The goal of this code is to implement the evaluation framework of interval forecasts proposed in Dumitrescu, Hurlin and Madkour (2011). This is a set of three univariate tests (unconditional coverage J_UC, conditional coverage J_CC, and independence J_IND) that can be implemented with any dataset. Besides, the results of Christoffersen (1998)'s tests (LR_UC, LR_CC and LR_IND) are also reported. To run my codes, you need (i) the Indicator Series (1 - if the realization of the variable to be forecasted does not belong to the interval forecast, i.e. violation; 0 - if it is in the interval) (ii) Risk level (5% or 1%) (iii) Block size N=T/nn, where nn is chosen from (4;10;50) and T is the out-of-sample size (iv) Number of moment conditions for the GMM test (from 2 to 10).
Working Paper (2021)
Couperier Ophélie and Leymarie Jérémy