Leveraging Machine Learning for Strategic Performance Management

This dissertation investigates the use of machine learning (ML) in strategic performance management. While ML applications have been widely explored in financial accounting, their use in management accounting remains relatively underexamined. This research aims to fill this gap by demonstrating how ML can provide in different stages of strategic performance management, including identifying strategic groups, performance measurement and resource allocation. The first chapter explores the potential of ML algorithms to mitigate cognitive biases that managers face when analyzing performance data and making strategic resource allocation decisions. Through a computer-simulated business game, this study compares the effectiveness of ML-based budget allocation against human decision-making. The findings indicate that ML algorithms significantly outperform human participants in optimizing budget allocations, leading to improved organizational value creation. However, the results also highlight the complementary nature of ML and human strategic reasoning. While ML efficiently processes large datasets and uncovers complex, nonlinear relationships, human expertise is needed to align the resource allocation with broader strategic objectives. The second chapter applies an unsupervised learning approach to develop a more nuanced classification of business strategies in the airline industry. Existing research typically categorizes airlines into either focused or full-service strategies. However, recent industry trends suggest that some airlines are adopting hybrid strategies that blend elements of both approaches. Using fuzzy clustering, this study identifies such hybrid airlines and evaluates their performance relative to pure strategic positions. The results reveal that hybrid airlines often achieve superior financial performance, but only when they effectively manage their capacity utilization. If they fail to leverage their increased complexity into a better use of their capacity, the benefits dissapear. The third chapter leverages supervised learning techniques to examine the relationship between nonfinancial performance measures and profitability in the airline industry. By applying ML methods, this study takes an exploratory approach to identify key performance indicators that predict airline profitability, taking into account interactions and nonlinearities. The findings suggest that operational efficiency measures, such as load factors, labor productivity, and fuel consumption , are the strongest predictors of financial success. Moreover, the study uncovers interaction effects, such as the moderating impact of capacity utilization on service failures and a U-shaped relationship between customer complaints and profitability. These results highlight the importance of considering both direct and indirect effects of performance metrics in strategic decision-making. By integrating ML techniques into strategic performance management, this dissertation contributes to the management accounting literature by showcasing ML's ability to uncover hidden patterns, enhance decision-making, and optimize resource allocation.