Error component model

Summary

• A common model for the error term in a LRM with panel data is the error component model . There are three versions of the error component model:
• The one-way error component model with individual specific effects

$$ε_{i,t}=α_i+μ_{i,t}$$

• The one-way error component model with time specific effects

$$ε_{i,t}=γ_t+μ_{i,t}$$

• The two-way error component model

$$ε_{i,t}=α_i+γ_t+μ_{i,t}$$

• $α_i$ is called the individual specific effect and $γ_t$ is called the time specific effect .
• The error terms $μ_{i,t}$ are typically assumed to be homoscedastic and not autocorrelated.