My dissertation aims at improving our understanding of what are the critical determinants for a choice of ownership structure of a firm and how the ownership structure affects a firm's performance. For this I examine manufacturing firms in the United States during the nineteenth century. The United States has excellent data for this research endeavor in the R.G. Dun & Co. Collection and the manuscript census of manufactures. The R.G. Dun & Co. Collection is a set of detailed credit reports recorded by a credit rating company, the R.G. Dun & Co. Starting in 1842 agents of the R.G. Dun & Co. gathered information on the firms in their various regions and regularly reported this information to the New York headquarters. The basic content of the record includes the owners of the firm and their status (age, family relation, personal aptitude as an entrepreneur, and wealth), firm performance (sales, profit), financial transaction (mortgages, failure, bankruptcy, etc.), and an evaluation of their creditworthiness. In the 1960s Dun & Bradstreet, the successor of the R.G. Dun & Co., made the firm's records collected from 1842 to 1890 available to the public at the Baker Library of the Harvard Business School. The collection consists of about 2,600 ledgers with 300–400 pages per volume with records on more than a million firms. The manuscript census of manufactures consists of records of firm-level information such as total product, capital, employment, and wages, on which published aggregate information in the census of manufactures was based. By combining the manuscript census of manufactures of 1850, 1860, 1870 and 1880 with the R.G. Dun & Co. Collection, it is possible to construct a firm-level panel data set covering the 1850 to 1880 period.

Due to large size of the two sources, sampling is necessary. I chose two leading industrial areas of the nineteenth century, Bristol and Essex County, Massachusetts, recorded every firm from these regions in the two data sources, and then linked them. Bristol County had three major industrial cities, Fall River, New Bedford, and Attleboro. Fall River and New Bedford had grown up as a center of the textile industry during the nineteenth century, and Attleboro was one of the national centers of the jewelry industry. Essex County was an important producer of textile, boots, and shoes. The boot and shoe industry of Essex County, led by Lynn and Haverhill, produced almost 50 percent of the total national output. These thriving and competitive industrial conditions are suitable for investigating determinants of firm ownership structures and their impact on firm performances.

I first explore the dynamics of this firm population by analyzing their exit behavior. Because the R.G. Dun & Co. Collection recorded whether a firm exited by closure, sales, or dissolution, etc, it is possible to comprehensively examine exit patterns and the relation between ownership structure and firm performance. In addition, this analysis can contribute to a better understanding of the business failure problem during the nineteenth century. Business failure was one of the most important social and political issues during the nineteenth century. Business journals frequently dealt with a business failure as their cover story, and the U.S. Congress had repeatedly enacted and repealed national bankruptcy laws until the introduction of a permanent one in 1898. Previous studies have investigated legal, social, or political aspects of default or bankruptcy during this period. However, my finding indicates that only 10 to 20 percent of the firms in my data set discontinued their production due to financial default. This implies that analysis of exit in general can shed new lights on this historical event.

For exploring why business failure was such a big social problem during the nineteenth century, I compare the exit rate for the nineteenth century with that of the post–Second World War period by Timothy Dunne, Mark J. Roberts, and Larry Samuelson.¹

The comparison shows that the exit rate for the nineteenth century was much higher than that of the post–Second World War period. This was mainly due to the fact that the average firm size in the nineteenth century was much smaller than that of the post–Second World War period. Out of various mechanisms generating the negative correlation between firm size and exit rate, I show how the indivisibility of entrepreneurial input increased the exit rate of small firms and consequently the overall exit rate for the nineteenth century. In addition, I show that the large firms of the post-Second-World-War period contributed to decrease in exit rate by internalizing introduction of new products, which were mainly performed by entry of new firms in the nineteenth century.

However, size-specific exit rates indicate that size differences alone do not completely explain the difference in the exit rates between the two periods. Medium size firms in the nineteenth century had higher exit rates than the average exit rate of the post–Second World War period. Although a case study of the boot and shoe industry implies that the high exit rate of medium size firms was partly due to rapid technological change, this pattern was evident in other industries as well. Why did these medium size firms have such a high exit rate? I study the ownership structure of the firm during the nineteenth century to discover the reason. Because only large firms were organized as corporations during this period and most of the medium size firms were either proprietorships or partnerships, I focus on the choice between proprietorship and partnership.

In the twentieth century, partnerships account for about 10 percent of all firms. However, this proportion was much higher in the nineteenth century, when it reached about 30 percent. One might believe that the importance of family relations in business during the nineteenth century drove the high share of partnerships. However, the data do not support this hypothesis. About 60 to 70 percent of the total partnership firms were nonfamily partnership firms. On the other hand, it is widely believed that partnerships are optimal for firms in special activities such as law and medicine.²

If we generalize and apply this view, we can expect industrial variations of ownership structures, that is, some industries were dominated by proprietorships whereas others were occupied by partnerships. However, partnership firms were not concentrated in a few industries as this industrial-variation view predicts but occupied from 20 to 40 percent of each industry.

I show that entrepreneurs adopted partnerships in order to grow quickly in an environment with high transaction costs. First, the distribution of firm sizes shows that whereas proprietorships ranged from small to medium size firms, partnerships were mainly adopted for medium size production. Second, the geographical distribution of ownership structures suggests that the share of partnerships in local industries was 20 percent higher in large cities than in other areas. We can infer that firms tried to grow faster by organizing partnerships when they had more demand. Third, the relation between the growth of the firm and adoption of co-partners is examined by estimating the correlation between the transition from proprietorship to partnership and the growth rate of the firm. Regression results show that adding a new co-partner accompanied an increase in employment and capital.

One question, therefore, is why did not all firms adopt a partnership structure for faster growth? The answer lies with three factors. First, a partnership was just one alternative to alleviating resource constraints that did not weight equally on all entrepreneurs. Many firms that did not suffer from resource constraints could operate their firms as proprietorships. Second, the cost of searching for good co-partners and the potential cost of unsuccessful relationships (e.g., costs due to forced reorganization or liquidation) could deter many entrepreneurs from organizing partnerships. In this case they would choose the slower incremental approach to growth through proprietorship. Lastly, duration of the partnership contract was short, which cancelled out increases in new partnerships. On the one hand, this short duration could reflect the weakness of partnership contracts. However, it could also be a result of voluntary dissolutions. Many partnerships might have been dissolved once members had built up their reputation and wealth, and could thus operate independently. These three factors help explain why all firms were not partnerships.

I also make a detailed case study on the ownership structure of the jewelry industry at the Providence/Attleboro area during the nineteenth century. Providence, Rhode Island and adjacent Attleboro, Massachusetts were renowned as a major center of jewelry production throughout the nineteenth century. The Manufacturing Jeweler, the trade journal of the jewelry manufacturers in the Providence/Attleboro area, asserted that “forming a business partnership is the next best thing to getting married.”³

Why did the Manufacturing Jeweler encourage entrepreneurs to organize partnerships? Did the jewelry manufacturers adopt partnership as the Manufacturing Jeweler recommended? If so, was it a general phenomenon of the jewelry industry or was it specific to the Providence/Attleboro jewelers? If partnership was popular, why did jewelers organize partnerships?

I show that the industrial conditions of the Providence/Attleboro area made its jewelry manufacturers adopt partnership during the nineteenth century. Due to the differentiated nature of the product, the jewelry manufacturers needed active promotion for sales. The promotion activities were made not directly to customers or retailers but to wholesalers located in large cities such as New York. There are two reasons why this industrial condition—the geographical separation from the wholesalers and the differentiated nature of product—made the Providence/Attleboro jewelry firms prefer partnership. First, partnership ameliorated the informational problem between producer and salesperson. Second, retaining a sales force for the promotion activities required a nontrivial cost. One way to optimize the cost given this sales force was to increase the total value of sales per transactions. This was possible when entrepreneurs had a greater variety of products through increasing production capacity. Some entrepreneurs might not have enough money or credit, and then they might organize partnerships to overcome the resource constraints and exploit economies of scale.

The data support this hypothesis. First, the share of partnerships for the Providence/Attleboro jewelry producers was about 60 percent, much higher than any other contemporary industries. Comparison with the producers in New York shows that this high ratio was not the characteristic of jewelry manufacturers in general but that of the Providence/Attleboro producers. Second, these partnership firms were more successful in terms of growth than proprietorship firms. Cross-sectional pattern shows that partnership firms were at least twice as large as proprietorship firms, and partnership firms had higher growth rates. Third, I collected about 200 biographies of Providence/Attleboro jewelry entrepreneurs from the Manufacturing Jeweler and analyze how they chose the ownership structure of their firms in their business life. I find that about 90 percent of these entrepreneurs organized partnerships when they first started their own business in their twenties or thirties for exploiting these advantages over proprietorship.

My dissertation highlights the importance of partnership as a firm ownership structure for understanding entrepreneurial activities in the United States during the nineteenth century. The share of partnerships started to decline from around 1900, and it was replaced by corporations. Changes in production technology, capital market conditions, or the introduction of corporate tax or bankruptcy laws, etc., could affect this transformation, which will be studied in the future research. In addition, I am currently extending the data set by connecting it with the manuscript population census. By analyzing this data, I will further investigate the relationship between individual wealth, financial market condition, and choice of ownership structures.