Patent laws are designed to create the optimal incentives for innovation, but we know little about how exactly this works.¹

The need to better understand the effects of patent laws is particularly urgent today, as industrialized countries lobby to introduce and strengthen patent laws in developing countries around the world. Although it is difficult to predict the results of such changes, historical data from the mid-nineteenth century may hold important lessons for patent policies today. The nineteenth century is an ideal period to study the effects of patent laws: Mid-nineteenth-century patent laws were adopted in a relatively ad-hoc manner, depending on legal traditions rather than economic considerations. Large differences in patent systems existed across countries, and patentees depended on domestic patent laws because patenting abroad was prohibitively expensive and almost all countries discriminated heavily against foreign patentees. As a result, domestic patent laws played a more important role in creating incentives for domestic invention than at any later stage in history.

This dissertation introduces a new data set that reveals the effects of patent laws on innovation across a wide range of countries in 1851 and 1876. I have constructed internationally comparable data from the exhibition catalogues for two nineteenth-century world fairs on technology: the Crystal Palace Exhibition in London in 1851 and the Centennial Exhibition in Philadelphia in 1876. The Crystal Palace was the most popular event of its age, and it became the first in a long series of world fairs, which allowed inventors and firms to exchange innovations across countries. More than six million visitors attended the Crystal Palace, and almost ten million visitors came to the Centennial Exhibition. For the Crystal Palace fair I have counted and classified 13,876 exhibits in 30 industries of use. For the Centennial I have counted and classified more than 19,076 exhibits in 344 industries.

A typical entry in the exhibition catalogue begins with the exhibitor's name, address, and country of origin, followed by a brief description of the innovation, as well as some information about the exhibitor's occupation and state of patent protection. From these records, I have constructed data for 32,952 innovations, their industry of use, and country of origin. I also determine whether exhibitors held a patent or not, distinguishing foreign from domestic patents. Information on domestic patents allows me to calculate proxies for patenting rates across industries. In addition, I construct measures for the quality of innovations based on awards for inventiveness. These awards ranked all exhibits according to their novelty and usefulness, and granted prizes to the most innovative exhibits.

Exhibition data are particularly useful for studying the effects of patent laws on innovation because they measure economically useful innovation in a way that is independent of changes in patent laws. As employed here, economically useful innovation—the commercial introduction of new or improved products and processes—is distinct from invention—the conception of such products and processes. Exhibition data measure economically useful innovations, whereas patent data count those inventions which inventors chose to patent. This distinction matters because only a fraction of innovations are patented and not all patents develop into economically useful innovations. Historical records bear witness that national committees selected their most innovative products to be exhibited at the fairs. Participation was competitive and a uniform system of selection admitted less than one-third of all applicants to exhibit at the fairs.²

Most importantly, exhibition data measure innovation independently from differences in domestic patent laws. Exhibitors displayed innovations regardless of whether they could be patented at home, including many innovations that they had chosen not to patent.

Exhibition data allow me to compare innovative activity across countries with and without patents. This represents an empirical advance over the theoretical literature on patent laws, which simply assumes that the existence of patent laws raises levels of innovation. Without exhibition data, empirical studies of nineteenth-century patent laws relied on patent data, but these studies cannot identify whether the existence of patents raised levels of innovation. Equally important, they cannot identify whether the nature of innovation differs in countries without patent laws. Nineteenth-century exhibition data allow me to analyze these questions empirically.

Previous studies of innovation have focused on the number of innovations that are created in an economy. Among the key predictions in this literature, such as those of William Nordhaus, Paul Klemperer, and Richard Gilbert and Carl Shapiro, is that strong patent laws raise the number of innovations that are made within a country.³

According to this argument, countries without patent laws should display few novel technologies. In contrast, the exhibition data show that countries without patent laws brought many important innovations to the fairs. Mid-nineteenth-century Switzerland, for example, had the second highest number of exhibits per capita among all countries that visited the Crystal Palace. Moreover, exhibits from countries without patent laws received disproportionate shares of medals for outstanding innovations. Countries without patent laws were exceptionally inventive in a small number of industries: food processing and the manufacture of scientific instruments.

Previous studies have ignored the effects of patent laws on the direction of innovation, although this may be a key determinant of economic growth. Simon Kuznets observes that technological innovation concentrates in a few fields at any given time; he argues that such concentrations determine countries' potential for economic growth.⁴

Economic history supports these claims: Germany's focus on chemical innovations is widely understood to have enabled Germany to replace Britain as the industrial leader in the late nineteenth century. Nathan Rosenberg argues that the United States overtook Europe at the beginning of the twentieth century because of its concentration on labor-saving innovations.⁵

This dissertation sets out to analyze the causes of such concentrations of innovative activity. More specifically, it asks the questions whether and how patent laws create or intensify a country's tendency to focus innovation in certain industries. A simple model predicts that weak patent laws encourage innovation in industries where alternative methods to protect intellectual property are particularly effective relative to patent protection. By this argument, weak patent laws may help to create differences in technological comparative advantage among industrialized countries and their followers.

Previous empirical analyses of the effects of patent laws had to rely almost exclusively on patent counts, despite some obvious limitations of these data.⁶

Economic research benefits from data on mid-nineteenth-century innovations, because this period is uniquely suited to study the effects of patent laws. At the time of the Crystal Palace Exhibition in 1851, most governments had adopted patent laws but they had not yet begun to modify these laws in response to pressures from domestic interest groups. Differences in domestic patent laws were large and mattered greatly to domestic inventors, because international treaties had not yet begun to harmonize patent laws or to protect the rights of foreign patentees. However, nineteenth-century patent data are available for less than a handful of countries. Even within these countries, they omit entire industries because patent laws did not extend to all sectors of the economy. In addition, many important nineteenth-century patents, such as various improvements in the steam engine, had to be dropped from patent counts because they could not be assigned to a specific industry. Patented inventions vary greatly in their commercial viability and quality; the quality of innovations is difficult to proxy based on patent data. Most importantly, the way in which patents measure innovation varies with changes in patent laws. The definition of what constitutes a patentable invention varies across countries. In the United States only first and true inventors are allowed to patent; in France first importers have equal access to exclusive patent rights.⁷ Differences in the length of patent grants, in patent costs, and in the security of patent rights create further differences in what inventors choose to patent.

Exhibition data offer a complement to nineteenth-century patent data that addresses many of the concerns about patent data. Uniform rules of selecting exhibits ensure that exhibits are comparable across countries, regardless of domestic patent laws. Exhibition data include information on three patentless countries: Switzerland and Denmark in 1851 and Switzerland and the Netherlands in 1876. No other data are available to study early innovation in these countries. Exhibition data cover innovations in all industries, including those that were excluded from the patent system. Awards to the most innovative and useful exhibits provide a measure for the quality of innovation. References to patents in the exhibition data allow me to distinguish patented innovations from those that are protected by alternative mechanisms.

The analysis of exhibition data suggests that patent laws may be an important factor in determining the direction of innovative activity. Exhibition data show that countries without patents share an exceptionally strong focus on innovations in two industries: scientific instruments and food processing. At the Crystal Palace, every fourth exhibit from a country without patent laws is a scientific instrument, whereas no more than one-seventh of other countries' innovations belong to this category. At the same time, the patentless countries have significantly smaller shares of innovation in machinery, especially in machinery for manufacturing and agriculture. After the Netherlands abolished her patent system in 1869 for political reasons, the share of Dutch innovations that were devoted to food processing increased from 11 to 37 percent.

Proxies for differences in the propensity to patent across industries, which I have constructed from the Crystal Palace data, document that innovators' propensity to patent varies strongly across industries. Exhibition data show that innovations in scientific instruments and food processing are less likely to be patented, especially compared to innovations in machinery. These proxies indicate that only 10 percent of exhibits in scientific instruments, but close to 40 percent of exhibits in manufacturing machinery are patented. Nineteenth-century sources report that secrecy was particularly effective at protecting innovations in scientific instruments and in food processing. On the other hand, patenting was essential to protect and motivate innovations in machinery, especially for large-scale manufacturing.

I use these differences in patenting rates across industries in combination with detailed information on the geographic location of exhibitors to examine whether patent laws fulfilled their second major goal, the diffusion of technical knowledge within countries. An analysis of geographic clustering and the patenting behavior of 4,465 British exhibitors at the Crystal Palace suggests that patents facilitated the spread of new ideas. Innovations in industries with high patenting rates were geographically dispersed, whereas innovations in industries with low patenting rates exhibited strong patterns of geographic concentration, not only across counties, but even within cities. These results suggest that patent laws may not only influence the direction of innovation, but also its geography.