This paper by James G. MacKinnon presents a method for calculating distribution functions for unit root and cointegration test statistics using response surface regressions based on simulation experiments. The key contributions include a set of data files containing estimated response surface coefficients and a computer program, urcdist, for calculating critical values and $P$ values for these tests. The program is freely available via the Internet and can handle both asymptotic and finite-sample distributions. The paper also discusses the asymptotic and finite-sample distributions of various test statistics, including Dickey-Fuller (DF) and Augmented Dickey-Fuller (ADF) tests for unit roots, and Engle-Granger (EG) and Ouliaris-Park-Phillips (OPP) tests for cointegration. The simulation experiments involve 200,000 replications for unit root tests and 50 experiments for cointegration tests, with sample sizes ranging from 20 to 700. The response surface regressions are estimated using Generalized Method of Moments (GMM) to account for heteroskedasticity. The paper provides empirical examples and discusses the differences between asymptotic and finite-sample distributions, emphasizing the importance of accurate critical values and $P$ values in applied research.This paper by James G. MacKinnon presents a method for calculating distribution functions for unit root and cointegration test statistics using response surface regressions based on simulation experiments. The key contributions include a set of data files containing estimated response surface coefficients and a computer program, urcdist, for calculating critical values and $P$ values for these tests. The program is freely available via the Internet and can handle both asymptotic and finite-sample distributions. The paper also discusses the asymptotic and finite-sample distributions of various test statistics, including Dickey-Fuller (DF) and Augmented Dickey-Fuller (ADF) tests for unit roots, and Engle-Granger (EG) and Ouliaris-Park-Phillips (OPP) tests for cointegration. The simulation experiments involve 200,000 replications for unit root tests and 50 experiments for cointegration tests, with sample sizes ranging from 20 to 700. The response surface regressions are estimated using Generalized Method of Moments (GMM) to account for heteroskedasticity. The paper provides empirical examples and discusses the differences between asymptotic and finite-sample distributions, emphasizing the importance of accurate critical values and $P$ values in applied research.