Research

with John Kramer

Stock market participation remains low even among households that can afford to invest. Building on Duraj et al. (2025), we test whether this gap is sustained by a specific misperception—that succeeding in financial markets requires expertise, continuous monitoring, and active trading, reflecting a broader belief that effort is rewarded and that experts beat the market. In a survey experiment conducted in May 2026 with a representative sample of Danish adults aged 18–55, unconstrained non-investors were randomized to watch either a video explaining the efficient market hypothesis, diversification, and index funds (treatment) or a general video about the stock market (control). We measured perceived time requirements, willingness to pay to outsource portfolio management, and investment intentions, linking responses to administrative registry data on market entry and income and wealth data. The treatment shifts beliefs about diversification, time requirements and the perceived unpleasantness of investing. We replicate Duraj et al.’s gap in perceived time costs between investors and non-investors, though even some investors report large time commitment, and we find non-investors are markedly more pessimistic about historical returns.

Data collected May 2026

with John Kramer

The Danish government introduced the Aktiesparekonto (ASK), a tax-advantaged stock savings account intended to make capital taxation more transparent and to lower effective tax rates on certain investments. Despite these benefits, take-up has remained limited among both existing and prospective investors. We study whether low uptake reflects an information problem. In a survey experiment with a representative sample of Danish adults aged 18–55, financially unconstrained investors who do not yet hold an ASK were randomized into a treatment group, shown concrete facts about the account—including its tax advantages—or a passive control group. We then elicit awareness of the account’s features and intentions to open one, and link survey responses to administrative data to observe whether respondents subsequently open an ASK. This design lets us separate whether limited adoption stems from unawareness of the ASK’s benefits or from other frictions, and to quantify how far a low-cost informational nudge can move both stated intentions and actual account openings. 

Data collected May 2026

with Daniel Riegels

Dynamic pricing promises to make consumers internalize the marginal social cost of their electricity use and shift it to off-peak hours. But this rests on two assumptions rarely tested together: that consumers perceive the price signal, and that they can act on it. When either fails, a price meant to improve efficiency can instead shift costs onto those least able to respond. Combining a large survey with high-frequency consumption data, we measure both how aware households are that they face time-varying prices and what it would cost them to shift consumption away from peak hours — separating non-response that reflects genuine inflexibility from non-response that reflects inattention. The project also examines how the rapid adoption of electric vehicles is reshaping the distribution of flexibility across households.

Work in progress.

with Mark A. Andor, Sven Hansteen and Eva Hümmecke

Meeting Denmark’s target of reducing emissions by 70\% by 2030 requires decarbonizing residential heating, and policymakers have signalled an ambition to phase out fossil-based systems entirely. While the transition has so far relied on subsidies, these have shown limited effectiveness and tend to benefit higher-income households. This raises the possibility of a shift toward mandatory heat pump installation — a policy whose distributional consequences remain poorly understood. This paper provides an ex-ante assessment of who would bear the burden of such a mandate. Rather than modelling behavioural responses, we use detailed administrative register data from Statistics Denmark on roughly 75,000 homeowners to simulate the upfront investment each household would face and relate it to their financial resources. We classify households along a gradient of financial resilience based on liquid resources, assets that can be liquidated, and access to credit. Our preliminary results indicate that heating technology varies more by location and housing type than by income, but that a non-negligible share of homeowners lack both the assets and the borrowing capacity to finance the transition without external support. This work is ongoing, and results are preliminary.

with Daniel Riegels

We evaluate a randomized information provision experiment with $N=509$ survey participants in Denmark
aimed at increasing everyday electricity demand flexibility by highlighting environmental and social benefits of more flexible consumption habits. Survey responses are linked to rich administrative data, including hourly household-level smart meter readings of electricity consumption. Participants were not aware that they were participating in an experiment, or that their electricity consumption was observed. We find our information treatment increases intentions to consider peak hours by $14.1$ percentage points $(p<0.01)$ among those not already paying attention to peak hours. However, we find no effects on plans to alter any specific electricity behaviors or on observed electricity consumption. We provide evidence that 1) pro-environmental preferences extend to electricity use, 2) demand flexibility is hindered by both an information and an intention gap, and 3) correcting just the information gap is insufficient to meaningfully alter behavior.

Data collected – Working Paper coming soon

I study consumer inertia in the Danish retail electricity market, where switching can
yield substantial savings and frictions are low. Using administrative smart-meter data
for 200,000 randomly-sampled households, a survey of 9,047 nationally-representative
consumers, and a randomized experiment, I test whether attention and reduced switching
costs increase switching. Targeted savings information and brokerage services raise
switching by 0.8–1.3 percentage points over three months, despite average potential
savings of $140–$360. Behavioral frictions—especially procrastination and distrust—drive
a large gap between switching intentions and actions, suggesting that overcoming inertia
may require active policies such as default enrollment into lower-cost contracts.

Working paper

We examine the impact of a carbon tax on consumer choices via a large-scale online randomized controlled trial. Higher taxes generally reduce the demand for high-carbon goods. Compared to an import tax, a carbon tax reduces demand when the tax is zero (i.e., announced but not levied) but shows relatively higher demand for high-carbon goods when a positive tax is introduced. This contradiction of basic price theory is entirely driven by climate-concerned consumers. Our findings suggest that carbon taxes can crowd out climate concerns, leading to important implications for policy.

Working Paper

with Martin Dufwenberg and Katja Görlitz

Peer evaluation tournaments are common in academia, the arts, and corporate environments. They make use of the expert knowledge that academics or team members have in assessing their peers’ performance. However, rampant opportunities for cheating may throw a wrench in the process, unless, somehow, players have a preference for honest reporting. Building on Dufwenberg & Dufwenberg (2018)’s  theory of perceived cheating aversion, we develop a multi-player model where players balance the utility of winning against the disutility of being identified as a cheater. We derive a set of predictions, and test these in a controlled laboratory experiment.

Working paper

If experimental subjects arbitrage against market interest rates when making intertemporal allocations of cash, the data will reveal nothing about subjects’ discounting, only uncovering subjects’ market interest rates. If subjects instead frame choices narrowly, they will plan to spend cash rewards when received, implying cash has properties similar primary rewards. We test arbitrage directly by forcing all transactions with subjects to go exclusively through their financial institutions via instant electronic transfers. If subjects wish to arbitrage, this should make it as easy and salient as possible. Our evidence contradicts arbitrage, and finds evidence of present bias, supporting the view that money payments in experiments can be treated as a primary reward.

Working paper