Shift‐Share Analysis and Multifactor Partitioning: What do Aggregated Data Hide?

ABSTRACT Shift‐share analysis (SSA) is a widely used tool for studying economic changes, particularly in employment, due to its simplicity and minimal data requirements. However, its reliance on crude growth rates and issues associated with aggregation can lead to biases, such as Simpson's Paradox, that may hide regional and industry‐specific insights. Multifactor Partitioning (MFP) addresses these limitations by standardizing growth rates in a way that disentangles industry and regional effects. This paper compares SSA and MFP using employment data from 10 U.S. states between 2005 and 2019. The analysis incorporates three levels of disaggregation: (1) aggregate employment and time, (2) disaggregated employment with aggregated time, and (3) both sectoral and temporal disaggregation. Results show that while SSA and MFP yield similar conclusions at an aggregate level, discrepancies emerge in disaggregated analyses, particularly in high‐growth regions. These findings highlight the importance of data disaggregation and MFP's capacity to provide nuanced insights for policymakers and researchers.