Table 3.

Mini-max absolute error and mean squared error comparing simulated moments of order 31 with numerical solutions using different moment approximation schemes, among the parameters given in Table 1. Lowest errors inicated in bold.

Approximation method	Renormalize	Parsimonious	Absolute error	MSE
Logit-linear	True	False	0.000027	$\mathbf{4.24}·\mathbf{1}{\mathbf{0}}^{-\mathbf{7}}$
Linear	True	False	0.000122	$8.88·{10}^{-6}$
Jackknife	True	True	0.000254	$1.07·{10}^{-4}$
Jackknife-unconstrained	True	False	0.000334	$2.92·{10}^{-5}$
Jackknife-unconstrained	True	True	0.000240	$1.01·{10}^{-4}$
Least-squares	True	NA	0.000201	$6.75·{10}^{-5}$