The relationship between social parameters and life satisfaction is inherently complex to estimate, as it is difficult to unambiguously conclude how the specific social parameter associated with well-being directly affects self-reported and subjective feeling. This causality challenge is well known across social science disciplines and is not unique to wellbeing economics.
Following Fujiwara & Campbell (2011)[1], we can express the subjective well-being function as follows:
SWB (SP, I, X)[1].
where SP is the social parameter we are interested in (the non-traded market good), I is income and X represents a vector of other factors that affect an individual's subjective well-being.
We can empirically estimate the function as follows:
SWBi= 𝛼 + ꞵSP SPi + ꞵIIi + ꞵXXi+ εi[2].
where i represents the individuals in our dataset, 𝛼 is a constant, ꞵSP, ꞵI and ꞵX are the coefficients of each input and εi is the error term.
This estimation can be carried out using various econometric approaches - the deciding factor is data availability (e.g. cross-sectional data, longitudinal data, experimental data). Estimation techniques that have been attempted include fixed effects analysis, natural experiments (two otherwise similar groups are exposed to different situations in an approximately random manner), real experiments (such as randomized controlled trials, RCT). However, the latter can be challenging to conduct in practice, as the situations we are typically interested in can be difficult to randomize.
In the existing literature, cross-sectional data is typically used to establish the correlation, as it will often be this type of data that is available when we need the above inputs in the same dataset. However, this can be challenging in terms of establishing a true causal relationship, and in such cases, a strong theoretical argument or previous evidence will be needed to support the relationship.
In the Open Social Value Bank, we will experiment with different approaches to establish this correlation, including drawing on the many good research experiences that exist in a Danish context and the opportunities offered by the strong Danish register data. We will thus test estimation strategies based on existing data sources from both panel surveys and cross-sectional data, but we will also experiment with data collection through choice experiments.
Regardless of the approach, there will be a consistent focus on whether the established relationship can be considered causal, as this will carry over to the next step and affect monetization. Specifically, the UK Treasury - Wellbeing Guidance for Appraisal[2] recommends that only evidence with a sufficiently convincing causal wellbeing effect should be considered for monetization.
The result of the estimation will be a quantification of the change in wellbeing resulting from a change in a social parameter. As described above, when we work with life satisfaction as a measure or indicator of subjective or assessed wellbeing, we operate with the concept of a WELLBY (wellbeing-years). A WELLBY represents a one-point change in life satisfaction for one person in one year (on the 0-10 life satisfaction scale).
.svg){kind=icon}