Inflation concerns and green product consumption: Evidence from a nationwide survey and a framed field experiment

We want to add a new perspective to the discussion on the determinants of individuals' consumption of sustainable products: expected inflation and, hence, the future value of income. Given that a meta-analysis on consumer demand for organic food has found that income is only a partial explanatory factor and superseded by psychographic variables (Aschemann-Witzel and Zielke, 2017), we posit that inflation expectations negatively affect organic food consumption ―which might explain the decrease in sales in the year 2022 (e.g. -3.5% in Germany).

We plan to conduct a framed field experiment (Harrison and List, 2004), where participants know that they take part in a study, but we observe actual decision making in a natural task with significant stakes, i.e. we ask participants to fill their shopping basket (up to a given budget) in an online supermarket. To incentivize participants to fill their basket according to their actual preferences, we will randomly draw 10 participants who will actually receive their chosen basket. Organic products can be easily identified using the filter option of the supermarket's online store.

To causally assess the impact of inflation expectations, we will use the priming technique to make individuals' fears about future price increases more salient by asking some price and inflation-related questions. To compare this rather psychological effect with a sizeable relaxation of the budget restriction, we will also have a third treatment group without the priming but with a 50% increase in their available budget.

Finally, we also want to investigate whether inflation expectations might even counteract any measures that could otherwise spur the consumption of organic products. Therefore, we will implement two additional treatment groups. In combination with a significantly larger budget, nudging participants with social norms (e.g. Demarque et al., 2015) ―which we obtained from a quasi-representative nationwide pre-experimental survey― should increase the consumption of organic products. Implementing the same treatment combined with the inflation priming and using a difference-in-difference estimator, we can analyze whether the fears about future inflation hinders the effectiveness of measures that should spur the consumption of sustainable products.

We posit the following hypotheses:

H1: Fears about future inflation will reduce the share of organic products.
H1a: The effect is driven by individuals with a low environmental attitude.
H2: The budget relaxation has (if at all) a lower effect (in absolute terms) on the share of organic products than the priming intervention.

H3: A higher budget joint with informing participants about the social norms increases the share of organic products compared to baseline consumption.
H4: Fears about future inflation will negate this positive effect.

2023-10-02

2023-11-25

Organic product consumption

The dependent variable will be the share of money spent for organic products.

We will do both the binary treatment testing using parametric or non-parametric tests as well as regression analyses. The latter does not only allow to control for important characteristics such as household income and size, gender, age, educational background and environmental attitude. Moreover, we will conduct a difference-in-differences analysis to test H4. To analyze potential heterogeneity as posited in H1a, we will conduct a median split regarding participants' environmental attitudes based on a simple 3-items survey measure. Furthermore, we will analyze whether effect sizes depend on the degree of concern regarding future inflation (somewhat concerned, very concerned) using subsample analyses.

We used the postal codes to apply a block design for randomization. This method ensures a (roughly) equal distribution of all treatment conditions within areas and, therewith, we account for socio-demographic differences between areas.

The unit of randomization is at the individual level (i.e. registered participants).

No

Not relevant.

We plan to have 125 observations for each group, resulting in a total of 625 participants.

The target sample size is 625 participants, with 5 different groups, this results in around 125 participants per group.
There will be five different conditions to which participants will be allocated to (using stratified randomization based on postcodes):
1) Baseline (125 participants)
2) Prime (inflation) (125 participants)
3) Money (the budget rises by 50%, no priming) (125 participants)
4) Money+Norm (as before but with information about the social norm) (125 participants)
5) Prime+Money+Norm (125 participants)

We collected receipts from larger supermarket shopping trips to calculate the average share and the standard deviation regarding the purchase of organic products. Power calculations based on this data (targeting a medium-sized effect size, Cohen's d=0.5) resulted in 63 observations per treatment group. Given that our pre-experimental survey results suggest that individuals with a rather low environmental attitude might drive the expected effects, we double the sample size and plan to have 125 observations each, resulting in a total of 625 participants.