Discussion Paper

No. 2018-21 | February 21, 2018
The effect of Double Taxation Treaties and Territorial Tax Systems on Foreign Direct Investment: evidence for Spain
(Published in Special Issue FDI and Multinational Corporations)


The present paper evaluates the effect of Double Taxation Treaties and the Territorial Tax System of countries on Spain’s inward and outward FDI for the period 1993–2013. Estimations produce a positive and statistically significant effect of Treaties for both samples when using a simple binary variable for measuring the effect of the mere existence of the same. These outcomes keep for old and new Treaties and for the sub-sample of developed partner countries of Spain. However, regarding developing countries, the positive result exists only for the outbound sample. Also for the global samples and the sub-samples of developed countries, there is an additional positive effect on investments for countries applying the Territorial Tax System for taxing foreign income.

Data Set

JEL Classification:

F21, F23, F68, H25, H32, H87


  • Downloads: 155


Cite As

Ángela Castillo-Murciego and Julio López-Laborda (2018). The effect of Double Taxation Treaties and Territorial Tax Systems on Foreign Direct Investment: evidence for Spain. Economics Discussion Papers, No 2018-21, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2018-21

Comments and Questions

Anonymous - Referee Report 1
March 19, 2018 - 08:04

Report on MS 2546 “The Effect of Double Taxation Treaties and Territorial Tax Systems on Foreign Direct Investment: Evidence from Spain” submitted to Economics.

The paper analyzes the effects of double taxation treaties and territorial tax systems on inward and outward FDI to and from Spain. Tax treaties are ...[more]

... found to have a positive effect. In additional specifications, the results suggest that (i) both old and new treaties have a positive effect; (ii) it matters what type of international tax system (territorial or worldwide) is implemented in home/host countries.

The paper addresses an interesting and important issue. However, I have two main concerns.

First, the paper is not well written. This includes issues of grammar, exposition, a number of inconsistencies (e.g. in the references) as well as typos.

Second, I have strong reservations regarding the empirical analysis. The empirical model presented in equation (1) cannot be correct as dist(sh) is not identified once eta(sh) is included. Including the sub-period indicators in columns (3) and (6) of Table 2 is superfluous, it just assumes a higher level of aggregation for the aggregate time shocks. There should be more rigor in the way how variables are defined. For example, there should be a more formal definition of the dtt indicator variable. How many countries are included in the sample? There are 21 years of data, the columns (1) to (3) report 860 observations. Hence, how many of the countries shown in Table 1 are really included in the estimates. On how many changes in dtt is identification actually based on? Shouldn’t the empirical analysis condition on a tax rate? There cannot be an endogeneity problem related to the old dtts if columns (1) and (3) are considered. The reason is that these are fully captured by the eta(sh). My point is not that the coefficients on odtt and ndtt should not differ. My point is that the arguments presented on page 10 concerning endogeneity are not correct. You can present the same arguments concerning the ndtt variable. In fact, selection into new treaties might be an issue when it comes to ndtt.

Additional comments:

I think calling Section 2 “The theory of Double Taxation Treaties” is misleading. The section provides some literature and a general notion of what a double taxation treaty does.

The introduction suggests that only few papers examine the effects of double taxation treaties. There are quite some studies investigating this question.

Page 6: The knowledge-capital model of Markusen is a general equilibrium model, it should be made clear that the econometric specific is a reduced-form specification as some of the arguments imply that the theory is directly tested by estimating equation (1).

Why is the distance measure indexed by “ij” and not by “sh” (“ij” is not even introduced); dq_sht_ijt is indexed twice.

Page 7: The “selected econometric specification is in absolute terms”. The specification cannot be “in absolute terms” (variables or measures can be).

I think the tables presenting the results should show the estimated coefficients of the control variables. The notes on the tables are incomplete.

Why should the territorial system have a positive effect? The paper only poorly explains what we would expect from a theoretical/taxation point of view.

The paper provides a number of reasons why previous findings are not always clear. This part does not really add to this debate, though.

As mentioned above, there are quite a few grammatical issues which have to be fixed.