Levin-Lin-Chu (LLC) Panel Unit-Root Test
Also known as: LLC Test, Panel Unit-Root Test (Homogeneous), Levin-Lin Unit-Root Test, Panel Birim Kök Testi (LLC)
The Levin-Lin-Chu (LLC) test, introduced by Levin, Lin, and Chu (2002), is a first-generation panel unit-root test that pools cross-sectional information to test whether all units in a panel share a common autoregressive unit root. It is widely used in applied economics and finance when researchers work with balanced or near-balanced panels and require a powerful test against a homogeneous stationary alternative.
Key highlights
- Substantially higher power than single-equation ADF or PP tests when N is large and the homogeneity assumption holds.
- Asymptotically standard-normal distribution after simple bias corrections simplifies inference without simulation.
- Accommodates individual fixed effects, linear time trends, and heterogeneous short-run dynamics across units.
- Seminal and well-understood benchmark; results are directly comparable across a vast applied literature.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Apply the LLC test when you have a panel with moderate to large N and moderate T, and theory or prior evidence suggests that the autoregressive parameter is homogeneous across units. It is appropriate for balanced or mildly unbalanced panels. The test assumes cross-sectional independence—a strong assumption that is often violated in macro panels where common shocks (business cycles, commodity prices) induce cross-unit correlation. When cross-sectional dependence is suspected, prefer second-generation tests such as the CIPS test (Pesaran, 2007) or the Moon-Perron test. If heterogeneous autoregressive roots are plausible, use the Im-Pesaran-Shin (IPS) test instead.
Strengths & limitations
- Substantially higher power than single-equation ADF or PP tests when N is large and the homogeneity assumption holds.
- Asymptotically standard-normal distribution after simple bias corrections simplifies inference without simulation.
- Accommodates individual fixed effects, linear time trends, and heterogeneous short-run dynamics across units.
- Seminal and well-understood benchmark; results are directly comparable across a vast applied literature.
- Assumes a common rho under the alternative, so the test cannot detect partial stationarity (some units stationary, others not).
- Requires cross-sectional independence; size distortions can be severe under strong cross-sectional dependence.
- Assumes a balanced or near-balanced panel; severely unbalanced panels may violate the asymptotic approximations.
- Finite-sample size can exceed nominal level when T is small relative to N, even with bias corrections.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How does the LLC test differ from the Im-Pesaran-Shin (IPS) test?
LLC assumes a common autoregressive coefficient rho under both null and alternative, yielding higher power when homogeneity holds but consistency only against the global alternative that all units are stationary. IPS allows unit-specific rho_i under the alternative, so it remains consistent even if only a fraction of units are stationary. Use LLC as a baseline and IPS when heterogeneous roots are plausible.
What deterministic specification should I choose for the LLC test?
Choose based on economic theory and visual inspection of the series. If the series appears mean-zero (demeaned data), use no deterministic term. If levels fluctuate around a non-zero mean, include individual fixed effects. If a secular trend is visible, add a linear time trend. Misspecification—especially omitting a trend when it exists—degrades the test's size and power properties substantially.
Does rejecting the LLC null mean all panel units are stationary?
Not necessarily. LLC rejection implies that the assumption of a common unit root for all N units can be rejected. It does not pinpoint which units are stationary nor does it imply uniform stationarity. If the researcher needs unit-by-unit conclusions, complement the LLC test with individual ADF tests or the IPS test and apply appropriate multiple-testing corrections.
Sources
- 1.Levin, A., Lin, C.-F., & Chu, C.-S. J. (2002). Unit root tests in panel data: asymptotic and finite-sample properties. Journal of Econometrics, 108(1), 1–24.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 2). Levin-Lin-Chu Test. ScholarGate. https://scholargate.app/econometrics/levin-lin-chu-test