Monday, December 21, 2020

Nash Bargaining, Splitting Factor, and Ex-Post Outcome Testing

A research that produces know-how or other intangibles often involve highly skilled personnel, and vary in terms of importance and success. If the parties involved are independent from each other, a combination of market force, bargaining power, self-interest and profit maximizing behaviour of both parties will result in return and remuneration that are at arm’s length.

If the parties are affiliated, pricing could be distorted by other factors. One of which is tax-motivated profit shifting. If the tax rates governing the parent and the subsidiary differ, taxpayer may be tempted to apportion income beyond what is arm’s length to the entity in low-tax jurisdiction. As enterprises become globalized, and capitals increasingly morphed into intangibles (Haskel and Westlake, 2017) this problem has grown in size and complexity (Borkowski and Gaffney, 2012).

The arm’s length principle purports that if conditions are made or imposed between affiliated enterprises that differ from those that would be made between independent enterprises negotiating at arm’s length, that difference must be included in the profits of that enterprise and taxed accordingly. Further, Para 1.40 of OECD Transfer Pricing Guideline 2017 states that “All methods that apply the arm’s length principle can be tied to the concept that independent enterprises consider the options realistically available to them.”

“Options Realistically Available”

“Options realistically available” implies that independent, economically-rational enterprises would strive for Pareto optimality, i.e. only enter into a transaction if it is not expected to make them worse off than their next best option. They will only enter into the transaction if they see no alternative that offers a clearly more attractive opportunity to meet their commercial objectives.

One example of the adoption of options realistically available in domestic regulation is the US Treasury Regulation Section 482, which cites “alternatives available” to the taxpayer in determining whether the terms of the controlled transaction would be acceptable to an uncontrolled taxpayer faced with the same alternatives and operating under comparable circumstances. The changes to Section 482 by the Tax Cuts and Jobs Act (TCJA) 2017 also stipulates “realistic alternatives” as a basis of valuation of intangible property transfer.

In the Indonesian tax regulation, the Minister of Finance Regulation no. 22/PMK.03/2020 concerning Advance Pricing Agreement also contains similar clause in its transfer pricing provision. Article 14 of PMK-22/2020 considers transactions involving service, use/the rights to use intangibles, cost of debt, transfer of property, business restructuring, and cost contribution arrangement (CCA) to be special transactions that warrant a preliminary step. For the transaction involving transfer of property and business restructuring, the regulation mandates that taxpayer must demonstrate that such transaction is “the best option out of other available options”. This clause also reiterates the ex-ante application of arm’s length principle in the Indonesian transfer pricing regime.

Parekh (2015) and Amici (2020) proposed various approach to identify, value, and determining options realistically available, inter alia:

• Capital budgeting, using internal calculation to value a project based on its internal rate of return or its net present value or its payback periods, or a combination of these methods ;
• Opportunity cost, using explanation of why the other options are not realistically available, e.g. due to being incompatible with the business of taxpayer as such, commercially unattractive, not available at the time of transaction, not acceptable by the other party, or not in accordance with the regulation
• Best alternative to a negotiated agreement (BATNA), using reservation point and alternatives if negotiation fails
• Risk simulation, using the expected value of the investment (weighted average sum of all the probabilities of the possible outcomes of an investment) and the variance (deviation of possible outcomes). Other techniques such as Monte Carlo simulation could also be used.
• Bargaining power theory and game theory, using hypothetical negotiation that can be simulated based on the bargaining power, the contributions made to the marginal benefit of the transaction, and the strategy of the other party to the negotiation
In the context of bargaining power theory and game theory, Sattertwaithe (2019) highlights the similarity of “bargaining problem” (where two players negotiating over the apportionment of something desirable) to the provision of “options realistically available”. Applying John Nash’s bargaining theory, Sattertwaithe propose that seeking a Nash equilibrium strategy will maximize the utility for the player regardless of what strategy the other player adopts. This Pareto optimality will imply the most efficient allocation of utility relative to the bidding parties’ respective alternatives to entering into the transaction – hence, an arm’s length condition.

Probably owing to its mechanistic and mathematical foundation, Nash’s theory rarely enters the legal world. Even its admissibility in patent infringement case is somewhat inconsistent and vary from judge to judge, depending on its ties to the facts of the case (e.g. Compare Robocast Inc. v. Microsoft Corp. (2014), Gen-Probe Inc. v. Becton Dickinson & Co. (2012)).

Economists, on the other hand, have noted the applicability of Nash’s bargaining in transfer pricing, for instance in interdivision negotiated price (Clempner and Poznyak, 2017) or vertically integrated supply-chain (Rosenthal, 2008). There have also been studies of game theory’s application for specific taxation purpose such as in applying profit-split method (Pogorelova, 2015; Voegele, Gonnet, and Gottschling, 2008) or a BEPS-related reporting game (Dorey, 2015). Indeed, Sattertwaithe noted at least 6 elements of similarity between Nash bargaining problem with transfer pricing in general: 1) players and subject of bargaining, 2) alternatives and rationality, 3) surplus utility maximization, 4) zero-sum competing interest, 5) utility value, and 6) allocation.

Sattertwaithe further notes that the profit-split method is readily applied to intangible property primarily because it offers a solution when each party provides unique and valuable contributions but comparable data are lacking, and that an allocation of value (e.g. residual profit) should be calculated anyway under profit-split. For a tax authority’s perspective, profit-split also overcomes the deficiency of one-sided methods where they automatically assume that the residual profit earned by the “untested party” as arm’s length, even though it is literally untested (Wells and Lowell, 2014). This opens up an avenue for abuse. In the case of intangibles, an MNE group can easily designate its intellectual property holder in low-tax jurisdiction, reducing its overall tax liability on royalty income and shielding it from adjustment.

We thus borrow some parts of profit-split into our calculation.

Nash Bargaining

Nash solution for bargaining problem is:

$\max x_{P}^{S}=(U(x_P)-V_P)(U(x_S)-V_S)$

where U denotes utility; subscript P denotes Player P; subscript S denotes Player S; and V denotes fall-back utility, which would be gained if the player chose the second-best option to bargaining that is realistically available to her. X is possible bargaining outcome, for which if U(X) > V implies a surplus utility.

This section is again motivated by Sattertwaithe’s paper. However, we differ in the sense that we use a tax administration’s perspective in doing arm’s length outcome testing (ex post) approach when doing the audit. We thus do not assume the existence of third party firm to value the (ex ante) price of intangible. At the very least, this could serve as a sanity check in intangible transfer pricing analysis. Para. 6.113 of OECD TPG 2017 is relevant here, as transferor would not be expected to accept a price for the transfer of either all or part of its rights in an intangible that is less advantageous to the transferor than its other realistically available options which includes making no transfer at all. In a similar vein, Para. 6.79 of OECD TPG 2017 noted that “Compensation based on a reimbursement of costs plus a modest mark-up will not reflect the anticipated value of, and the arm’s length price for the contribution of the research team in all cases.”

The Setup

Assume a parent company P and subsidiary S, where – besides doing its routine functions – also conducts R&D activity to enhance P’s intangibles. (Despite its “contract research” features, Para. 7.41 OECD TPG 2017 note that the consideration of options realistically available may also prove useful in this situation.) Assume that the R&D results in a material, identifiable structural advantage in the market (thus valuable). This synergistic benefit increases the combined profit of P and S, either via increased sales and/or reduction of costs.

Following the residual approach for profit split, we first calculate the residual income of P and S after deducting the remunerations for their respective routine functions, hereinafter denoted as Π, such that:

$\Pi=(Y_P-C_P-r_P)+(Y_S-C_S-r_S)$

where rP and rS denote the routine function of P and S, respectively. Suppose S is remunerated for its R&D function using a cost-plus method, where some mark-up µ is added to R&D costs borne by S (denoted Cr). The Pareto-optimum fall-back positions of P and S are thus
$V_P=\frac{(1+\mu)C_r}{\Pi}$

and
$V_S=\frac{C_r}{\Pi}$

where there are additional profits of (1 + µ) Cr for P for not having to remunerate S; and Cr for S for not having to bear the R&D costs. The Nash bargaining problem is thus used to solve:

$\max x_{P}^{S}=(X-V_P)-((1-X)-V_S)$

which is done by first taking the derivative with respect to X,

$\frac{d}{dx}(X-(\frac{(1+\mu)C_r}{\Pi})((1-X)-\frac{C_R}{\Pi})$

then setting it to zero. Resulting in:

$2X=1-\frac{C_r}{\Pi}+\frac{(1+\mu)C_r}{\Pi}$

which then could be solved for X, the proportion of Π for P; and (1 – X), which is the proportion of Π for S.

The usage of Nash bargaining as “splitting factor” indeed differs from the usual asset/capital/cost-based allocation keys. We note that while R&D expense may be suitable for manufacturers, it may be insufficient given that: 1) it is only independently born by S, and 2) it may not be a reliable measure of the relative value of the transaction, taking into account options realistically available.

Illustration:

P is the parent of S, a contract manufacturing companies which sources raw material from third party suppliers. S manufactures, and subsequently sells the finished goods to P. P then sells the product to third party customers. S’s cost of goods sold and P’s sales are of independent transactions.

Assume that 100% of P’s inventory comes from S, and 100% of S’s sales are sold to P. S also conducts research for improving the products. P will remunerate S by S’s costs of doing research with a mark-up of 10%. P will also remunerate S for its contract manufacturing function with a full-cost mark-up of 10%.

Suppose the cost of research is 2, thus remuneration from P to S related to research is 2.2. The combined profit of P and S, after remunerating their respective routine functions is 16. Suppose it is sufficiently established that there is simply no justification from the taxpayer as for why the residual profit should all be automatically attributed to P, in the absence of their internal arrangement. (For example, P neither performs nor controls the research.) Hence the relative bargaining position is:

$2X=1-\frac{2}{16}+\frac{2.2}{16},X=\frac{81}{160}$

Therefore, out of profit of 16, 8.1 could be attributable to P, while 7.9 could be attributable to S. (This is almost similar to if the taxpayer use profit-split with 50:50 splitting factor). Nevertheless, this shows that S is inadequately remunerated since – viewed in totality – it is doing functions beyond mere contract manufacturer-contract research service provider.

A Partial Contribution Approach

The above-mentioned residual approach necessitates tax administration to calculate the routine function of P and S. We note that this does not take into account a marketing function done by P that may be similarly valuable. One obvious approach is to take into account marketing expense in calculating the relative bargaining position, which results indeed in higher proportion for P. Alternatively, we may calculate both parties’ contributions simultaneously.

Consider that the P’s sales is affected by S’s contributions in the form of R&D, or further development or enhancement of manufacturing know-how (which is accounted in S’s cost of goods sold or costs of employee). These contribution enables P to either increases price or maintains a desired profit margin given a determined price (via reduced costs). We thus propose – combining P’s effort to market the products – that a log-log function of:

$\ln(Y_P)=\alpha+\beta_1\ln(C_r)+\beta_2\ln(C_m)+\epsilon$

where Cm is P’s marketing expense and ε is error term, may be appropriate. The model could of course be expanded to include other costs, or to use profit in lieu of sales. The error term, which may reflect the residual, “non-routine” profit, may also be employed. (These alternatives warrant further explorations, which is the limitation of this post.)

Exploiting approximation that (1 + x)a ≈ 1 + ax for small a, then % Δ YP ≈ β1% Δ Cr, i.e. a 1% change in S’s R&D cost result in β1% change in P’s sales. Π is then obtained by approximation of β1% * YP, and the Nash bargaining as mentioned above will follow.

Second alternative, considering that not every R&D could be expected to always increase sales – especially if the research is “blue-sky” – we may instead be interested in R&D expense variability in relation to the variability of sales. Using variability instead of correlational direction may also alleviate the drawback that using historical data would arguably constitute a hindsight.

Following Shorrocks (1982), the proportional contribution of factor Cr to the decomposition of the variance of YP could be computed as:

$s_{C_r}^{*}=\frac {Cov(C_r,Y_P)}{\sigma^2(Y_P)}$

where cov(Cr, YP) is the joint variability of Cr and YP, and σ­2(YP) is the variance of YP. While this is not intuitively easy to associate, via Scherrer (1984), coefficient of determination R­2 could be computed as:

$R^2=\sum\limits_{j=1}^{k}a_jr_{xy_j}$

where aj is standardized regression coefficient of j-th explanatory variable, and ry,xj is the Pearson correlation coefficient of y and xj. Assuming we regress Cr and other variables mentioned above to YP, then the contribution of Cr to the variance of YP equals to:

$a_{C_r}r_{{Y_P}{C_r}}=a_{C_r}\frac {Cov(C_r,Y_P)}{\sigma^2(Y_P)}$
which sums to unity.

It should be noted that this function is not monotonic. A more exact approach would be to use Shapley-Owen decomposition. But for k parameters this requires k! combinations, hence 2k possible models to be calculated, and this is computationally expensive.

Illustration:

Assume P as the parent of a contract manufacturing subsidiary S similar to the illustration above. The table below gives the sales of P, R&D expense of S, and marketing expense of P, all in natural log, for the last 10 fiscal years:

The log-log regression gives result:
So even though the coefficient of R&D expense is higher than marketing, given that marketing expense is more statistically significant than R&D in explaining P’s sales, the contribution of R&D is lower than marketing expense. In this case, 33% and 62% of P’s variability of sales may be approximately contributed by S’s R&D activity and P's own marketing activity, respectively (5% of sales variability is due to other factors). Subsequently, the current pricing policy could be tested against 35:65 split  (0.33:0.95 = 0.35 and 0.62:0.95 = 0.65) similar to the Nash bargaining problem as outlined above, and see if their ex-ante testing is appropriate.

Conclusion

Nash bargaining could be used in ex-post outcome testing whether a transaction involving research resulting in valuable intangibles would have been entered into by independent parties, taking into account options realistically available. It should be noted that this analysis is presumptive, and should not be taken as prima facie proof of transfer mispricing. OECD cautions that better-than-expected result may not be reasonably foreseeable ex-ante by taxpayers. Indeed, assuming taxpayers do not provide adequate justification of ex-ante pricing, insofar as the remuneration for transferor and the actual outcome do not deviate by 20% from the projection, Para. 6.194 OECD TPG 2017 discourages the usage of ex-post facto rationalization, as it is considered as hindsight.

Reference

Minister of Finance Regulation no. PMK-22/PMK.03/2020

OECD Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations, 2017 Edition

Amici, D. (2020) In-Depth Analysis of the Concept of Options Realistically Available in Transfer Pricing. 27 Intl. Transfer Pricing J. 2, pp. 112-122

Borkowski, S. and M. A. Gaffney (2012) “Uncertainty and Transfer Pricing:(Im)Perfect Together?” J. Int’l Acct. Auditing & Taxation Vol. 32 (2012). IBFD Journal Articles & Papers

Brealey, R. A., S.C. Myers, and F. Allen. (2019) Principles of Corporate Finance ch. 5 (13th ed.). McGraw-Hill

Bullen, A. (2011) Arm’s Length Transaction Structures: Recognizing and Restructuring Controlled Transactions in Transfer Pricing. IBFD

Clempner, J. B. and A. S. Poznyak (2017) “Negotiating Transfer Pricing Using the Nash Bargaining Solution,” 27 Int’l J. Applied Mathematics & Comput. Sci. 853

Dorey, M. (2015) “To Audit or Not to Audit: Applying Game Theory to a Post-BEPS World,” 24 Transfer Pricing Rep. 404 (Aug. 6, 2015)

Haskel, J. and S. Westlake. (2017) Capitalism Without Capital: The Rise of Intangible Economy. Princeton University Press

Hafkenscheid, R. P. F. M. (2011) De bepaling van een zakelijke risicoallocatie in een business restructuring, Weekblad voor Fiscaal Recht 2011/660 (12 May 2011), at 660-668

Fisher, R. & W. Ury (1981) Getting to Yes: Negotiating Without Giving In. Penguin Group

Parekh, S. (2015) The Concept of “Options Realistically Available” under the OECD Transfer Pricing Guidelines. 22 Intl. Transfer Pricing J. 5, pp. 297-307 (2015). IBFD Journal Articles & Papers

Pogorelova, L. (2015) “Transfer-Pricing and Game Theory,” 43 Intertax 395

Rosenthal, E. C. (2008) “A Game-Theoretic Approach to Transfer Pricing in a Vertically Integrated Supply Chain,” 115 Int’l J. Production Econ. Oct. 2008

Sattertwaithe, B. M. (2019) Nash Bargaining Theory and Intangible Property Transfer Pricing. Tax Notes Federal, September 30 2019 Issue

Scherrer, B. (1984) Biostatistique. Quebec, Canada: Gaetan Morin

Shorrocks, A. F. (1982) “Inequality Decomposition by Factor Components”. Econometrica. Vol. 50 No. 1 (Jan. 1982)

Voegele A., S. Gonnet, and B. Gottschling. (2008) “Transfer Prices Determined by Game Theory,” Tax Planning Int’l Transfer Pricing

Wells, B and C. Lowell (2014) “Tax Base Erosion: Reformation of Section 482’s Arm’s Length Standard”. Florida Tax Review. Vol. 15 no. 10

Monday, September 28, 2020

An Ode to the Little Giant

There is not so much changes with you, with us.

We’re still the same, weather-beaten, rain-sodden us. We’re red dwarfs awaiting supernova implosion, burning through our last hydrogen atoms. We sigh into the onslaught by identical days. Far too exhausted for tomorrow morning stroll, just enough for waking up. Bound by tyranny of boredom and the confinement forced into us from some foreign, voracious, microscopic agents of chaos.

We are tired and sullen, the skeletal, spectral us.

As we get older, we drag our feet more. We started losing sight. A bleary vision. Stanzas of worry etched on our foreheads. The wound from ramshackle dreams latched onto our shoulders. Fragmented plans, shredded like the Oxyrhyncian trash heap.

The sun goes up and down like the most despicable clockwork. Older you be and you doubt more. (We all do, holding to that frail strand of hope that made thinner by day.)

Yet while I started giving up, you were always a snowdrop flowering at the first day of spring. Frigid wind above you, the small giant, adorable in your fortitude. Dawn gives you rain and you unfurl your petals, forever my raincoat for the storm.

Amidst all this cacophony you endure, you always manage to find some notes of symphony.

Today is the day when you get one year older, and us a year older. These days are dark, and there is no candle on the cake (there is no cake at all).

So come, let’s try be human once more. This I can offer you. Let’s sit together on the floor, imagining foreign places. Our banters be like fireworks. Let’s stay away from the edges of our bed. Let’s always fight for blanket, our nightly annexation game. Let’s exchange stories before dreams, before the day is done. Let’s eat badly, raw fish and third-rate rice wine and all.

Let’s cry, let’s curse, let’s... grow up together.

Let’s embrace senescence like an old friend. Let’s create a makeshift compass and a dim torch that guide us to each other.

For the day is dark, and there is no candle.

But you, you are always the only light that matter.

Tuesday, July 14, 2020

Through the CbCR Looking-Glass, and the Profit Shifting Found There

Last week, the OECD published for the first time its aggregated and anonymised Country-by-country Reporting (CbCR) data. Despite its limitation, CbCR is quite a significant move in the fight against corporate tax abuse, as a part of Action Plan on Base Erosion and Profit Shifting (BEPS).

Information asymmetry has always been the bane of tax authority, especially in dealing with multinational enterprise. Prior to transfer pricing documentation regime, tax authority relies much on domestic tax return and financial statements in order to understand MNE's structure.

Those aforementioned sources, understandably, do not paint a complete picture of the functions performed, assets used, and risks assumed by each of MNE's constituent entities. The closest analogy that we often use is "blind men and elephant". A group of blind men heard that a strange animal, called an elephant, had been brought to the town, but none of them were aware of its shape and form. The first person, who touched the trunk, said, "This animal is like a snake". Another one whose hand reached its ear, it seemed like a kind of fan. Another person, whose hand was upon its leg, said that the elephant is like a tree-trunk, or a pillar. They are not entirely incorrect, of course, but they are missing the big picture.

CbCR attempts to bridge this asymmetry by requiring MNE to report its economic activities in each jurisdiction it operates. CbCR's first main forms, CbC-1, lists MNE's financial information (revenues, profit/loss, taxes, assets, capital, and employees) in per country basis. The second form, CbC-2, lists MNE's entities alongside its functions (e.g. manufactures, procurement, distribution/marketing, R&D, service, financing, etc.)

CbC-1 and CbC-2

Nevertheless, this trove of information comes with a trade off.

Firstly, CbCR's functions are limited to transfer pricing and other BEPS-related risks assessment, as well as for statistical purposes. It cannot be used as conclusive proof that there exists incorrect transfer pricing. It also cannot be used for global formulary apportionment purpose. In short, it cannot be used directly for transfer pricing correction in tax audit, but only to initiate further inquiries.

Lastly, CbCR is not public, rendering its use for public accountability limited. (Some MNEs voluntarily made their CbCR public, however, but this is not a requirement.) For the purpose of measuring BEPS risks, OECD collects aggregated and anonymised CbCR from many jurisdictions. By aggregating and anonymising the CbCRs, it make them difficult to trace back to certain MNEs, which once again serve CbCR's private characteristic. The part of OECD's report on Indonesia serves as the basis for our profit shifting measure here.

Global Profit Shifting and Revenue Loss Measures

Profit shifting to country, $i$, denoted $S_i$, is measured from the profit booked $\pi_i$, and theoretical profit $P_i$, i.e.

$S_i = \pi_i - P_i$

The profit booked is taken from profit (loss) before tax in CbC-1.

We employ 2 estimation strategies. First, we follow Tax Justice Network (TJN)'s formula in measuring share of economic activity.

$P_i$ is calculated by multiplying the total profits by the share of economic activity. The share of economic activity is calculated on the basis of unrelated party sales, $R_i$, and number of employees, $E_i$, i.e.

$P_i = \sum\limits_{i}{\pi_i}\cdot (\frac{\frac{1}{2}\cdot R_i}{\Sigma_i R_i}+\frac{\frac{1}{2} \cdot E_i}{\Sigma_i E_i})$

We also employ the so-called "Massachusetts formula", where instead of using half-revenue half-employee as weights, we use weighting from unrelated party revenue, employees, and tangible assets $A_i$, each with a third of weight.

$P_i = \sum\limits_{i}{\pi_i}\cdot (\frac{\frac{1}{3}\cdot R_i}{\Sigma_i R_i}+\frac{\frac{1}{3}\cdot E_i}{\Sigma_i E_i}+\frac{\frac{1}{3} \cdot A_i}{\Sigma_i A_i})$
Massachusetts formula is used in US to apportion income or cost of a corporation among US states, by placing equal weights on sales, payrolls, and assets.

TJN’s choice of two weights – revenues and employees only, but not tangible assets – follows their rationale that tangible assets are biased by profit shifting. On their analysis, tangible assets in Luxembourg are greater than Germany, France, and African countries combined, despite the former is not renowned for their real sector economy compared to the latter.

Arguably, the same reasoning could be used to suspect that revenues might be biased. For example, if the MNEs employ third party to act as distribution hub in Singapore to perform sales in Asia-Pacific region and consequently books massive unrelated party revenues in Singapore. We thus include Massachusetts formula that incorporate tangible assets to still account for MNEs that have real economic activities that have low level of labour but significant capital investment in properties, plants, or equipments.

To obtain estimates of profit shifting, $S^c$is then defined as the sum of positive values of $S_i$ for countries where the effective tax rate is below 15%, such that

$S^c = \sum\limits_{i}{S'^c_i} - P_i$, where
$S'^c_i = \begin{cases} S^c_i & \text{if } S^c_i > 0 \text{ and } ETR^c_i < 15\%\\ 0 & \text{otherwise } \end{cases}$ TJN argues that this correction allows us to remove some resource-rich countries with large profits and high tax rates, therefore obtaining a more conservative estimate of profit shifting.

The choice of break-out threshold in 15% follows TJN's methodology, based on their finding that every country with an effective tax rate above 15% loses tax revenue through profit shifting.

The global tax revenue loss, $TRL^c$ is calculated by multiplying shifted profit $S^c$ with the average effective tax rate in countries that is higher than 15%, weighted by the measures of real economic activities, i.e.

$TRL^c = S^c \cdot \widehat{ETR^c}$

Effective tax rate (ETR) is calculated as follows:

$\widehat{ETR^c}=\sum\limits_{i\in{\{N-c\colon ETR_i \geq 15\%\}}}ETR_i(\frac{\frac{1}{2}\cdot R_i}{\Sigma_i R_i}+\frac{\frac{1}{2} \cdot E_i}{\Sigma_i E_i})$ , for TJN weight, and

$\widehat{ETR^c}=\sum\limits_{i\in{\{N-c\colon ETR_i \geq 15\%\}}}ETR_i(\frac{\frac{1}{3}\cdot R_i}{\Sigma_i R_i}+\frac{\frac{1}{3}\cdot R_i}{\Sigma_i R_i}+\frac{\frac{1}{3} \cdot A_i}{\Sigma_i A_i})$ , for Massachusetts formula weight.

However, their definition of effective tax rate is taxes paid divided by profit before tax, which is closer to the definition of cash ETR in tax avoidance literature. We supplement this analysis with our measure of effective tax rate which is taxes accrued divided by profit before tax. From this point on, CETR will be used to denote TJN's measure, and ETR to denote "traditional" ETR.

Results

We estimate around 1-3 trillion IDR of shifting out, depending on the measure used.

To these following countries.

Plotting the distribution into box plot, there are significant variants if cash ETR is used.

How much revenue is lost? Depending on the measure of ETR and weighting, this translates to 200-600 billion IDR of global tax revenue loss.

NOTE: Due to its aggregated nature, this is not the tax revenue lost of Indonesia only, but collective tax revenue lost for some countries where Indonesian MNEs operate.

Conclusion and Further Remarks

Based on the aggregated and anonymised CbCR, there are indications that MNEs engage in global profit shifting, resulting in significant revenue loss.

Aggregated and anonymised CbC Report released by OECD is a neat tool to monitor global profit shifting lost. This type of analysis could even be conducted on MNE level, and with the indicators outlined in the OECD Handbook of Effective Tax Risk Assessment, this may be used to complement current compliance risk management analysis. Incorporating the information of CbC-2 as weights and employing different ETR threshold (e.g. using MNE's own group ETR) could also improve future analysis. It may be even useful for the purpose of Pillar 2, Global Anti-Base Erosion proposal to address digital economy and other economy of scale without mass.

Reference

INFINITE (2020), Comments on Public Consultation Document: Review of Country-by-Country Reporting (BEPS Action 13) https://www.infinite-tax.org/2020/05/19/comments-on-public-consultation-document-review-of-country-by-country-reporting-beps-action-13/ (Accessed 13 July 2020)

OECD (2020), New corporate tax statistics provide fresh insights into the activities of multinational enterprises http://www.oecd.org/tax/new-corporate-tax-statistics-provide-fresh-insights-into-the-activities-of-multinational-enterprises.htm (Accessed 13 July 2020)

OECD (2017), Country-by-Country Reporting: Handbook on Effective Tax Risk Assessment, OECD, Paris.
www.oecd.org/tax/beps/country-by-country-reporting-handbook-on-effective-tax-risk-assessment.pdf (Accessed 13 July 2020)

Tax Justice Network (2020), Watershed data indicates more than a trillion dollars of corporate profit smuggled into tax havens, https://www.taxjustice.net/2020/07/08/watershed-data-indicates-more-than-a-trillion-dollars-of-corporate-profit-smuggled-into-tax-havens/ (Accessed 13 July 2020)

Tax Justice Network (2020), Methodology: Analysis of OECD country by country reporting data, https://www.taxjustice.net/wp-content/uploads/2020/07/Methodology-Analysis-of-OECD-country-by-country-reporting-data-July-2020.pdf (Accessed 13 July 2020)

Tuesday, October 15, 2019

On the Comparability of Pan-Asian to Local Companies

Transfer pricing is not necessarily done for the purpose of avoiding tax, but it is one of the most common methods to do so.

In essence, affiliated companies can set up intragroup pricing that differs from market mechanism. Suppose the cost of product P is 80, and the price of product P in the market is 100. In independent setting, selling P will result in 20 of profit.

Suppose A is selling P to B (A’s affiliate) for a transfer price of 85. B then sells it to an independent party for 100. A will pocket a profit of 5, and B 15. If the transfer price is 99, A will get 19 and B only 1; and so on.

Unlike independent parties’ behavior that each party will maximize its own interest/profit, affiliated companies can set price that may “hurt” one of them. In the example above, setting transfer price to 99 may indeed hurt B, for B only gets a meager profit of 1. The price of 99 may be too expensive for an independent party that it may not have entered such transaction.

[The above method uses price as a point of comparison. If there is no comparable transaction between independent party, for instance if the price of products comparable to product P is not available, we can compare the profit level instead. The party that conducts less complex functions – the one that only conducts routine production or distribution, bears risks that are not economically significant, and possess no unique and valuable intangibles – is designated as "tested party". The choice of simpler company as tested party is due to the consideration that the more complex a company is, the less likely it has comparable companies similar to it. If both parties assumes economically significant risks or contributes unique and valuable intangibles, then the profit is split to both parties.]

If A and B are in the same country and taxed with same rate, arguably, such arrangement does not pose a significant base erosion problem for tax authority. If tax authority failed to do a transfer pricing adjustment to A, they could do it to B with similar net gain in tax revenue.

But if A and B are in different countries with different tax rates/facilities, then the 20 profit can be arbitrarily shifted in whichever company that will pay smaller amount of tax. Or no tax at all. This is probably the reason of US, Australia, or India to deny transfer pricing benefit only in cross-border transactions. But other country, like Indonesia, allows tax authority to make a transfer pricing adjustment even when the affiliated parties are in Indonesia. There may be the case that two Indonesian companies may be affected by different tax regimes (for instance if one of them has a reduction of tax rate, tax holiday, loss carryforward, final income tax, etc.)

Tax authorities of the world thus have incentive to adjust transfer pricing done by their taxpayers. They will look up at transaction done by independent parties according to sound business practice, see how much the difference in price or profit is observed, attribute the difference, and then tax accordingly.

Accurately delineating the transaction, and then comparing affiliated transaction with independent transaction – so-called comparability analysis – is therefore the “heart” of transfer pricing adjustment. A comparable independent transaction (often called “comparable” for brevity) can be found within the company if one of the companies under scrutiny also transacts with independent party. This is called internal comparable, and generally considered to be more reliable than external comparable.

To find the appropriate comparable, there are 5 comparability factors that must be considered:
a. contractual terms;
b. functions, assets, risks;
c. characteristics of goods/services;
d. economic circumstances; and

A reliable comparable must be similar. Or, at least must not be materially different from the transaction under scrutiny so that reliable adjustment – if necessary – can be made.

Given that information in the market is imperfect – barring commodities or internal comparable, for instance – sometimes the comparables found are “inexact”. This most often happens with external comparables, such as those found via commercial database. Adjusting transfer price in Indonesian taxpayer may sometimes invoke comparables found in other countries, due to the lack of Indonesian comparables. Now recall that economic circumstances is one of comparability factor that must be considered. How to take this into account?

One of most common practice is disregarding such difference insofar as the comparables selected come from similar region. This assumes that the economic circumstances in Indonesia is more similar with other countries in Far East and Central Asia compared to, say, European Union. Of course, disagreement may arise whether this assumption holds.

The issue is therefore whether Far East and Central Asian comparables can be used, or whether there is significant difference with Indonesian comparables that, strictly speaking, only Indonesian comparables should be used in adjusting transfer price in Indonesia. There is also the issue whether some adjustments that may be implemented to improve reliability, such as working capital adjustment or country-risk adjustment, are warranted. (The latter adjustment is especially of our interest here.)

The case for European comparables has been tested by Meenan et al. (2004) and then updated by Peeters et al. (2016) as well as by Platform for Collaboration on Tax (2017). They all found that there are no material difference that comparables from other European countries can be used without adjustment.

What about Indonesia vis-a-vis Far East and Central Asian comparables? This short post is a rudimentary attempt to answer.

The Set-up

The procedure is similar to the works cited above: select companies with NACE v. 2 industry codes following Peeters, et al. (2016) and then group them into 4 industries (automotive manufacturing, electronics manufacturing, chemical distribution, and electronics distribution). We subsequently extract necessary financial information (Sales, Costs of Goods Sold/COGS, Other Operating Expenses/OPEX, Operating P/L, Tangible Fixed Assets, Stocks, Debtors, Cash and Cash Equivalents, and Total Assets) to compute some profitability ratios, which are:
a. Gross Profit Margin (GPM): (Sales – COGS) / Sales
b. Cost Plus Markup (CPM): (Sales – COGS) / COGS
c. Operating Profit Margin (OPM): Operating P/L / Sales
d. Return on Assets (ROA): Operating P/L / (Tangible Fixed Assets + Stocks + Debtors + Cash and Cash Equivalents), and
e. Full Costs Mark-up (FCMU): Operating P/L / (COGS + OPEX)

The periods used are 2014 – 2018 (5 years) using Orbis data. All are computed using five-year weighted average to smooth the earning profile that may be affected by business cycle.

Among the indicia above, only OPM and ROA are computed in the original studies by Meenan et al. (2004) and Peeters, et al. (2016). These indicators are on net profit level, which is only useful for transactional method such as Transactional Net Margin Method (TNMM). I expand the choice to include GPM and CPM to account for traditional methods such as Cost-Plus and Resale Price methods. I also include FCMU to account for the manufacturing remuneration.

We, however, relaxed some of the screening criteria because in some industries there is no Indonesian company that fits in. We, for example, do not limit based on arbitrary threshold of sales. We also do not limit the samples to only include companies whose operating profit margin is around -5% to 15% and ROA around -10% to 20% (by assuming the five-year weighted average and the usage of interquartile range will smooth out the outliers). On the other hand, we implement stricter criteria in other place, such as excluding companies that do not have financial data for at least 3 years.

After we compute the indicia, we divide the Indonesian samples and Asian samples (which include Indonesia) in each industry group by median. Then, we apply Chi-square test to each lower quartile and upper quartile.

The aggregated cumulative distribution function of comparables' profitability is denoted by 𝐹(𝑟), which gives the share of comparables with a profitability ratio of our choice smaller than or equal to r. Our interest is interquartile range, and subsequently 𝐹(𝑟) for each industry is measured using the critical values defining the 1st and 3rd quartile of the cumulative distribution as 𝐹(𝑟∗) = 0.25 and 𝐹(𝑟∗∗) = 0.75.

Given the values 𝑟∗ and 𝑟∗∗, the number of firms in Indonesia with profitability ratios below and above this benchmark profitability were recorded. Formally, the analysis defines 𝑜𝑖1=𝐹𝑖(𝑟∗)𝑁𝑖, 𝑜𝑖2= [𝐹𝑖(𝑟∗∗) − 𝐹𝑖(𝑟∗)]𝑁𝑖, and 𝑜𝑖3= [1 − 𝐹𝑖(𝑟∗∗)]𝑁𝑖

where i denote Indonesian specific comparable. Under the null-hypothesis, 25% of the Indonesian-specific comparables in both 𝑜𝑖1and 𝑜𝑖3, and 50% in the middle group 𝑜𝑖2 are expected.

Accordingly, a joint test statistic defines the variable

where 𝑒𝑗𝑛 denotes the expected number of firms in each country-specific group.
The chi-square statistic 𝑋2 increases as 𝑒𝑗𝑛 - 𝑜𝑗𝑛 increases, which indicates that the observed distribution of comparables is significantly different from expected.

The Result and Conclusion

No significance difference exists between Indonesian comparables' lower and upper quartiles and Far East and Central Asian comparables' lower and upper quartiles in the industries tested. We repeat the calculation with Fischer exact test which is more conservative than chi-squared test in small samples (even in sample size less than 5) and the results still hold. Changing the denominator of ROA with Total Assets also does not change the result.

What does this mean? First, this does not mean that comparables from outside of Indonesia are as reliable as Indonesian comparables. This only means that the profitability level of Indonesian and other Far East and Central Asian comparables are drawn from similar distribution function, in this case a Chi-squared. Again, a thorough comparability analysis – with consideration to all five comparability factors – is more paramount than a simple similar-country-therefore-more-reliable approach.

So, does this mean that country risk adjustment is unnecessary?

In our opinion, any kind of adjustment to increase reliability must ultimately be proven to be justifiable. If, for instance, taxpayer proposes Working Capital Adjustment, she must first demonstrate that the comparables she obtains do not have a high degree of reliability. But then again, why does she knowingly choose less reliable comparables to begin with? If, for example, the level of inventory-to-total assets of the tested party is significantly different from the comparable companies the taxpayer choose, and that difference may significantly affect price/profit, then it begs the question of why the taxpayer does not apply a quantitative screening using inventory-to-total assets as a criterion.

The similar reasoning is used in multiple-year data vs. single year data. OECD Transfer Pricing Guideline 2017 para. 3.75 does not put emphasis on the usage of multiple year data as a default or systematic requirement. But only when the usage of multiple year data adds value to the analysis (for example if there is an effect from business life cycle) then it may be justifiable.

Ultimately, referring to OECD Transfer Pricing Guideline 2017 para. 3.59, at least in the industries referred here there are no substantial deviation in the interquartile range that indicates unreliability due to country-level difference. Therefore, unadjusted comparables may be used if other comparability factors are indeed similar enough that the comparables are suitable for inclusion.

Reference:

Meenan, P., Dawid, R., and Hülshorst, J. (2004) Is Europe One Market? A Transfer Pricing Economic Analysis of PanEuropean Comparables Sets. European Commission, Brussels Taxud/C1/LDH/WB

Peeters, R., Noben, S., and Laurent, I. (2016) Study on Comparable Data Used or Transfer Pricing in the EU. European Commission, Brussels TAXUD/2014/CC/126 doi: 10.2778/657328

Platform for Collaboration on Tax (2017) A Toolkit for Addressing Difficulties in Accessing Comparables Data for Transfer Pricing Analyses. IMF - OECD - UN - World Bank

OECD (2017) OECD Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations 2017. OECD Publishing, Paris. http://dx.doi.org/10.1787/tpg-2017-en

Saturday, September 28, 2019

One

“ I found the one my heart loves. I held, and I would not let go...” (Songs of Solomon 3:4)

My thought wanders to a time long passed, almost a decade ago. A time when I started to be in love with this woman, Nitha. A time half spent in silent longing, while she had been another’s.

In today’s vernacular, I would’ve been called a “bucin.” A slave of love. Forgetting is so long, to quote Pablo Neruda. And five years of failing to forget was very enslaving to the soul.

There were nights when I think of her, or, to be more precise, the absence of her. Thinking that I was not with her, and to feel that I had lost everything that was beautiful to me.

There were days I masochistically watched Shinkai Makoto’s film, “5 Centimeters per Second.” The protagonist, Takaki Tohno, was a reflection of mine (perhaps also shared by many others). An image cruelly reflected by the black mirror. It was not the lost of love that broke me every time I watched that film. It was the lost of the ability to love again; the lost of the ability to move on.

It made me shudder: how many years should I spend before I am free from this torment?

After all, the heart wants what the heart wants. And the heart wants nothing less.

Nothing is perfect, that is probably true. Perfection, if any, is in the eye of the beholder. And Nitha is close enough to perfection for me.

When Nitha said to me that she broke up with her ex, my immediate reaction was befuddlement, instead of joy. After all, I have spent many a year constructing a temple around the idealized, imaginary persona of her. She was the idol, beyond my horizon. I was thus unafraid of failure and heartbreak, because how can I fail a relationship that does not exist?

But now, having a relationship with her, to be disappointed (or to be a disappointment), and altogether failing the relationship seemed like an imminent possibility.

But I decided to try. I demolished my metaphorical temple, to lay ground for a new reality full of imperfections.

Nitha is a strong-willed woman, often labelled herself as egoist. She knows what she wants, the way she wants it to be. If faced with 10 options, she will comprehensively evaluate the pros and cons of all 10 of them, while I will evaluate three of them and pick the best. She’s sometimes annoyed at me for not being thorough or lacking of initiative. While I am often annoyed at her picky nature –bordering on obssessive – and inefficiency.

For the last 5 years of having this free-form arrangement, we upped the ante. We wanted to see if we could tolerate with each other’s personal quirks, habits, and difference in opinions.

We started a trial we weren’t sure how it would end. We even weren’t sure how it started. Our “anniversary” was arbitrarily chosen at 20 October because I never properly asked her to be my girlfriend, and she never properly replied a “yes”. 20-10 seemed like a cute number and we went with it for years.

We tested ourselves whether we can set aside preferences and desire. We tried giving up individuality and the convenience of being alone.

It is of my great surprise that she tolerates my jittery and erratic decision-making ability. Without admitting it, she has sacrificed her ego for far too often.

Thankfully, we pretty much survived. We then braved ourselves to be tied in engagement, and then planned to marry.

We asked our friends why they decide to marry – especially for reasons beyond mere legality and religiosity. There wasn’t very satisfying answer. To be honest, we had no particular reason to get married.

One pragmatic reason is to satisfy the state apparatus and our families, thus keeping them from intruding what should’ve been our business and our business only. Indeed, had this country been like Sweden (or had there been no unconscionable draft of Criminal Code penalizing cohabitation), we’d just probably choose to live together. Tethered by hearts but remain unattached by the law.

So, why I think I ultimately agree to marry?

“Marriage”, wrote G. W. F. Hegel, “is a contract to transcend the standpoint of all contract”. Unlike regular contract in which both parties retain their abstract freedom, marriage is instead a mutual surrender of abstract freedom and autonomy to a higher organic ethical unity.

Gracy Olmstead expounded that marriage was never meant to be a vehicle for self-fulfillment, convenience, or pleasure. It’s an ethical bond to the other, for the good of the other. Marriage was never meant to satisfy every desire and longing of the heart. On the contrary, marriage provides discomfort. Heck, even preparing for marriage is already a frustrating and tiresome endeavour. Marriage indeed goes against every individualistic instincts.

On its surface, marriage is just a declaration made on the altar, officiated by the representative of God, confirmed by the state, witnessed and recognized by family, and sometimes followed by superfluous celebration. They are the “Big Others” as Slavoj Zizek wrote when he invoked his Lacanian legerdemain.

But marriage should not simply be taken as constraining. When the knot is tied, two hearts are liberated from the transient, fickle and purely subjective aspects of love. It is a liberation through self-restraint, which is rendered substantial by the something other than the self. Marriage is not merely a union of egoists.

These – the reference of marriage to something other than me, to selflessly love someone other than me – make a marriage is at least defensible for me.

And if I have to marry, I am sure as hell wanting to marry someone close enough to my perfection.

So here we are, after all these times, plunging ourselves deeper into the unknown future.

On 28 September 2019, Nitha’s birthday, we officially start our beautiful ride. I’ve made my vow to carry her home if she falls sick. To love her at the best of times and the worst of times. To stay till the world turns to oblivion and time unwinds to apocalypse. Till death do us part.

All these would not be possible without Nitha, who has shouldered an unequal share of burdens, who makes everything more beautiful than I’ll ever do. I solemnly promised I will do the dishes and the laundry.

It is of course unfortunate, however unintentional, that we’re married in one of the bleakest moments in Indonesian democracy. I personally hope the government and its citizens may unite once again in democracy, as Nitha and I unite in holy matrimony.

Please pray for all the good things in the best possible world for us. May the good things and our happiness today also extend to you all. You have our utmost gratitude.

Lastly, wish us luck. We’re sure as hell going to need it.

Friday, August 9, 2019

Viability of Commercial Database to Measure BEPS Risks

There have been various studies and anecdotal evidence showing that multinational enterprises (MNEs) engage in what is termed as base erosion and profit shifting (BEPS) activity. Their multinational nature afford them to utilize mismatches and gaps in domestic tax rules and tax treaties. Among other modus operandi are transfer pricing, avoidance of permanent establishment, thin capitalization, deferring tax via controlled foreign corporation, hybrid entities/instrument mismatch, et cetera.

Measuring the scale of BEPS, quantitatively speaking, proves very challenging. BEPS is complex and has numerous variations of arrangement. Corporations react to regulations, tweaking their schemes to escape taxing ambit. Current audit may not necessarily conform to its previous result. Court verdicts may be inconsistent, especially in civil law country. So BEPS is essentially irreducible to mere numbers or variables.

Nevertheless, measuring BEPS impact is of utmost importance. It gives picture to the scale of taxation abuse, as well as providing tax authorities with tool to gauge a regulation’s efficacy in preventing BEPS. This is why measurement of BEPS is included among 14 other action plans in the 2013 OECD BEPS Action Plan. Specifically, it is designated BEPS Action Plan 11.

But there is the issue of limitation of data. Available data may not be representative. There are also mismatch between real economic effect and BEPS, and between financial (accounting) and fiscal information. There are issues of timing, accessibility, and adequacy of details.

In this regard, tax authority may combine or separately analyze macro-level data and micro-data. Micro-data, in particular data sourced from published financial statements (either from public companies which are legally required to submit financial report, or from commercial database) may supplement tax authorities with information to measure BEPS. Aside from the issues mentioned above, however, financial statements data may be problematic. There is no distinction between related party and independent party transactions, effect of different accounting standards and consolidation, as well as low coverage in developing countries in particular, including Indonesia.

How well this type of data will fare in the light of BEPS Action Plan 11? Using ORBIS database, we gather data from Indonesian public and private companies. We search active companies, with known operating turnover for at least one year among 2009-2018 (this is to prevent much of “junk” data containing the name of company only but nil financial information). We further exclude financial companies (NACE code: K) following numerous past profit shifting studies. We obtain 575 companies for 10 fiscal years of 2009-2018. (n = 5750)

Testing Effective Tax Rate

We opt to test the the propensity of member of an MNE to engage in tax planning (Box 3.A1.3) and manipulation of the location of external debt (Box 3.A1.5). Our variable of interest are therefore effective tax rate and leverage. For this we need the information about profit before tax and income tax to calculate effective tax rate, as well as total equity and total liabilities and debt to calculate leverage. If all four are not available (n.a.) they are excluded them from sample, resulting in n = 4496 in unbalanced panel data.

Out of these, we extract information about their assets, employees, existence of patent and trademarks, and the locations of their global ultimate owner, controlling shareholder, immediate shareholder, headquarter, subsidiaries, and branch. These locations are of particular importance because they are used to determine whether an Indonesian entity is acting as headquarter, and whether it is a part of an MNE group. We also use these location to calculate its average headline statutory corporate income tax rate. For example, if an entity is known to have subsidiary in Singapore (statutory tax rate: 17%) and parent in Hong Kong (16.5%) then – taking into account Indonesia’s statutory tax rate of 25% – its average statutory tax rate will be (25% + 17% + 16.5%)/3 = 19.5%.

The average statutory rate of MNE, plotted against non-MNE is as follow:

On paper, it looks like Indonesian MNE indeed have higher BEPS risk on average. But if we plot their effective tax rate (calculated by dividing profit before tax with income tax as reported on ORBIS), we obtain this:

The results are widely different depending whether weighted average (sum of profit before tax divided by sum of income tax) or simple average is used. In both cases, however, MNEs often produce higher ETR compared to non-MNEs. Our unpaired t-test confirmed that in the case of weighted average, the effective tax rate of MNE is higher than non-MNE (in case of simple average, they are not statistically different). This runs in contrast to our theoretical understanding of BEPS.

Indeed, when we apply regression using equation 3.A1.3 to our data, it produce no statistically significant result. ETR is neither affected by the large size of a company nor its multinationality.

Benchmarking and Bunching the DER

Testing the debt-to-equity ratio, we also run to similar problem.

The purple line is the maximum debt-to-equity ratio (DER) to limit interest deduction as regulated by Minister of Finance Regulation no. PMK-169/2015, which is 4:1. As we can see, different counting method resulting in drastically different measurement. If we use weighted average, it seems that Indonesian entities are, on average, still within the allowed DER. But using simple average will show that in 2016-2017, MNEs are (on average) going over the 4:1 threshold. Not only that, but 2014-2015 saw double-digit level of DER. In both cases, however, t-test shows that the difference between MNE’s DER and non-MNE’s DER are not statitically significant. The result of regression using equation 3.A1.5 further confirms that difference of average statutory tax rate to Indonesian tax rate does not significantly correlate with DER.

Indeed, when evaluate whether Saezian bunching exists post-PMK 169/2015, ORBIS data did not show the evidence of debt-to-equity ratio bunching around 4 (which is the maximum allowed by PMK 169/2015). Plotting the binned frequency before and after PMK-169/2015 shows that (at least according to ORBIS data) taxpayers' behavior remain unchanged in the light of new limitation.

In fact, utilizing "bunch_count" by Chetty, et al. (2011), bunching occurs around DER = 1 instead of 4:

Probability of Being Flagged for Audit

Lastly, for a much immediate application. Suppose we flag Indonesian MNE for audit based on their ETR and DER. We flag an MNE that has lower-than-statutory tax rate ETR for BEPS-related audit. Suppose we also flag an MNE that has higher than 4:1 DER or has negative equity (which is not eligible for interest expense deduction) in the fiscal year 2016-2018 where PMK-169/2015 applies. The probability of an Indonesian MNE flagged for audit based purely from ORBIS database is:

Only 50-60% of Indonesian MNEs will be flagged for audit based on their ETR in our sample. For DER, it is even less, only around 16% will be flagged.

Conclusion

Does that mean Indonesia is safe from BEPS-related risks? Probably quite the contrary. This piece means that BEPS-related risks cannot be captured using third party commercial database alone. It requires a more holistic compendium of data from tax return, other institutions, agencies, association, and other parties, as well as data from automatic exchange of information in the form of financial accounts data and Country-by-Country Report.

References

OECD (2015a) Measuring and Monitoring BEPS, Action 11 - 2015 Final Report. OECD/G20 Base Erosion and Profit Shifting Project, OECD Publishing, Paris.

OECD (2015b) Transfer Pricing Documentation and Country-by-Country Reporting, Action 13 - 2015 Final Report. OECD/G20 Base Erosion and Profit Shifting Project, OECD Publishing, Paris.

Chetty, R., Friedman, J., Olsen, T., Pistaferri, L. (2011). “Adjustment Costs, Firm Responses, and Micro vs. Macro Labor Supply Elasticities: Evidence from Danish Tax Records”, Quarterly Journal of Economics, 126(2).

Saez, E. (2010). “Do taxpayers bunch at kink points?" American Economic Journal: Economic Policy vol. 2, no. 3, August 2010 (pp. 180-212)

"Beggar Thy Neighbor? Or Thyself?" On Inequality and Tax Policy Spillover

Few months ago, my colleague Rizmy wrote a piece on Investor Daily about tax competition and inequality. As you may know, government sometime lowers tax rate in order to attract investments. Lower tax rate means government imposes less tax to the rich, or gets less revenue for redistributive function. Hence, it may induce inequality.

Inequality, it could be argued, could be induced even though domestic policy does not change. Lower tax rate in neighboring country may create inequality via profit shifting in the domestic country (Baker and Murphy, 2019) or by capital inflow surges to the neighboring country which can worsen inequality (Azis and Shin, 2015).

Current version of Stata (Stata 15) can seamlessly create inverse-distance weighting matrix using shapefile, which is basically a map file used in mapping software. Stata 15 also has sp- prefix to enable spatial autoregressive model to be combined nicely with regress, ivregress, or xtregress. Basically, my life would be much easier if I got a hand on this back then.

This post is intended as personal exercise in new Stata feature to answer whether neighbor’s tax policy induce inequality.

Measuring Tax Policy

Our parameter of interest is tax policy, but which one? Headline corporate income tax rate is traditionally used in tax policy research notwithstanding its shortcoming of being too uni-dimensional. Keller and Schanz (2013)'s Tax Attractiveness Index is promising, as it takes into account the existence of important regulation such as transfer pricing and controlled foreign corporation (CFC) rules. Heritage Foundation's "Fiscal Freedom" is less complex than Tax Attractiveness Index, but it takes into account total tax burden to GDP ratio as measurement of the overarching effectiveness of fiscal policy to tax. In here, we use all of them as well as headline personal tax income tax rate as addition, following Duncan and Gerrish (2014).

As you can see above, Asia-Pacific countries on average are becoming less progressive. Fiscal Freedom scores show increasing trend (more freedom = less taxed), while corporate income tax rates are decreasing. Tax Attractiveness Index and personal tax rates, however, are more or less stable.

Measuring Inequality

We use Gini index as computed by Solt (2019)'s Standardized World Income Inequality Database as a basis. Since sp- demands balanced panel data, we complete missing observations using data from UNU-WIDER's World Income Inequality Database. If there are still missing observation, data from World Bank and/or domestic statistics are added.

Plotting Gini to our tax policy measure shows the following relationship:

As is quite expected, inequality is inversely related to the progressiveness of tax policy. The more "free" or "attractive" tax policy, the higher is Gini index (and thus inequality). The reverse happens with tax rate. Whether inequality is more affected by domestic tax policy, or neighbor's, or both, are what we're trying to test here.

Model

We slightly modify the model in Martinez-Vazquez, et al. (2012) by including neighbor's tax policy measure (headline corporate income tax rate, Tax Attractiveness Index, Fiscal Freedom, and headline personal income tax rate) instead of lagged Gini index. Population growth, percentage of young population age 0-15 to total population, percentage of old population age > 65 to total population, GDP growth, GDP per capita growth, and unemployment are used as control variables.

Results

None of the control variables are significant, except for GDP growth. So the full regression result is omitted for the sake of brevity.

Now, the question whether spillover exists:

Neighbor's tax policy indeed affects domestic inequality though not in the way we expected. Firstly, such in the case of Fiscal Freedom (FISCALFREEDOM) and corporate income tax rate (CITR), it turns out that foreign tax policy is more strongly correlate to inequality than domestic tax policy.

Secondly, it completely reversed what we can infer from the scatterplots above. For example, in the case of Tax Attractiveness Index (TAXATTRACT), the coefficient is negative. This means that the more attractive our neighbor, the less unequal is our domestic economy. In the case of corporate income tax rate (CITR), the lower our neighbor tax rate, the less unequal is our domestic.

This is still consistent with Azis and Shin (2015). If what capital inflow surges increase inequality, and if lower tax rate increase capital inflow in a country, that country's Gini index rises. For example, if Hong Kong lower its tax, capital would flow there and Hong Kong citizen become less equal. Yet this warrants further study, because this one does not test if indeed capital is mobile. As an aside disclaimer: some specification error/bias may exist that I am unaware of.

Conclusion

This post is just an exercise in STATA 15 new feature. The result, in a counterintuitive way: it appears that domestic tax rate does not correlate with GINI, but neighbors tax rates correlate negatively with GINI. That means when the neighbors lower their taxes, it reduce inequality at home insofar as domestic tax rate stays the same. Does this mean because capital flows abroad then inequality is reduced at home? Does this mean that we can lower our tax rate, but still keeping it higher than our neighbors' so as not to increase inequality? Probably. This exercise is not designed to answer that question, nor can it examine the dynamics of GINI (including the persistence of tax rate effect to GINI).

Take this with an ocean worth of salt.

References:

Azis, I. J. and Shin, H. S. (2015) Capital Flows and Income Distribution. In: Managing Elevated Risk. Springer, Singapore

Baker, A. and Murphy, R. (2019) The Political Economy of ‘Tax Spillover’: A New Multilateral Framework. Glob Policy, 10: 178-192. doi:10.1111/1758-5899.12655

Duncan, D. and Gerrish, E. (2014) "Personal Income Tax Mimicry: Evidence from International Panel Data." International Tax and Public Finance, Springer; International Institute of Public Finance, vol. 21(1), pages 119-152

Keller, S. and Schanz, D. (2013) Measuring Tax Attractiveness across Countries. arqus-Working Paper No. 143

Martinez-Vazquez, J., Moreno-Dodson, B. and Vulovic, V. (2012) The Impact of Tax and Expenditure Policies on Income Distribution: Evidence from a Large Panel of Countries. Andrew Young School of Policy Studies Research Paper Series No. 12-30

Novastria, R. O. (2019). "Kompetisi Pajak Picu Kesenjangan". Investor Daily https://investor.id/archive/kompetisi-pajak-picu-kesenjangan (accessed 10 June 2019)

Solt, F (2019) Measuring Income Inequality Across Countries and Over Time: The Standardized World Income Inequality Database. SWIID Version 8.1, May 2019.