Statistics, experimentation, causal reasoning · insight-0022
Never read price elasticity off observational fares and bookings when prices respond to demand
- Evidence type
- practitioner heuristic
- Difficulty
- medium
- Roles
- Data scientist, Data analyst
- Industries
- aviation, retail, e-commerce
Summary: situation, action, reason, boundary
Situation
A regression of bookings on average fare across flights shows a positive slope, and someone proposes raising fares to sell more seats.
Action
Treat fare as confounded by expected demand. Estimate response from randomized price tests, from exogenous price variation such as cost-driven or competitor changes with a stated identification argument, or from conditioning on the demand signals the pricing system used.
Reason
A coefficient is a conditional association; without breaking the link between demand and price, it cannot answer "what if we change price".
Boundary
Dynamic pricing, promotions, discount targeting and credit-limit decisions.
Symptom or risk
Revenue management raises fares when demand is high, so high fares and high bookings co-occur; the regression measures the pricing policy, not customer response.
Exceptions
Prices fixed independently of demand for the period studied, verified from the pricing process rather than assumed.
Trade-offs
Price experiments risk revenue and customer fairness concerns; observational designs carry untestable assumptions.
Pitfalls
Adding more demand features to a forecast model and reading its fare coefficient as causal.
Example
Synthetic airline simulator: true elasticities are negative (-0.6 to -1.35) yet the raw fare-bookings correlation is positive.