A Speedy and Seamless Stationarity Analysis via causfinder Package in R

Abstract

Stationarity analysis is a must for various studies in many fields that employ time series analysis. adfcs function in causfinder package in R performs Augmented Dickey-Fuller (ADF) test that takes into account the usage of same (i.e., common) sub-sample for all of the lag orders for the autoregressive process when stationarity is investigated. As is known, in all of the lag selection procedures in econometrics, same sub-sample must be used to determine the correct optimal minimum lag. We bouqoueted adfcs functions in adfcstable function whose functional value is a table that reveals all of the needed stationarity analysis of all the variables in a given system in a couple of seconds. In the returned ADF table, the results of all of the three standard cases (“both drift and time trend”, “drift without time trend” and “no drift, no time trend”) are presented for all of the variables in question. Whether the drift and time trend coefficients in the ADF regressions is significant is specified. adfcstable reveals the inconclusivities of ADF tests (the coefficient of the 1st lag of the dependent variable in the right of ADF regression is not “<0”; in the left, the dependent variable appear with the differenced form) whenever there appears such cases. It also presents optimal minimum lag order for the ADF regressions. We used three datasets from various fields: a dataset of functional integration of brain, a dataset for the determinants of foreign direct investment in Turkey, and a dataset for the determinants of current account deficit of Turkey.