Working papers
Rivalry in consumption generates variation in the choice sets decision-makers face. Neglecting such variation generates biased demand estimates. Whereas previous research used information from periodic inventory systems to correct this bias, this study instead uses novel data from a real-time inventory system that records all public charging sessions by electric vehicles in a residential area of Amsterdam.
For each transaction I can reconstruct a user's exact set of available alternatives at the time of arrival which allows me to impose deterministic choice set constraints. I show that this significantly improves demand forecasts for local charging facilities, with estimated differences between observed and latent demand reaching up to 40 percentage points at some locations.
For the current demand and infrastructure, consumption rivalry however remains a local phenomenon limited to peak hours. A policy analysis suggests that it is not cost effective to install more capacity just to mitigate consumption rivalry. Instead, it may be better to implement policies aimed at increasing the utilization of existing capacity.
Thesis series
Previous studies based on US data have shown that people respond to lower fuel prices by substituting to a premium brand. This thesis investigates whether this "fungibility of money" assumption is also violated in the Dutch context. No evidence for this effect is found. This suggests that: 1) results obtained with US data do not automatically translate to the Dutch context; 2) to have Dutch consumers switch to different, cleaner fuel-types, price incentives may have to complemented with efforts that emphasize differences in product characteristics.
E-bikes have become an increasingly popular mode of transport among Dutch households. This thesis convincingly shows that e-bike ownership adoption has taken a flight, and whereas the group of early adoptors predominantly consisted of retired people, e-bike ownership has become increasingly more common among younger persons as well. Households living in rural areas who need to cover longer distances are found more likely to own an e-bike, and in more recent years, e-bike ownership seems to positively correlate with household income. From a public health perspective, an important question is whether the e-bike is used for trips that were previously completed by car, or for trips that were previously completed using a conventional bike. The research shows that as a mode of transport the e-bike increasingly replaces the conventional bike and not the car.
This study investigates the impact of price promotions (happy hours) within the gasoline industry. Building on existing sales promotion literature, this study examines the influence of happy hours on total sales, decomposed by the number of transactions (transaction effect) and average purchase value (spending effect). I find evidence that happy hours result in 41% more transactions and 57% more liters sold across all stations. In contrast, no evidence is found of an increase or decrease in the average purchase value. One explanation could be that customers find it difficult to time their purchases efficiently (e.g., empty tank) due to the short period of happy hours. Furthermore, contrary to my expectations, this study finds that happy hours have a greater impact on sales at smaller stations. To conclude, my findings show that price promotions are beneficial in the gasoline industry, due to their substantial impact on transaction volume.
Analyses of gasoline demand have thus far relied on aggregate data, which is not able to capture consumer decisions at the individual level. In this thesis, a rich dataset with almost 6,000 unique individuals, accompanied by transactional and customer information, is utilized to overcome this limitation. The first empirical specification estimates the price elasticity of gasoline demand. The resulting estimates are consistently negative and inelastic but do show considerable variation across distinct groups. Next, the effects of time-limited offers in the form of happy hours on gasoline demand are analyzed. The model estimates an increase of around 4.6 transactions per hour and 160 more liters of gasoline sold per hour during a happy hour at a specific filling station. The estimated coefficients suggest an increase in the amount of gasoline sold during happy hours through the extensive margin.
This thesis focuses on the effect of a one-hour discount at tank stations on fuel sales. Analyses, in line with the inelastic demand of fuel, suggest no increase in the liters purchased by a customer. However, during a happy hour, an increase is observed in the total number of liters purchased, suggesting that sales increase through the extensive margin. Some evidence is also found for anticipation effects and post-promotion dips (customers purchase less fuel in the days prior to -or after- a happy hour).
Work in progress
Carbon offsets and information requests: A field experiment with gas station visitors - Anouk Schippers, Gert-Jan Romensen and Adriaan R. Soetevent
We ask gas station visitors to estimate the carbon offset cost of their fuel purchase and to indicate an offset donation to a charity involved in reforestation projects. Using on-site pump displays, we experimentally vary whether visitors can observe or request information about the true carbon offset cost while making the donation decision, and consider the impact of a monetary incentive and social nudge on the demand for this type of information. We find that virtually all visitors (95%) overestimate the cost of offsetting the carbon emissions of their fuel transaction. The estimates are on average more than a factor four larger than the true costs. Donations are strongly associated with consumer beliefs about carbon offset costs: a one euro increase in expected costs increases pledged donations with about €0.60. Information on the true offset cost has no impact on donations, and neither is there any interest from consumers in requesting such information. Monetary incentives or social nudges do not change the demand for information. An additional analysis with payment data shows that only 2.6% of the subjects follow through on their pledged donations and transfer the amount.
This paper also appears as Chapter 5 in Anouk Schippers PhD Thesis Prosocial Behavior in Markets (Defense date: 14.11.2023)
focuses on the effect of a one-hour discount at tank stations on fuel sales. Analyses, in line with the inelastic demand of fuel, suggest no increase in the liters purchased by a customer. However, during a happy hour, an increase is observed in the total number of liters purchased, suggesting that sales increase through the extensive margin. Some evidence is also found for anticipation effects and post-promotion dips (customers purchase less fuel in the days prior to -or after- a happy hour).
P(l)ay-As-You-Go: Large-Scale Periodical Risk-Elicitation in the Field - Gert-Jan Romensen and Adriaan R. Soetevent
We use digital technologies in business-to-consumer (B2C) communication for the large-scale, longitudinal measurement of individual risk attitudes. We use of the on-site pump displays of a fuel company to create a unique panel data set of over 6,500 visitors who over the course of four years repeatedly and frequently complete an incentivized (Bomb Risk Elicitation Task, BRET) and non-incentivized (SOEP risk question) risk elicitation task. Our approach overcomes the challenges of user time constraints and limited control that commonly hinder the use of incentive-compatible experimental risk-elicitation methods in the field.
We find that more than 80% of the decisions in the BRET-game are consistent with risk aversion. No gender differences are observed in the BRET, nor do other demographics explain some of the variance. The correlation with responses to the SOEP question is positive, but small in magnitude. Neither COVID-19 nor the fuel price-hikes following the invasion of Ukraine seems to have fundamentally changed general risk preferences among the sampled population.
Don’t wait on the world to change! How pernicious technophilia causes group inaction – an experiment - Lennart Stangenberg and Adriaan R. Soetevent
Many forms of individual consumption create negative group externalities, the cost of which accumulate over time. We show that introducing a chance that technological progress reduces this cost has a pernicious effect on a group’s ability to coordinate on the individually and socially optimal solution of settling the cost immediately. Our 2x2 experimental design extends the game in Walker et al. (2000) to a multi-period setting with a consumption stage. We vary whether there is a possibility of exogenous reductions in the abatement cost and how group members share the accrued cost: equally or in proportion to each person’s consumption share. We find that, independent of the cost-sharing mechanism, fewer participants vote for settlement in the progress treatments, thereby reducing social welfare. The hope for technological breakthroughs leads to costly inaction.
A field experiment on financial incentives, social preferences, and social comparisons to improve parking by users of bike-sharing services - Duan Su, Yacan C. Wang and Adriaan R. Soetevent
Reference dependence in voting behavior: Experimental evidence - Oliver Herrmann, Richard Jong-A-Pin and Lambert Schoonbeek
This paper will also appear as a chapter in the PhD Thesis of Oliver Herrmann.