Xavier X. Sala-i-Martin of Columbia University and Universitat Pompeu Fabra analyzes the empirical growth literature, challenging the "Extreme Bounds" method for identifying robust relationships. Instead of focusing on extreme bounds of coefficients, he examines the full distribution of estimates. His analysis reveals that many variables are strongly related to economic growth, contradicting the "Nothing is Robust" conclusion of previous studies. The paper discusses the limitations of economic growth theories in specifying which variables are important for growth. Empirical economists often test various variables, but results are inconsistent due to multicollinearity and the difficulty of measuring theoretical determinants like human capital or efficient government. Sala-i-Martin proposes a method that assigns confidence levels to variables based on the distribution of their estimates. He considers two cases: normal and non-normal distributions. By weighting regressions based on likelihood, he finds that many variables are significantly correlated with growth, with 21 out of 59 showing high significance. Key findings include the importance of political rights, civil liberties, rule of law, and certain religious groups (Confucius, Buddhist, Muslim) for growth. Market distortions, types of investment, primary sector production, openness, and economic organization also show significant relationships. Conversely, variables like government spending, inflation, and ethno-linguistic fractionalization are not strongly correlated with growth. When the investment rate is included as a fixed variable, the results remain largely consistent, though some variables lose significance. The investment rate itself is significant in most regressions, though not robust in the extreme bounds sense. The paper concludes that a substantial number of variables are strongly related to economic growth, challenging the notion that nothing is robust in the empirical growth literature. The analysis highlights the importance of considering the full distribution of estimates rather than relying solely on extreme bounds.Xavier X. Sala-i-Martin of Columbia University and Universitat Pompeu Fabra analyzes the empirical growth literature, challenging the "Extreme Bounds" method for identifying robust relationships. Instead of focusing on extreme bounds of coefficients, he examines the full distribution of estimates. His analysis reveals that many variables are strongly related to economic growth, contradicting the "Nothing is Robust" conclusion of previous studies. The paper discusses the limitations of economic growth theories in specifying which variables are important for growth. Empirical economists often test various variables, but results are inconsistent due to multicollinearity and the difficulty of measuring theoretical determinants like human capital or efficient government. Sala-i-Martin proposes a method that assigns confidence levels to variables based on the distribution of their estimates. He considers two cases: normal and non-normal distributions. By weighting regressions based on likelihood, he finds that many variables are significantly correlated with growth, with 21 out of 59 showing high significance. Key findings include the importance of political rights, civil liberties, rule of law, and certain religious groups (Confucius, Buddhist, Muslim) for growth. Market distortions, types of investment, primary sector production, openness, and economic organization also show significant relationships. Conversely, variables like government spending, inflation, and ethno-linguistic fractionalization are not strongly correlated with growth. When the investment rate is included as a fixed variable, the results remain largely consistent, though some variables lose significance. The investment rate itself is significant in most regressions, though not robust in the extreme bounds sense. The paper concludes that a substantial number of variables are strongly related to economic growth, challenging the notion that nothing is robust in the empirical growth literature. The analysis highlights the importance of considering the full distribution of estimates rather than relying solely on extreme bounds.