Add IV to Table 5

[paul-opheim]

It seems to me that we should include the results of the IV regression in Table 5 (instead of Table 6) now that we are also including IV in the simulations.

[ijyliu]

I agree, but what about the univariate OLS? Idk if we should drop it or move it to table 6.

Also, what order should we present results in? The simulations and the empirical section are ordered differently.

[paul-opheim]

How about we include the univariate OLS but then order the application results in the same order as the simulation ones?

[ijyliu]

Should we add univariate OLS to the sims? Since we are going to do this all in one sweep anyway

Anyway, for the empirical section I guess it's possible to fit 6 columns of regression results but not sure

[paul-opheim]

Up to you on the first question. I personally don't see a need to add the univariate OLS to the sims (it wouldn't really tell us anything that we would need to discuss in the paper).

I would think that we could fit 6 columns of regression results given that we've made a 7 column table before (the summary statistics table), but true we would have to look into that.

[ijyliu]

For now I removed univariate OLS. I think we will want to add it back though. I really think we should do it in both the sims and empirical section or not do it in both though.

If we add it back we maybe can put it in the additional stuff table. So then the main table exactly corresponds to the sims. Maybe then there's no need to add it to the sims

[ijyliu]

@marionoro do you like the current setup? We kept the univariate ols in the empirical section but I moved it to the additional items table. That way the main table is directly comparable to the simulations. I guess there's no need to do univariate ols in the sims.

Also, what order should we present results in? The simulations and the empirical section are ordered differently.

Are there any additional models we should add to the additional items table?

[paul-opheim]

I like the current setup!

I figure the order doesn't matter too much, so we should just follow the order from the simulation tables. I changed the empirical table so that it aligns with this order.

I can't think of any additional models. It seems like the list that you have there is good.

[ijyliu]

I think what we have is probably fine.

In terms of additional models, I think if we were to do all the models in the first table with fixed effects, we wouldn't really learn anything because they would probably all be insignificant/similar. And of course the number of PCs doesn't matter because the first one explains so much.