City of Atlanta E-bike Rebate Program, Pass-through and Additionality Implications

• In Event: Critical Minerals, EV Adoption, and Clean Energy Demand

Abstract

E-bike adoption is rapidly growing with over 500,000 bikes sold annually. Looking to spur adoption, there have been several e-bike subsidy programs across the country – statewide programs include those in California, Colorado, Connecticut, and Minnesota; city-wide programs in Tampa Bay, Denver, and Raleigh. These programs could allow for the transition from motor vehicles to e-bikes for local trips, reducing congestion and transportation-related pollution, and could be at least part of the solution to the chicken-egg problem of bike infrastructure and bikers using it. Yet, there is sparse peer-reviewed research evaluating the effectiveness of these programs.We examine The City of Atlanta E-bike Rebate Program, administered by the Atlanta Regional Commission (ARC). This rebate program provides an ideal setting for causal analysis of program effectiveness due to its random selection of rebate recipients from a pool of applicants. Of note, the program was designed to distribute 75% of the funds to income-qualified applicants and 25% to non-income qualified applicants. Randomization of selection was conducted within those groups. In our analysis, we focus on the question of additionality of the rebate program. That is, how did receiving a rebate through the program impact e-bike purchasing behavior relative to what it would have been in the absence of a rebate?To get at this question, we leverage three surveys covering a total of 1,750 individuals: an initial pre-rebate survey of all applicants, a post-rebate survey of all rebate recipients, and a post-rebate survey of those who were not selected to receive a rebate. From these surveys, we empirically examine how receiving a rebate affects (1) the probability of purchasing an e-bike, (2) the expected expenditure on an e-bike, and (3) how these differ between income-qualified and non-income-qualified applicants as defined by the program.          
 
To analyze the question of how receiving a rebate affects the probability of purchasing an e-bike, we first compare purchasing behavior of rebate recipients to non-recipients. Leveraging the randomization of the program, we find that 93% of income qualified recipients purchase a bike while only 6% of non-recipients purchased an e-bike. This implies that receiving the rebate increased the probability of purchasing a bike by 87pp. For non-income qualified applicants, around 63% of recipients purchased an e-bike while 15% of non-recipients still purchased an e-bike. Thus, the rebate increased the probability of purchasing a bike by 48pp, smaller than for income-qualified applicants.           

We plan to use econometric modeling to analyze how individual observable characteristics, such as age and income, influence purchasing behavior and the impact of receiving a rebate. Using a logistic regression model, we will analyze the probability of purchasing an e-bike. Using an OLS regression, we analyze how receiving a rebate affects e-bike expenditures.

Our analysis provides insights that are valuable to the development of future e-bike subsidy programs and for subsidy programs more broadly. Preliminary results indicate a sizeable impact of the Atlanta Rebate program on e-bike adoption. Further, we find higher impact for income-qualified recipients, indicating the value in targeting this population with subsidy programs.

Author

• Derek A McNutt, Georgia Institute of Technology
  Presenting Author