Panel Cointegration Test

A panel cointegration test checks whether nonstationary variables observed across many units over time share a stable long-run equilibrium relationship, pooling information across units to increase power.

패널 공적분 검정은 여러 단위에서 시간에 걸쳐 관측된 비정상 변수들이 안정적인 장기 균형관계를 공유하는지를 단위 간 정보를 결합해 검정하는 방법이다.

How do I test for a long-run relationship between nonstationary variables in panel data?

패널 자료에서 비정상 변수들 사이의 장기 관계를 어떻게 검정하나요?

Panel cointegration tests, such as the Pedroni, Kao and Westerlund tests, examine whether integrated variables move together in the long run across cross-sectional units. They fit macro panels with a moderately long time dimension, after panel unit root tests indicate that the variables are nonstationary.

When to use Panel Cointegration Test

• Country, region or firm panels with long annual or quarterly series
• Variables found to be I(1) by panel unit root tests
• Testing a long-run relationship before estimating panel FMOLS, DOLS or error correction models

When to use something else

• Variables are stationary in levels: use standard panel regression with fixed or random effects
• Only a single time series is available: use the Engle–Granger or Johansen cointegration test

Assumptions and requirements

• Variables integrated of the same order, typically I(1)
• Sufficiently long time dimension for each unit
• Appropriate treatment of cross-sectional dependence (or a test robust to it)
• Correct deterministic terms (constant, trend)

What the analysis reports

• Panel and group test statistics
• p-values or bootstrap p-values
• Lag and deterministic specification used
• Preceding panel unit root test results

How to report it

The Westerlund test rejected the null hypothesis of no cointegration, Gt = [value], p = [value], indicating a long-run relationship between [Y] and [X].

Related methods

• Panel Unit Root Test
• Vector Error Correction Model (VECM)
• Panel ARDL

Run Panel Cointegration Test on Quorant