On the appropriate measure of tax burden on foreign direct investment to the CEECs

Abstract

We argue from a conceptual and empirical point of view that tax-rate elasticities of foreign direct investment (FDI) to Central and East European Countries (CEECs) derived from statutory tax rates (STRs) are likely to be flawed. STRs are problematic measures of tax burden as they capture neither tax base effects, nor effects of the home country or international and supranational tax laws. From an empirical point of view STRs are questionable as their behavior over time and between country-pairs may differ from that of the conceptually superior bilateral corporate effective average tax rates (BCEATRs) of the Devereux-Griffith type. The variability of host-country STRs and BCEATRs of seven major home countries of FDI in eight major CEEC host countries is compared via Levene-tests for 1995--2005. Results indicate that using STRs instead of BCEATRs in empirical investigations of FDI is likely to result in tax-rate elasticities which are too low in absolute value.

Notes

¹ Mintz and Tsiopoulos (1997), Beyer (2002), Edmiston et al . (2003), Alfano (2004), Carstensen and Toubal (2004), Javorcik (2004), Benassy-Quere and Lahreche-Revil (2005) and Clausing and Dorubantu (2005).

² The elasticity is derived by multiplying the semi-elasticity by the average STR of the host countries considered in the surveyed articles. As it is probably below 69%, the derived elasticity is likely to be below 1.

³ Backward-looking rates, based on Revenue Statistics and National Accounts data are likely to suffer from endogeneity and hence, are inferior to both STRs and BCEATRs in econometric analysis. For the CEECs, backward-looking rates have been used e.g. by Benassy-Quere and Lahreche-Revil (2005).

⁴ Home countries are Austria (AUT), France (FR), Germany (GER), Italy (IT), the Netherlands (NL), United Kingdom (UK) and the United States (US). Host countries are Slovenia (SL), Hungary (HU), Poland (PL), Czech Republic (CZ), Slovak Republic (SK), Bulgaria (BU), Romania (RO) and Croatia (CRO). The assumptions made follow other studies (e.g. Devereux and Griffith, 1999; European Commission, 2001) except that we give inventories less and buildings more weight.

⁵ A significance level of 10% is appropriate here as the median version of the Levene-test is conservative.