The retail store is a ubiquitous feature of the modern American economy. Over 15 million Americans, 11.3 percent of the work force, work in retail trade.Footnote ¹ Retailers employ more workers than all other private employers except professional/business services and the healthcare industry. Retail sales account for over 43 percent of personal consumption and 30 percent of GDP. Only manufacturing and wholesale sales are a larger share of the economy.Footnote ² While these figures provide an indication of the size of the industry, they do not capture the full impact of retailing on the American economy. No industry serves as many consumers as frequently as retailing. Only banking has a customer base and frequency of use close to that of the retailer, yet according to the Survey of Consumer Finance nearly 10 percent of Americans do not have a checking account. Meanwhile every American is a customer of a retail establishment, often daily. This wide scope means changes in the sector have profound and immediate impacts on the entire American economy and populace. My dissertation begins the task of documenting and analyzing the retail revolution of the early twentieth century and how that revolution was influenced by demand-side consumer technological innovations.

The first chapter describes how retail services were produced at the turn of the last century and how those methods changed over the course of the century. I model a retailer as an establishment that produces a service, which reduces a consumer's cost of acquiring goods. In 1900 this was done in large part by the proverbial “Mom and Pop” retailer. They were small, situated near the consumer and had low levels of stock turnover. Its small size was manifest in nearly every measure. Data on stores around 1900 is sparse, but as late as the 1920s over 10 percent of stores were opened on less than $100 in capital and over 30 percent with less than $300. This is low when compared with the average annual income of a full-time retail employee of $1,311.Footnote ³ A customer seeking to buy an item requested it from the proprietor, who then retrieved the item from stock and packaged it for the consumer. By the 1930s for many stores this whole procedure was altered in a variety of ways that led to the methods we observe today. I focus on seven key factors that were involved in this alteration including the introduction of nationally branded goods, the advent of self-service, the spread cash-and-carry, the penetration of chain stores, the growth in store size, new location choices, and shifts in labor utilization. These seven innovations were not necessarily isolated or exogenous events. Rather they interacted in a dynamic fashion, resulting in many stores that looked vastly different in 1930 than they did in 1900. In fact, a customer today might well find many stores of 1930 more familiar than would a 1900 shopper.

An important change that occurred at the beginning of the twentieth century was a rise in the size of retail establishments. In the early 1920s a Ralph's store was around 6,000 square feet. In 1929 they constructed a 10,000 square foot store. The grocer Henke & Pillot opened a 30,000 square foot store in 1923. While extremely large for the time, only three years later a 40,000 square foot store was opened.Footnote ⁴ The growth in size was not limited to independents. Although the large groceries that came to be known as supermarkets were established by independents, chains soon appreciated their advantages and also began constructing larger stores. The grocery chain, Safeway, began carrying a full range of food products during the early 1920s. A&P entered the Los Angeles market around 1930 with their standard store size around 4,000 square feet. They soon found, however, that in southern California this format was not the most competitive. In only a couple of years they began constructing stores between 6,000 and 8,000 square feet, much larger than their previous format.Footnote ⁵ The process of deepening and widening inventories was no doubt slowed for chains by their existing investment in smaller store formats. However, by 1937 most leading chains were closing small stores and opening supermarkets.Footnote ⁶ The growth in size was also not limited to food stores. A drug store that had a relatively large selection of products was typically around 1,000 square feet in 1930. In 1936 the Sontag Drug Company opened a 16,000 square foot store. While larger than most, by this time the size of the typical store had risen to over 6,000 square feet.

The rapidly increasing size of stores renewed debate about the competitive conduct of retailers. Many believed retailing to be uncompetitive and inefficient in great part due to what was viewed as large and rising gross margins. Lawrence Mann noted persistent statements that middlemen were wasteful and had too high margins.Footnote ⁷ Even with large selections of retailers from which to choose, these margins led many to conclude that the industry was not competitive either because of collusion, product differentiation, or some combination of the two.Footnote ⁸

Chapter two explores these issues of competition as well as the increase in store size by examining two new demand-side consumer technologies: the automobile and radio. The car changed consumers' cost of reaching more distant stores and gathering information. My results suggest that as more consumers bought automobiles, and became more mobile locally, competition between stores increased, which in turn cut margins for retailers. Consequently, the number of customers required to support a store was larger in markets with more automobiles.

Although the radio might have increased competition by providing consumers with more information about the range of stores, advertising can also reduce competition when stores use it to differentiate themselves. Empirical results suggest that during the 1920s the radio was associated with higher margins and less competition, because the primary form of advertisement was sponsoring of programming. This highlighted the name recognition of the stores without increasing competition among stores because the ads said little about prices or store characteristics.

Expanding store sizes and reductions in the number of stores led many to assume that the previous system of retail distribution had been uncompetitive and inefficient. The analysis here shows that retailers were subjected to competitive pressure as the number of stores in a market increased in response to greater automobile ownership. It appears then, that the key factor changing the structure of stores was technological change. Thus, stores at the beginning of the century when the technology was not available, might well have been making efficient decisions given the constraints they faced.

The Industrial Revolution of the twentieth century brought with it fundamental changes in the use of labor and capital. Over the last 100 years the size of stores has increased as measured by sales or customers. For example, between 1929 and 2002 the number of stores per capita fell over 40 percent while real sales per store rose more than fourfold. These statistics might be explained in part by the further penetration of the automobile over the century. However, while the size of stores grew as measured by sales per store, the amount of labor increased more slowly, such that the number of employees per dollar sold has fallen over 60 percent.Footnote ⁹ This illustrates that there has been a significant shift away from many small, labor-intensive operations towards fewer larger stores utilizing less labor inputs per dollar sold.

This shift corresponds with some of the changes in how retail stores produced their output. By 1929 retailing had begun the shift towards cash-and-carry, and self-service models. Both were less labor- but more capital-intensive methods for producing retail service. Chapter three examines why these format changes began prior to the Depression and the roles played by the automobile and the radio in their adoption. Starting with a model of the production function and cost structure of a retail store, the demand for retail labor is derived and used to estimate the structural parameters of a retail store's production function. Similar to models of skill-biased technological innovation, these parameters can vary with the adoption of technologies such as the automobile. I show that in markets where a larger share of consumers owned cars, retail stores used relatively fewer labor inputs per customer served. In the grocery segment, a 10 percentage point increase in the mean auto penetration would have resulted in an 11.7 percent increase in capital inputs per customer served and only a 1.5 percent increase in labor input. When these parameters are used to predict the use of capital and labor in the mid 1950s, the increase in car penetration and rises in input prices account for a significant portion of the observed changes in labor and capital between 1930 and 1954.

The growing size of retailers had a profound impact on the amount of and the type of labor hired. Using data from the population census, the last chapter of my dissertation documents a number of interesting patterns. The ratio of proprietors to employees declined over 77 percent between 1910 and 1970 as chain stores overseen by employee managers spread. The distribution of skill levels for workers displayed greater variance and the reported occupations of retail workers changed from predominantly generic sales positions to more specialized jobs with titles such as buyer and cashier.

I document that the timing of these changes and their cross-sectional variation show that the increase in specialization was tied to the growing size and chain affiliation of retail stores. As stores grew larger and became part of chains, retailers began to take on more of the wholesale responsibility. The chain format allowed shared aspects of a branded store to be specialized. Specialized buyers were hired to research available products and better screen inventory to create the width and depth that customers demanded. A larger operation also permitted specialization in the accounting and bookkeeping aspects of an operation.

Labor specialization could increase inside the store as well. A small store within a chain would still have been unable to support a manager, sales staff, stock boy, and cashier. The growing size of any given store was also important in the specialization of retail labor. The proprietor or store manager was freed to oversee the operations of the store while lower skilled and less expensive labor could be used to stock shelves and ring up merchandise; and people skilled at sales could devote 100 percent of their time on the floor working with customers.

The findings in my dissertation not only inform us about the past, but they are also important for understanding the changes in the modern era. Today, similar innovations are again changing the retail sector. Consumer access to the internet has spread rapidly over the last ten years. Like the car and radio, the internet lowers the cost of acquiring information about retail products and services. These results provide clues on understanding how the internet and other new technologies may affect the future of the retail industry.