Economics of smart products with machine learning
研究了机器学习如何影响智能产品企业的定价策略和利润，发现消费者可能因预期未来降价而推迟购买，从而影响数据收集；通过两期博弈模型比较了响应式定价与预先宣布定价两种策略的优劣。
Driven by advances in machine learning (ML), smart products improve over time through data-driven insights as ongoing user interactions generate usage data that enable the training, evaluation, and refinement of underlying algorithms. However, when firms implement strategies to collect more data to enhance product quality and profits, they must also consider strategic consumer behavior that may lead to unintended negative consequences. Specifically, consumers may intentionally postpone purchases in the early stages of a product’s development, anticipating future quality improvements and price reductions, which in turn complicates data collection during this critical period. Considering advancements in disruptive technologies, this study examines the critical yet underexplored economic impact of ML on pricing strategies. Few studies have focused on how ML influences profit maximization in the presence of strategic consumers. To address this gap, we develop two-period game-theoretic models that employ two dynamic pricing strategies, responsive versus preannounced pricing, to investigate how firms developing smart products adapt to the disruptive impact of ML, considering the behavior of strategic consumers. Our study provides several significant implications. First, we find that ML impacts firms’ profits by impacting consumers’ strategic behaviors in opposite directions. Second, under both dynamic pricing strategies, prices may initially be low and may either rise or decline over time. Third, we demonstrate that, different from findings in the existing literature on strategic consumer behavior, preannounced pricing policies are generally not optimal for the firm when its ability to leverage ML is relatively limited, and consumers are less strategic. Overall, this study makes three contributions to the literature. First, we clarify the impact of ML on a smart product firm’s profit. We find two effects in the application of ML: (1) a positive effect associated with ML (the “ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mi>M</mml:mi> <mml:msup> <mml:mi>L</mml:mi> <mml:mo>+</mml:mo> </mml:msup> </mml:math> effect”) and (2) a negative effect (the “ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mi>M</mml:mi> <mml:msup> <mml:mi>L</mml:mi> <mml:mo>−</mml:mo> </mml:msup> </mml:math> effect”). Second, this study highlights a fundamental economic mechanism for smart products in the presence of strategic consumers. Finally, it provides a decision-making tool for smart product firms to select an optimal dynamic pricing strategy.