![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
ABSTRACT
Firms that import intermediate goods choose between outsourcing and vertical integration. When corporate tax rates differ between the home country and the foreign country, the possibility of shifting income and reducing overall tax payments through transfer pricing makes integration more attractive than outsourcing. This paper develops an incomplete-contracting model in which an international firm facing tax rate differentials chooses whether or not to internalize intermediate transactions in order to trade off production efficiency and tax minimization. By shifting economic activities across borders, an integrated multinational enterprise establishes a proper transfer price and reaches the optimal profit-splitting arrangement that maximizes its total after-tax profit. This paper finds that cross-country differences in corporate tax rates and product intangibility play important roles in affecting firms’ internalization decision. Empirical analysis employing the US data also supports the theoretical findings. The positive correlation of the integration level of US firms and tax rate differentials between the US and foreign countries remains in the sample excluding tax havens.
Acknowledgments
I am grateful to Professor James Harrigan, John McLaren, and Leora Friedberg for their invaluable guidance and unwavering support. I thank Zhengyu Bao for his excellent research assistance.
Notes
1. See Section 3 for details of variable construction and data sources. is drawn using data in 2004. To diminish endogeneity concerns, I also employ tax rates in 2004 and imports in 2005 to utilize the one-year lag of tax rates. The graph looks similar to .
2. More tests addressing this issue will be conducted in Section 4.4.
3. As in Antrãs (Citation2003), in a world of incomplete contracts where the production of intermediate goods requires a combination of non-contractible and relationship-specific investments from both the headquarters (F) and the foreign subsidiary (M), F and M bargain over the ex post gains from trade. The key insight is that F or M will invest less if too much profit is obtained by the other party.
4. International Transfer Pricing 2009, published by PricewaterhouseCoopers (London), has detailed description on the criteria of the establishment of the transfer price.
5. Tax havens are low-tax jurisdictions that provide investors opportunities for tax avoidance. Examples of such tax havens exist in Ireland and Luxembourg in Europe, Hong Kong and Singapore in Asia, and various Caribbean island nations in the Americas. I will list the tax havens in the sample used in this paper in Section 4.5.
6. According to the US credit and deferral system, for profits realized in a foreign country, a US MNE not only pays taxes to the foreign tax authority, but also takes on a tax liability in the US. However, it receives a credit from the US government for the taxes paid abroad which can be subtracted from the domestic tax liability. If the tax rate is lower in the foreign country, the credit will be smaller than the domestic tax liability, and the MNE has a ‘deficit foreign tax credit.’ The MNE is still taxed at the ‘differenced rate,’ and net taxes paid should be the same. In this case, no tax avoidance arises from shifting income to the foreign country. However, domestic taxes on foreign income can usually be deferred until the income is remitted in the form of dividends.
7. Besides the time value of money, the deferral system could yield other benefits. In practice, the incentives to repatriate are weak because funds retained offshore can be used in various ways, such as purchasing foreign companies and securing loans; firms sometimes bring back a fraction of their overseas profits, and others might do so in the future; studies show that repatriations from low-tax jurisdictions are small today (Zucman Citation2014). It is also possible that companies can be taxed at a lower rate during a tax holiday in the future.
8. As mentioned in Section 1, there are three main business responsibilities (functions, risks, and the OIP) in transfer pricing practice, which serve as the criteria to establish a proper transfer price. The idea that MNEs can shift income by planning the allocation of the OIP can also be generalized to functions and risks when they can be easily moved. However, to simplify the logic into a model, this paper focuses on the OIP.
9. I restrict from being equal to 1, because if
, there would be no costs associated with firing M. In this case, F would always have an incentive to seize all intermediate outputs ex post, and M would have no incentive to invest mk(i) ex ante. See Appendix 1 for more details.
10. is a function of
and β, where
is the endogenous arrangement of responsibility delegation and β is the exogenous bargaining power.
11. An implicit transfer price can also be defined as a function of and
. Since I will not empirically test the transfer price, it is unnecessary to compute it in the model. The fundamental interest is to model the income-shifting behavior.
12. Although this model does not consider the possible expenses of income shifting (for example, the adjustment costs), efforts have been made to specify the function of income adjustments (or double taxation). Although MNEs prepare documentation annually to demonstrate that the transfer price is established based on business responsibilities, it is still possible that they are subject to income adjustments after a tax audit, namely double taxation, under which the company is then required to compensate for the tax underpayment by shifting income back from the foreign country. In principle, the tax overpayment to the foreign country should be refunded to the MNE, but this may well not be possible after a tax liability in the foreign country has become final meaning that the shifted income will, in full or part, be taxed twice. In the former version of the paper, such expenses of income shifting has been built into the model where I assumed that the probability of double taxation depends on the transaction volume. Numerical exercises have shown that incorporating the expected adjustment costs will not change the numerical solution displayed in . Therefore, to avoid distraction, this element is not presented in the current model.
13. See http://sasweb.ssd.census.gov/relatedparty/. This database reports the total imports and related-party imports since 2002. Related-party trade refers to shipments between US companies and their foreign subsidiaries as well as trade between US subsidiaries of foreign companies and their affiliates abroad. Firms are ‘related’ if either party owns, directly or indirectly, 10 percent or more of the other party.
14. Though brand value and advertising investments are also regarded as marketing intangibles in general, I do not expect that advertising intensity is positively related to the level of integration because advertising depends heavily on market conditions and the age of the product. For instance, when a new product enters a new market, the company tends to spend more on advertising. However, such an unestablished brand may not be able to bear the high fixed cost associated with integration as generally assumed. Due to factors that are not considered in this study, I do not include advertising intensity in the regressions.
15. This is a service for academic research and teaching provided by Professor Werner Antweiler at the Sauder School of Business at the University of British Columbia. The website provides access to current and historic daily exchange rates, which are available for approximately 200 countries. See http://pacific.commerce.ubc.ca/xr/data.html.
16. I caution that there could be other explanatory controls that are not included in the specification, e.g. training costs embedded in human capital of the host country (Grover Citation2012). Therefore, country-fixed effects are controlled in the industry--country-level regressions whenever it is applicable.
17. To save space, all tables of regressions using 2004 data of the effective tax rate and 2005 data of all other variables in Sections 4.1– 4.5 are not reported but are available upon request.
18. Roelfsema and Zhang (Citation2012) confirm the positive relationship between institutional quality (which is related to contract enforcement) and outsourcing in lower-middle-income countries. However, their proxy of institutional quality is an indicator of the quality of the government, rather than the ‘rule of law’ measure.
19. To save space, estimates of control variables are not presented in (as well as in Section 4.4 and in Section 4.5), but are available upon request.