Joakim Westerlund (2006) proposes new error correction-based cointegration tests for panel data. These tests are based on structural rather than residual dynamics, avoiding the common factor restriction that can reduce the power of residual-based tests. The paper derives the limiting distributions of the tests and provides critical values. Simulation results show that the new tests have good small-sample properties with small size distortions and high power relative to other popular residual-based panel cointegration tests. In an empirical application, the paper presents evidence suggesting that international health care expenditures and GDP are cointegrated once the possibility of an invalid common factor restriction has been accounted for. The paper discusses two broad approaches to testing for cointegration in panel data. The first approach takes cointegration as the null hypothesis, while the second approach takes no cointegration as the null hypothesis. The second approach is based on the methodology of Engle and Granger (1987), where the residuals of a static least squares regression are subjected to a unit root test. The most influential theoretical contributions within this category are those of Pedroni (1999, 2004), who generalize the work of Phillips and Ouliaris (1990) by developing several tests appropriate for various cases of heterogeneous dynamics, endogenous regressors, and individual-specific constants and trends. The paper also discusses the implications of weak exogeneity and common factor restrictions. Weak exogeneity is the assumption that the model of interest is the conditional error correction model, and that the marginal model can be ignored. Common factor restrictions are assumptions that the error correction model is based on a common factor that is not necessarily valid. The paper shows that the new tests are based on structural rather than residual dynamics, and therefore do not impose any common factor restriction. The new tests are panel extensions of those proposed in the time series context by Banerjee et al. (1998). They are designed to test the null hypothesis by inferring whether the error correction term in a conditional error correction model is equal to zero. The paper presents four new panel tests of the null hypothesis of no cointegration. Two of the tests are based on pooling the information regarding the error correction along the cross-sectional dimension of the panel, while the other two test the alternative that there is at least one individual that is cointegrated. All four tests are shown to be very straightforward and easy to implement. The asymptotic results reveal that the tests have limiting normal distributions, and that they are consistent. Simulation evidence is also provided to evaluate and compare the small-sample performance of the tests relative to the performance of the popular residual-based tests by Pedroni (2004). The results suggest that the new tests maintain good size accuracy, and that they are more powerful than the residual-based tests provided that the conditions laid out in the paper hold. In our empirical application, we provide evidence suggesting that international healthJoakim Westerlund (2006) proposes new error correction-based cointegration tests for panel data. These tests are based on structural rather than residual dynamics, avoiding the common factor restriction that can reduce the power of residual-based tests. The paper derives the limiting distributions of the tests and provides critical values. Simulation results show that the new tests have good small-sample properties with small size distortions and high power relative to other popular residual-based panel cointegration tests. In an empirical application, the paper presents evidence suggesting that international health care expenditures and GDP are cointegrated once the possibility of an invalid common factor restriction has been accounted for. The paper discusses two broad approaches to testing for cointegration in panel data. The first approach takes cointegration as the null hypothesis, while the second approach takes no cointegration as the null hypothesis. The second approach is based on the methodology of Engle and Granger (1987), where the residuals of a static least squares regression are subjected to a unit root test. The most influential theoretical contributions within this category are those of Pedroni (1999, 2004), who generalize the work of Phillips and Ouliaris (1990) by developing several tests appropriate for various cases of heterogeneous dynamics, endogenous regressors, and individual-specific constants and trends. The paper also discusses the implications of weak exogeneity and common factor restrictions. Weak exogeneity is the assumption that the model of interest is the conditional error correction model, and that the marginal model can be ignored. Common factor restrictions are assumptions that the error correction model is based on a common factor that is not necessarily valid. The paper shows that the new tests are based on structural rather than residual dynamics, and therefore do not impose any common factor restriction. The new tests are panel extensions of those proposed in the time series context by Banerjee et al. (1998). They are designed to test the null hypothesis by inferring whether the error correction term in a conditional error correction model is equal to zero. The paper presents four new panel tests of the null hypothesis of no cointegration. Two of the tests are based on pooling the information regarding the error correction along the cross-sectional dimension of the panel, while the other two test the alternative that there is at least one individual that is cointegrated. All four tests are shown to be very straightforward and easy to implement. The asymptotic results reveal that the tests have limiting normal distributions, and that they are consistent. Simulation evidence is also provided to evaluate and compare the small-sample performance of the tests relative to the performance of the popular residual-based tests by Pedroni (2004). The results suggest that the new tests maintain good size accuracy, and that they are more powerful than the residual-based tests provided that the conditions laid out in the paper hold. In our empirical application, we provide evidence suggesting that international health