I'm not sure if this is 100% the best forum to post this but I figured I would give it a shot.
I'm currently researching the consistency between the European Central Bank's (ECB) 'words and deeds'. In order to do that, I constructed a simple index to quantify monthly press releases by the ECB. Each press release is rated on a scale of -2 to +2. -2 denotes a strong inclination to cut the bank's interest rate in the near future, 0 denotes tendencies towards no change, and +2 denotes strong inclinations to raise interest rates.
At the same time I have monthly data on the changes (if any) that the ECB made each month to the overall interest rate. Obviously the interest rate is not changed frequently, so I feel an ordered probit analysis to compare the two variables would be appropriate (possible results in the dependent variable are only raise/lower/no change to the rates).
So far this was fairly straightforward for me. I do have two questions however and I was wondering if someone could help me out:
1) Should I include other variables in this analysis? I figured this is merely a measurement of the consistency of communications to policy, and as such it is a bit of a unique independent variable as I'm not really looking for causality. I also thought it might get a bit weird to include other variables since it would end up being something like "Assuming inflation and unemployment don't change/ceteris paribus, the relationship between communications and policy = x", which wouldn't really tell me a whole lot.
2) As part of my analysis, I want to compare a lagged version of the communications index as well (i.e. how much do communications predict interest rate changes in the following month? In 2 months? In 3?). Would I have to run a separate ordered probit for each of these lagged index variables (so regular index vs. rate changes, then 1 month lagged index vs. rate changes, then 2 month lagged index vs rate changes etc.)? Or can I include all of them in one single ordered probit as several independent variables?
Thanks for any replies and I understand that these questions are probably quite dumb! (Sorry, still very new to statistics!)
Advertise on Talk Stats