“…LPs collect new estimates for each forecast horizon by regressing the dependent variable (vector) at horizon
on the information set at
; thus, the projections of forward values of the dependent variable (vector) on the information set are local to each horizon. Initially, it was argued that VARs and LPs are conceptually different, but the more recent literature demonstrates that the two methods in fact estimate the same population and sample IRFs (see also Mendieta‐Muñoz et al.,
2020, Section 5), and thus provide a suitable robustness check for VAR results. Figure 3 shows the key IRFs obtained from the models that considered
in the top row and
in the bottom row, respectively.…”