Regresses the squared (demeaned) series on a constant and its own q
lags and forms \(LM = m R^2\) with \(m = n - q\), asymptotically
\(\chi^2_q\) under the null of no ARCH. Rejection says the
variance is predictable from its own past while the level may stay
serially uncorrelated – a KS normality check cannot detect this at
all, which is what the Python placeholder this mirrors replaced.
Usage
morie_arch_lm_test(r, q = 1L, demean = TRUE)
Arguments
- r
Numeric series (returns or residuals).
- q
Lags in the auxiliary regression.
- demean
Subtract the sample mean before squaring.
Value
Named list with statistic, p_value, df,
r2, n, q, method.
Details
Mirrors morie.fn.volengle.
References
Engle, R. F. (1982). Autoregressive conditional
heteroscedasticity with estimates of the variance of United
Kingdom inflation. Econometrica, 50(4), 987-1007. Sec. 8.
Examples
set.seed(1)
morie_arch_lm_test(rnorm(200), q = 2)$p_value
#> [1] 0.2566383
