economies suffer, so it is not hyperbole to claim that problems generated by urban traffic congestion rise to the level of national importance and that autonomous vehicles may benefit an entire economy by ameliorating those problems.
By reducing congestion, reducing travel time, and improving travel-time reliability, the adoption of autonomous vehicles could also yield other significant economic benefits:
Employment benefits. Individuals’ choices of employers and employers’ choices of workers could expand and result in greater employment. Chetty and others (2014), for example, finds that longer commuting times are strongly and negatively related to the probability that a household will escape poverty, implying that reductions in congestion could expand job opportunities and increase employment, especially for the least affluent members of society.
Agglomeration benefits. Firms and urban residents benefit from the spatial concentration of economic activities, known as agglomeration economies (Glaeser and Gottlieb 2009). Puga (2010) summarizes the evidence that urban density contributes to agglomeration economies and higher earnings. Thus reductions in congestion and improved travel times may improve those economies because people in all walks of life could reach their destinations more quickly and thereby share information more easily, finish certain tasks sooner, and so on.
Trade benefits. Anderson and van Wincoop (2004) shows that travel distance and travel time represent important components of the cost of international trade, which accounts for some of the freight flows over the U.S. transportation network. Those factors are also an important cost of the freight flows generated by intracity and intercity trade. Truck transportation carries a large share of U.S. intracity, intercity, and international freight flows. Reductions in congestion on urban and intercity highways lower the cost of shipping goods and increase freight flows, resulting in more production and consumption throughout the U.S. economy.
Productivity benefits. Improving the speed and reliability of freight traffic enables firms to reduce their inventories and to improve productivity (Shirley and Winston 2004). Productivity is further enhanced if all inputs in a firm’s production process can reach their destinations faster. For example, Prud’homme and Lee (1999) estimates that a region’s productivity increases 1.3 percent when the area that can be reached in a given time period increases by 10 percent. Thus reductions in congestion can broadly affect an entire economy’s productivity.
How big an impact could autonomous vehicles have on the economy? No one can know for certain at this early stage. However, our calculations suggest that the economic benefits could be quite substantial and that they are likely to be broadly consistent with the impact of other significant improvements in mobility in the United States.
In the next three chapters, we report on an econometric model developed to estimate the causal effects of highway congestion on the growth rates of gross domestic product (GDP), employment, wages, and commodity-freight flows for congested counties in California. Lacking a natural experiment to measure congestion’s effects, we use a plausible instrumental variable for congestion that reflects exogenous political considerations in highway spending, and we perform several tests that indicate that our estimates are not likely to be biased by omitted variables or reverse causality. Our estimation results for California are corroborated by findings in the literature on the effects of reducing congestion in specific sectors and by circumstantial evidence on the effects of reducing congestion in nontransport sectors. We then use the econometric estimates to simulate how widespread adoption of autonomous vehicles would impact the economic performance measures by reducing congestion levels in California. Finally, we extrapolate those findings to the nation by presenting a base-case and more-conservative estimates for sensitivity analysis.
We summarize our main findings here so nontechnical readers may pass over the technical chapters and skip to chapter 7, which discusses other effects of autonomous vehicles in a nontechnical manner. To predict the effects of autonomous vehicles on the California economy, we assume that autonomous vehicles would lead to a 50 percent reduction in freeway delays (see chapter 6 for details). Strikingly, we estimate that if California motorists had been using self-driving vehicles in 2010, the state would have created nearly 350,000 additional jobs that year, increased its real GDP by $35 billion, and raised workers’ earnings by nearly $15 billion.1 Using available 2007 data on freight flows between California counties as a measure of trade benefits, we find that autonomous vehicles would have increased the value of intercounty shipments of commodities by $57 billion.
Extrapolating those effects to the nation, we predict that autonomous vehicles in 2010 would have added 3 million additional jobs to the U.S. economy, raised the nation’s annual growth rate by 1.8 percentage points from a 2010 baseline GDP of about $14.6 trillion, and increased annual labor earnings by more than $100 billion. In other words, the economic benefits to the United States are nearly eight times greater than those to California.
Naturally, an intuitive back-of-the envelope calculation of the potential economy-wide benefits that autonomous vehicles could generate by reducing congestion would be helpful. However, this type of calculation is difficult here because we are estimating benefits in nontransport sectors, where the benefits depend in a complex way on how consumers and firms respond to travel-time savings from reduced congestion by adjusting labor supply, production, and the like to reach new equilibrium levels of wages, employment, trade flows, and GDP growth.
Intuitively, it is well known that a given reduction in travel costs generally increases the value of a transportation network and the social gains that it generates by manyfold because the cost savings exponentially increase accessibility to more parts (nodes) of the network. Thus although our estimates may seem implausibly large, the significant improvements in mobility attributable to autonomous-vehicle use positively affect a considerable quantity of the nation’s inputs (labor and capital) and outputs that are transported on the extensive integrated network of local, state, and federal roads.2 Sensitivity analysis also indicates that even if very conservative assumptions are made about the effects of autonomous vehicles on reducing congestion, such as they are 50 percent lower for the United States as a whole than they are for California, then autonomous vehicles would still generate significant improvements in the nation’s annual rate of growth that approach 1 percentage point.
In sum, substantial reductions in congestion and improvements in travel time and travel-time reliability for automobiles and trucks have the potential to generate macroeconomic (supply-side) stimulative effects because more efficient transportation can facilitate favorable improvements in the labor, urban, trade, and industrial sectors that result in more people working, shopping, trading, and producing goods. The additional employment and better job-matching attributable to faster and more reliable commutes, the increased freight flows attributable to the reduction in transport costs, and the higher productivity attributable to reduced transit time for both capital and labor could combine to significantly increase the U.S. growth rate. Moreover, the findings we summarize from our simulations to preview those effects account only for the economic effects of self-driving cars through reductions in congestion, and so almost certainly understate the benefits of autonomous vehicles. As explained in chapter 7, there are a myriad of other ways in which autonomous vehicles could improve social welfare.
4
Estimating the Effects of Congestion on Economic-Performance Measures
Only fragmentary evidence exists of the effects of congestion on an economy. For example, Hymel (2009), Sweet (2014), and Angel and Blei (2015) find that highway congestion is associated with slower job growth in U.S. metropolitan areas, while Light (2007) uses the Bureau of Labor Statistics’ American Time-Use Survey to estimate reductions in workers’ productivity and income that are caused by traffic delays from highway congestion. This chapter provides a more comprehensive empirical picture of the effects of highway congestion on the U.S. economy as a basis for estimating the potential benefits from autonomous vehicles’ effect on congestion.
This picture reflects estimates of the causal effect of highway congestion on the growth rates