This paper examines the recent literature on the identification of production functions, focusing on structural techniques proposed by Olley and Pakes (1996) and Levinsohn and Petrin (2003). While these papers offer solid and intuitive identification ideas, the authors argue that the techniques, particularly those of Levinsohn and Petrin, suffer from collinearity problems, which cast doubt on their methodology. They suggest alternative methodologies that build upon the ideas in these papers but avoid the collinearity issues. The proposed approach uses investment or intermediate inputs to "proxy" for productivity shocks, making it easier to compare with dynamic panel data literature and providing more stable estimates across different proxy variables. The paper also reviews the key assumptions of Olley and Pakes and Levinsohn and Petrin, highlighting the collinearity issues in their first-stage estimating equations. It discusses potential data generating processes (DGPs) that could break the collinearity problem, including optimization error in labor inputs and timing assumptions about input choices. The authors conclude that their proposed approach is more robust and reasonable than the existing methods, offering a more reliable way to estimate production function parameters.This paper examines the recent literature on the identification of production functions, focusing on structural techniques proposed by Olley and Pakes (1996) and Levinsohn and Petrin (2003). While these papers offer solid and intuitive identification ideas, the authors argue that the techniques, particularly those of Levinsohn and Petrin, suffer from collinearity problems, which cast doubt on their methodology. They suggest alternative methodologies that build upon the ideas in these papers but avoid the collinearity issues. The proposed approach uses investment or intermediate inputs to "proxy" for productivity shocks, making it easier to compare with dynamic panel data literature and providing more stable estimates across different proxy variables. The paper also reviews the key assumptions of Olley and Pakes and Levinsohn and Petrin, highlighting the collinearity issues in their first-stage estimating equations. It discusses potential data generating processes (DGPs) that could break the collinearity problem, including optimization error in labor inputs and timing assumptions about input choices. The authors conclude that their proposed approach is more robust and reasonable than the existing methods, offering a more reliable way to estimate production function parameters.