dcce

Dynamic Common Correlated Effects Estimation for Panel Data

dcce is an R package implementing the family of Common Correlated Effects (CCE) estimators for heterogeneous coefficient panel data models with cross-sectional dependence. It is an R port of Jan Ditzen’s xtdcce2 Stata package and provides the standard estimators of Pesaran (2006), Chudik & Pesaran (2015), and related contributions, together with a comprehensive cross-sectional dependence (CD) test suite.

Features

Estimators

All three long-run estimators produce a three-block output: short-run coefficients, the adjustment (speed of return to equilibrium), and long-run elasticities with delta-method standard errors.

Cross-sectional dependence tests

Other diagnostics

Extensions

S3 methods and ergonomics

• broom-compatible tidy() and glance() (tidy includes short-run, adjustment, and long-run rows for LR estimators)
• confint() with type = c("mg", "lr", "adjustment")
• plot() for unit-level coefficient histograms and residual diagnostics
• update() for refitting with modified arguments
• coef(fit, type = "unit") for unit-level coefficient extraction
• Native support for L(), D(), and Lrange() operators in formulas (xtdcce2-compatible syntax)
• Unbalanced panel handling

Installation

The package is in active development and not yet on CRAN. Install the development version from GitHub:

# install.packages("remotes")
remotes::install_github("Mustapha-Wasseja/dcce")

To build the vignette during installation:

remotes::install_github("Mustapha-Wasseja/dcce", build_vignettes = TRUE)

System requirements

• R >= 4.1.0
• Imports: stats, Matrix, collapse (>= 2.0.0), sandwich, generics, rlang (>= 1.1.0), cli (>= 3.0.0), tibble
• Suggests: broom, ggplot2, lifecycle, plm, testthat (>= 3.0.0), knitr, rmarkdown

Quick start

library(dcce)

# Load the bundled Penn World Tables 8 dataset (93 countries, 1960-2007)
data(pwt8)

# Fit a Dynamic CCE growth regression with 3 lags of CSAs
fit <- dcce(
  data               = pwt8,
  unit_index         = "country",
  time_index         = "year",
  formula            = d_log_rgdpo ~ L(log_rgdpo, 1) + log_hc + log_ck + log_ngd,
  model              = "dcce",
  cross_section_vars = ~ log_rgdpo + log_hc + log_ck + log_ngd,
  cross_section_lags = 3
)

print(fit)

# Verify that DCCE has removed cross-sectional dependence
pcd_test(fit, test = "pesaran")

# Tidy output (broom compatible)
tidy(fit)
glance(fit)

For a complete walkthrough including motivation, theory, all estimators, and the Ditzen (2018) replication, see the package vignette:

vignette("dcce-introduction", package = "dcce")

Verify your installation

The fastest way to confirm the package works on your system is to run a few of the worked examples on the bundled datasets:

library(dcce)

# Example 1: Mean Group on the simulated dataset
data(dcce_sim)
fit_mg <- dcce(
  data = dcce_sim, unit_index = "unit", time_index = "time",
  formula = y ~ L(y, 1) + x,
  model = "mg", cross_section_vars = NULL
)
coef(fit_mg)

# Example 2: DCCE with CD test on residuals
data(pwt8)
fit_dcce <- dcce(
  data = pwt8, unit_index = "country", time_index = "year",
  formula = d_log_rgdpo ~ L(log_rgdpo, 1) + log_hc + log_ck + log_ngd,
  model = "dcce",
  cross_section_vars = ~ log_rgdpo + log_hc + log_ck + log_ngd,
  cross_section_lags = 3
)
pcd_test(fit_dcce, test = "pesaran")  # Should be insignificant after DCCE

# Example 3: Bootstrap inference
set.seed(42)
boot <- bootstrap(fit_dcce, type = "crosssection", reps = 199)
print(boot)

Note: broom::bootstrap conflict

If you load broom in the same session, broom::bootstrap will mask dcce::bootstrap (they share a name but have completely different signatures). Two workarounds, pick either:

# Option A: use the namespace prefix
dcce::bootstrap(fit, type = "crosssection", reps = 199)

# Option B: use the conflict-free alias exported by dcce
dcce_bootstrap(fit, type = "crosssection", reps = 199)

dcce_bootstrap() is identical to dcce::bootstrap() and cannot be masked by any other package.

Optional: validate against plm

If you have plm installed, the package’s static CCE estimator matches plm::pmg(..., model = "cmg") to three decimal places on the Produc dataset. This is checked automatically by the bundled tests/testthat/test-produc-validation.R file.

Usage notes

Formula operators. The package extends standard R formulas with three panel-aware operators:

• L(x, k) — k-th lag of x within each unit
• D(x, k) — k-th difference of x within each unit
• Lrange(x, k0, k1) — lags k0 through k1 (used in CS-ARDL)

Cross-section variables. Use cross_section_vars = ~ . to include all regressors plus the dependent variable as CSAs (the default), or provide an explicit one-sided formula such as ~ log_rgdpo + log_hc.

CSA lags. For dynamic models the Chudik-Pesaran rule p_T = floor(T^(1/3)) is the standard recommendation (cross_section_lags = 3 for T ≈ 30-50).

References

• Bai, J. (2009). Panel data models with interactive fixed effects. Econometrica, 77(4), 1229–1279.
• Bailey, N., Kapetanios, G., & Pesaran, M. H. (2016). Exponent of cross-sectional dependence: estimation and inference. Journal of Applied Econometrics, 31(6), 929–960.
• Chudik, A., & Pesaran, M. H. (2015). Common correlated effects estimation of heterogeneous dynamic panel data models with weakly exogenous regressors. Journal of Econometrics, 188(2), 393–420.
• Chudik, A., Mohaddes, K., Pesaran, M. H., & Raissi, M. (2016). Long-run effects in large heterogeneous panel data models with cross-sectionally correlated errors. In Essays in Honor of Aman Ullah, 36, 85–135.
• De Vos, I., Everaert, G., & Sarafidis, V. (2024). A method to evaluate the rank condition for CCE estimators. Econometric Reviews, 43(2–4).
• Ditzen, J. (2018). Estimating dynamic common-correlated effects in Stata. The Stata Journal, 18(3), 585–617.
• Dumitrescu, E.-I., & Hurlin, C. (2012). Testing for Granger non-causality in heterogeneous panels. Economic Modelling, 29(4), 1450–1460.
• Eberhardt, M., & Teal, F. (2010). Productivity analysis in global manufacturing production. Economics Series Working Papers 515, University of Oxford.
• Fan, J., Liao, Y., & Yao, J. (2015). Power enhancement in high-dimensional cross-sectional tests. Econometrica, 83(4), 1497–1541.
• Im, K. S., Pesaran, M. H., & Shin, Y. (2003). Testing for unit roots in heterogeneous panels. Journal of Econometrics, 115(1), 53–74.
• Juodis, A., & Reese, S. (2022). The incidental parameters problem in testing for remaining cross-section correlation. Journal of Business & Economic Statistics, 40(3), 1191–1203.
• Kao, C. (1999). Spurious regression and residual-based tests for cointegration in panel data. Journal of Econometrics, 90(1), 1–44.
• Margaritella, L., & Westerlund, J. (2023). Using information criteria to select averages in CCE. The Econometrics Journal, 26(3), 405–421.
• Pedroni, P. (2004). Panel cointegration: asymptotic and finite sample properties of pooled time series tests. Econometric Theory, 20(3), 597–625.
• Pesaran, M. H. (2006). Estimation and inference in large heterogeneous panels with a multifactor error structure. Econometrica, 74(4), 967–1012.
• Pesaran, M. H. (2007). A simple panel unit root test in the presence of cross-section dependence. Journal of Applied Econometrics, 22(2), 265–312.
• Pesaran, M. H. (2015). Testing weak cross-sectional dependence in large panels. Econometric Reviews, 34(6–10), 1089–1117.
• Pesaran, M. H., & Smith, R. (1995). Estimating long-run relationships from dynamic heterogeneous panels. Journal of Econometrics, 68(1), 79–113.
• Shin, Y., Pesaran, M. H., & Smith, R. (1999). An autoregressive distributed-lag modelling approach to cointegration analysis. In Econometrics and Economic Theory in the 20th Century, 371–413.
• Westerlund, J. (2007). Testing for error correction in panel data. Oxford Bulletin of Economics and Statistics, 69(6), 709–748.

License

GPL (>= 3)

Citation

If you use dcce in published work, please cite the package and the relevant methodological references above.

citation("dcce")

Issues

Bug reports and feature requests are welcome at https://github.com/Mustapha-Wasseja/dcce/issues.