Define and compute surge pricing metrics
Overview
This question evaluates a data scientist's competence in defining and computing marketplace pricing and operational metrics—such as demand and supply indices, surge multipliers, conversion and acceptance rates, ETA, and price elasticity—and in applying causal inference to estimate pricing effects; it belongs to the Statistics & Math domain with ties to econometrics and marketplace analytics and is commonly asked to assess the ability to operationalize metrics and quantify price-driven behavior in two-sided platforms. It tests both conceptual understanding (precise metric definitions and causal identification) and practical application (hand calculations on a toy panel and designing guardrail thresholds), including consideration of randomized or quasi-experimental approaches for estimating causal elasticity rather than relying on simple correlations.
Surge Pricing Metrics, Formulas, and Causal Estimation
You are evaluating surge pricing in a two-sided marketplace (customers place requests; drivers/couriers accept offers). For a single zone with 15‑minute slots, define each metric precisely and give a computation formula:
- Demand index
- Supply index
- Surge multiplier
- Order conversion rate
- Driver acceptance rate
- ETA (expected time of arrival)
- Price elasticity of demand (at a point and between two price bands)
Then, using the toy panel below, compute by hand: the average surge multiplier, overall conversion rate, overall acceptance rate, and the arc elasticity of demand from slots with multiplier 1.0 to 1.3.
Assume price = base_price × surge_mult, and quantity demanded is measured by orders.
slot_start | base_price | surge_mult | demand_requests | orders | offers_sent | offers_accepted | avg_ETA_min
- 18:00 | 10.00 | 1.0 | 50 | 30 | 40 | 32 | 29
- 18:15 | 10.00 | 1.3 | 55 | 28 | 42 | 30 | 33
- 18:30 | 10.00 | 1.3 | 48 | 24 | 38 | 27 | 35
- 18:45 | 10.00 | 1.0 | 46 | 29 | 36 | 31 | 28
Finally, propose a guardrail metric (with a clear threshold) that prevents unacceptable ETA inflation while using surge, and explain how you would estimate causal elasticity using randomized or quasi‑experimental variation (e.g., geo‑matched difference‑in‑differences) rather than simple correlations.
Solution
Login required