Levine and Renelt examine the robustness of cross-country growth regressions by testing how sensitive conclusions are to changes in independent variables. They find that most macroeconomic variables are not robustly correlated with growth, and many commonly used fiscal and trade indicators are not robustly related to growth. They confirm a positive correlation between investment share and growth, but note that the relationship between exports and growth is not robust, as it can be replicated using trade or import indicators. They also find that fiscal indicators, such as government consumption and deficits, are not robustly correlated with growth. The paper highlights the importance of considering alternative specifications in cross-country growth regressions and suggests that many findings are fragile to small changes in the list of variables. The authors conclude that while "policy" broadly defined may be related to growth, there is no strong independent relationship between most macroeconomic indicators and growth. The paper emphasizes the need for caution in interpreting cross-country growth studies and suggests that future research should focus on more comprehensive measures of policy and economic performance.Levine and Renelt examine the robustness of cross-country growth regressions by testing how sensitive conclusions are to changes in independent variables. They find that most macroeconomic variables are not robustly correlated with growth, and many commonly used fiscal and trade indicators are not robustly related to growth. They confirm a positive correlation between investment share and growth, but note that the relationship between exports and growth is not robust, as it can be replicated using trade or import indicators. They also find that fiscal indicators, such as government consumption and deficits, are not robustly correlated with growth. The paper highlights the importance of considering alternative specifications in cross-country growth regressions and suggests that many findings are fragile to small changes in the list of variables. The authors conclude that while "policy" broadly defined may be related to growth, there is no strong independent relationship between most macroeconomic indicators and growth. The paper emphasizes the need for caution in interpreting cross-country growth studies and suggests that future research should focus on more comprehensive measures of policy and economic performance.