When you output IRFs to a monetary shock using a VAR, are IRFs to one standard deviation monetary shock just differences between path of y with a one-time shock and and path of y without the shock?

For example, if the shock occurs at say $t=1$, what does $\frac{\partial y_{t+1}}{\partial \epsilon_{t}}$ really mean?

So before the monetary shock, is there a path of $y_{t}$ t=1,2,3,...?


1 Answer 1


The impulse response function simply tells you the path that $Y_t$ follows if kicked by a single unit shock $\epsilon_t$. I think that your notation is misleading. Consider the following:

If a shock ($\epsilon_t$) occurs at time $t$ (say today), then $\frac{\partial y_t}{\partial \epsilon_t}$ denotes the impact multiplier, that is, how $y_t$ responds to the impact (immediate response) of that shock. On the other hand, $\frac{\partial y_{t+j}}{\partial \epsilon_t}$, with $j>0$ and denoted as dynamic multiplier, tells you which is the effect of a shock occurred at time $t$ on the variable at time $y_{t+j}$.

For instance, if you have quarterly data, and you simulate a shock $\epsilon_t$ today, and you want to see how $y_t$ responds to that shock over time, $\frac{\partial y_{t}}{\partial \epsilon_t}$ simply tells you today's response of $y_t$ to the simulated shock $\epsilon_t$. Then, $\frac{\partial y_{t+1}}{\partial \epsilon_t}$ tells you what the persistence of today's shock is after 3 months, that is, how that shock still affects $y_t$ after 3 months.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.