Kónya Bootstrap Panel Granger Causality
Also known as: Bootstrap Panel Causality Test, Kónya Panel Granger Causality, SUR-Based Bootstrap Causality, Kónya Önyükleme Nedensellik Testi
Introduced by László Kónya in 2006, this method tests Granger causality in heterogeneous panels by estimating a Seemingly Unrelated Regressions (SUR) system and deriving country-specific critical values through bootstrapping. Unlike pooled panel tests, it delivers a separate causality verdict for each cross-section, making it particularly valuable in applied macroeconomics and international economics when panel units are expected to behave differently.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use Kónya bootstrap causality when working with macro panels—typically 10–50 countries or regions—where cross-sectional dependence is suspected and units are heterogeneous. The test is appropriate when T is moderate (roughly 20–40 periods) and N is small enough for SUR to be feasible. It requires stationarity or pre-testing for integration; if series are I(1) and cointegrated, error-correction specifications should be considered instead. Alternatives include the Dumitrescu–Hurlin test (suited to larger N) or Toda–Yamamoto if integration order is uncertain.
Strengths & limitations
- Allows fully heterogeneous slope coefficients and lag lengths across cross-sections, unlike pooled panel Granger tests.
- Accounts for contemporaneous cross-sectional dependence through the SUR estimator without requiring a separate CD pre-test.
- Bootstrap critical values are valid in small and finite samples where asymptotic distributions are unreliable.
- Delivers unit-level causal conclusions, enabling country- or firm-specific policy interpretations.
- SUR estimation becomes computationally problematic and potentially singular when N is large relative to T.
- Requires pre-testing for stationarity; applying the test to non-stationary series without appropriate modification yields spurious results.
- Bootstrap size and power properties depend on the number of replications and the residual resampling scheme, which must be chosen carefully.
- Country-specific results reduce power compared to pooled tests when the true causal structure is homogeneous across units.
Frequently asked
Does the Kónya test require a prior cross-sectional dependence test?
Not formally—the SUR estimator inherently accommodates contemporaneous correlation across units, so the test remains valid whether or not cross-sectional dependence is present. However, running a CD test such as Pesaran's beforehand informs the choice between SUR-based and pooled estimators and strengthens the justification for the Kónya approach in a research paper.
Can the test be applied to I(1) variables?
The original Kónya (2006) framework assumes stationarity. Applying it directly to I(1) series risks spurious causality findings. Practitioners typically difference the series, test after applying a Toda–Yamamoto augmentation, or use an error-correction form when cointegration is confirmed before proceeding with the bootstrap Granger test.
How many bootstrap replications are sufficient?
Kónya (2006) used 10,000 replications in the original application. Simulation studies suggest that 1,000 replications are the practical minimum to stabilise critical values and control size, while 5,000–10,000 replications are recommended when the panel is small or results are borderline, to reduce Monte Carlo noise in the empirical quantiles.
Sources
- Kónya, L. (2006). Exports and growth: Granger causality analysis on OECD countries with a panel data approach. Economic Modelling, 23(6), 978–992. DOI: 10.1016/j.econmod.2006.04.008 ↗
How to cite this page
ScholarGate. (2026, June 2). Kónya Bootstrap Panel Granger Causality. ScholarGate. https://scholargate.app/en/econometrics/konya-bootstrap-causality
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Dumitrescu-Hurlin CausalityEconometrics↔ compare
- Granger CausalityEconometrics↔ compare
- Pesaran CD TestEconometrics↔ compare