Hostname: page-component-745bb68f8f-l4dxg Total loading time: 0 Render date: 2025-02-11T01:55:52.123Z Has data issue: false hasContentIssue false

03.5.1. A Concise Derivation of the Wallace and Hussain Fixed Effects Transformation—Solution

Published online by Cambridge University Press:  01 October 2004

Badi H. Baltagi
Affiliation:
Texas A&M University
Rights & Permissions [Opens in a new window]

Extract

A concise derivation of the Wallace and Hussain fixed effects transformation.

Type
PROBLEMS AND SOLUTIONS
Copyright
© 2004 Cambridge University Press

In vector form the disturbances can be written as

where Zμ = IN [otimes ] ιT, IN is an identity of dimension T, and ιN is a vector of ones dimension N, Zλ = ιN [otimes ] IT, μ is of dimension N × 1, λ is of dimension T × 1, and ν is of dimension NT × 1. In general, if Δ = [X1,X2] then PΔ = PX1 + P[QX1 X2], where PΔ = Δ(Δ′Δ)−1Δ′ denotes the projection matrix on Δ and QΔ = IPΔ. Applying this result to Δ = [Zμ,Zλ] , one gets

where P = IN [otimes ] JT with JT = ιTιT′/T and Q = IN [otimes ] ET with ET = ITJT. Using the fact that QZλ = ιN [otimes ] ET, ZλQZλ = NET, (ZλQZλ) = (1/N)ET, one gets PQZλ = JN [otimes ] ET. Hence

which means that

as required. Here QΔ is the fixed effects transformation derived by Wallace and Hussain (1969). Note that the order does not matter; i.e., one could have orthogonalized on Zλ.

Footnotes

An excellent solution has been independently proposed by Francisco J. Goerlich, University of Valencia.

References

REFERENCE

Wallace, T.D. & A. Hussain (1969) The use of error components models in combining cross-section and time-series data. Econometrica 37, 5572.Google Scholar