department of Economics, Hunter College and the Graduate Center, CUNY, New York, NY 10065, United States Department of Economics, University of Connecticut, Storrs, CT 06269, United States

a r t i c l e

i n f o

Article history: Received 27 September 2011 Received in revised form 24 July 2012 Accepted 26 July 2012 Available online 6 August 2012 JEL classification: D82 E51 G21 K42

a b s t r a c t Debtors’ prisons have been commonplace throughout history, including in the United States. While imprisonment for debt no doubt elicited some repayment by benefactors of the debtor, we argue that its primary function was to deter default in the first place by giving borrowers an incentive to disclose hidden assets. Because of its cost, however, imprisonment was destined to be replaced by more efficient ways of preventing borrowers from sheltering assets. Empirical analysis of state laws banning imprisonment for debt provides some support for this argument. In particular, the results suggest that states in which the publishing industry developed sooner (thus facilitating the flow of information) were more likely to enact early bans on imprisonment for debt. © 2012 Elsevier B.V. All rights reserved.

Keywords: Debtors’ prison Default Imprisonment

“Any society that admits the concept of debt has to develop some means of dealing with those who default on their obligations.” -Coleman (1999, p. 3). “Annual income twenty pounds, annual expenditure nineteen nineteen and six, result, happiness. Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.” -Mr. Micawber’s advice to David Copperfield on how to avoid debtors’ prison (Dickens, [1850] 2006). 1. Introduction The availability of credit is an important component of a growing economy. A fundamental problem facing any system that relies on credit, however, is how to ensure timely repayment, and also how to deal with those unable to repay. These questions present society with a difficult trade-off because, while an unbending policy toward defaulters is generally necessary to enforce repayment terms, it can result in harsh treatment of those down on their luck. In early English law, the ultimate penalty for defaulters was imprisonment, often at the debtor’s own expense, until the debt was paid. Of course, this raises the obvious question of how one could repay debt while in prison, but the underlying presumption was that the threat of

∗ Corresponding author. Tel.: +1 212 772 4217; fax: +1 212 772 5398. E-mail addresses: [email protected] (M.J. Baker), [email protected] (M. Cosgel), [email protected] (T.J. Miceli). 1 Tel.: +1 860 486 4662; fax: +1 860 486 4463. 2 Tel.: +1 860 486 5810; fax: +1 860 486 4463. 0167-2681/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jebo.2012.07.010

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

217

imprisonment would induce borrowers to avoid default in the first place. In other words, the prospect of prison would act as a deterrent. Additionally, Mann (2002, p. 79) notes that creditors “hoped that the rigors of imprisonment would induce debtors to disclose concealed wealth or to part with assets that were exempt from attachment or, perhaps, that family members might step into the breach. . .” Holton (2007, p. 43) similarly observes that the system contained a “cruel logic, since it forced the delinquent debtor to reveal hidden assets. . .” Still, some debtors were bound to default, and others apparently were willing to sacrifice their freedom in order to protect their assets (Coleman, 1999, p. 9). As a result, debtors’ prisons flourished as a costly method for enforcing repayment, both for society and for prisoners. The fact is that, throughout history, imprisonment for debt was the norm rather than the exception to the rule. Ford (1926) describes the codification of imprisonment for debt in early Roman law, which evidently allowed debtors to be arrested and placed in jail for a prescribed period of time. References to imprisonment for debt can also be found in the bible, where it is suggested that if debts went unpaid by the end of the imprisonment term, the debtor could be killed or sold into slavery.3 While the tendency to imprison debtors apparently abated as alternative means of dealing with delinquent debtors developed, it reemerged in full force in the Middle Ages, waned again in the era of feudalism, only to return to prominence in the later Middle Ages, partly through the approval of the Catholic Church. For a variety of reasons discussed by Ford (1926), debtors’ prisons surged in medieval England and spread throughout Europe. In fact, Ford (1926, p. 30) cites the 1834 report of a British parliamentary commission asserting that at the time imprisonment for debt was legal in every country in continental Europe except Portugal. The American colonies imported the practice, and “[b]y the end of the seventeenth century the debtors’ prison had become an established colonial institution” (Coleman, 1999, p. 249). The system remained little changed until the beginning of the nineteenth century when most colonies began to enact reforms, driven in part by humanitarian concerns. And when Massachusetts abolished imprisonment for petty debts in 1811, the process of eliminating the practice altogether had begun. “Between 1811 and the end of Reconstruction most but not all of the eastern states gradually prohibited the imprisonment of defaulters except in cases of fraud and in damage suits for alimony, child support, and wrongful behavior” (Coleman, 1999, p. 256). The demise of the debtors’ prisons, however, was not driven entirely by compassion for debtors; it also reflected changes in lending practices. As credit became more impersonal, lenders began to require some form of security up front as protection against default. In addition, the legal system facilitated the seizure of secured assets in lieu of payment. Borrowers therefore had a harder time sheltering assets. Debtors’ prisons thus became an anachronistic and costly way to enforce payment in this world, and undoubtedly would have disappeared even without legal action (Coleman, 1999, p. 268). The goal of this paper is threefold. The first is to provide an historical picture of debtors’ prisons in the United States as they existed in the early years of the Republic. The second is to develop an economic theory of debtors’ prisons, focusing on their role in facilitating efficient lending and repayment of debt, and on the emergence of alternative enforcement measures. Finally, the paper offers some historical and empirical evidence on those factors that led to the demise of debtors’ prisons in the United States during the mid-nineteenth century.

2. An overview of debtors’ prisons in the United States While much anecdotal evidence suggests that imprisonment for debt was an important part of life in the United States at the beginning of the 19th century, how prevalent, really, was the practice, and how exactly did it function? At least one group, the “Prison Discipline Society,” a philanthropic society formed with the aim of improving prison conditions in the United States, kept careful records of debtors in prison. The Societies’ 5th report, published in part in 1831, described the results of a detailed survey of the nature and extent of debtors’ prisons across the United States.4 We present portions of this report in Tables 1–3, as a way of giving readers a feel for the importance of imprisonment for debt in the early Republic. The immediate lesson learned from these tables is that debtors’ prisons were in full use in the early part of the 19th century, and the number of people imprisoned for debt was large both in an absolute and a relative sense. In fact, from the data reported in Table 1, it can be seen that in many places, the majority of prison populations was composed of debtors. The Prison Discipline Society’s estimate is that five out of every six prisoners were in fact in prison for debt in the sampled states, and that this put the total number of people imprisoned for debt, at least across the northeastern United States, well into the tens of thousands. In spite of the apparent widespread use of the debtors’ prison, the data in Table 2 suggest that a relatively large fraction of debts for which people were imprisoned were small, with the majority being for $20 or less. This is borne out by the Prison Discipline Society’s own estimate of the value of a days’ time at $1. It is difficult to obtain readily comparable information about the distribution of wealth in the United States in 1830, which might be a better yardstick against which to compare the debt levels in Table 2. However, in 1850 the per capita value of personal estate across the United States was $122.59 (in

3 See Matthew 18:29–31 (New International Version) on imprisonment for debt. Matthew 18:24–26 describes an episode in which a debtor is to be sold into slavery to repay a debt. 4 While we refer to “debtors’ prisons,” generally speaking, imprisoned debtors were usually just kept in States’ prison or jail facilities along with the rest of the general prison population.

218

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

Table 1 Imprisonment for Debt in 1829. Cities, towns, and states

No. imprisoned for debt Prisoners

Concord, NH Taunton, MA Worcester, MA Boston, MA Massachusetts East Greenwich, RI Newport, RI Rhode Island Pennyan, NY Courtland Village, NY Buffalo, NY New York City, NY New York Belvidere, NJ Flemington, NJ Philadelphia, PA Pennsylvania Baltimore, MD Maryland Northern and Middle States combined

Ratio, debtors to other prisoners Statewide totals

31 126 271 1211

3 to 1 ∼3000

80 78 4 to 1 5 to 1 8 to 1

103 112 338 3000 ∼10,000

5 to 1 6 to 1 817 ∼7000 944 ∼3000 ∼5 to 1

Notes: Philadelphia numbers are for the 8 months ending February 25, 1830. Information compiled from the 5th report of the Prison Discipline Society, as published in the North American Review vol. 32, Gray and Bowen, Boston 1831. Table 2 Magnitude of Debt, for a sample of prisons and the city of Boston, 1829. Debt size

Less than $1 $1–$5 $5–$20 $20–$100 More than $100 Total

All prisons

Boston

Prisons reporting

Total imprisoned

12 30 32 32 53

62 595 2184 902 416 4159

Percent 1% 14% 53% 22% 10% 100%

Total imprisoned 30 233 314 142 98 817

Percent 3% 29% 38% 17% 12% 100%

Notes: Boston data pertains to June 6, 1829 to February 24, 1830. Information compiled from the 5th report of the Prison Discipline Society, as published in the North American Review vol. 32, Gray and Bowen, Boston 1831. Table 3 Sentence length for debtors, 1829. Sentence length

Number (15 prisons)

Less than 1 day 1–5 days 5–10 days 10–20 days 20–30 days More than 30 days Total

269 323 203 154 83 431 1463

Percent 18% 22% 14% 11% 6% 29% 100%

Notes: Information compiled from the 5th report of the Prison Discipline Society, as published in the North American Review vol. 32, Gray and Bowen, Boston 1831.

1850 dollars), while the per capita value of real estate was $225.16.5 As for the length of sentences for those imprisoned for debt, most were short, typically less than a month, and frequently only for a couple of days. However, some stayed in prison for substantially longer periods of time (see Table 3). In New York in the mid-eighteenth century, debtors who owed more incurred longer sentences (Coleman, 1999, p. 108), but unfortunately we do not know whether that was also true in other states. In addition to this information, the 1831 report of the Prison Discipline Society tells us that few of those imprisoned for debt actually wound up paying off their debts. According to data from 17 prisons (p. 493 of the 1831 report), out of the

5 While these numbers are not corrected for inflation over the time period, price indices spanning the time period seem to indicate that some deflation had occurred over the time period 1830–1850. See Historical Statistics of the United States: Colonial Times to 1970, Part 1, pages 201–207.

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

219

Table 4 Evolution of state laws, imprisonment for debt. State

Year

Notes

Alabama Arkansas* Connecticut Delaware Florida Georgia* Illinois* Indiana* Iowa (Terr.) Kentucky* Louisiana Maine* Maryland* Massachusetts Michigan Mississippi Missouri* New Hampshire New Jersey New York North Carolina* Ohio Pennsylvania Rhode Island* South Carolina* Tennessee Vermont Virginia* Wisconsin (Terr.)

1839 1843 1842 1841 <1850 1858

1826 – Women excluded 1837 – no petty debts 1841 – Only debts > 50$? 1824 – Women excluded 1847 – Women excluded

1842?

1838 – Prison bounds coextensive with the county

1821? 1840

1820 – Women excluded

1851 1857 1839 1839 1843 1840 1842 1831 >1850 1838 1842 >1900 >1850 1840 1838 1873 1841

1822 – Only debts > 5$ 1824 – Women excluded 1811 – Only debts > 5$ 1831 – only debts > 10$, women excluded

1818 – Only debts > $13.33 1831 – women excluded 1818 – Women excluded 1819 – Only debts > 10$ 1828 – women excluded 1823 – Women excluded 1844 – proof of concealment or transfer 1819 – Women excluded, 1833 – no petty debts 1825 – Only debts > 20$ 1841 – judicial district boundaries 1819 – Only debts > 15$ 1834 – women excluded 1849 – “Partial abolition”

Notes: The table was constructed through consultation of Coleman (1999), Kent (1848, 1866), Kinne (1842), McMaster (1920), and Prison Discipline Society (1841). * Marks states which Kent (1848, 1866) asserts still allowed some form of imprisonment for debt as of the writing of his book.

2057 persons imprisoned for debt, only 294 (14.3%) actually paid the debt and were discharged, while 1019 (49.5%) were discharged by the creditor, and 744 (36.2%) were discharged via a poor debtor’s oath. The society estimated that the total amount paid out of these debts was $7992, while it put the value of time lost due to imprisonment over the same time period at $19,987 (as noted above, the society used an estimate of $1 a day to value time lost). These pieces of information indicate that collectively, those in prison generally wound up not paying their debts, and that the costs of maintaining debtors in prison were substantial. In interpreting these data, however, it is important to remember that if the chief function of debtor’s prison was to prevent default in the first place (as we shall argue in the next section), then the overall gain from imprisonment cannot be reckoned solely in terms of recovered debt. States began responding to the moral pressures of philanthropic societies such as the Prison Discipline Society, and perhaps the costs incurred in keeping debtors in prison, in the third and fourth decades of the 19th century. As previously noted, the initial impetus for abolition of imprisonment for debt was part of the larger societal movement against cruel punishment.6 Indeed, in the decade of the 1840s, states began enacting legislation that banned imprisonment for debt, except in cases where outright fraud had been committed. Table 4 gives some estimates culled from various sources as to the year in which a state ultimately banned imprisonment for debt.7 Imprisonment for debt was banned in the United States for federal court actions in 1841 (Kent, 1848, p. 399). Interestingly, the majority of states abolished imprisonment for debt without clear alternative laws for dealing with bankrupts and insolvents in place.8 In the early part of the 19th century, some states, notably New York and Pennsylvania,

6 The Prison Discipline Society writes “It appears to us impossible, that it can ultimately be maintained under the growing intelligence and liberality of the times, and we confidently trust, that the present generation will not readily yield to their successors the honor of erasing this remnant of barbarism from the civil code of the country” (Prison Discipline Society, 1831, p. 508). 7 This table was constructed through consultation of Coleman (1999), Kent (1848, 1866), Kinne (1842), McMaster (1920), and Prison Discipline Society (1841). There is in fact considerable difficulty in deciphering whether or not, and when, a given state actually banned imprisonment for debt. The confusion arises from multiple dimensions. Some states modified their laws to allow imprisonment for cases in which debtors intended to flee, abscond with, and/or conceal property. This is a different thing than allowing imprisonment only in cases of fraud. Other states – particularly but not only southern states – passed laws that defined the boundary of the debtors’ prison to coincide with an area that still allowed prisoners some latitude. These boundaries ranged from a couple of acres near the prison wall, to the entire county or jurisdiction, and in some cases, the entire state; see footnote 8 for more details. 8 In most states, a debtor did have one remedy available: petition to the state legislature for relief. If favorably disposed, the legislature might fashion an arrangement with creditors and release the debtor from prison. In most cases, however, debtors were still required to pay their debts. See Coleman (1999, pp. 10–14). Many states provided other alternatives as well, such as indentured servitude, although this fell out of fashion in the early colonial period.

220

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

experimented with bankruptcy laws, but the constitutionality of these laws almost immediately came into question (Coleman, 1999, pp. 32–33). In two important decisions, the courts ruled that the bankruptcy laws were unconstitutional: Pennsylvania’s 1812 relief law was ruled unconstitutional in 1814 in Golden v. Prince (1814), and the relatively liberal practices for dealing with insolvents in New York – in place since colonial times – were challenged successfully on constitutional grounds in Sturgis v. Crowninshield (1819). The controversy derived from Article 1, Sections 8 and 10 of the Constitution. Section 8 endowed Congress with the power of enacting uniform bankruptcy laws, while Section 10 decreed that states could not pass laws that inhibited contracts between individuals. While the courts later reversed course on the constitutionality question in 1827 in Ogden v. Saunders (1827), the prior rulings seem to have cooled states to the idea of enacting bankruptcy laws, and the majority of states did nothing on the subject of bankruptcy until the latter half of the 19th century, if at all.9 The notes in Table 4 provide some additional details as to how the law evolved over time in various states. Most of the law pertains to northern and middle states; apparently, in Southern States, there was very little imprisonment for debt, in spite of active laws in many states’ constitutions.10 We speculate that this might be due to the fact that in the South, many prison boundaries, at least for purposes of debt, were defined to be circumscribed areas around the prisons.11 Coleman (1999) describes several reasons for the apparently informal notion of imprisonment for debt in much of the South. Many rural communities in the South did not have formal prisons for debtors, so in many cases, debtors were discharged to the sheriff, who could be held responsible for the debts of absconding prisoners. It also appears that many southern states had a historical tradition of reliance on alternative institutions where possible. As one example, Virginia traditionally relied upon elegit and replevin systems, through which creditors could attach returns derived from debtors’ property, or debtors could have bonds posted reflecting intent to pay off a debt. There also appears to have been a greater reliance on servitude as a substitute for formal imprisonment in the south. Coleman (1999, p. 243) hypothesizes that these North–South differences might be due to the “. . .lack of economic diversification in the South and the resulting dependence on single-crop staples and distant markets.” In spite of these differences in how imprisonment was actually applied, the fact remains that many states, southern or northern, maintained formal laws allowing imprisonment for debt into the middle of the 19th century. In our empirical analysis, we therefore include a dummy variable for southern states in some models. Short of an outright ban of debtors’ prisons, the Prison Discipline Society noted that states enacted several measures that served to reduce the number imprisoned for debt – measures such as minimum debt amounts for imprisonment, and holding the creditor responsible for jailing fees. In fact, the Society reported that one prison manager estimated that this latter measure reduced the number imprisoned for debt by roughly one-third (Prison Discipline Society, 1831, p. 493).

3. An economic theory of debtors’ prisons The economic theory of debtors’ prisons to be developed in this section is based on the notion of deterrence. In particular, lacking collateral, borrowers could only pledge their freedom as security for the loan. As a result, borrowers faced the choice of either paying back the loan, or going to prison. The argument is an application of the economic theory of prison as a deterrent for ordinary crimes when offenders either lack sufficient wealth to pay the optimal fine (Becker, 1968; Polinsky and Shavell, 1984), or seek to shelter their assets from seizure (Levitt, 1997). To be explicit, consider a borrower who has taken a loan of L dollars and invests in a project that yields an uncertain return.12 Specifically, suppose that the borrower’s return takes a value of XH with probability p and XL with probability 1 − p, where XH > L > XL ≥ 0

(1)

Thus, in the low-return state, he is unable to repay the loan.13 This need not present a problem if the lender can observe the borrower’s realized return. In particular, suppose that the lender required the borrower to pay R > L in the high return state and XL in the low return state (i.e., he forgives the unpaid balance). One interpretation we offer below is that XL is collateral offered by the borrower as (partial) security for the loan.

9 Coleman (1999, Chapter 3) provides an illuminating discussion of these cases and surrounding developments. Massachusetts was one exception to the rule, passing a bankruptcy law in 1838, well before imprisonment for debt was banned in the state in 1857 (Coleman, 1999, p. 50). 10 This fact was noted by the Prison Discipline Society, who, while noting that “In seventeen prisons heard from in the Northern States, the number of persons imprisoned during the year ending December 30, 1829 was two thousand seven hundred and forty-two,” (Prison Discipline Society, 1831, p. 494), added that “In the same number of Prisons in the Southern States, only thirty-five” (Prison Discipline Society, 1831, p. 494). 11 Coleman (1999) describes this tendency in several Southern States. For example, Coleman (1999, p. 186) notes that in South Carolina in 1841, the boundaries of the prison were defined to be coterminous with the prison’s judicial district. In Georgia an act passed in 1820 allowed imprisoned debtors the privilege of the jail yard, which amounted to an area of 10 acres around the jail. This area was increased to 100 acres in an 1840 act (Coleman, 1999, p. 235). 12 We treat L as fixed. See Gale and Hellwig (1985) for a model in which the loan amount is endogenous. 13 We assume that the borrower has no other assets besides the return on the investment. Relaxing this assumption would not alter our results provided that the borrower’s total assets in the low-return state are insufficient to pay off the loan.

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

221

Suppose the lender has an opportunity cost of the loan equal to K. For example, K = L(1 + r), where r is the return on the lender’s next best investment (implying that K > L). The lender will therefore expect to just cover his opportunity costs if pR + (1 – p)XL = K, or if R=

K − (1 − p)XL p

(2)

The borrower will take the loan in the first place if he expects a positive return, or if p(XH − R) ≥ 0.14 After substituting for R from (2), this condition becomes pXH + (1 − p)XL − K ≥ 0

(3)

Thus, the borrower will accept the above “implicit contract” if the expected return from the investment exceeds the opportunity cost of the funds. Under this contract, the borrower agrees to pay a premium in the high return state to compensate for the lender’s expected losses in the low return state. Now suppose that the lender cannot observe the realization of the borrower’s return from the project. In this case, the above contract will not work because the borrower will always have an incentive to report low wealth. In other words, the contract is not incentive compatible. The problem is that if lenders expect all borrowers to report low wealth, they will not be able to cover their opportunity costs (given XL < L < K), and the loan market will fail.15 We suggest that debtor prisons were a response to this potential source of market failure. The idea is that the threat of prison makes it costly to default (report low wealth), so it becomes incentive compatible for borrowers to repay the loan in the high wealth state.16 The outcome is second-best, however, because prison imposes a deadweight loss on society. Thus, debtor prisons are an efficient response to the above market failure only if the net gain from making funds available to borrowers outweighs the expected social cost of imprisonment. To examine this function of debtor prisons formally, let z be the length of the prison term imposed on defaulters, z ≥ 0, and let ˛ be the per unit disutility (or opportunity cost) of prison to borrowers (measured in dollars). Borrowers thus expect to incur a non-pecuniary cost of ˛z if they default.17 The advantage of a non-pecuniary cost in this context is that it is not constrained by the borrower’s wealth (Gale and Hellwig, 1985, pp. 651–652). As before, a borrower who realizes a low return on his investment defaults and pays XL to the lender, but he now must also serve a prison term of z periods. A borrower who realizes a high return, however, now has a choice. He can either pay off the loan and receive a return of XH − R, or he can claim a low return and default, which yields a return of XH − XL − ˛z. Since the sole purpose of imprisoning defaulters is to induce truthful revelation by borrowers in the high return state (Levitt, 1997), the length of the prison term must satisfy the following incentive compatibility constraint: XH − R ≥ XH − XL − ˛z or z≥

R − XL ˛

(4)

Now consider lenders. Since they may have to incur some pecuniary costs of maintaining the prison (for example, lenders might be taxed to finance the maintenance of debtor prisons), the condition for them to participate in the contract is pR + (1 − p)(XL − ˇz) ≥ K,

(5)

where ˇ is the unit cost of prison maintenance.18 Writing (5) as an equality (given competition among lenders) and rearranging yields R=

K − (1 − p)(XL − ˇz) p

(6)

The expected return to the borrower under the debtor-prison contract is EUDP = p(XH − R) − (1 − p)˛z,

14 We assume that both the borrower and lender are risk-neutral. Gale and Hellwig (1985) consider the impact of risk aversion. The qualitative nature of our results would be unaffected by such an extension. 15 The assumption that lenders cannot observe the realization of X is obviously extreme. More realistically, borrowers will be able to conceal a fraction  of their “excess” return XH − XL in the high return state. In that case, the expression in (2) becomes R = {K − [1 − p(1 − )]XL }/[p(1 − )], which is increasing in . (Note that (2) is a special case of this expression when  = 0.) Thus, the more borrowers can conceal, the higher will be the interest rate, which will drive some borrowers out of the market. In the extreme case where  = 1, R is infinite and the market completely fails. 16 Alternatively, we could have shown that the threat of imprisonment induces borrowers to work harder to avoid default (a moral hazard argument). The implications of this specification, however, would be the same. 17 Rea (1984) similarly examines the use of a non-pecuniary cost of default (“arm-breaking”) that lenders can impose on defaulting borrowers as a way to induce truthful reporting. 18 Both Massachusetts and Connecticut enacted reforms in the seventeenth century explicitly requiring creditors to pay the jail fees of their imprisoned debtors (Coleman, 1999, p. 251).

222

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

or, after substituting for R from (6), EUDP = pXH + (1 − p)XL − K − (1 − p)(˛ + ˇ)z.

(7)

Clearly, the optimal contract minimizes the prison term, z, subject to the incentive compatibility constraint in (4). It follows that the constraint will be binding. Thus, writing (4) as an equality and substituting for R from (6), we obtain the optimal prison term: z∗ =

K − XL p(˛ + ˇ) − ˇ

(8)

which is positive if p > ˇ/(˛ + ˇ). Thus, the probability of repayment must be sufficiently high to make imprisonment for debt profitable.19 It follows from (8) that ∂z*/∂p < 0 and ∂z*/∂XL < 0. Thus, the optimal prison term is decreasing in the probability of a high return on the borrower’s investment, and also in the borrower’s wealth in the low-return state. Given the above interpretation of XL as collateral, the latter result suggests that better collateral can substitute for prison when borrower wealth is unobservable. It cannot completely solve the incentive compatibility problem, however, unless XL > K, in which case there is no risk of default.20 Although we have solved for the optimal (second-best) debtor-prison contract, there is no guarantee that it yields a net social gain. The assumption of competition among lenders awards any surplus to borrowers, so the condition for a net gain ∗ ≥ 0. is that (7) be non-negative when evaluated at the optimal contract; that is, EUDP As an alternative to the threat of debtor prisons to enforce repayment, lenders can devote resources to learning the state of nature, which amounts to discovering the borrower’s ability to repay the loan. This costly state verification approach to debt contracts was first examined by Townsend (1979) and was extended by Gale and Hellwig (1985).21 The key finding of this literature is that if the borrower can learn the state of nature with certainty by an expenditure of resources, then, under reasonable conditions,22 the optimal contract will resemble a standard debt contract under which the lender only audits the borrower’s resources when the latter claims default. In the current model with only two states, this implies that the borrower will pay R > L in the high-return state and not be audited, but will declare default and pay XL after being audited in the low-return state. This contract is incentive compatible for the borrower as long as R > XL , since in that case he will be unwilling (and unable) to falsely claim the high return state in order to avoid an audit. (He cannot falsely claim the low return state since it will trigger an audit that will reveal his true return to be XH .) As for lenders, they will participate in the contract if pR + (1 − p)(XL − C) ≥ K,

(9)

where c is the cost of an audit. Writing this as an equality (again, due to competition) and solving for R yields R=

K − (1 − p)(XL − c) p

(10)

from which it follows that R > XL as required. Finally, we can write the borrower’s expected return under the costly-state-verification contract as ECCSV = p(XH − R)

(11)

which, unlike the debtor-prison contract, entails no explicit cost in the default state. However, after substituting for R from (10) this becomes EUCSV = pXH + (1 − p)XL − K − (1 − p)c.

(12)

Comparison of (7) and (12) immediately shows that borrowers will prefer the costly-state-verification contract, when feasible, to the debtor-prison contract, when c < (˛ + ˇ)z ∗

(13)

or when the cost of an audit is less than the cost of imprisonment. The abolition of debtors’ prisons therefore becomes more likely as the cost of information acquisition decreases and/or as prison costs rise.23 Overall, the model shows that there was a role for debtors’ prisons in facilitating the development of credit markets prior to the emergence of more efficient ways to enforce repayment of debt. The primary function of imprisonment was to

19 Note that z* will always be positive if ˇ = 0; that is, if there is no pecuniary cost of imposing the punishment (e.g., if it takes the form of arm-breaking), or if the cost is external to the contractors (e.g., if the pecuniary cost of imprisonment is paid out of general tax revenues). 20 Collateral serves a different purpose in the current model as compared to Bester (1987). Specifically, Bester considers the use of collateral to separate borrowers with unobserved risks of default rather than to provide individual borrower’s an incentive to accurately report their return. 21 See Bolton and Dewatripont (2005, pp. 190–197) for a useful overview. 22 For example, risk-neutrality of the contractors. 23 A requirement for both types of contracts to be feasible is that the threat either to put defaulting debtors in prison, or to audit their wealth, must be credible (Bolton and Dewatripont, 2005, p. 196). Of course the same can be said of criminal penalties involving prison.

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

223

discourage debtors from concealing assets from creditors, but credibility of the threat necessitated imprisonment of those debtors who truly lacked sufficient resources to repay their loans. The resulting cost to society was high, however, and so the practice was gradually phased out. The next section describes the forces, both economic and social, that led to that decline. 4. The demise of debtors’ prisons in the United States This section offers some evidence on the factors leading to the eventual decline of debtors’ prisons in the United States. We first offer some qualitative historical evidence, and then, against this background, we attempt a more rigorous empirical analysis. 4.1. Historical evidence Imprisonment for debt was inherited by the American colonies from England, but it was not adopted by colonial governments as a matter of course; rather, legislatures and courts “created the institution deliberately as acts of public policy” (Coleman, 1999, p. 247). The preceding theoretical analysis shows that there was indeed an economic case to be made for the practice, and colonial lenders and lawmakers apparently understood this. Early on, however, critics pointed to the inhumaneness of the system as well as to its practical drawbacks. As to the latter, imprisonment of defaulters rarely induced repayment of the debt while at the same time imposing substantial costs on both the creditor (who often had to finance the prison) and the community, which was deprived of the debtor’s labor and in many cases had to support his dependents. Such concerns brought about the gradual decline of the system, though remarkably it was not until the early twentieth century that the last vestiges of it were repealed in the U.S.24 By the time of the revolution, the institution still existed in all colonies, but its enforcement varied greatly, and few debtors who were arrested actually remained in prison for extended periods of time (though few were induced to repay either). Formal steps toward eliminating the imprisonment of debtors accelerated after independence. These typically first involved imposing the costs of maintaining jails on creditors, which discouraged their use of the institution, but then the categories of defaulters who could be jailed were reduced (see Table 4). For example, most states first abolished imprisonment of female debtors, then petty debtors, and eventually truly impoverished debtors could be released upon taking a “poor debtors’ oath.”25 Many states, when they finally abolished prison for all debtors, still allowed imprisonment of those who committed fraud or concealed property, and expanded the ability of creditors to attach property in lieu of payment. Both of these provisions reflect the logic of the above model. Throughout the reform period, critics relied as much on humanitarian arguments against imprisonment as they did on practical considerations. Prison, they said, essentially branded defaulters as immoral, if not criminal, rather than as what they truly were – unfortunate victims of circumstances. (Though some small fraction of borrowers consciously chose prison rather than turning over their wealth.) But by the 1820s, the actual use of the system had so declined owing to its apparent inefficiency that newspaper accounts decrying the practice seemed to have somewhat overstated its importance (Coleman, 1999, pp. 254–255). Such negative publicity, exaggerated or not, undoubtedly contributed to the formal efforts to abolish imprisonment for debt altogether, and may have led to its abolition sooner than strict economic logic dictated. In the end, however, it was the development of better ways of dealing with the problem of enforcing repayment of debt that sealed the fate of debtors’ prisons. Important innovations in lending included the creation of various means of collateralizing loans and/or the requirement of co-signers for large debts, and the creation of rights of garnishment and pawn broking for petty debts. Also, the increasing efficiency of the legal system made it easier for lenders to collect on their debts, while at the same time, economy-wide institutional innovations made it easier for creditors to ascertain the true underlying reasons for default (Allen, 2012). Therefore, as Coleman (1999, p. 268) notes, The debtors’ prison disappeared because it was obsolete. Reformers accelerated the contraction of creditor rights by pressing first for modifications in the imprisonment system and then for abandonment of the debtors’ jail, but in the long run the institution would have disappeared anyway. The formal repeal of debtors’ prison laws in most states was therefore a formality, reflecting their economic obsolescence. 4.2. Empirical analysis This section undertakes a more formal empirical analysis of debtors’ prisons, focusing on those factors that led to their eventual demise. Table 4 provides some details on the actions many states took to restrict or ban imprisonment for debt in the early decades of the 19th century. A useful reference in tracking the ultimate progress of laws over time is James Kent’s Commentaries on American Law, which was first published in 1826 and went through twelve editions running through 1901 (Kent, 1848, 1866). (In the process, it picked up additional authors, including such notable legal scholars as Oliver Wendell Holmes, Jr. and John M. Gould.) Although it did not give explicit dates as to when imprisonment for debt in every state was

24 25

The discussion in this section is based largely on Coleman (1999). Some states also exempted Revolutionary War veterans from imprisonment.

224

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

Table 5 1840 Census data and model hypotheses. Variable

Relationship to model

Hypothesized impact

Year of statehood

Year of statehood may influence the state of development of capital markets or inertia in previous state laws Proxy for the development of capital markets Proxy for the development of capital markets Difference in prison sentencing and the efficacy of debtors prison Population preference for interests of commerce Better flow of information about borrowers, more anti-imprisonment propaganda Same as above, but measure may better control for scale

Ambiguous

Same as above, but measure may better control for scale

Earlier ban

Proxy for the ease of flight from creditors or importance of commerce Proxy for the importance of commercial interests Proxy for the importance of commercial interests

Later ban Later ban Later ban

Same as above, but measure may better control for scale

Later ban

Proxy for the importance of commercial interests

Later ban

Population density Fraction population urban Southern state dummy Fraction voting Whig, 1840 presidential election Number of newspapers and periodicals per 10,000 pop. Employment in publishing, newspapers, and binding per 1000 pop. Dollars invested in printing, newspapers and periodicals per cap. Emp. in navigation per 1000 pop. Grocery and dry goods stores per capita Commercial foreign trade and commission houses per 1000 pop. Capital invested in grocery and dry goods stores, $ per cap. Capital invested in commercial houses, $ per cap.

Earlier ban Earlier ban Ambiguous Later ban Earlier ban Earlier ban

banned, the Commentaries do list, as of 1848 (and perhaps a little bit earlier allowing for research and publication lags), the states that still allowed imprisonment for debt in some form, and those that had banned it. The 1866 list in the Commentaries reproduces the same text. Kent’s typology of states allows us to investigate the characteristics of those states that had banned imprisonment for debt by the 1840s – we will call these “early ban states” – with states that did not ban imprisonment for debt until much later. Obviously, an econometric analysis of the debt-banning decision cannot be too detailed, and the evidence presented here will necessarily be more suggestive than definitive. It is important to remember that there was some variation across states in who could be imprisoned for debt, and what exactly it meant to be imprisoned for debt. We noted above that Southern states, in particular, defined “imprisonment” more loosely compared to Northern states, where, more often than not, “imprisonment” can be taken literally. In spite of this caveat, we include all states in the analysis, where any state with an active debt imprisonment law at one time in the sample is included. Where possible, we shall augment the analysis to control for possible differences in the nature of imprisonment. To identify independent variables that might be used to explain the incidence of early versus late banning of imprisonment for debt, we will rely primarily on state-level data from the 1840 census. Although this greatly limits the set of available independent variables that we might use as there is very little data in the 1840 census, it nevertheless represents a considerable improvement in both quality and quantity of data over earlier U.S. censuses. The nature of the available data and their relationship to the hypotheses described earlier in the paper are presented in Table 5. The variables in the table can essentially be broken into categories – those that we think proxy for the importance of commercial interests, and those that may function as proxies for the costs of being informed and maintaining information about borrowers. Thus, we may view variables that have something to do with the dissemination of information as variables that measure the state’s ability to come up with an alternative to the debtors’ prison, which should rely on the capacity of borrowers (or the state) to maintain and develop information about potential borrowers. As we shall see, one of the variables that functions as a good predictor of whether or not a state banned the use of debtors’ prisons earlier rather than later is the state of the publishing industry, whether measured by the number of publications – daily, weekly, or semi-weekly newspapers and periodicals per capita – or by statewide investment in the printing industry. On one hand, this may in fact be a good proxy for the presence on the part of lenders of the ability to maintain stores of information about potential borrowers and their assets, but on the other, it might just as well be measuring the ability of anti-imprisonment propagandists to reach and convince the population of their case. (As we noted above, newspaper accounts were an important vehicle for reformers.) While we cannot distinguish which of these factors is really at work, we can say a bit more about the form and content of the typical newspaper of the period. To this end, we investigated the content of a sample of 126 available newspapers published in the period 1830–1850.26

26 We took this course of action because available secondary sources (e.g., Emery et al., 2000) have almost nothing to say about the format and content of early newspapers, instead focusing on other issues such as the quality and nature of journalism. The sample we used is unscientific. All newspapers are weighted equally regardless of importance or circulation, and the factors that led to their preservation relative to other newspapers are unknown. The sample was obtained by looking at actual copies of the newspapers via the website www.newspaperarchive.com. We checked all states and towns east of the Mississippi river for available issues of newspaper from the period. If a newspaper published at least one issue between the years 1830 and 1850 that was available on the website, we selected the issue closest to January 1st, 1839 and looked at its content. Only one southern state – Virginia – had newspapers available from this time period. A Microsoft Excel spreadsheet containing this data is available from the authors on request.

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

225

Table 6 Newspaper advertising relevant to debtors and creditors (1830–1860).

Classified advertising Public auctions Mortgage sales Sheriff’s auctions Notice of attachment, chancery, or assignment Notice of insolvency or bankruptcy At least one of the last three categories

Full sample (n = 130)

Wisconsin excluded (n = 63)

126 96.9% 29 22.3% 27 20.8% 9 6.9% 15 11.5% 11 8.5% 46 35.4%

62 98.4% 23 36.5% 15 23.8% 5 7.9% 11 17.5% 7 11.1% 29 46.0%

Notes: Sample was obtained by looking at all available historical newspapers appearing on the website www.newspaperarchive.com. One issue of all papers publishing at least one edition between the years 1830 and 1850 was examined for content.

The typical 1840 newspaper, almost without exception, was four pages long. While some papers were unapologetic about their political leanings (five papers in the sample contained the word “Whig” in their title), content was not that much different than a modern newspaper – the typical period paper carried stories about politics, supreme court rulings, wars, editorials, and news from other states. Papers also carried extensive classified advertising, which occupied anywhere from one-half to two pages of the paper. Notably, classified ads often contained information directly relevant to borrowers or lenders.27 We have tabulated some rough content measures in Table 6. Summary statistics for these measures are broken into two categories on the table. The first column contains data from the full sample, while the second excludes data from Wisconsin newspapers; for unknown reasons, a rather large number of Wisconsin papers were available on to the website. From the table, one can see that most papers carried classified advertising. Classified advertising could, if nothing else, help a creditor in possession of debtors’ assets dispose of them. Papers in the sample also frequently carried news about public auctions and mortgage sales. Perhaps more surprisingly, newspapers apparently carried much information on who had defaulted on debts. For example, sheriff’s auctions, which occurred at the behest of the court when property needed to be liquidated to satisfy a court judgment, were sometimes announced in papers, as were announcements that an individual’s property had been attached or assigned as the result of debt default. Papers also published notices of individuals’ insolvency or bankruptcy. Overall, roughly 35% of the papers in the sample contained some reference to a sheriff’s auction, an attachment notice, or a notice of insolvency or bankruptcy.28 T o give an example of a typical attachment notice, consider the following item, published in the Guernsey Jeffersonian on June 2nd, 1844, a Washington, Ohio newspaper: “At my residence, a writ of attachment was this day issued by J.P. Cunnard, a Justice of the Peace. . .against the goods. . .and effects of William Stewart. . .” An example of a typical insolvent’s notice appeared in the Phoenir, published in Green Bay, Wisconsin on October 8, 1841: “On petition of Jacob McKinney praying to be discharged under his debts under an act entitled ‘an Act for the Relief of Insolvent Debtors. . .”’ The September 16, 1829 edition of the Onandaga Standard, of Syracuse, New York published approximately 20 notices of insolvency. A typical example is: By order of Otis Bigelow, Esq. a judge of the court of common pleas in and for the county of Onandaga – notice is hereby given to all the creditors of Jonathan Holden. . .an insolvent debtor, to shew[sic] cause, if any they have before the judge at his office. . .why an assignment of the said insolvent’s estate should not be made for the benefit of all his creditors, and his person be exempted from imprisonment pursuant to an act entitled ‘an act to abolish imprisonment for debt in certain cases,’ passed April 7, 1819. The act cited in the above passage refers to a modification of the law of imprisonment for debt, barring imprisonment for debtors owing less than $10 in New York State. While it appears that newspapers carried a bevy of information relevant to creditors and debtors, there were undoubtedly other factors that contributed to a state’s decision of whether or not to ban debtors’ prisons. We therefore introduce additional available information that might also correlate with imprisonment for debt. Table 7 presents some summary statistics for the indicated variables, both in aggregate, and broken down by whether or not the state banned imprisonment for debt earlier or later, according to Kent’s criterion. One can begin to get a feel from Table 7 which of the variables described in Table 5 appear to vary systematically between states that were deemed by Kent to have enacted relatively earlier bans on

27

The most commonly advertised item by far was patent medicine; some papers even had classified sections wholly devoted to patent medicines. To cite one rather stark example, roughly half of the four pages of the July 2nd, 1842 edition of the Wiskonsan Enquirer, published in Madison, Wisconsin, were occupied by bankruptcy notices. 28

226

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

Table 7 Summary statistics. Variable Year of statehood Population density Fraction population urban Southern state dummy variable Percent voting Whig (n = 25) Newspapers and periodicals per 10,000 pop. Engineers, professionals per 1000 pop. Illiterate whites 25 or older, per 100 pop. Employment in publishing per 1000 pop. Per capita investment in printing industries Employment in navigation per 1000 pop. Groceries, dry goods stores per 10,000 pop. Per capita investment groceries, dry goods Trade and commission houses per 1000 pop. Employed in commerce, per 1000 pop. Per capita investment, trade and commission Employment in navigation per 1000 pop.

Early adoption (n = 17) 1807.41 (23.57) 28.72 (25.89) 9.55 (11.17) 0.29 (0.47) 53.44 (4.85) 1.17** (0.38) 4.64*** (0.44) 3.27** (2.78) 0.71** (0.37) 0.39*** (0.21) 6.27 (8.68) 38.24 (15.27) 15.92 (7.99) 2.88 (3.14) 8.12* (5.40) 8.23 (12.23) 6.27 (8.68)

Late adoption (n = 12) 1802.75 (17.87) 23.89 (27.06) 12.80 (8.73) 0.42 (0.51) 52.69 (6.61) 0.83** (0.38) 3.17*** (0.25) 6.14** (3.06) 0.41** (0.28) 0.18*** (0.09) 4.86 (7.07) 33.19 (19.71) 13.63 (5.93) 2.04 (2.46) 5.19* (0.79) 4.14 (5.31) 4.86 (7.07)

Full sample (n = 29) 1805.48 (21.17) 26.72 (26.01) 9.21 (11.66) 0.34 (0.48) 53.11 (5.58) 1.03 (0.41) 4.03 (1.64) 4.46 (3.19) 0.59 (0.36) 0.30 (0.20) 5.69 (7.95) 36.15 (17.10) 14.97 (7.19) 2.54 (2.87) 6.91 (4.66) 6.54 (10.04) 5.69 (7.95)

Notes: Asterisks denote significance levels in two-sample T-tests. Significant differences appear in bold type. All T-tests were performed assuming unequal variances. * 0.10 significance level. ** 0.05 significance level. *** 0.01 significance level.

imprisonment for debt. Two things stand out from the table. First, it appears that states with better-developed and more important publishing industries were quicker to abolish imprisonment for debt. This is evidenced by the fact that states that banned imprisonment for debt earlier apparently tended to have relatively larger newspaper, publishing, binding, and printing sectors, regardless of how one measures these variables. Table 7 further reveals a tendency for states with relatively more professionals and engineers, and smaller illiterate populations, to favor earlier bans on imprisonment for debt. This could likely be the case for two reasons. Most plainly, more educated populations are more likely to be up on current events and debates, and debates over imprisonment for debt were common in the early part of the 19th century. An additional tendency is for the relative numbers of people employed in commerce to correlate positively with early banning of imprisonment for debt. This is likely because the development of commerce is strongly correlated with the overall development of the economy and of more effective means both for screening debtors’ ability to repay and for ensuring timely repayment. Table 8 presents some results of cross-sectional logistic regressions in which the dependent variable is whether or not the state banned imprisonment for debt in or around the year 1840, or whether it banned imprisonment much later, once again according to the criterion of Kent. These results largely bear out what we learned from inspection of Table 7. The first thing to notice from Table 8 is that all the regressions include the per capita investment in publishing, printing, and binding. This is because a little experimentation reveals that, of the several proxies we described in Tables 5 and 7 to capture the state of the publishing industry, this one consistently outperformed the others. Moreover, in a situation in which degrees of freedom is an issue, parsimony is important. The logistic regression results continue to include other kinds of proxies in light of this, and the lesson is that most of them – even those that Table 7 suggest might help in predicting the incidence of early imprisonment bans – do not bring much to the empirical model that is not already captured by the per capita investment in the publishing industry. In every model we estimated, this variable was strongly positive and significant, indicating that those states with a more active and developed publishing industry were more likely to enact early bans on imprisonment for debt.29 As suggested, this result makes sense in light of our hypothesis. We hasten to add, however, that they are also consistent with other explanations for the demise of debtors’ prison. As noted, it could reflect the use of newspapers for lobbying against imprisonment by reformers. Additionally, the development of the publishing industry could be correlated with the degree of urbanization or the level of income, both of which may be related to the abolition of imprisonment for debt.30

29 One might be concerned about the potential for endogeneity in these specifications; that is, the possibility that the nature of the debtors’ prison law influenced the development and use of the publishing industry. Of course, given the small sample size and the lack of available instruments, it is difficult to formally rule out this possibility. In spite of these misgivings, we did estimate several models with GMM using as instruments some of the variables listed in Table 5. The results were essentially unchanged. 30 A further possibility is that there may have been some contagion effects across state borders. To control for this possibility, we attempted estimation of a spatial logit model using GMM as outlined in Klier and McMillen (2008), in spite of our small sample size. The results of this effort were inconclusive.

Per cap. investment, publishing Population density Fraction population urban Per cap. engineers, professionals Southern state dummy Year of statehood Per cap investment, groceries Per cap. navigation employment Per cap. investment, trade % Whig vote, 1840 Constant Log-likelihood Psuedo-R2 N Notes: Z-statistics in parenthesis. * Significant at 10%. ** Significant at 5% *** Significant at 1%

A

B

C

D

E

F

14.19*** (2.79) −0.03 (−1.46)

23.77** (2.46)

8.66** (2.55)

12.77** (2.43)

21.54*** (2.98)

12.72*** (2.87)

10.92*** (2.58)

−0.20* (−1.70) 0.49* (1.85) 0.45 (0.39) 0.04 (1.23) −0.23** (−2.41) −0.00 (−1.02) −0.05 (−0.86) −2.35** (−2.19) −13.16 0.33 29

−3.64** (−2.29) −10.99 0.44 29

−3.75** (−2.43) −12.95 0.34 29

−66.97 (−1.27) −13.07 0.34 29

−1.41 (−1.39) −12.17 0.38 29

−2.47** (−2.50) −13.74 0.302 29

−0.01 (0.15) −1.86 (−0.41) −11.82 0.311 25

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

Table 8 Logistic regression results: dependent variable = early abolishment of imprisonment for debt.

227

228

M.J. Baker et al. / Journal of Economic Behavior & Organization 84 (2012) 216–228

5. Conclusion The idea of debtors’ prisons, on its face, seems irrational. How, one asks, can a borrower be expected to pay back a loan when imprisoned? Yet debtors’ prisons have flourished throughout history and were even imported into the United States, where they endured until the middle of the nineteenth century. This suggests that they must have served some useful function in the operation of early credit markets. The analysis in this paper has sought to explain the economic reasons for their existence, and ultimately, for their demise. While the imprisonment of debtors no doubt elicited some amount of repayment of delinquent loans by benefactors of the debtor, we argued that its primary function was to deter default in the first place by giving borrowers an incentive to disclose hidden assets that could be used for repayment. Imprisonment was costly, however, and was therefore destined to be replaced by a more efficient means of preventing borrowers from sheltering assets. Although most states repealed debtors’ prison laws in the decades prior to the Civil War, by then such actions were largely symbolic since the practice had already substantially declined under its own economic weight.31 Empirical analysis of state laws banning imprisonment of debtors in the United States provides some support for the economic argument. In particular, the results suggest that those states in which the publishing industry developed sooner, thus facilitating the flow of information, were more likely to enact early bans of imprisonment for debt. Acknowledgement We acknowledge the helpful comments of two anonymous reviewers. References Allen, D.W., 2012. The Institutional Revolution: Measurement and the Economic Emergence of the Modern World. The University of Chicago Press, Chicago and London. Becker, G., 1968. Crime and punishment: an economic approach. Journal of Political Economy 76, 169–217. Bolton, P., Dewatripont, M., 2005. Contract Theory. Cambridge, MA, MIT Press. Bester, H., 1987. The role of collateral in credit markets with imperfect information. European Economic Review 31, 887–899. Coleman, P., 1999. Debtors and Creditors in America: Insolvency, Imprisonment for Debt, and Bankruptcy, 1607–1900. Beard Books, Washington, DC. Dickens, C., 2006 [1850]. David Copperfield. Bantam, New York. Emery, M., Emery, E., Roberts, N., 2000. The Press and America: An Interpretive History of the Mass Media. Allyn and Bacon, Boston. Ford, R., 1926. Imprisonment for debt. Michigan Law Review 25, 24–49. Gale, D., Hellwig, M., 1985. Incentive-compatible debt contracts: the one-period problem. Review of Economic Studies 52, 647–663. Golden v. Prince, 1814. 10 Fed. Cas. 542, no. 5509 C.C.D.Pa. Holton, W., 2007. Unruly Americans and the Origins of the Constitution. Hill and Wang, New York. Kent, J., 1848. Commentaries on American Law, vol. 2., 6th ed. William Kent, New York. Kent, J., 1866. In: James Comstock (Ed.), Commentaries on American Law, vol. 2, 11th ed. Little, Brown and Co., Boston. Kinne Asa, 1842. The Laws of the Different States and Territories of the United States on Imprisonment for Debt. W.E. Dean, New York. Klier, T., McMillen, D., 2008. Clustering of auto supplier plants in the United States: generalized method of moments spatial log it for large samples. Journal of Business and Economic Statistics 26, 460–471. Levitt, S., 1997. Incentive compatibility constraints as an explanation for the use of prison sentences instead of fines. International Review of Law and Economics 17, 179–192. Mann, B., 2002. Republic of Debtors: Bankruptcy in the Age of American Independence. Harvard Univ. Press, Cambridge, MA. McMaster, J., 1920. A History of the People of the United States, From the Revolution to the Civil War, vol. 8. D. Appleton and Co., New York/London, pp. 1841–1850. Ogden v. Saunders, 1827. 25 U.S. 213. Newspaper Archive, 2011. http://www.newspaperarchive.com. (accessed February 19–21, 2011). Polinsky, A.M., Shavell, S., 1984. The optimal use of fines and imprisonment. Journal of Public Economics 24, 89–99. Prison Discipline Society, 1831. Imprisonment for debt: fifth annual report of the board of managers of the prison discipline society. North American Review 32, 490–508. Prison Discipline Society, 1841. 16th Annual Report of the Prison Discipline Society. Prison Discipline Society, Boston. Rea, S., 1984. Arm-breaking, consumer credit and personal bankruptcy. Economic Inquiry 22, 188–208. Sturgis v. Crowninshield, 1819. 17 U.S. 122. Townsend, R., 1979. Optimal contracts and competitive markets with costly state verification. Journal of Economic Theory 21, 265–293. United States Census Bureau, 1975. Historical Statistics of the United States: Colonial Times to 1870, Bicentenial Edition, Part 2. U.S. Census Bureau, Washington, DC.

31 An alternative explanation for the need for states to formally repeal laws is that reliance on debtors’ prisons may bear some aspects of coordination failure. That is, it might have been the case that abolition of the laws served as a means to coordinate lenders on the socially more desirable enforcement mechanism based on better public information.