Splitting dummy variable into subgroups

[Ismoil Ismoilov]

Dear Statalist, I need help splitting the dummy variable into two subgroups.

I have three dummy variables: Fixed, Floating and Floating (Exchange Rate Regimes) where Fixed is the reference group in the regression. So now I need to analyze the effects of each Regime with and without the Monetary Policy Anchor for a partic, i.e split each regime into two subgroups, for a particular period.

1. What is the easiest way to split them by code?
2. How to control splitting the reference group (Fixed)?
3. How to interpret the results? (Before I interpreted only according to the reference group)

Thank you.

[Carlo Lazzaro]

Ismoil:
I find your question unclear.
If -Fixed- is the reference group:
1) you should omit that 0/1 categorical variable if you actually have a set of three categorical variables, such as Fixed, Floating and Floating. BTW: I would not sponsor this approach and prefer the one described at point 2);
2) if you have benefitted from -fvvarlist- notation, you have a three-level categorical variable (let's call it -Exchange Rate Regime-). When plugged in as -i.Exchange Rate Regime- in the right-hand side of your regression equation, Stata omits the refererence category (that you can select according to your preference) automatically;
In addition, form yoiu description I'd guess that you're considering an interaction between Monetary Policy Anchor and the abovementioned categorical variable(s). Again, -fvvarlist- can help you out in this respect, too;
Last but by no means least, following the FAQ about posting what you typed and what Stata gave you back (and/or sharing an excerpt/example of your dataset via -dataex-) is a receommendation that I cannot help from kindly remind you about. Thanks.

[Ismoil Ismoilov]

Ismoil:
I find your question unclear.
If -Fixed- is the reference group:
1) you should omit that 0/1 categorical variable if you actually have a set of three categorical variables, such as Fixed, Floating and Floating. BTW: I would not sponsor this approach and prefer the one described at point 2);
2) if you have benefitted from -fvvarlist- notation, you have a three-level categorical variable (let's call it -Exchange Rate Regime-). When plugged in as -i.Exchange Rate Regime- in the right-hand side of your regression equation, Stata omits the refererence category (that you can select according to your preference) automatically;
In addition, form yoiu description I'd guess that you're considering an interaction between Monetary Policy Anchor and the abovementioned categorical variable(s). Again, -fvvarlist- can help you out in this respect, too;
Last but by no means least, following the FAQ about posting what you typed and what Stata gave you back (and/or sharing an excerpt/example of your dataset via -dataex-) is a receommendation that I cannot help from kindly remind you about. Thanks.

Dear Carlo, thank you for your quick response and my apologies with the unclear expression of my analysis.

I am analyzing the effects of Exchange Rate Regimes (Fixed, Intermediate and Floating) on the Economic Growth of countries for a particular period of time. Particularly:

DV: GDP per capita growth

IV: Gross Capital Formation, Government Consumption, FDI, Initial per capita GDP, Population Growth, Secondary School Enrollment, Terms of Trade, Inflation, Domestic credit to the private sector, Rule of Law Banking crisis and Exchange Rate Regime dummies (where Fixed Regime is the reference group).

I ran regression and got some results

xtreg  gdp_pcg grosscapform lagGovconsump fdigdp gdpppc_ini pop_g sec tradegdp lagcpi tot domcredit RuleofLaw Bankingcrisis IMFdejureinter IMFdejurefloat i.year, fe rob cluster(id)

And as the next step of my research, I want to see the effects of ER regimes on economic growth with/without Monetary Policy Frameworks. Thank you.

[Carlo Lazzaro]

Ismoil:
1) you would be better off with creating a single three-level categorical variable for -Exchange Rate Regime- via -fvvarlist- notation;
2) you have to interact 1) with a categorical variable (0/1) for Monetary Policy Frameworks (see -fvvarlist- notation again).

[Ismoil Ismoilov]

Ismoil:
1) you would be better off with creating a single three-level categorical variable for -Exchange Rate Regime- via -fvvarlist- notation;
2) you have to interact 1) with a categorical variable (0/1) for Monetary Policy Frameworks (see -fvvarlist- notation again).

Thank you very much!