Profit shifting: drivers of transfer (mis)pricing and the potential of countermeasures

Abstract

In trying to explain the drivers of global profit shifting by multinational enterprises (MNEs), we investigate firm-specific determinants and their variation across major industries. Using the ORBIS database, we show that intangible asset endowment of subsidiaries and the supply-chain complexity of MNE groups explain aggregate profit-shifting trends. According to our estimates, subsidiaries with no intangibles react to an incremental increase of the tax rate by reducing reported profits by 0.76 %, while subsidiaries with above median intangible endowment decrease their profits by 1.2 %. This difference is significant at the 5 % level. We find an even more pronounced difference in the observed semi-elasticities comparing affiliates belonging to simple ($-$0.52) and more complex MNEs ($-$1.92), suggesting a significantly larger sensitivity to CIT rate changes of the latter group. Moreover, we incorporate country-specific transfer pricing mitigation measures (documentation requirements) into our analysis. We find significant mitigation effects, which vary depending on the drivers identified in our analysis. On average, estimated profit shifting among MNE subsidiaries in our sample is reduced by 52 % 2 years after the introduction of mandatory documentation requirements. We do, however, not find a significant effect on affiliates with high intangible endowments, whereas documentation reduces profit shifting of subsidiaries within complex MNE groups. Our analysis suggests that complexity poses less of a challenge to effective domestic enforcement than the appropriate pricing of intangible assets. These findings thus provide additional insights on profit-shifting risks and mitigation effects conditional on firm attributes, which may support the design of anti-avoidance approaches and help guide the allocation of scarce analytical and enforcement resources.

Appendices

Appendix 1: Measuring supply-chain complexity

To quantify cross-border supply-chain complexity, we build on the notion of inter-industry linkages introduced by Rasmussen (1956).²⁷

Using Input–Output tables, provided by Eurostat, we calculate the $N \times N$ matrix

$$\begin{aligned} A=\left\{ a_{ij}\right\} =\left\{ \frac{X_{ij}}{x_j}\right\} , \end{aligned}$$

(2)

of direct requirement coefficients $a_{ij}$, with economy-wide output across $N$ sectors denoted by the $N \times 1$ vector $x$. The variable $X_{ij}$ depicts intermediate use of sector $i$’s outputs for the production process of sector $j$. Assuming a linear production function, the direct input requirements thus expresses the share of good $j$ that is produced by intermediate inputs from industry $i$. The column sums of $A$ are thus a simple measure of direct linkages between industry $j$ and the rest of the economy.²⁸

A measure that captures the cascading demand effect of increasing production, first developed by Leontief (1936), is given by so-called total linkages. In the static Leontief model, the most common application of Input–Output tables, the total output is related to the vector of final demand $y$ by

$$\begin{aligned} {x}=(I-A)^{-1}y=By, \end{aligned}$$

(3)

where the $N \times N$ matrix $B=\{b_{ij}\}$, called the Leontief-inverse, captures the economic dynamics in response to a change in final demand.²⁹ Specifically, additional economy-wide activity that is induced by a demand shock in sector $j$ is given by the $j$’th column sum $B_{j}=\sum _{i=1}^{n} b_{ij}$ of the Leontief inverse.

As discussed above, we hypothesize that transfer (mis)pricing risks of a subsidiary are increasing in the scale of the MNE group’s cross-border linkages. Denoting the set of different industries within MNE $k$ that are located in country H by $\mathcal {I}_{H,k}$, and the set of different industries within the same group that are located abroad by $\mathcal {I}_{-H,k}$, we define total backward cross-boarder linkages for affiliates in country H and MNE $k$ by

$$\begin{aligned} CB_{H,k}=\sum _{i\in \mathcal {I}_{H,k}} \sum _{j\in \mathcal {I}_{-H,k}} b_{ij}. \end{aligned}$$

(4)

Given that cross-border activity may also be driven by supply shocks, we calculate forward linkages, which are based on the Ghosh-model (Ghosh 1958), a supply-driven analog to the Leontief model. Our calculations follow the approach illustrated above, but rely on the direct sales coefficients $\tilde{a}_{ij}=X_{ij}/x_i$. Forward linkages are thus represented by row sums $\tilde{B}_{i.}=\sum _j \tilde{b}_{ij}$ of the Gosh-inverse $\tilde{B}=(I-\tilde{A})^{-1}$. We define total forward cross-border linkages within MNE $k$ and country H, analogously, by

$$\begin{aligned} CF_{H,k}=\sum _{i\in \mathcal {I}_{-H,k}} \sum _{j\in \mathcal {I}_{H,k}} \tilde{b}_{ij}. \end{aligned}$$

(5)

Our proxy for the complexity of the supply chain is given by the sum of forward and backward cross-border linkages

$$\begin{aligned} \textit{Complex}_{H,k}=CF_{H,k}+CB_{H,k}. \end{aligned}$$

(6)

Appendix 2: Descriptive statistics

See Table 7.

Table 7 Descriptive statistics

Appendix 3: Mitigation regression

See Table 8.

Table 8 Mitigation effect: Intangible intensity and complexity

Appendix 4: Documentation requirements: detailed description

See Tables 9 and 10.

Table 9 Documentation (1/2)

Table 10 Documentation (2/2)