Bringing Democracy to Finance: A Field Experiment on Pension Fund Members' Preferences for Impact Investing

Experiment setup:
In this field experiment, we want to assess whether participants desire to expand impact investing in a democratic setting and whether recommendations from a deliberative forum help shape their decisions. To this end, we first inform participants about impact investing and assess their priors regarding (sustainable) investing knowledge, return expectations, risk perception, impact beliefs, and preferences for sustainable investing. Subsequently, we treat them with two different forms of information and ensure they understand the information or assign them to a control group without information (see below). Next, we ask participants whether to expand, keep unchanged, or reduce impact investing in their pension fund with a binding vote. We also ask participants to vote with commitment on the topics and location of impact investing. Last, we assess participants’ trust in their pension funds, the provided information, and commitment. We also collect demographic information and ask a control question to determine whether they understood the definition of impact investing.

Interventions:
Alongside participants’ general tendencies to vote for expanding impact investing, we study whether they follow impact investing recommendations from a deliberative forum when provided with this information. This allows us to test the generalizability of forum outcomes in representing the general population. To this end, we introduce a forum information treatment covering a random sample of 30% of participants.
The key information participants receive in this treatment is the share of forum participants who want to expand impact investing, keep impact investing unchanged, or reduce impact investing; their expectations about the effect of impact investing on returns; and their beliefs regarding the environmental and social consequences of impact investing. We communicated that this information was collected during a deliberative forum involving 43 participants from Pensioenfonds Detailhandel. We elaborated on the forum's expert-led informational sessions, deliberative nature, and collective recommendation-building aspects. Further, we emphasize that these recommendations were diligently made as they result from a three-day intervention, for which participants traveled throughout the entire country. Last, we convince participants that participants like them make these recommendations, as the forum was representative of the average Pensioenfonds Detailhandel demographics. We provide the exact treatment information at the end of this pre-registration form.
One concern with the above setting is that our results might be driven by peer effects rather than being specific to our deliberative forum. We address this concern by introducing an additional peer effects treatment covering 30% of the sample. To adequately and truthfully measure this, we provide participants with identical information from the forum on participants’ willingness to expand impact investing, return expectations, and impact beliefs. We explain that this information was collected through a 15-minute survey (the method of collecting data during the deliberative forum). As with the forum treatment, we emphasize the representativeness of respondents. The sole difference in treatments is the knowledge that these recommendations stemmed from a deliberative forum. The remainder of the participants (40%) serve as a control group and provide no additional information.

2025-05-12

2025-06-30

• Participants’ vote on impact investing (expanding, keeping unchanged, or stopping)

• Participants’ vote on the topic and location of impact investing
• Impact investing knowledge, preferences, return and risk expectations, and impact beliefs

Experiment setup:
In this field experiment, we want to assess whether participants desire to expand impact investing in a democratic setting and whether recommendations from a deliberative forum help shape their decisions. To this end, we first inform participants about impact investing and assess their priors regarding (sustainable) investing knowledge, return expectations, risk perception, impact beliefs, and preferences for sustainable investing. Subsequently, we treat them with two different forms of information and ensure they understand the information or assign them to a control group without information (see below). Next, we ask participants whether to expand, keep unchanged, or reduce impact investing in their pension fund with a binding vote. We also ask participants to vote with commitment on the topics and location of impact investing. Last, we assess participants’ trust in their pension funds, the provided information, and commitment. We also collect demographic information and ask a control question to determine whether they understood the definition of impact investing.

Interventions:
Alongside participants’ general tendencies to vote for expanding impact investing, we study whether they follow impact investing recommendations from a deliberative forum when provided with this information. This allows us to test the generalizability of forum outcomes in representing the general population. To this end, we introduce a forum information treatment covering a random sample of 30% of participants.
The key information participants receive in this treatment is the share of forum participants who want to expand impact investing, keep impact investing unchanged, or reduce impact investing; their expectations about the effect of impact investing on returns; and their beliefs regarding the environmental and social consequences of impact investing. We communicated that this information was collected during a deliberative forum involving 43 participants from Pensioenfonds Detailhandel. We elaborated on the forum's expert-led informational sessions, deliberative nature, and collective recommendation-building aspects. Further, we emphasize that these recommendations were diligently made as they result from a three-day intervention, for which participants traveled throughout the entire country. Last, we convince participants that participants like them make these recommendations, as the forum was representative of the average Pensioenfonds Detailhandel demographics. We provide the exact treatment information at the end of this pre-registration form.
One concern with the above setting is that our results might be driven by peer effects rather than being specific to our deliberative forum. We address this concern by introducing an additional peer effects treatment covering 30% of the sample. To adequately and truthfully measure this, we provide participants with identical information from the forum on participants’ willingness to expand impact investing, return expectations, and impact beliefs. We explain that this information was collected through a 15-minute survey (the method of collecting data during the deliberative forum). As with the forum treatment, we emphasize the representativeness of respondents. The sole difference in treatments is the knowledge that these recommendations stemmed from a deliberative forum. The remainder of the participants (40%) serve as a control group and provide no additional information.

Computer - outsourced to an external party (Flycatcher)

Person

No

We send emails to 220k participants. Bauer et al. (2021) found that the same pension fund's response rates were 6.3% to 6.7%. We take a conservative lower-bound response rate of 3% and expect 3%* 222k = 6,660 participants.
Bauer, R., Ruof, T., & Smeets, P. (2021). Get real! Individuals prefer more sustainable investments. The Review of Financial Studies, 34(8), 3976-4043.

We send emails to 220k participants. Bauer et al. (2021) found that the same pension fund's response rates were 6.3% to 6.7%. We take a conservative lower-bound response rate of 3% and expect 3%* 222k = 6,660 participants.

We expect 6,660 participants. We assign 40% of them to the control group (2,664), 30% to the deliberative forum treatment (1,998), and 30% to the peer effects treatment (1,998).

Minimum detectable effect size for primary outcomes (accounting for sample design and clustering) • For the first main question, we assess the share of participants voting for impact investing. Since it is a fraction rather than a difference in effect sizes, power calculations are not needed. • For the second main question, whether the effectiveness of forum information is higher than peer effects, we compare the share of participants voting to expand impact investing across treatments, each with an N of 4,158. Assuming a power of 80% and a significance level of 5%, we can detect effect sizes of 0.063 standard deviations.

Democraticing Finance for Pension Members - AEA RCT registry