Transfer Pricing Comparables: Preferring a Close Neighbor over a Far-away Peer?

Bert Steens*, Thibaut Roques, Sébastien Gonnet, Christof Beuselinck, Matthias Petutschnig

*Corresponding author for this work

Research output: Working paper / PreprintWorking paperProfessional

Abstract

In a globalized economy, transfer pricing estimations are key in valuing international transactions between related entities of multinational corporations (MNCs), and the use of uncontrolled peer group comparison methods are widespread. In the absence of uniform guidelines on the optimal identification for comparable companies, however, it remains a concern that poor selection choices may lead to biased estimates. This, in turn, may systematically bias cross-jurisdictional revenue flows. The current approach employed by tax practitioners and implicitly endorsed by several tax administrations worldwide commonly relies on comparable selection from neighboring countries. We employ a large global sample of over 11,000 manufacturing firms located across 84 countries over the period 2012-2016. We find evidence that the risk level of the country of incorporation of the companies is highly correlated with their profitability and that geographical closeness is less relevant for explaining profitability when controlling for country risk. Our findings suggest that the search for foreign independent comparable companies is better guided by country risk rather than geographic proximity. From a taxation perspective, our study also suggests that insufficiently controlling for country-level sovereign risk on average biases high-risk countries’ corporate tax revenues downwards.
Original languageEnglish
PublisherSSRN e-library
Pages1-42
Number of pages42
DOIs
Publication statusPublished - 3 Dec 2020

Keywords

  • Transfer pricing; Corporate tax; Comparables; Country risk; Geographical proximity

Fingerprint

Dive into the research topics of 'Transfer Pricing Comparables: Preferring a Close Neighbor over a Far-away Peer?'. Together they form a unique fingerprint.

Cite this