Mock M5 · Fast-Fashion's Hidden Water Bill

Impact accounting Behavioral model Policy design

The problem

Fast fashion (Shein, Temu, Zara, H&M) sells clothes designed to be worn a handful of times. The water footprint — from cotton irrigation, dyeing, finishing, and laundering — is massive but invisible to consumers. The European Commission is considering a mandatory water-impact label for all apparel sold in the EU. Your team has been asked to model the system and predict whether labeling will change behavior enough to matter.

Requirements

Build a per-garment water-footprint model accounting for fiber type (cotton, polyester, viscose), country of origin (water-stress weighted), dyeing process, and lifetime laundry. Apply to 5 garment categories (basic tee, jeans, dress, athleisure top, sweater).
Estimate the EU's total apparel-driven water footprint (or "withdrawal" — be explicit about which) using 2024 sales data.
Build a behavioral model: given a label that shows "L of water" per garment, how does consumer purchase probability change as a function of price and label? Use any reasonable framework (logit, linear utility, etc.) and justify your parameter choices with cited consumer-research studies.
Simulate three policy regimes:
- Voluntary label (adopted by 30% of brands).
- Mandatory label (all garments).
- Mandatory label + water tax (€0.005 per liter on the labeled footprint).
Report the projected 5-year change in total water footprint.
Sensitivity: which uncertain input most determines whether the policies actually move the needle?
One-page briefing for an EU Commissioner.

Solution sketch

Footprint model

$W_{\text{garment}} = W_{\text{fiber}} \cdot \omega_{\text{country}} + W_{\text{dye}} + W_{\text{finish}} + L \cdot W_{\text{wash}}$

Typical values: cotton tee ≈ 2,700 L; jeans ≈ 7,500–10,000 L; polyester athleisure top ≈ 60–150 L of process water but ~500 L over 50 washes (microfiber wastewater issues separate). $\omega$ is a water-stress multiplier per WRI Aqueduct (Pakistan ≈ 4×, EU average ≈ 1×).

Behavioral model (logit)

Utility of garment $j$ to consumer: $U_j = -\alpha p_j - \beta W_j + \gamma \text{brand}_j + \varepsilon_j$. Purchase probability $\propto e^{U_j}$. Calibrate $\beta$ to studies on eco-label effect size (a recent meta-analysis suggests labels reduce purchase probability of high-impact items by 10–20%, but effect halves for low-information consumers).

Three regimes — likely results

Regime	5-yr water Δ vs. baseline
Voluntary label (30% coverage)	−3 to −6%
Mandatory label	−10 to −15%
Mandatory + €0.005/L tax	−25 to −35%

Sensitivity finding

The dominant uncertainty isn't the footprint numbers — those are pretty well established. It's the price elasticity of fast-fashion consumers, which is largely unknown for the sub-25 EUR segment that dominates the market. Recommendation in the briefing: the EU should fund a randomized field experiment before committing to a regulatory path.

Self-grading focus

Did you distinguish "water consumption" from "water withdrawal" from "blue/green/grey water"?
Is your behavioral model based on cited studies, not just "we assume people care"?
Did the three regimes actually produce different results, or did you flatten them?
Is the briefing focused on the decision, not the methodology?

← All mocks