ARCH (Autoregressive conditional heteroskedasticity)

ARCH refers to an autoregressive equation in which the variance of the errors terms is heteroskedastic (i.e., error variance is not constant).

Testing for an ARCH process

1. Take the residuals from the original autoregressive model
2. and then square them
3. in determining whether an ARCH process exists is to regress the squared residuals from this period against the squared residuals from the previous period as follows:
ε_t² = b₀ + b₁ * ε_t-1²
4. If b₁ is statistically different from zero, then we conclude that the regression model contains an ARCH process.