Green Premium Puzzle in Food & EV Markets

• Green Premium Puzzle is defined as the discrepancy where lower-emission products do not command the higher prices predicted by traditional sustainability theory.
• In climate-friendly foods, empirical studies from Sweden and Switzerland show that improved climate scores are associated with neutral or even negative retail prices.
• For electric vehicles, while manufacturing and acquisition premiums are positive, full life-cycle cost analysis reveals a negative premium that drives market diffusion.

The green premium puzzle denotes an empirical and conceptual tension in which lower-emission products fail to exhibit the positive price or cost premium predicted by standard sustainability arguments. In climate-friendly food markets, the puzzle is defined as the finding that products with better climate performance show no statistically significant average retail price uplift and are often associated with lower prices. In electric vehicles, the puzzle takes a different form: positive manufacturing or acquisition premiums can coexist with a negative full life-cycle premium and continued diffusion, implying that the sign of the premium depends on whether the relevant unit of analysis is shelf price, acquisition cost, or Total Cost of Ownership (TCO) (Nakavachara et al., 14 Jul 2025, Li et al., 2023).

1. Definitions and analytical scope

In consumer goods, the green premium is defined as a price difference associated with products that have lower life-cycle greenhouse gas emissions. In the food setting, the focal environmental attribute is climate-related performance communicated through front-of-package climate impact scores. In the EV setting, green premiums are defined as “the difference in cost between emissions-emitting technology and zero-emissions or emissions-reducing technology,” and the analysis distinguishes among manufacturing cost, acquisition cost, and full life-cycle cost (Nakavachara et al., 14 Jul 2025, Li et al., 2023).

Standard theory and conventional wisdom predict a positive premium. The food study identifies three main channels behind that expectation: willingness-to-pay for sustainability, signaling and differentiation through credible eco-labels, and nudging via simple and trusted climate scores. In that framework, a subset of consumers assigns utility to environmental attributes, firms can differentiate and extract higher margins from green segments, and salient labels can steer choices toward lower-emission products even at higher prices (Nakavachara et al., 14 Jul 2025).

The two papers imply that the phrase “green premium” is not invariant across domains or metrics. In food, the relevant outcome is retail price. In EVs, the relevant outcome can be acquisition cost or TCO. A plausible implication is that the puzzle is partly a measurement problem: an apparent premium in one accounting layer can disappear or reverse in another (Li et al., 2023).

2. Climate-friendly food and the absence of a positive retail premium

The food evidence is based on product-level data from a Swedish supermarket, as of March 2025. The study uses all food categories for which product size can be expressed in grams; items measured by volume are excluded, beverages are dropped, and some categories were removed due to insufficient information. The final Swedish sample contains 5,458 products, with climate scores available for 2,470 products; the main Swedish regressions use the subset with non-missing climate scores, such as 2,314 observations. The climate score is sourced from the RISE Food Climate Database, which provides LCA-based emission intensities in kg CO$_2$eq per kg of product. The original scale is inverted so that a higher score means better climate performance, with 5 as best and 1 as worst; the average score in the Swedish data is 3.89. The outcome variable is $\ln(\text{price})$, and controls include package size, carbohydrates, fat, protein, salt, energy in kcal, country-of-origin dummies, and Level 3 category fixed effects. Standard errors are clustered at Level 3 categories (Nakavachara et al., 14 Jul 2025).

The hedonic specification is

$$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$

With the inverted climate scale, a positive $\beta$ would indicate higher prices for better climate performance, while a negative $\beta$ would indicate a negative premium. For a one-point increase in the climate score, the percent change in price is approximately $100 \times \beta$ percent. Because the data are cross-sectional and brand and promotions are not observed, the estimates are descriptive hedonic associations rather than causal effects (Nakavachara et al., 14 Jul 2025).

The main Swedish result is a climate coefficient of $-0.0730$ with standard error $0.0471$, which is statistically insignificant and is interpreted as no average green premium. For Europe-only products, the coefficient is $-0.0868$ with standard error $0.0499$, significant at 10%, implying that a one-point improvement in the climate score is associated with an 8.68% lower price. For non-Europe products, the coefficient is $\ln(\text{price})$0 with standard error $\ln(\text{price})$1, which is insignificant. Category-level disaggregation at Level 1 yields significant negative premiums in Freezer, with $\ln(\text{price})$2 and standard error $\ln(\text{price})$3; Meat, Poultry & Charcuterie, with $\ln(\text{price})$4 and standard error $\ln(\text{price})$5; and Ready Meals & Snacks, with $\ln(\text{price})$6 and standard error $\ln(\text{price})$7. Other categories are statistically insignificant, and there are no robust positive premiums (Nakavachara et al., 14 Jul 2025).

A separate Swiss dataset, as of February 2025, serves as a robustness sample. There the climate scale is a five-star rating with different thresholds, where 5 stars is best and 1 star is worst, and the same “higher is better” convention applies. The full-sample climate coefficient is $\ln(\text{price})$8 with standard error $\ln(\text{price})$9, significant at 1%. The Europe-only coefficient is $$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$0 with standard error $$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$1, also significant at 1%, while the non-Europe coefficient is $$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$2 with standard error $$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$3, which is insignificant. Significant negative premiums also appear in Bread, Pastries & Breakfast; Frozen Food; Meat & Fish; and Pasta, Condiments & Canned Food. Across both Sweden and Switzerland, the recurring empirical pattern is the absence of a robust positive green premium and the frequent appearance of negative associations between climate-friendliness and price (Nakavachara et al., 14 Jul 2025).

3. Mechanisms proposed for the food-market puzzle

The food study interprets the negative or zero premium through both demand-side and supply-side mechanisms. On the demand side, climate outcomes are characterized as diffuse across space and time, creating psychological distance and reducing urgency and willingness to pay relative to more salient ethical attributes. Climate labels also compete with taste, health, convenience, and other labels such as organic, fair trade, and nutrition scores. Label proliferation and varying verification standards may reduce trust and generate confusion, limiting the effectiveness of climate labeling (Nakavachara et al., 14 Jul 2025).

The paper summarizes these forces with the utility representation

$$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$4

If $$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$5 is small because climate attributes are low-salience or distrusted, the climate attribute need not translate into higher observed prices. The same outcome can arise if firms price strategically or if promotions weaken price pressure. On the supply side, the paper invokes the Porter Hypothesis: environmental regulation and innovation can lower the costs of cleaner production, so firms may offer greener products at the same or lower prices, thereby erasing the expected premium (Nakavachara et al., 14 Jul 2025).

Illustrative category results are consistent with this interpretation. In Sweden, the coefficient of $$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$6 for Meat, Poultry & Charcuterie implies that moving one climate score point higher lowers price by about 22%, all else equal; moving two points higher would be associated with roughly a 40–45% lower price using $$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$7. The paper states that this is consistent with greener meat options such as poultry versus beef having lower emission intensity and often lower unit prices. For freezer categories, the negative premium is described as plausibly reflecting plant-based frozen items and vegetables having lower emission intensity and typically lower prices than frozen meat or seafood (Nakavachara et al., 14 Jul 2025).

A common misconception is that climate-friendly food should necessarily command higher prices in sustainability-aware markets. The reported coefficients directly contradict that expectation. The puzzle, in this formulation, is not weakly positive pricing but rather the absence or reversal of the expected signal in large food assortments (Nakavachara et al., 14 Jul 2025).

4. Electric vehicles: from apparent premium to life-cycle discount

The EV study models the green premium as a relative cost difference and makes TCO the key analytical object:

$$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$8

Here, $$\ln(p_i) = \beta \,\text{ClimateScore}_i + \boldsymbol{\gamma}' \mathbf{X}_i + \theta_{\text{origin}(i)} + \delta_{\text{catL3}(i)} + \varepsilon_i.$$9 denotes a positive lifetime premium, $\beta$0 parity, and $\beta$1 a “green discount.” The TCO framework discounts annual operating costs and residual value and incorporates production cost, government subsidy, fuel or electricity cost, maintenance cost, and residual value. The benchmark comparison is a Tesla Model 3-like B-class EV with a 75 kWh pack and 480–668 km NEDC range against a B-class ICEV with a 2.0–2.4 L supercharged engine. Key assumptions include $\beta$2, $\beta$3 years, annual mileage of 15,000 km, electricity at 1.2 RMB/kWh, gasoline at 7.5 RMB/L, and a battery pack cost trajectory from RMB 7,500/kWh in 2010 to RMB 820/kWh in 2021, with a forecast of RMB 650/kWh by 2025 (Li et al., 2023).

The 2021 cost decomposition distinguishes sharply among accounting layers. The manufacturing green premium is $\beta$4, meaning EV production cost is higher than that of comparable ICEVs. The acquisition cost premium is $\beta$5 after accounting for credits, tax exemption, and subsidies. The usage-phase advantage is $\beta$6, and the TCO green premium is $\beta$7, implying that EV full life-cycle cost is already lower than ICEV full life-cycle cost in 2021. Levelized cost per km is reported as RMB 1.52/km for EVs and RMB 1.80/km for ICEVs. Annual operating savings are approximately RMB 7,000–10,000 per year for EVs, and the ordering of parity transitions is stated explicitly as “life-cycle cost $\beta$8 acquisition cost $\beta$9 production cost” (Li et al., 2023).

The same study embeds the TCO green premium in a generalized Bass diffusion model through

$\beta$0

The fitted relationships are qualitative but substantive: $\beta$1, so imitation dominates innovation; $\beta$2, so higher premiums reduce adoption; and the magnitude of $\beta$3 is described as large, indicating strong price sensitivity. With annual China EV diffusion data from 2010–2021 and a Genetic Algorithm implemented in Matlab using population size 800, crossover probability 0.8, mutation probability 0.1, and max iterations 500, the generalized model improves fit from $\beta$4 for the vanilla Bass model to $\beta$5 (Li et al., 2023).

The paper therefore treats EV diffusion as a case in which positive production or acquisition premiums do not prevent adoption, because the TCO premium is already negative. Short-range EVs are reported to have achieved full life-cycle cost parity by 2018, acquisition cost parity by 2025–2026 after a rebound following subsidy withdrawal in 2022, and production cost parity by 2027–2028. Long-range EVs also achieve life-cycle green discount by 2018, while acquisition and production cost parity occur around 2030. Under baseline assumptions, annual BEV sales are forecast at approximately 10.97 million by 2030 in the case-study figure, while the abstract reports 10.77 million units, with an overall penetration rate of about 39% (Li et al., 2023).

5. Cross-domain interpretation of the puzzle

The food and EV results point to structurally different manifestations of the same problem. In food, the relevant observed market signal is the shelf price of a fast-moving consumer good. In EVs, the relevant signal for diffusion is full life-cycle economics rather than production cost or acquisition price alone. Taken together, the studies suggest that the green premium puzzle is highly sensitive to temporal horizon, salience of operating savings, and the extent to which consumers or firms internalize lifecycle information (Nakavachara et al., 14 Jul 2025, Li et al., 2023).

The food study explicitly places its findings in a broader literature in which green premiums are more often documented for energy-efficient housing, appliances, and electric vehicles, where labels are well-established, future operating cost savings are salient, and the purchase is infrequent and deliberative. By contrast, food purchases are frequent and habit-driven, small price differences loom large, climate attributes are less tied to immediate personal benefits than health or taste, and competing certifications can reduce the marginal impact of climate scores. The EV study fits this contrast: it emphasizes lifetime savings, policy support, and imitation effects as the mechanisms that sustain diffusion despite positive acquisition premiums (Nakavachara et al., 14 Jul 2025, Li et al., 2023).

This comparison also clarifies a frequent misunderstanding. A positive acquisition premium does not imply a positive TCO premium, and a negative retail association in food does not imply that climate attributes are irrelevant. In the EV case, the puzzle is resolved by showing that the appropriate cost measure is already negative. In the food case, the puzzle remains, because even after controlling for size, nutrition, origin, and fine-grained category heterogeneity, there is no robust positive price premium. A plausible implication is that “green premium” should be interpreted as a domain-specific empirical object rather than a universal market law (Nakavachara et al., 14 Jul 2025, Li et al., 2023).

6. Limitations, external validity, and operational implications

The food evidence is limited by its single-supermarket Swedish setting, albeit with a Swiss robustness sample. It is cross-sectional and therefore cannot study dynamic pricing or promotional cycles. Brand and promotions are not available, creating omitted-variable concerns; store placement is also unobserved. Climate scores depend on LCA assumptions, and although RISE is described as authoritative or reputable, any LCA-based measure entails methodological uncertainty. Not all products have climate scores, automated translation from Swedish to English may introduce minor classification errors, and exclusions such as beverages and categories with insufficient information may limit representativeness (Nakavachara et al., 14 Jul 2025).

The EV analysis also rests on strong assumptions. The discount rate is fixed at $\beta$6; insurance, parking, and the “shadow value” of license plate restrictions are not monetized; maintenance values conflict between table and text, although the analysis relies on EV maintenance being lower; and only 12 annual observations from 2010–2021 are available for fitting. Regional heterogeneity in tariffs, fuel prices, and policies is not modeled explicitly, and NEV credit prices are volatile. Robustness is addressed primarily through sensitivity analysis and through the comparison of $\beta$7 values between the generalized and vanilla Bass models (Li et al., 2023).

The operational implications follow directly from these empirical patterns. For food markets, the recommendations are to standardize climate impact metrics and bins across retailers and countries, strengthen third-party verification, communicate absolute CO$\beta$8eq values alongside color or traffic-light cues, and avoid label overload. Retailers are advised to consider targeted promotions or shelf placement for high-score items and to pair climate scores with taste or quality cues. Public policy recommendations include education campaigns, support for standard-setting and accreditation, and, where appropriate, fiscal tools such as differentiated VAT, subsidies for low-emission categories, or taxes on high-emission items (Nakavachara et al., 14 Jul 2025).

For EVs, the most effective levers are reported to be battery cost reductions, purchase tax exemptions, stricter CAFC standards, gasoline prices or fuel taxation, targeted subsidies, and NEV credits. Sensitivity magnitudes identify battery unit price, purchase tax policy, ICEV fuel economy standards, and gasoline prices as especially consequential, while electricity price and discount rate matter less. This suggests that the EV version of the puzzle is less about consumer indifference to environmental attributes than about the sequencing of cost parity across manufacturing, acquisition, and use phases (Li et al., 2023).

7. Synthesis

Within the evidence considered here, the green premium puzzle has two distinct but related meanings. In climate-friendly food products, it refers to the empirical finding that better climate performance does not command a higher retail price on average and often coincides with lower prices. In EVs, it refers to the apparent inconsistency between positive production or acquisition premiums and strong market diffusion, an inconsistency that disappears when TCO and imitation dynamics are modeled explicitly (Nakavachara et al., 14 Jul 2025, Li et al., 2023).

The broader significance lies in the mismatch between environmental performance and market signals. In food, that mismatch appears as a missing or negative premium at the point of purchase. In EVs, it appears as divergence between short-run acquisition price and full life-cycle cost. The combined evidence suggests that climate-friendly goods cannot be evaluated through a single pricing lens. Retail price, acquisition price, usage-phase savings, TCO, label salience, trust, and policy design all condition whether a premium is observed, suppressed, or reversed (Nakavachara et al., 14 Jul 2025, Li et al., 2023).