Organizational ambidexterity and investment efficiency: a machine learning approach

This paper aims to investigate the relationship between organizational ambidexterity (AMBI) and investment efficiency. While prior research has explored drivers of efficient corporate investment, limited attention has been paid to the strategic role of ambidexterity. We aim to address this gap by examining how firms' ability to balance exploration and exploitation contributes to more effective investment decisions. We measure AMBI using a semi-supervised machine learning approach that analyzes earnings call transcripts from 2005 to 2019. Investment efficiency is assessed using the residuals from a regression of investment on sales growth. To ensure robustness, we apply a range of empirical strategies, including multivariate regressions, propensity score matching, entropy balancing and instrumental variable techniques. Our results show a positive and statistically significant relationship between ambidexterity and investment efficiency. The effect is stronger in firms led by experienced and capable managers or where managerial incentives are well-aligned with shareholder interests and weaker among firms facing financial constraints. We also find that real option intensity, organizational capital and innovation play mediating roles in this relationship. The measure of ambidexterity depends on a specific machine learning technique, which may not capture all dimensions of strategic behavior. Future work could explore how ambidexterity influences other corporate outcomes, such as financing policies or risk-taking behavior. This study is among the first to leverage neural language models to quantify AMBI. By grounding ambidexterity in real options theory and introducing real option intensity as an empirical mechanism, we show how managerial flexibility under uncertainty improves investment efficiency.