In the history of pension funds, the goal of their investment policy was capital growth within a limited investment risk. This goal was part of their fiduciary duty, the legal responsibility to act in the beneficiary's best interest. In recent years, the purpose of pension funds has shifted to aim for long-term social gains in addition to focusing on financial returns. Pension funds have increasingly integrated social, environmental, and ethical considerations in their investment process. This has been attributed to various factors, including the positive correlation between ESG and financial performance but also external pressures from non-governmental groups fuelling public opinion through the media and actions to act more responsibly. The participants must support decisions relating to a responsible investment policy for the pension funds. However, pension fund beneficiaries' preferences for integrating responsible criteria have not been sufficiently investigated.

The aim of this study can be split up into three parts. First, to understand the preferences on the sustainable criteria from the beneficiaries of ADP. Secondly, based on research findings and interviews, find a suitable approach to implementing this in the investment strategy of ADP. Lastly, recommendations will be made on other stakeholders' steps for a successful SRI implementation.

First, this research sets up a choice experiment. It then uses discrete choice modelling to determine participants’ preferences and willingness to trade financial profits against SRI goals using Random utility maximisation (RUM). With this method, we find that participants are willing to contribute part of their pension income to implementing SDGs. Furthermore, we can see that participants think it is especially important that pension funds focus on reducing inequality. There are several motives for integrating ESG into the decision-making process. However, the execution of policy is different per motive. When participants want to integrate ESG to achieve higher returns, the implementation differs from when the motive is to impact the world positively. Therefore, additional analysis has been executed.

To determine whether the participants consider an equal contribution between the step from no attention to do no harm and do no harm and do good. The utility contribution of each attribute level is also measured; we can observe that for most SDGs, there is little willingness to put in extra money to make a positive contribution to the SDGs, except for reducing inequality. The RUM model is very good for calculating the weights given to the attributes that best fit all participants' answers. However, the model does not reflect whether there is heterogeneity in the population. Therefore, interaction effects on individual characteristics are estimated; this method allows heterogeneity between different groups based on their characteristics. However, when using this methodology, an issue could arise. Observable covariates may not produce some heterogeneity. Therefore, latent class analysis was also used.

Through the analysis of interaction effects, we find that there are already many differences in preferences based on individual characteristics. In the latent class analysis, we see three distinct groups. In these classes, we find a large group that gives little utility to pension income and finds all the SDGs important. The 2nd and 3rd groups are smaller, containing about 18% of the population. Group 2 especially finds retirement income very important, and group 3 finds reducing inequality very important. In this analysis, we find that the strong preference of groups 2 and 3 greatly impacts the RUM model. From these analyses, we find that there is heterogeneity in the population. We can use this input when we look at how participants’ preferences can be implemented in the investment policy.

By first analysing the factors and stakeholders involved, we can find which obstacles to overcome when implementing policy. When analysing stakeholders, we find three different groups. First, we find the societal actors; these mainly influence the shaping of policies. Secondly, we find the political actors who influence the rules of the game. Lastly, we find the economic actors who influence the policy's implementation. In this analysis, several challenges have been found.

Lastly, we have looked broadly at options for implementing participants' preferences. In conclusion, the analysis shows that there are different ways in which a pension fund can develop an SRI policy. This can be done differently, depending on the participant's wishes and preferences and the pension fund's contract form. The pension fund first needs to develop a vision of how they want to achieve a sustainable world. After this, they can create an investment strategy and investment mandate. Policymakers also have a role to play in changing the financial sector. Besides implementing policies to improve transparency, they could also implement financial incentives so it will be more financially attractive to adopt ESG practices.

We can conclude that participants are willing to contribute part of their pension income to the implementation of SDGs. Furthermore, we can see that participants think it is especially important that pension funds focus on reducing inequality. Based on the latent class analysis results, we can see that different groups within the population have other preferences. This is important to take into account when formulating policies.

The model results and our analysis are subject to limitations. The model could be improved by allowing participants in a choice experiment to allocate money between different investment options, including policies only focusing on returns. Furthermore, the results could be influenced by self-selection and the population focus. Therefore, we recommend studying a more diverse population with participants from different pension funds. Lastly, we focused on how the results could be implemented and what tools the pension fund could use to accomplish an impact, as most of these recommendations are based on qualitative research. It may be interesting to conduct a quantitative analysis of the established objectives based on implemented policies.