Hello,

It is not clear what is your goal when you try to model a function like:

GDP=f( IT/GDP, DT/GDP )

This will give you an option to test a hypothesis that an expansion of the macro tax burden (i.e. share of taxes in GDP) can predict (potentially) an expansion (or, shrinkage?) of the GDP itself - is your intention something like this, no?

This is a serious problem in economics and economic policy, by my opinion, cannot be answered in such a simple way - "more /or less?/ relative fiscal intervention could accelerate GDP growth"...

As a start, I would suggest that you get familiar with the concept of Wagner's Law, e.g.:

http://questjournals.org/jrbm/papers/vol4-issue6/D463345.pdf

In your case, see 2 particular forms: Musgrave version; Peacock-Wiseman "share" version.

However, these models swap your dependent and independent variable - they seek answers to questions related to the growth of the public sector - is this happening systematically along with an expansion of the income per capita? For example: G/GDP=b0+b1GDPpc , hypothesis: b1>0.

Your "aggregate taxes" are closely related to the macro-size of the government G.

Hoping this will be of some help to you...

Hello Venelin Boshnakov,

I want to see whether tax revenue (direct tax and indirect tax) can contribute to the growth of GDP? I am not sure the way I have model it GDP=f(IT/GDP, DT/GDP) , is enough to serve my purpose?

or should I just model GDP=f(IT, DT)

Please suggest.

The model GDP=f(IT, DT) includes so called "endogenous" variables - their values are determined simultaneously by the economic system, e.g. in most simple form we have IN THE SAME TIME:

IT=c0+c1*GDP or log-log (to get elasticity of IT to GDP):

ln(IT)=c0+c1*ln(GDP)

However, you might be interested in equation like:

IT=c0+c1*GDP+c2*(100*IT/GDP) - here "c2" should introduce a kind of "policy" variable that reflects the tendency of government to extract more by the "indirect" fiscal channel. And this is "ceteris paribus" - holding the level of gross income constant, c2 should measure the marginal change in IT only due to the increase of [IT share in GDP] by 1 percentage point.

For example, there is a hypothesis that liberal governments (that tend to keep low taxes, as a share of GDP) succeed to induce growth - here you can use a "GDP growth" dependent variable - e.g. y[t]=ln(Y[t]/Y[t-1]). But there are many other determinants of growth, so you need to expand the equation - it might happen that growth is due to other factors, not much to the governmental policy to "tax more or less".

You are interested more in:

GDP=f(IT/GDP, DT/GDP)

( or alternatively, GDPgrowthrate=f(IT/GDP, DT/GDP) )

for example: GDP=12.5 -0.80*(IT/GDP) -0.40*(DT/GDP)

So, you could find that 1 percentage point increase in the share of IT is associated with (lets say) $0.8 Bn lower lewel of GDP, holding constant the burden of direct taxation. The marginal effect of the burden of direct taxes here is twice lower (-0.4 Bn).

But these equation require much more comprehensive elaboration - there are also many other variables that participate in the formation of GDP (or GDP growth rate), and too much endogeneity... This requires development of a system of simultaneous equations. Also, with time series you need to check issues of stationarity, integration, cointegration, etc. - things that will complicate the work substantially. But this depends on the goals of your analysis - one thing is a subject course work, another is BA thesis, ot MA thesis, or PhD thesis and related journal articles with required scientific level...

Wishing you success!