VAR models have become increasingly popular in recent decades. VAR provides empirical evidence on the response of macroeconomic variables to various exogenous shocks or impulses. Within the framework of a Vector Auto-regressive model (VAR), I want to conduct a robustness test. Specifically, I want to study the impact of policy rate on lending rates and examine the impact of a positive and negative shock on lending rates.
A positive shock (e.g. increase in policy rate) can only allow me to see the impact of a contractionary monetary policy on lending rates. But, I am more interested to look at the impact of a negative shock (e.g. decrease in policy rate or expansionary monetary policy) on lending rates.
So, how to determine the impact of a negative shock via the ‘Impulse Response Function’ using Eviews or other statistical packages?
Administer the proper variables that have direct and indirect influence on lending-rate; work-out the associate relation between policy-rate and lending rate; multiple correlation co-efficient statistical method may help to find-out the impact on each other.
In EViews, use the "user specified" option for inputting the magnitude of the shock (this will be new variable vector that you create with zeroes in all but one element. Include a negative value of your choosing in one period only for the negative shock)
The impulse responses for positive and negative shocks, however, will be mirror images of one another unless you specify some form of non-linearity in your VAR equations (interact the endogenous variables with a dummy indicating a negative change in lagged policy variable, for example).
hi there, if you assume log-normality, u wont get the negative impact you require. Normality, will allow nominal rates to be negative.
You may select a dummy to control for that, say 1 for negative effects and 0 for positive effects.
Positive shock effects and negative shock effects in impulse response functions are the mirror images of each other. When one is known , the other is automatically known.. All statistical functions work in that way. A positive shock increases say variable X by 1 % , negative shock is supposed to decrease variable X by 1 %.
Isolating impact would be especially complicated by the delay in transmission of any change in lending rates and a prelim study to select the period of delay that has significance could probably be a first milestone? Also study only negative impact may thus make any data trend invisible as the inelastic market means the change in lending rates is probably much smaller..
what if there is theoretically an asymmetry in the response to neg vs. pos shocks? How can I capture the asymmetry in responses? I am studying the response of output to oil price shocks in oil exporters. conditional on the relative size of the government, output should respond stronger to negative oil price shocks in state-dominant economies, while its response to positive oil price shocks should be smaller. - thanks in advance
- Badges
- Science topic
- Similar topics
- Biological Science
- Anatomy
- Tissue