Do legal origins matter? The case of bankruptcy laws in Europe 1808-1914

Since the early 1997 paper by La Porta et al. , a growing body of research has argued that ‘legal origins’ have a country-speciﬁc, time-invariant effect on property rights and economic development. Following the methodology of La Porta et al. , an original database of 51 bankruptcy laws has been built: it ranges over 15 European countries and more than a hundred years ( 1808 – 1914 ), and summarises how the rights and incentives of the parties were deﬁned as the procedures unfolded. The ﬁrst conclusion is that, over the entire period, all legal traditions strongly protected creditors’ rights; only English law comes out prima facie as less protective. Second, evidence suggests that the evolution of these laws was inﬂuenced less by their past than by continent-wide trends, arguably linked to capitalist development. An early nineteenth century model thus saw heavy repression of failed debtors and highly regulated judicial procedures. After a transition period from the late 1860 s to the late 1880 s, prison for debt was abandoned, rehabilitation became easier, and the parties were given much more room to recontract on property rights.


Introduction: 'legal origins' and the evolution of bankruptcy law
Over the last ten years, a growing body of research has emphasised the impact of legal and judicial structures on economic development. The basic intuition stems from standard neo-institutionalist economics: property rights, the integrity of contracts, and the security of transactions matter for financial contracting, and hence for investment and growth. The main innovation brought about inter alia by La Porta, Lopes-de-Silanes, Shleifer and Vishny (hereafter LLSV) was to rely upon large cross-country databases in order to test these propositions empirically. However, in this attempt, the authors were confronted with a standard problem of endogeneity: institutions, including legal ones, could be shaped by the process of economic development, rather than being a shaping factor, so that they would not help in accounting for variations across countries. The solution was to use each country's 'legal origin' as an exogenous variable: that is, their belonging to either the common law tradition or to various currents of continental law (French, German, Scandinavian, or Socialist). As they measured the quality of legal institutions, LLSV indeed observed strong covariance within these subgroups of countries. And when included in regression equations, this variable came out as a significant factor in explaining different economic outcomes: bank intermediation, stock market capitalisation, the availability of equity finance, and so forth. 1 Already, in their first joint paper, LLSV (1997) concluded that, other things being equal, common law countries protect property rights better and draw economic benefits from this; they are followed by, respectively, Scandinavian, German, and finally French law countries. Comparable conclusions were reached in later papers, and by other authors, that focused on an ever-increasing array of variables, though they generally came up with the same hierarchy of 'legal origins '. 2 There is no question that bringing legal history into the economic debate was an important and welcome innovation. Neither will anybody contest that legal history belongs to the very long run; as stated by Glaeser and Shleifer (2002), the canonical opposition between English and French law was already observable in the Middle Ages. 3 Yet the argument's main difficulty derives from the econometric and analytical interpretation of this very notion: 'legal origin' is used as a country-specific, time-invariant parameter that is expected to have permanent effects on institutions and on economic performance. This assumption is actually required if the endogeneity problem is to be solved neatly. 4 Hence the underlying paradox. Whereas the overall approach reeks of Northian historiography, the actual use of history, or duration, is profoundly ahistorical. It does not and cannot account for phases or cycles in economic or political development, and neither can it be assimilated to an extreme form of path dependency. Nor are these propositions consistent with the standard assumptions of the law and economics school which state that the law is shaped by market forces. On the contrary, 'legal origins' are supposed to have emerged in a given historical context and to have then crystallised: they are interpreted as some essential hard-core identity, which would lie beyond the reach of either economic or political competition. 5 This interpretation obviously raises concerns. To start with, no empirical evidence has been presented that would support it. Historians have certainly identified different legal and institutional patterns, over subperiods, that may have had diverging economic consequences. But this is different from identifying permanent biases that would have tangible and predictable economic effects over all countries and centuries. A second objection is that the actual indices used as a proxy for those rights in the econometric regressions are very narrow. 6 For instance, LLSV in their early seminal papers choose bankruptcy laws as a representative creditors' rights institution. 7 And within these laws, the authors identify four critical items that are supposed to reflect the guarantees generally offered to creditors under the respective procedures. This highly discriminating approach certainly goes against the grain of mainstream legal theory, which defines property rights as a complex bundle of rights, rather than a neatly defined, positive endowment. It also makes such rights more vulnerable to countryor period-specific patterns. Can, for instance, the criteria chosen by LLSV be applied to other historical periods? Or should they be viewed merely as a present-day expression of a more fundamental, underlying reality? How should that reality then be identified?
This article presents an original attempt at testing the proposition that legal traditions are a time-invariant, country-specific variable that can actually work in a purely exogenous way in economic development. Following on LLSV, this article focuses on bankruptcy law in Europe during the nineteenth century: 51 legal acts, statutes and codes adopted between 1808 and 1914 have been collected and coded. They offer a lot of a priori variation across time and countries: all Western legal traditions are represented in the sample; the period under review witnessed large-scale economic and institutional 5 A number of authors have presented this thesis in a less straightforward manner. Berkowitz et al. (2003), for instance, argue that the impact of 'legal origins' is contingent on whether their adoption is voluntary or not (that is, colonial). Djankov et al. (2003b), though arguing from the standard LLSV viewpoint, emphasise that there is room for a country-specific trade-off depending upon, say, its degree of development. From a mostly theoretical viewpoint, Ayotte and Hayong (2004) defend a comparable idea though with a different conclusion: developing countries should adopt more regulated or formalised law than developed ones (see also Berkovitch and Israël, 1999). Standard opponents to the students of 'legal origins' often argue that political (rather than legal) institutions are the key when defending property rights; see, for example, Rajan and Zingales (2003), Acemoglu (2003), and of course Marx and Engels (1848). 6 See Graff (2005) for a critique of LLSV methodology. 7 Note that, historically, bankruptcy law in England and the United States stems from statutory law, whereas case law has never produced a coherent body of rules on this issue: the only major exception in this respect is the US equity receivership, which emerged in the late nineteenth century (see Skeel 2001, Martin 1974 changes; and many reforms have been adopted in all countries. In the majority of cases, the primary text has been consulted (often in translation), and in the other cases, nineteenth century legal treaties and textbooks were used. 8 The contemporary legal literature also helped to identify the main themes in the policy debates of the time. A series of simple, LLSV-type indices could thus be designed that reflect actual trends in the evolution of bankruptcy laws over time and across countries. The notion of time-invariant patterns may thus be thoroughly tested against alternate hypotheses. For instance, nineteenth century lawmakers may have primarily attempted to address pragmatically emerging problems or the demands formulated by social actors. In so doing they would, of course, be forced to deal with existing legal institutions and professions, as with a specific legal grammar: this is where path-dependency stems from. But it may be as well that these variables did not have a tangible impact on how the eventual solutions actually worked. Perhaps evolutions observed across countries are less reflective of past institutional legacies than of the explicit attempts by lawmakers to affect microeconomic behaviours.
This approach, which has no equivalent in terms of method or scope, brings about two main conclusions. First, under LLSV's own limited criteria, creditors' rights during bankruptcies were strongly protected by law during the whole century and in all countries, whatever their legal tradition; only English law comes out prima facie with a weaker performance. As a rule, the literature of the time indeed suggests that a bankruptcy code which did not aim primarily at protecting creditors' rights would have just been pointless. Second, when one goes beyond the specific features selected by LLSV, the evidence does not confirm the notion that national traditions have exerted an exogenous and permanent (that is predictable) effect on successive, national bankruptcy reforms. In fact, policymakers did not fight much about the balance between the respective rights of creditors and debtors. A more serious issue was the trade-off between, on the one hand, the judicial guarantees to existing, pre-bankruptcy rights, and on the other the constraints on the parties entailed by judicialisation, when recontractingfor instance when negotiating on a continuation agreement.
Two successive, continent-wide models are thus identified, that cover respectively the first half of the period (1810s-1860s) and the latter decades under review (1880s-1910s). The first model was strongly repressive vis-à-vis the failed debtor while imposing strict procedural rules on the parties when negotiating. The second paradigm was then characterised by a much more relaxed approach on both counts: apparently, the threats to debtors' and creditors' rights had become less pronounced. Legal traditions show up only against the backdrop of these continent-wide trends.
Section 2 briefly discusses some analytical issues at stake in bankruptcy laws and Section 3 presents the dataset. Section 4 discusses how European countries performed under the LLSV criteria of creditors' right and then presents an enhanced index, adjusted to nineteenth century realities. Section 5 and 6 follow with the evolution of, respectively, the status of the failed debtor and the rules of renegotiation offered to the parties. Section 7 identifies the main hypotheses that could account for cross-country convergence patterns. Section 8 concludes.

On rights, procedure, and environment
Bankruptcy is a complex and multifaceted institution that is not easy to describe with a limited set of discrete variables. What market outcome could actually be more confusing and dangerous than one where all debt contracts are broken, where the time horizon of actors may shorten dramatically, and where the disciplining effect of ongoing market transactions is extinct? Default and insolvency can actually be thought of as a paradigmatic example of a contractual failure that immediately raises major threats of a breakdown in collective action. If creditors are able to run on the physical assets or to freely pursue individual judicial remedies, the problems of common pool, or prisoners' dilemma may rapidly become uncontrollable: a tragedy of the commons is never far away when a private business with more than one creditor fails. The end result may be altogether highly inequitable as regard the distribution of capital losses among individual creditors, and economically sub-optimal for them as a whole, especially if there are 'going concern' issues. 9 Hence the call for the intervention of a benign, Lockean sovereign, who would guarantee an orderly and efficient allocation of capital losses and help restore the conditions for secure contractual exchange. But in order to support such a settlement it will have to infringe upon the rights the parties will have and to redefine them, both individually and collectively. 10 Binding rules of interaction will be imposed upon them, as they will be transferred into a non-market, procedural framework for collective negotiation; agents and officials will have to be dealt with and possibly monitored; remaining assets will be seized and possibly managed by an interim agent; lastly, individual choices will be substituted by qualified majority votes and judicial confirmation, that may forcibly reallocate property rights against the will of dissenting, minority creditors.
How far the Lockean state enters into contractual matters and regulates the post-default, non-market interactions between the parties can be thought of as the first main issue in the history of bankruptcy laws. Though wide variations are observed in this respect, at no point in modern times has the sovereign been taken completely out of the picture. In fact, there has never been such a thing as a fully privatised bankruptcy procedure, a point that suggests that sanctioning insolvency is not endogenous to markets per se, but should be the eventual act of an agent of the sovereign -say a bankruptcy judge.
Of course, here as elsewhere, rent-seeking behaviours by stakeholders, officials and legal professions have been quite common. During the twentieth century, non-contractual, special interests have often held sway over the rights of creditors. But the defining element of this institution is neither its quasi-fiscal dimension, nor its use as an instrument to affect market outcomes. Rather bankruptcy is part and parcel of any competitive, openmarket, capitalist economy and tends to emerge as a policy issue when the economy is being liberalised, rather than when evolving in the reverse direction.
The interaction between this institution and its overall environment is thus a second, defining issue. The main underlying theme here is probably the more or less extended risks of moral hazards and opportunistic behaviours borne by this environment. Take for instance the quality of the signal for entering the procedure -that is, when the debtor firm becomes illiquid and should stop trading. If the payment system is poorly regulated and has a limited capacity to manage aggregate liquidity, due for instance to the absence of a modern Central Bank armed with a lender of last resort function, then viable firms may be pushed into default too rapidly; conversely, overextended firms may also survive too long when creditors are exposed to large asymmetries of information.
Exiting the procedure also raises serious problems, that bear on how bankruptcy works and the incentives it creates for agents. Since the Middle Ages, for example, liquidation has typically called for an open auction of assets against cash, which might not be readily available. Well into the nineteenth century, owners of large plantations in the American South could survive insolvency for decades, just because of the impossibility of liquidating their stock of assets (lands and slaves) at a viable price; Thomas Jefferson is a well-known example. Conversely, the success of the present-day Chapter 11 in the United States cannot be accounted for unless the huge development and sophistication of capital markets is brought into the picture (buy-out finance, markets for mergers and acquisition, distress firm finance, and so on). By the same token, the slow resolution of the insolvency crisis in Japan, since the early 1990s, was also conditioned by the creation of a market that would help to dispose of large stocks of real assets. 11

A database on European bankruptcy laws, 1808-1914
In order to analyse the evolution of national bankruptcy laws over the nineteenth century, 51 legal texts adopted in fifteen European countries have been collected (Appendix 1). 12 The series starts with the 1808 French Code de commerce, which was the main source of influence on the continent until mid-century, and persisted until World War I. 13 Some countries, such as Belgium, the German Reich and Italy did not exist at the beginning of the period; but Prussia and the Kingdom of Piedmont are included in the database. Other countries did not have a unified bankruptcy law before midcentury, such as Norway, Finland, and a large part of the Northern German Confederation (outside Prussia); and in some cases only partial information was available. Two Europe-wide cycles of bankruptcy reforms can then be identified: first a minor upswing during the 1840s, then much more activity during the 1870s and 1880s.
The primary legal texts were used when accessible (often as a translation); as a second-best, a substantial number of nineteenth century legal treaties, commentaries and textbooks were used. Features of each law have been coded in a series of 0/1 digits reflecting how it defined both the rights of the players and the rules of the game. This database, however, reflects exclusively a formalistic and comparative history of the legal texts. It does not include any material relating to the social or economic history of bankruptcy, or on how agents actually interacted with this institution. Though this is certainly a restrictive approach, there is no doubt that bankruptcy laws were widely 12 The history of bankruptcy laws has attracted increasing attention since the 1980s, though this literature deals mostly with the US experience. At one end of the spectrum are various trends in cultural history that often centre on the 'moral economy' of debt and default (Weiss 1986, Finn 2003, Mann 2003, Anderson 2004, plus various approaches in the social history of failure (Duffy 1985, Hoppit 1987, Lester 1995. Other authors have focused on the actual working of the institution during specific episodes: for instance, the short life of the second American federal law in 1841-1842 (Balleisen 2001), the role of the third federal law (1867-1878) in the economic reconstruction of the South (Thompson 2004), or the political economy history of the more recent 1978 US law (Posner 1978, Carruthers andHalliday 1998). Before that, some early (mostly descriptive) works contributed to opening the field and identifying the main issues at stake; see Warren (1935) and Coleman (1974). A classic example is the conflict on bankruptcy reform and land exemption that opposed the rural West and the financial centres of the Northeast. In this perspective, Howard Rosenthal and his colleagues have provided new insights into the political economy determinants of bankruptcy reforms in the United States (Domowitz and Tamer 1997;Rosenthal 2000, 2004;Nunez and Rosenthal 2002;Berglöf, Rosenthal and von Thadden 2001 which also extends to some European experiences). This approach has been extended by David Skeel (2001) to twentieth century trends and with an often close analytical language; Hansen and Hansen (2005) also follow these lines, though emphasising as well changes in the perception of the law, over time. 13 Austria, Belgium, Denmark, England, France, Finland, Prussia/Germany, Hungary, Italy, The Netherlands, Norway, Portugal, Russia, Spain, and Sweden. Greece, Malta, and Switzerland are mentioned occasionally but are not included in the dataset. debated at the time and that reforms generally mobilised a large array of interests and opinions. In other words, agents certainly did not consider the law as unconsequential, whatever their actual practices. The starting point is the four items chosen by LLSV (1998) to measure the protection of creditor's rights during bankruptcy. Namely, they consider that creditors' rights are better protected if: (1) The management does not stay during the reorganisation procedure; (2) The management cannot seek protection from creditors unilaterally; (3) Reorganisation procedures are not associated with an automatic stay; (4) The rights of secured creditors are protected during reorganisation.
A primary aim was thus to replicate or approximate this simple additive index, which ranks the protection of creditors' rights from 0 (minimum) to 4 (maximum). When working on nineteenth century texts, however, problems rapidly arise from the differences in economic and institutional contexts: many policy issues of the time fall just outside LLSV's quality criteria. Most obviously, bankruptcy laws concerned mostly personal entrepreneurs and small firms, rather than incorporated businesses, and this naturally affects many agency problems.
Take the case of going concern issues -for instance whether and how the firm can operate during the procedure. This point emerged only in the late nineteenth century, typically in the case of railway companies. 14 Before that, the principal/agent problem during the procedure was secondary: the priority for creditors, when a default was known, was to immediately take control of the remaining assets so that the bankrupt would not hide them, secretly transfer them or agree on side-arrangements with preferred creditors. Indeed, seizing the assets or the body of the debtor were long considered as two principal avenues to protect creditors against opportunism. On the other hand, how reorganisation plans were decided was certainly a major issue. But the core question in this respect was not whose voice would be decisive -there was no uncertainty on this -but rather the balance between the judicial guarantees offered to the parties and the relative autonomy they were left with when negotiating.
Thus, two main analytical issues are documented by the database, beyond those directly derived from LLSV's analysis. First is how the law addressed the risks of moral hazards, on the side of both the debtors and the creditors, in a ratherà la LLSV, literal sense. Second, the emphasis is on the rules of interaction between the parties and the autonomy allowed to them, when bargaining or recontracting; or, again, on the balance between the guarantees offered by the judicial procedure and the constraints it imposed on the parties.

LLSV and the quality of nineteenth century bankruptcy laws
If LLSV's four criteria are taken literally, then measuring the quality of European bankruptcy laws during the nineteenth century is neither difficult nor very informative: creditors' rights were strongly protected in all countries during the whole period. As already stated, a bankruptcy law that did not have this primary aim would have been considered pointless. On that basis, almost all countries would have probably received either a 3 or 4 mark, depending upon how the criteria are interpreted. Let's look in more detail at how they applied in the nineteenth century context. First, as a universal rule, a trader would lose control over, and often the legal possession of, his assets (personal and commercial) on the day his bankruptcy was declared. But he would also lose the legal capacity to trade -that is to sell and buy, pay and borrow, and so forth; even a fresh start would not be possible unless he were rehabilitated (often a difficult feat). 15 In this sense, there is no way that entering bankruptcy would have protected managing rights. Only two partial exceptions can be observed: first some early-century cases of judicially controlled individual moratoria; and a series of out-of-bankruptcy frameworks adopted at the end of the century that allowed the parties to negotiate under some judicial oversight though without incurring the financial and social costs of entering a full-blown bankruptcy procedure. If LLSV criteria are taken literally, these options actually weakened creditors' rights. But at the time, as will be discussed later, they were explicitly envisaged as efficiency-enhancing options, that would support co-operation and early entry into negotiations.
Orthodox conclusions are reached again on the matter of secured creditors: nineteenth century laws emphatically protected their rights during bankruptcy. As expected, the sole contentious issue is that of the privileged (that is, statutory, non-contractual) claims of third-parties on the bankrupt's estate: claims owned by the Treasury and churches, workers and servants, doctors and pharmacists, et cetera. But this was mostly a threat for junior creditors, who are not the focus of LLSV. Even though measuring the extent of such privileged claims is hard, 16 the legal and policy literature of the time does not suggest that they were instruments for funnelling large 15 In other words, the opposition between manager-driven and manager-displacing bankruptcy law did not work during the nineteenth century; see Skeel (1998) as well as Armour, Cheffins and Skeel (2002). 16 The main difficulty encountered in this survey is that these clauses are most often to be found not in the actual bankruptcy text but in many different bodies of law -the fiscal and civil code, the laws on tenants and land lease, the emerging labour law, and so on. However, provisions for wages and rents would typically be limited to a year or eighteen months, with a ceiling on the amount being reimbursed. Debt moratoria, a common US practice during the nineteenth century, is also unknown in Europe (Alston 1984, Bolton and.
extra-contractual interests into the distributive machine that is bankruptcy. More generally, as regards the principle of absolute priority between investors, an interesting, albeit marginal, option was to offer some percentage in the proceeds of the liquidation to the benevolent, co-operative debtor: England, Würtemberg, and Malta had such a clause. 17 Literally taken, this prescription represented an explicit infringement of the creditors' rights, though again it was conceived as a pro-creditor incentive device.
The fourth LLSV argument, on stays, is more problematic. By definition, bankruptcy is a collective instrument of debt collection that substitutes for individual remedies when they threaten a loss of value for creditors as a whole. Bankruptcy is an extra-contractual institution that necessarily suspends or rewrites some private rights. Two common features of bankruptcy laws come up at this point. 18 (1) Stays on individual remedies during the procedure were general practice on the Continent, with some qualification in early century Austrian law, and some undocumented periods in Denmark for instance. Otherwise, as a rule, foreclosures were stayed when bankruptcy was declared, generally until liquidation was decided; when senior interest payments were dealt with, however, the law excluded them from the stay (that is, they were served during the procedure). England did not adopt the principle of a stay on private remedies until 1869.
(2) Qualified majority voting by all non-senior creditors was a standard feature of all reorganisation agreements. Typically, the vote would be counted both in terms of number of creditors and sums of claims, with majority thresholds of three-quarters and two-thirds respectively; judicial confirmation was conditional upon positive votes and was a prerequisite for the agreement to become binding. 19 A different issue arose when debt discharge could be imposed upon (some) creditors, as had been the case in England. Since 1702 a qualified majority of creditors had been able to relieve the honest and co-operative debtor of her debt after the assets had been auctioned off. This clause has generally been presented as reflecting a remarkable pro-business bias in English law, 20 though continental creditors could generally make the same decision. The 17 This incentive varied within 5-10 per cent, 5-8.3 per cent and 5-10 per cent brackets respectively. The reference in the case of Malta is a 1815 ordinance on civil procedure; for Würtemberg, see Saint Joseph (1844). 18 As a simplification, both options are considered here together. 19 No statute provided the judge with the right to impose an agreement on creditors if they failed to agree (as with the 'cram down' provision of the present US Chapter 11). 20 'Thus the bankrupt becomes a clear man again; and (. . .) may become a useful member of the commonwealth' (Blackstone 1811, p. 488). On the English debt discharge, see Holdsworth (1925), Tabb (1991), andMcCoid (1996). Note that courts also had discretion when confirming discharge. difference is that it would be part of a generally more comprehensive arrangement of a rather private nature, instead of being addressed at the end of the official procedure as a single-issue vote. 21 It may well be that eventual discharge was easier to obtain in England in the eighteenth and early nineteenth centuries; probably the rights to trade were likewise easier to recover. But as far as creditors' rights are concerned, the major divergence with continental practices only emerged in 1843: discretion on discharge was then transferred to the courts, with no veto power to the assembly of creditors. This step toward weaker property rights, as a counterpart to an easier fresh start for failed entrepreneurs, was never taken by any other country during the whole period under review.
In order to summarise these various elements, an index has been calculated that tries to adjust the LLSV variables to nineteenth century rules, while remaining as close as possible to their view of what should, and should not be a bankruptcy procedure. Selected items are the following (Appendix 2): 22 • Regulations on stays explicitly preserve interest payments on secured debt; • Opening negotiations on a reorganisation plan is not associated with a stay; • No money incentive to the debtor (rule of absolute priority); • No capacity by the court to declare a debt discharge. Figure 1 shows that creditors' rights were strongly protected during the whole period in all countries, whatever the respective legal traditions. Only English law comes out as an exception under the adjusted criteria, as its performance declined over time and ended the period clearly below average.

Fighting moral hazard: prison for debt and 'la mort civile'
Bankruptcy procedures aim at enforcing a rule-based distribution of residual assets, at a time when the incentives on all actors are to run and grab or to escape the consequences of commercial failure and capital losses. In past centuries the challenge arguably was magnified by acute problems of information and communication: contracts, accounting books, property titles, instruments of payment and judgements were all much less formalised than today and circulated much more slowly. The room for opportunistic 21 A corollary issue is the common law-specific debate on whether bankruptcy should be only involuntary (that is, initiated exclusively by creditors) or also voluntary (initiated by the debtor); see McCoid (1987McCoid ( , 1988. This dispute has no equivalent on the Continent, where both parties have traditionally had the capacity to initiate the procedure. 22 Items close to LLSV intuition, but that are not much differentiated across nineteenth century statutes, have not been included (for example, on issues of senior creditors). behaviour was thus arguably larger than in more recent periods, on the side of both the debtor and the creditors. Starting with the former, one can hypothesise that high transaction risks were a factor behind the strong repressive features observed in all early statutes, which indeed defined bankrupts as outright criminals: publicos ladrones y verdaderos robadores. 23 Apparently the protection of commerce and debt markets could not do without heavy-handed instruments of social discipline, whatever the costs for the proverbial 'honest but unlucky trader'. 24 The 1808 Napoleonic Code de commerce probably marked the high point in the reliance upon repressive instruments as a substitute for apparently insufficient market institutions: all failed debtors were jailed at least for a short exemplary period, and rehabilitation was highly conditional. Shame and infamy were part and parcel of contractual discipline. Remarkably, however, this bias was not specific to France or even to civil law countries: in all Europe (as in America), many debtors ended up in jail in the early nineteenth century. 25 This early pattern of convergence across countries was followed after a short transition period by a second one: whereas in 1866 no country had yet suppressed prison for debt, thirteen countries, of all legal traditions, had taken that step in 1877. At the time, many argued (in today's language) that moral hazard would become uncontrollable and that credit markets would decline. But they lost the argument, inter alia to humanitarian militants and to those who believed that prison unduly increased the risk of entrepreneurship, at a net loss for the economy. In other words, in most countries it was finally agreed that the unconditional, hard-headed defence of creditors' rightswhatever the instruments -might not always be consistent with economic development. Investors may be indeed willing to take a calculated risk if they assume that the underlying economic opportunity is worth it. A corollary is that one could indeed become insolvent in a fast-growing Darwinian market economy without one's civil and political rights being heavily affected. Rules governing the market place and the public space became more independent one from the other, as the market institutions strengthened.
The decline of repression as an instrument of market discipline is illustrated by an index that reflects the reliance upon prison for debt and the conditions for rehabilitation (see Appendix 3). The average, cross-country index indeed confirms the account of a twice-in-a-century convergence pattern across Europe (Figure 2) as in the case of the autonomy to renegotiate, discussed later. After having shared a repressive approach, most countries eventually agreed that commercial failure should not cause 'la mort civile' -provided the law had been respected. This evolution was, however, not unanimous; differences between legal families arose within this broad trend (Figure 3).
• England already had a tradition and reputation for allowing bankruptcies to have limited social costs (at least for entrepreneurs). As a rule, however, commercial rights were easier to recover than civic or political ones, which in England were still affected by bankruptcy at the end of the nineteenth century. On the other hand, prison for small debtors -now called consumer debtors -was still widespread in this country before World War I, in stark contrast to the situation in most Western countries. 26 25 See inter alia Sacré and Oudin (1874), Thaller (1887), Coleman (1974), Hoppit (1987), and Lester (1995). Note that imprisonment could be the penal consequence of fraud, but also an instrument in the hands of creditors in order to pressure debtors so that they would release their assets -that is, 'prison on mesne'. Imprisoned debtors then insisted on not being mixed with convicts and were often sent to a specific prison. See Mann (2003) on the late eighteenth century US experience. 26 Lester (1995).  • The landmark German 1877 code then provided a model for a bankruptcy law without any repressive feature, modelled as an almost purely procedural and problem-solving instrument. Its belonging to a civil law tradition and its large influence in neighbouring countries made this, in the eyes of many commentators, the true successor of the 1808 French code. 27 • France (together with Austria) was the first to put an end to prison for debt, but it then had more difficulty than others distancing itself further from the repressive Napoleonic turn: only in 1888 did France adopt a new status for lawful debtors with limited professional costs, but some political stigma remained until the early twentieth century. 28 More generally, it took more time on average for countries with French and Scandinavian law to evolve towards a more liberal model, though they did eventually converge.

Arrangements and the contractual autonomy of parties
Entering new contractual commitments with a once-failed entrepreneur clearly requires a leap of faith. But recontracting may also be highly beneficial if there is a 'going concern' value in the stock of assets; or, for instance, if the expected return of liquidation is brought down by the poor liquidity of the markets for property and capital goods -a common feature of early-capitalist economies. Given these type of incentives, one might expect that the tradeoff between recontracting and liquidation should be entirely for the parties to settle. 29 Yet, historically, transaction risks have always called for at least a degree of regulation. Some creditors may be better or earlier informed than others, or some may agree on discriminatory, hidden arrangements with the debtor, or the latter may reimburse some creditors before others, or social leverage may bear on transactions. Until the end of the eighteenth century, arrangements between debtors and creditors on the Continent had been mainly private affairs, subjected to qualified majority voting and rather light supervision (though often an increasing one, as in France); 30 this was a direct legacy of medieval Italian communal practices exported by Italian traders via the major European 27 '. . . une manifestation très sérieuse et probablement durable du génie juridique allemand' (Thaller 1887, p. 87). 28 Percerou (1935). 29 Jackson (1982). The terms 'arrangements', 'composition', 'continuation agreement' and 'reorganisation' are being used synonymously. 30 See Savary (1749), Denisart (1771), Renouard (1857), and Hilaire (1986)  fairs, such as in Lyons. 31 The 1808 Code then arrived as a countermodel: arrangements, now called concordats, were to be negotiated and concluded within a minute judicial process. The parties could certainly negotiate on reorganisation plans and then obtain confirmation, but the law stated most exactly when, where, and under what conditions they could speak up, and the creditors had limited oversight on how the assets were managed. This, however, was not perceived as an undue state interventionimprisonment for debt apparently raised much more protest. The Code de commerce, just like the more famous Code civil (1804), was indeed the legal foundations of a bourgeois, liberal society based on private property and contract, that would thrive in the forthcoming decades. In this context, the rationale for a highly regulated bankruptcy law was again to offer strong guarantees against fraud, dissimulation, and corruption -which were seen as the hallmarks of work-out techniques during the ancien régime and the Revolutionary years. There was indeed a demand for procedural formalism as a way to address inter alia inter-creditor equity concerns when recontracting. 32 The striking fact, however, is that again, in most other European countries, private parties interacted in procedures that did not leave them with much flexibility. As a rule, European traders in the first six decades of the nineteenth century could only bargain under the close control of judges and officials, within the long, costly and shameful single-entry process of bankruptcy. The only exceptions were the above-mentioned judicial stays, or individual moratoria, that aimed at addressing short-term liquidity problems. 33 Path-breaking reforms emerged only during the last quarter of the century, in a context marked inter alia by the emergence of big corporations, developed financial markets and growing problems of 'going concern'. In order to account for this evolution, an index of creditors' autonomy has been designed that adds six variables (Appendix 3). Its main aim is to reflect how the rules of the game between judicial institutions and private interests evolved, hypothetically toward a less intrusive and less constraining model. A first issue is whether judicial confirmation of arrangements was contingent 31 Hilaire (1986). 32 '. . . if, by a fatal negligence, the bankrupt and the debtors are allowed to cast off any [legal or procedural] provision, the aim of the lawmakers will be missed: fraud will come together with the impudence of impunity, it will seize the sanctuary of justice and flout its authority' (Laurens 1806, p. 152). On this period see also Renouard (1857), Picard (1910), and Percerou (1935). 33 As a rule, the rejection of this option was grounded again on risks of opportunism by either the debtor or some creditors; see for example Renouard (1857) or Füger and Wessely (1841). Legislation allowing such moratoria was introduced in France both in 1848 and 1871, but then rapidly withdrawn, though in the later case some jurisdictions in trading cities apparently kept sanctioning the practice for some years: see Silvian (1915).  upon substantive pre-conditions or the discretionary judgement of the court. Another is the possible shift toward a more differentiated institutional set-up, where agents are offered a menu of options, under more or less judicial oversight, as opposed to a single-entry bankruptcy process. Finally, the possibility of actively managing the assets during the procedure is added. The average, cross-European index of contractual autonomy again offers a bipolar view. The average country offered a low degree of contractual autonomy until mid-century before allowing more discretion to the parties after 1870 (Figure 2). Three countries that belonged to each, main legal tradition ( Figure 4) come out as major innovators: England, Belgium, and Germany.
England's emergence as a legal innovator is surprising. Although commonlaw countries are generally considered more supportive of market forces and institutional innovation, no statutory guarantee to arrangements was possible in England before the mid-nineteenth century. 34 Until that time, the law offered much less support to recontracting than both the pre-Napoleonic 34 A 1697 English act, allowing for qualified majority votes on arrangement, was abandoned one year later as a result of apparently extensive fraud and dissimulation (for quotations see Holland 1864, pp. 14-17). This implies not that private arrangements were unusual, but rather that ex post judicial confirmation seems to have raised considerable difficulties in England. Information however is scarce because these agreements explicitly aimed at avoiding publicity. For indications on eighteenth century practices see Hoppit (1987), and Lester (1995) as regards the nineteenth century. This anti-arrangement bias is also present in the United States, where the option was not introduced until the third federal and the Concordat approaches favoured on the Continent. 35 English creditors could choose only between unanimity accord under private seal and bankruptcy, which would exclusively lead to liquidation; as already stated, the only open question at that point was whether the bankrupt would be eventually offered a discharge on his residual debt. In both 1825 and 1849, two attempts at supporting concordat agreements had been failures: the majority threshold was too high (9/10 in sums in the former case), and the debtor had to relinquish most of his goods, despite the arrangement's aim of avoiding undue liquidation. The breakthrough came in 1861, when these restrictions were abandoned so that bankruptcy could be used also as an instrument for restructuring balance sheets. 36 The key policy question at that point was whether such accord should be negotiated within or outside the bankruptcy process per se, that is whether softer forms of judicial oversight could be envisaged. After some trial and error, agents were left in 1883 with three options: (1) A full bankruptcy procedure leading to liquidation, with the possibility of the judge granting a discharge; (2) A reorganisation or self-liquidation plan decided outside formal bankruptcy, with limited personal costs to the debtor but still under tight judicial oversight (the debtor lost the control over assets, he was publicly interrogated, the judge had substantial power to reject the voted plan, and so on); (3) A high-majority, low-oversight formula that was similar to the past, unanimity deeds of arrangement, which appeared to be the favourite option (as became clear once registration and some publicity rules were introduced in 1887).
On the Continent, Belgium took in the same year a comparable route toward a 'menu' approach to renegotiation. The 1883 Concordat préventif allowed distressed debtors to negotiate wide-ranging plans under some judicial oversight, though without supporting the many costs of entering formal bankruptcy. The debtor would not lose control over his assets, social and symbolic costs were intended to be much smaller, and his obligations in terms of providing information were more limited. But he was put under some control by the court and the creditors, and if he failed to obtain qualified support, then he would be shifted into bankruptcy proper: the Concordat préventif was not a soft option to protect the debtor and/ or the manager. Finally, Germany took exactly the opposite way to that adopted by bankruptcy law in 1867. Tellingly, Coleman (1974) does not mention the terms 'arrangement' or 'composition' in his index. See also Mann (2003). 35 On eighteenth century and early nineteenth century English law, see Davies (1744), Cullen (1800), and Cooper (1801). 36 See Holland (1864) and Robson (1888  Italy 1903Switzerland, 1874Russia 1903Sweden, 1877Denmark, 1905Scotland, 1877Italy, 1882Netherlands, 1893Greece, 1900 England and Belgium: after due consideration, lawmakers decided that the modern approach to bankruptcy adopted in 1877, which explicitly aimed at incurring minimal economic and civic costs to the debtor, should remain the sole entrance to any form of collective negotiation, whether they would lead to an arrangement or a liquidation. In this respect, Germany remained closer to LLSV's preference for maximal procedural guarantees against their optional relaxation (see Table 1).
Convergence at the end of the century thus took place around two models ( Table 1): by 1914, ten countries had adopted a multiple-track, Anglo-Belgian approach while the remaining five had opted for the integral, German one. A major innovation -restructuring assets and liabilities without liquidation -had thus been adopted in most countries, though without the 'legal origin' issue having here a clear impact. Legal traditions were not a serious obstacle for inventive lawmakers.

Why convergence? Some hypotheses
What were the driving forces beyond the successive patterns of Europe-wide convergence? Though providing a complete answer to this question is beyond the limits of this article, some indications are readily available. A first point is that there was a lot of communication and exchange between lawyers of different countries, especially during the second half of the period. In particular, the French legal profession and public administration developed a consistent effort at collecting and translating foreign statutes (see the bibliography of this article). Given the centrality of French civil law, French language and the Sorbonne at the time, there is good reason to believe that this was the support for a de facto co-ordination between legal professions, across a substantial part of Europe. 37 A wholly different question is how communication and influence made the legal texts converge: what reasons induced the French 1889 lawmakers, or the 1903 Italian ones, to take a serious look at earlier English and Belgian innovations and to follow their lead? A first, polar hypothesis is that strong market actors, increasingly engaged in cross-border exchanges, explicitly pressured national governments to reduce discrepancies between national laws and procedures. Convergence would then reflect a demand for crosscountry legal co-ordination, as a response to the regulatory challenges of trade integration. In that case, however, one would expect these issues to have also emerged as an important field of legal expertise, then as a topic for international negotiations and eventual treaty-making. But interestingly that point is missing.
Although the subject had been approached before, legal research on crossborder bankruptcies, as a sub-field of international private law, emerged only in the latter decades under review. The founding contribution was published in Italian by Guiseppe Carle, in 1870: it established unity and universality as the key principles that should govern cross-border proceedings, against their being fragmented between competing, territorial procedures. 38 But what the latter legal literature then states, again and again, is that national laws remained closely aligned on a territorial doctrine: rules of co-ordination across borders remained very weak, and the attempts at improving them were limited till 1914, and indeed well after that date. 39 Basically, the norm for each country was to defend the integrity of its own procedures, though they generally offered equal rights to nationals and foreigners, domestically. Some would then state that after local creditors had been reimbursed, the balance of the proceeds of a liquidation could be transferred to a foreign bankruptcy court; but even that was conditional upon 37 Saint-Joseph (1844) was an early, quite interesting attempt at comparing commercial laws across Europe. But the Annuaire de Législationétrangère, published every year by the Société de Législation Comparée from 1871 onward, is the most consistent outcome of that investment. Goldschmidt (1875) is also a landmark, though bankruptcy is not included as such: following upon the German tradition, it is classified as a part of civil procedure. A later though remarkable endeavour was the project of a global encyclopaedia of trade law, launched jointly in France, Germany and the United States in 1911, but aborted in 1914: a series of volumes, edited on a country by country basis, offered a detailed presentation of each country's trade law, both in its original language and, respectively, in French, German and English; see the collection 'Lois commerciales de l'univers', in Aström et al. (1911). 38 See Carle (1875) for the French translation, which adds a lot of complementary indications on inter alia., French, Belgian, Italian and German case law on conflicts of law. 39 The difficulty of present-day European countries in achieving a reasonable degree of co-ordination between national procedures can also illustrate the point; see Béghin (1994). reciprocity and generally subjected to case by case, bureaucratic confirmation. Beyond this, few of the steps discussed by Carle were taken. No country accepted for instance that a foreign judgement opening a bankruptcy process could have direct legal effects domestically, on contracts and the execution of guarantees for instance, not so speak of its triggering a domestic procedure. The rule was that foreign judicial decisions should first be examined and confirmed by domestic courts, under their own terms. By the same token, the principle of a cross-border insolvency being subjected to a lead-procedure, possibly foreign, to which secondary ones would report, was not at all on the horizon. One exception was a bilateral agreement reached by France and Belgium in 1899, and a more limited accord with Switzerland in 1869. Otherwise, no bilateral or multilateral treaty on bankruptcy was ever negotiated or signed, in a period famous for the number of trade-supporting agreements being agreed upon. A series of international conferences on private law, taking place in The Hague, actually put the subject on the agenda in 1894, but reached a clear stalemate in 1900 and ended up in 1909 with only a blueprint for possible bilateral conventions -without any practical consequence. 40 These elements suggest that the patterns of convergence observed between European bankruptcy laws did not reflect primarily a demand for crossborder, trade-driven co-ordination. Nor can they be interpreted as the effect of underlying 'legal origins', shared by sub-groups of countries. Most probably, then, convergence was the ex post effect of two factors: pure intellectual influence working across national legal professions; and the pragmatic decisions of national lawmakers within a 'territorial' game-set. Practically, they would have reached comparable legal solutions to comparable problems, arising from comparable, parallel processes of economic and social development. And in this case, copying each other or copying a leadinnovator is rational. 'Why re-invent the wheel, if the Belgians have already done it?'. The intuition is that things indeed happened this way.
Specifically, three main institutional innovations have been mentioned, that may account for the mid-period, joint-change in bankruptcy paradigm. First is the possible reduction in informational problems, due to the rationalisation of, inter alia, accounting rules, property titles, collateral and mortgage registrars, payments flows and so forth; these standardised tools were then mobilised by a new class of intermediaries (universal banks, credit bureaux, specialised journals, and so on). Second is the emergence of large publicly-traded firms that raised new issues of going concern and governance (the extension of limited liability and the differentiation between managers and shareholders). Finally, the better regulation of the money and capital markets would have increased the quality of the signal for entering 40 See Conférence de la Haye chargée de réglementer diverses matières de droit international privé (1894,1900,1904,1909).
bankruptcy, while making easier the disposal of assets and the restructuring of liabilities. Though hypothetical, this broad interpretation of convergence patterns is consistent with the propositions of Rajan and Zingales (2003) who underline the impact of domestic, political economy factors in the long-term evolution of institutions over the twentieth century. During that century, the main trend as regard the regulation of insolvencies was of course, after the 1930s, the declining relative strength of creditors against non-contractual stakeholders (specifically workers and tax administrations). This is clearly the background of the LLSV approach to bankruptcy, as of the large differentiation across countries reflected by their database. As important, however, is probably the long-term decline of capital markets as an instrument for sanctioning capitalist failure and reallocating assets, with parallel evolutions observed on the labour markets. There is actually no point in enforcing a strong bankruptcy law in a corporatist regime, that is explicitly built on the premise that market forces should not have the upper hand in the (re-) allocation of factors. Conversely, the renewed interest in bankruptcy reforms since the 1980s should be seen as a reflection of the reverse trends: the conflicting reinstatement of creditors' rights against non-contractual claims, 41 and the evolution toward more competitive economies and much freer markets for factors (flexible labour markets and liquid capital markets). The present success and influence of the US Chapter 11 would then resemble the emergence of the Anglo-Belgian liberal solution to re-contracting, in the early 1880s. And of course, today as a century ago, some countries innovate, some others adjust rapidly, and others lag.

Conclusion
A dataset of 51 European laws and statutes has documented the evolution of bankruptcy procedures between 1808 and 1914. A first conclusion is that, throughout the period, the protection of creditors' rights was a core feature of all statutes, whatever legal tradition they belonged to; only England may be considered a partial exception. The claim that 'legal origins' have a permanent, country-specific impact on creditors' rights, as defended in LLSV (La Porta et al. 1998) and elsewhere, is thus not warranted. 42 This certainly does not imply that all countries were equally efficient in actually protecting stated rights; but as far as the structure of the law is concerned, there is not much room for doubt.
Beyond this, the main lesson is that broad, continent-wide evolutions, arguably linked to the process of capitalist development, are much more important than country-specific features. Two bankruptcy models were identified across the Continent. A first one, best represented by the 1808 Napoleonic Code, was characterised by heavy threats and repression vis-à-vis the debtors and by limited contractual autonomy; at the time, the rationale put forward was the need to control moral hazard and opportunism. After a transition period between the late 1860s and the late 1880s, an alternate and more liberal model emerged: prison for debt was abandoned, rehabilitation became easier, and the parties were given much more room to recontract on property rights. If the explanation for the first model is correct, then transaction risks should have declined sharply in the latter part of the century.
The empirical evidence does not contradict the observation of covariance within legal traditions. Instead it underlines the extent of joint changes across countries as well as the pattern along which traditions may evolve: they can endure for centuries, but they can also adjust rapidly to a changing environment. The shift of English statutory law toward court-based debt discharge and multiple-entry procedures is a remarkable example. Yet, it is not possible to point out any occasion when 'legal origins' might have been at work, against or in support of creditors' rights. No essential or ahistorical hard core of legal institutions could be observed that would predict how realworld institutions are designed and how they bear on economic outcomes. 'Legal origins' are a proxy for a social entity whose shape, structure, and quality remain elusive.