Cumulative Impulse Responses

[nreigl]

Hi,

are there any theoretical considerations regarding a possible implementation of cumulative Impulse Response Functions for the linear local projection method?

[AdaemmerP]

Thank you for your interest in the package!

I haven't considered implementing cumulative multipliers for the other functions besides the panel ones. The wording "cumulative multipliers" in the package might be somewhat misleading, as it does not sum up the values of the endogenous variable but rather estimates the differences in regards to the first lag of the endogenous variable. The formula for computing them is given in the description of the functions. However, I'll consider renaming the option in a future version.

[nreigl]

Dear Philipp,

thank you for the response. Maybe I was not clear in my questions. I was thinking if it is possible to compute the cumulated impulse response coefficients to get an idea of the overall long-run effect of the shock.

[AdaemmerP]

No, this option is not included in the package. What functions are you currently using and what paper(s) are you referring to that use it? I might consider the option for a future version.

[nreigl]

I'm currently using lp_lin to compute linear impulse responses.

For example: Enders et al. (2018) seem to use from my understanding local projects to compute cumulative impulse response of output growth to a nowcast error (Figure 2: right Panel).

Z. Enders, M. Kleemann, and G. J. Muller, “Growth expectations, undue optimism, and short-run fluctuations,” p. 52., July 2018

[AdaemmerP]

Thanks for the paper. I do not see any equation how they estimate the local projections. I'll keep you posted regarding future implementation.

[alemello14]

Hi, Have you developed a cumulative option for lp_lin()? I got it the cumulative IRFs, since it's just do a cumsum() in irf_lin_mean. However, that's not the case for the CIs. In fact, even the Enders paper above seems to just sum the CIs, which is wrong. Many thanks!