Should We Trust Clustered Standard Errors? A Comparison with Randomization-Based Methods /
Paz, Lourenço S.
Should We Trust Clustered Standard Errors? A Comparison with Randomization-Based Methods / Lourenço S. Paz, James E. West. - Cambridge, Mass. National Bureau of Economic Research 2019. - 1 online resource: illustrations (black and white); - NBER working paper series no. w25926 . - Working Paper Series (National Bureau of Economic Research) no. w25926. .
June 2019.
We compare the precision of critical values obtained under conventional sampling-based methods with those obtained using sample order statics computed through draws from a randomized counterfactual based on the null hypothesis. When based on a small number of draws (200), critical values in the extreme left and right tail (0.005 and 0.995) contain a small bias toward failing to reject the null hypothesis which quickly dissipates with additional draws. The precision of randomization-based critical values compares favorably with conventional sampling-based critical values when the number of draws is approximately 7 times the sample size for a basic OLS model using homoskedastic data, but considerably less in models based on clustered standard errors, or the classic Differences-in-Differences. Randomization-based methods dramatically outperform conventional methods for treatment effects in Differences-in-Differences specifications with unbalanced panels and a small number of treated groups.
System requirements: Adobe [Acrobat] Reader required for PDF files.
Mode of access: World Wide Web.
Should We Trust Clustered Standard Errors? A Comparison with Randomization-Based Methods / Lourenço S. Paz, James E. West. - Cambridge, Mass. National Bureau of Economic Research 2019. - 1 online resource: illustrations (black and white); - NBER working paper series no. w25926 . - Working Paper Series (National Bureau of Economic Research) no. w25926. .
June 2019.
We compare the precision of critical values obtained under conventional sampling-based methods with those obtained using sample order statics computed through draws from a randomized counterfactual based on the null hypothesis. When based on a small number of draws (200), critical values in the extreme left and right tail (0.005 and 0.995) contain a small bias toward failing to reject the null hypothesis which quickly dissipates with additional draws. The precision of randomization-based critical values compares favorably with conventional sampling-based critical values when the number of draws is approximately 7 times the sample size for a basic OLS model using homoskedastic data, but considerably less in models based on clustered standard errors, or the classic Differences-in-Differences. Randomization-based methods dramatically outperform conventional methods for treatment effects in Differences-in-Differences specifications with unbalanced panels and a small number of treated groups.
System requirements: Adobe [Acrobat] Reader required for PDF files.
Mode of access: World Wide Web.