Forecasting Americans’ long-term adoption of connected and autonomous vehicle technologies

The introduction of connected and autonomous vehicles (CAVs) represents a significant shift in the automotive and transportation sectors. CAVs promise various benefits, including reduced crash rates and congestion. However, concerns exist regarding security, safety, privacy, and potential negative economic consequences during the transition to automation. Policymakers, industry professionals, and researchers require accurate forecasts of CAV adoption to guide their decisions. Previous predictions have relied on extrapolations of past trends, expert opinions, or supply-side factors, lacking a comprehensive understanding of demand-side factors. This study addresses this gap by employing a simulation-based fleet evolution framework that integrates both supply-side (technology costs) and demand-side (WTP, vehicle transaction decisions, government regulations) variables to predict long-term CAV adoption in the United States.

The literature review highlights the scarcity of long-term CAV adoption forecasts conducted by academic researchers. Most existing studies come from consulting firms or private enterprises, often lacking methodological transparency. The review summarizes forecasts from various sources regarding future shares of self-driving vehicles, sales of autonomous vehicles (AVs), and vehicle-miles traveled by self-driving cars. These forecasts vary widely. For example, estimates of AV fleet penetration in 2050 range from 30% to 100%. The authors note that most previous research has focused on understanding public perception and WTP for CAV technologies rather than forecasting long-term adoption rates. This study differentiates itself by incorporating demand and supply-side factors, including the impact of government regulations.

The research methodology involved a US-wide online survey of 2167 Americans using Qualtrics and the Survey Sampling International (SSI) continuous panel. The survey collected data on household vehicle ownership, vehicle transactions, WTP for various CAV technologies, opinions on CAVs, and travel patterns. The sample was weighted to reflect the 2013 American Community Survey's Public Use Microdata Sample (PUMS) demographics, correcting for overrepresentation of certain groups. Respondents' home locations were geocoded to incorporate the impact of built-environment factors on vehicle transaction and technology adoption decisions. The data were then used in a simulation-based framework to forecast the long-term adoption of CAV technologies. The simulation involved several stages, starting with a multinomial logit (MNL) model estimated using BIOGEME to predict households’ annual vehicle transaction and technology adoption decisions (sell, buy, replace, add technology, do nothing). Separate binary logit models were used to predict the likelihood of buying one or two vehicles and the choice between new and used vehicles. These probabilities, along with WTP data and technology price forecasts, were then used in Monte Carlo simulations to predict yearly technology adoption rates under eight different scenarios.

The survey revealed interesting insights into public attitudes toward CAV technologies. Most respondents perceived themselves as good drivers, enjoyed driving, and preferred a wait-and-see approach to new technologies. While a majority found AVs useful, a significant portion expressed fear. WTP for CAV technologies varied significantly, with a considerable proportion of respondents unwilling to pay for advanced automation features. The average WTP for Level 3 and Level 4 automation was considerably lower than the projected prices, even after five years. The simulations, conducted under eight scenarios that varied technology prices (5% and 10% annual drops), WTP (0%, 5%, and 10% annual increases), and government regulations (mandating ESC and connectivity), showed substantial differences in long-term adoption rates. Under the scenario with a 5% annual price drop and constant WTP, Level 4 AV penetration reached 24.8% by 2045. This share jumped to 87.2% with a 10% annual price drop and a 10% annual WTP increase. Mandated adoption of ESC and connectivity significantly accelerated the adoption of those technologies. The scenarios demonstrated the substantial influence of both technology pricing and WTP on the adoption of CAV technologies, particularly advanced automation features.

The findings highlight the complex interplay of factors influencing CAV adoption. Public perception, WTP, technology costs, and government regulations all play crucial roles. The results emphasize the critical need for policies that encourage WTP for CAV technologies, alongside rapid reductions in technology costs. The study's findings underscore that the absence of significant WTP growth or supportive policies could significantly hinder the widespread adoption of CAVs in the near future. The model's sensitivity to different scenarios demonstrates the need for further research into the dynamic evolution of WTP and the potential impact of unforeseen events on adoption rates.

This study provides valuable insights into the potential future of CAV adoption in the United States. The simulation-based framework effectively integrated both supply and demand side factors, providing a comprehensive forecast under diverse scenarios. The results suggest that government regulations play a crucial role in accelerating adoption of some technologies, while substantial price reductions and WTP increases are necessary for widespread adoption of advanced automation. Future research could refine the model by incorporating household evolution dynamics and more sophisticated modeling of WTP evolution over time.

The study's limitations include the reliance on a cross-sectional survey, which may not perfectly capture the dynamic changes in public perception and WTP over time. The assumed constant annual rates of WTP increase might not reflect the complex reality of changing social and economic factors. The model also does not explicitly account for potential impacts of autonomous vehicles on vehicle ownership patterns. Further research could address these limitations by employing longitudinal data, developing more nuanced models of WTP dynamics, and incorporating the effects of CAVs on vehicle ownership.