Comparative lawyers are interested in similarities between legal systems. Artificial intelligence offers a new approach to understanding legal families. This chapter introduces machine-learning methods useful in empirical comparative law, a nascent field. This chapter provides a step-by-step guide to evaluating and developing legal family theories using machine-learning algorithms. We briefly survey existing empirical comparative law data sets, and then demonstrate how to visually explore these using a data set one of us compiled. We introduce popular and powerful algorithms of service to comparative law scholars, including dissimilarity coefficients, dimension reduction, clustering, and classification. The unsupervised machine-learning method enables researchers to develop a legal family scheme without the interference from existing schemes developed by human intelligence, thus providing a powerful tool to test comparative law theories. The supervised machine-learning method enables researchers to start with a baseline scheme (developed by human or artificial intelligence) and then extend it to previously unstudied jurisdictions.

Chapter 10 identifies 21 variants of the good-faith purchase doctrine, which are often different combinations of several key factors. That said, none of the 21 schemes are the most efficient. Among the forms of good-faith purchase doctrine currently in use, the market overt rule comes closest to ex ante efficiency because original owners, merchant dealers, and consumers all have incentives to spend close to optimal costs on verification and prevention, and the movables in question are more likely to be in the hands of higher valuers. Drawing on mechanism design literature, Chapter 10 argues that when both an original owner and a consumer are non-negligent, the two parties can be assigned 50% shares of the movable in question, and an ensuing internal auction between them can ex post tease out who values the resource more. This internal auction design is inexpensive to administer.

The first book of its kind, Property Law: Comparative, Empirical, and Economic Analyses, uses a unique hand-coded data set on nearly 300 dimensions on the substance of property law in 156 jurisdictions to describe the convergence and divergence of key property doctrines around the world. This book quantitatively analyzes property institutions and uses machine-learning methods to categorize jurisdictions into ten legal families, challenging the existing paradigms in economics and law. Using also other cross-country data, this book empirically tests theories about property law and comparative law. Using economic efficiency as both a positive and a normative criterion, each chapter evaluates which jurisdictions have the most efficient property doctrines, concluding that the common law is not more efficient than the civil law. Unlike many prior studies on empirical comparative law, this book provides detailed citations to laws in each jurisdiction. Data and documentation are released with the book.

Chapter 12 discusses the specificatio (mistaken improver) doctrine. About two-thirds jurisdictions have this doctrine, and the doctrinal structure is highly convergent. Most of these jurisdictions limit the application of the doctrine when the nonconsensual improvement is irreversible, and most assign sole ownership to either original material owners or improvers. Almost all jurisdictions adopt the disparity-of-value test and/or the transformation test, but there are eight ways that bad-faith improvers are treated. The disparity-of-value test, in and of itself, does not tend to assign ownership to higher valuers, however. While no ex ante rule-making can ensure allocative efficiency ex post, requiring both the disparity-of-value test and the transformation test is more likely to increase efficiency. Lawmakers looking for a radical reform proposal may also adopt the internal auction mechanism to resolve the problem in specificatio. Besides, even good-faith improvers should not be compensated, as the non-transformative, low-value-increasing improvements are unlikely to be what material owners want. A clear rule of no compensation also decreases litigation cost.

Chapter 7 summarizes the various partition approaches used around the world, finding that partition in kind is often preferred, with selling co-owned things through public or internal auctions a common back-up plan. Perhaps as a result, the existing literature has focused on partition in kind and partition by sale, while ignoring intermediate partition approaches like partial partition that are prevalent in practice. Little attention has been paid to the use of revelation mechanisms such as self-assessment, nor to how judicial partition rules affect co-owners’ pre-judicial-partition behaviors. Chapter 7 brings partial partition into the theoretical framework and proposes a new and feasible partition method that utilizes private information among co-owners and makes partition more efficient.

Chapter 1 makes use of two empirical approaches. Its first part uses property law from 136 jurisdictions in an unsupervised machine-learning method (hierarchical clustering) that divide these jurisdictions into 10 legal families. Unlike the traditional wisdom that highlights the difference between common law and civil law, this chapter finds that, in terms of property doctrines, a trichotomy better describes the legal systems: one big group is jurisdictions affected by French property law; another big group is composed of jurisdictions that follow or resemble German property law; and the final group contains common-law jurisdictions, Nordic countries, and a number of socialist jurisdictions. The second part of Chapter 1 re-combines 156 jurisdictions into 149 countries, and computes the correlation coefficients among each country pair, to show dyadic similarities in property law.

Chapter 4 focuses on how ownership of immovables and movables are transferred (that is, whether registration is not needed, necessary, or creating opposability to third parties), whether registration creates absolutism (public faith principle), whether a real agreement is conceptually separate from a sale contract, and whether an invalid sale contract always leads to the invalidity of a real agreement (non-causa principle), and whether delivery or certain intentions are required to transfer ownership of personal properties or the sale contract itself is sufficient. This is where the traditional idea of legal families is conspicuous. Transfer doctrines involve how notice is given. The choice of registration system demonstrates how states, given path dependence, trade off transaction costs and third-party information costs. Which type of conveyance doctrine regarding immovables is efficient is contingent on factors outside of the law. It is easier to reform conveyance doctrine regarding movables, and lawmakers should provide alternative default rules (“menus”) more frequently and establish clear opt-out procedures (“altering rules”).

Chapter 13 focuses on the accessio and confusio doctrines, traditionally sibling doctrines to the specificatio doctrine. The accessio doctrine includes three types of combinations: immovables and immovables, movables and movables, and movables and immovables. Confusio concerns only mixture of movables. The big picture of these doctrines is that there is little sign of convergence, except perhaps in confusio. From an economic standpoint, it is quite clear when two things should be considered combined (thus the accessio doctrine applies) rather than separable: If (the value of attached thing) > (the value of the two post-separation things combined) + (the cost of separation), the two things should remain combined. The next question is who should own it. The key concern behind my analysis is still to deter opportunistic or careless interference with other’s property.

Chapter 5 analyzes acquisitive prescription, a broader concept than adverse possession, and argues that registration-based acquisitive prescription with title and good-faith requirements can be justified by efficiency under certain conditions—Possession, however, is redundant, and may even give rise to undesirable outcomes. Given that boundary disputes can be left for another doctrine, possession-based acquisitive prescription—no matter whether possessors act in good or bad faith—can hardly be justified on an economic basis in countries with well-functioning registrars if possessors do not have title. The possession-based acquisitive prescription can only be justified in jurisdictions with dysfunctional registrars.