Dynamic Pricing with Elastic-ARIMA Demand

Dynamic pricing is a fundamental problem in operations research, requiring accurate demand estimation for revenue optimization. This study introduces Elastic-ARIMA Demand, a novel stochastic demand model that integrates price elasticity with autoregressive integrated moving average (ARIMA) processes to more accurately reflect real-world demand behavior. We investigate the sufficiency of linear demand models in dynamic pricing scenarios, comparing their performance against Elastic-ARIMA demand across Our findings demonstrate that, despite their misspecification, linear models approximate optimal pricing decisions with bounded regret and near-optimal revenue outcomes. The research further quantifies the ‘price of misspecification’ in dynamic pricing contexts, evaluating how quickly regret converges under linear assumptions and whether such models are practically viable. We show that while Elastic-ARIMA models offer a more accurate structural representation of demand, their performance gains over linear models are marginal in finite time horizons, reinforcing the robustness of simplified linear pricing strategies. The study concludes with implications for revenue management and open questions for future research on competitive pricing dynamics and real-world demand forecasting.