xtdhcoint

Overview

The xtdhcoint package implements the Durbin-Hausman panel cointegration tests of Westerlund (2008). The tests are robust to cross-sectional dependence through common factor extraction using principal components.

Installation

Install from CRAN:

install.packages("xtdhcoint")

Or install the development version from GitHub:

# install.packages("devtools")
devtools::install_github("muhammedalkhalaf/xtdhcoint")

Usage

library(xtdhcoint)

# Load example data
data(fisher_panel)

# Run Durbin-Hausman panel cointegration test
result <- xtdhcoint(inflation ~ interest, data = fisher_panel,
                    id = "country", time = "year")

# Print results
print(result)

# Detailed summary with decision table
summary(result)

Test Statistics

The package provides two test statistics:

• DHg (Group-mean statistic): Tests the null of no cointegration against the heterogeneous alternative that at least some units are cointegrated.

• DHp (Panel statistic): Tests the null of no cointegration against the homogeneous alternative that all units are cointegrated with a common autoregressive parameter.

Both statistics are right-tailed tests. Large positive values indicate evidence of cointegration.

Options

• kmax: Maximum number of common factors; the search runs over k = 0, …, kmax (default: 5)
• nfactors: Fix the number of factors (0 means no defactoring) and bypass the criterion
• criterion: Information criterion for factor selection (“ic”, “pc”, “aic”, “bic”)
• penalty: Penalty function 1, 2 or 3 for “ic” and “pc”; not used with “aic” and “bic”
• bandwidth: Bartlett kernel bandwidth; -1 (default) uses floor(4 (T/100)^(2/9)) as in Westerlund (2008, Section 3.4)

Missing values are rejected with an error. Unbalanced panels are aligned on the time variable and reduced to the common time window; see ?xtdhcoint.

References

Westerlund, J. (2008). Panel cointegration tests of the Fisher effect. Journal of Applied Econometrics, 23(2), 193-233. doi:10.1002/jae.967

License

GPL-3