A Formula for Investing in the Semiconductor Industry : An Empirical Study Using Firm Fundamentals and System GMM

Sammanfattning

This thesis investigates whether firm-specific financial metrics can predict stock returns in the semiconductor industry and whether these predictors can form the basis of a systematic investment strategy. Using a panel of fifty global semiconductor firms from 2006 to 2022, the analysis applies a two-step System Generalized Method of Moments estimator to address endogeneity and firm heterogeneity. Return on assets emerges as the most robust and statistically significant predictor of future stock returns, reflecting the central role of asset efficiency in this capital-intensive sector. Portfolio tests compare buy and hold and annually rebalanced strategies against the NASDAQ benchmark. The results show that ROA based portfolios achieve superior raw performance, particularly when rebalanced annually, although excess returns largely reflect higher market exposure once risk adjustment is applied. Within the framework of the Adaptive Market Hypothesis, the findings suggest that profitability-based strategies may occasionally capture time varying inefficiencies, but their effectiveness depends on evolving market conditions. The results overall show that, in a semiconductor-specific dynamic panel that accounts for unobserved firm effects and endogeneity, return on asset is the only robust fundamental predictor of stock returns. This motivates an investment formula that can narrow the set of firms investors need to evaluate when making investment decisions in this sector.