4. Investment Impacts of Pillar One and Pillar Two

4.1.1. New tax rules impact incentive structures of MNEs and governments

291. The Pillar One and Pillar Two proposals would introduce significant changes to the international tax rules, affecting global investment through their impacts on the incentives faced by MNEs and governments. Amount A of Pillar One involves the creation of a new taxing right and the reallocation to market jurisdictions of a share of residual profit determined at the MNE group level, based on a formulaic approach. Pillar Two addresses remaining BEPS challenges and is designed to ensure that large internationally operating businesses pay a minimum level of tax regardless of where they are headquartered or the jurisdictions they operate in. Without prejudging the final design and parameter choices, which are still the subject of discussions among members of the Inclusive Framework on BEPS, the structural changes embedded in these new rules could have substantial direct and indirect effects on investment and economic output.

292. The new rules will not affect all MNEs, but are targeted to large, highly profitable MNE groups (Pillar One) and large MNE groups with low effective tax rates (Pillar Two). The inclusion of several scope restrictions would imply that the effects of Pillar One and Pillar Two on investment costs would not affect all firms, but would be targeted to a subset of MNEs. For example, although it remains subject to political decision, the scoping criteria currently being discussed for Amount A of Pillar One only includes businesses or business segments that perform activities within the definitions of the Automated Digital Services (ADS) and Consumer Facing Businesses (CFB) under discussion. Importantly, it includes a profitability threshold that is determined at the MNE group level, which targets the impact of the rules to highly profitable MNEs. Pillar One will also incorporate a global revenue threshold, ensuring that only larger MNE groups are targeted. Pillar Two, on the other hand, is focussed on large MNEs with low effective tax rates, which will mainly affect firms reporting large profits in low-tax jurisdictions. This will especially be the case if a formulaic substance-based carve-out is provided.

4.1.2. Investment responses at the MNE entity level

293. At the MNE entity level, increases in investment costs could lead to some relocation of investment away from low-tax jurisdictions. Increases in investment costs are mostly driven by Pillar Two, notably affecting MNE entities that would otherwise have realised effective tax rates (ETRs) below the minimum tax rate. Confirming the established literature, new empirical research on the tax sensitivity of investment suggests that affected MNE entities would be expected to respond to potential cost increases by decreasing or relocating investment, in the absence of other commercial considerations.

294. Investment relocation across jurisdictions does not necessarily lead to a decrease in global investment if it is linked to an investment increase in another location. Relocation of MNE activities can increase the efficiency of capital allocation across jurisdictions, in situations where investment decisions were previously driven mostly by tax considerations. MNEs make investment decisions based on a range of commercial considerations, including taxation. To the extent that taxation has an influence, investment decisions are driven by post-tax rather than pre-tax returns. Given that ETR differentials can be substantial, MNEs could potentially be induced to choose locations where investments yield lower pre-tax, but higher post-tax returns. The Pillar One and Pillar Two proposals reduce ETR differentials across jurisdictions, thus reducing the scope for tax-induced distortions of investment decisions and potentially leading to a more efficient capital allocation and higher global output.

4.1.3. Government responses can have indirect effects on investment

295. Additional tax revenues can have positive indirect effects on the economy, for example, supporting domestic resource mobilisation in developing countries. While the optimal level of tax revenues, the optimal tax mix, and the optimal use of public funds depend on the country-specific context, additional tax revenues could have positive effects on the economy, as they can be used, e.g., to support public or private investment, to reduce public debt levels or to finance reductions in other taxes. Lower ETR differentials reduce the potential gains from profit shifting, implying that corporate tax bases may become less elastic and public funds may be able to be raised more efficiently. These positive indirect impacts on fiscal capacity are particularly important for developing countries, which often face more stringent capacity constraints.

296. Governments’ responses to changes in the international tax system could have important indirect effects on the attractiveness of their jurisdiction for foreign investment. Pillar Two may reduce the effectiveness of certain tax incentives provided through the corporate income tax (CIT) system. Notwithstanding this, governments will continue to be able to draw upon a wide range of tax and non-tax instruments to support policy objectives such as increased innovation or economic development. However, in response to structural changes, some jurisdictions could adapt their policies to improve attractiveness for foreign investment over and above simple tax-cost considerations. For example, a more limited and cost-efficient use of investment tax incentives could strengthen domestic resource mobilisation in developing countries. Both responses would reinforce positive effects on international capital allocation.

4.1.4. Investment responses at the MNE group level

297. At the MNE group level, the GDP-weighted average increase in effective marginal tax rates could be around 1.4 percentage points, suggesting only limited impacts on global investment levels. The global GDP-weighted average increase in effective average tax rates (EATRs) is estimated to be around 0.3 percentage points, representing a small impact compared to the weighted average EATR in the sample (24%) or the six percentage point reduction in the unweighted average EATR that was observed between 1999 and 2017. The corresponding increase for effective marginal tax rates (EMTRs) is estimated to be around 1.4 percentage points, again representing a small change compared to the weighted average EMTR of around 25%. These effects represent global averages across all MNEs; the corresponding increase in EATRs and EMTRs would generally be higher for MNEs that are within the scope of Pillar One or Pillar Two. These effects are mostly driven by Pillar Two and are larger for projects located in investment hubs, most of which currently face comparatively low effective tax rates.1

298. The response of investment is expected to be weak for entities in MNE groups affected by the reforms, as new evidence suggests that MNE entities in more profitable MNE groups are less sensitive to tax increases than entities in MNE groups that are out of scope. Specifically, entities in MNE groups with a profitability rate between 0% and 10% would, on average, reduce their domestic investment rate by around 0.15 percentage points following a one percentage point increase in the jurisdiction’s EMTR. The size of this effect is almost half as large for entities in MNE groups with profitability ratios above 10% and more than three times smaller for entities in MNE groups with profitability rates above 15%.

299. The proposals are expected to produce a more level playing field among MNEs, and vis-à-vis their smaller and domestic competitors. Pillar One would be targeted towards large and profitable MNEs, while Pillar Two would ensure that all MNEs pay a minimum level of tax in each jurisdiction in which they operate irrespective of where they are based.

4.1.5. Impacts on the global economy in case no consensus is reached

300. These results must be compared to the implications for the global economy in a counterfactual scenario where a multilateral consensus-based solution cannot be secured. It would be incorrect to assume that the counterfactual scenario looks like the status quo. Equally important, the evaluation has to acknowledge that the baseline scenario has changed due to the COVID-19 pandemic.

301. Failure to secure a consensus-based solution would increase the domestic pressures on governments to address the tax challenges arising from digitalisation unilaterally, and would likely lead to the introduction of digital service taxes or similar measures in a growing number of jurisdictions. An increasing number of jurisdictions have implemented, or are considering the introduction of digital services taxes; other unilateral measures include alternative applications of the permanent establishment threshold, withholding taxes or diverted profit taxes. The introduction of such measures, as well as potentially more aggressive actions by tax administrations, could affect investment indirectly by leading to a rise in disputes between countries and MNEs which generates significant administration and compliance costs.

302. A proliferation of digital services taxes (DSTs) would lead to economic inefficiencies, reducing global investment and output. DSTs are not designed as taxes on corporate profits, but as taxes on total revenues associated with specific types of digital transactions. They could lead to positive tax liabilities imposed on loss-making firms as well as to economic double taxation, and are thus more distortive than profit-based taxes, potentially leading to higher prices, lower sales, and less investment in the affected sectors. Since they are typically levied on intermediate services, e.g., online advertisement, these effects are expected to flow through to a much larger number of firms and sectors, an effect that could be larger if intermediate services are provided by online platforms with some degree of market power.

303. More widespread adoption of DSTs would likely give rise to an increase in trade disputes, in addition to the immediate negative impacts on economic efficiency.

304. The negative economic implications of protracted tax- and trade-related conflicts are likely to be significant, ranging between around -0.1% to -1.2% of global GDP. Based on different DST rates and retaliatory tariffs, global GDP is simulated to be reduced by -0.1% to -0.2% if a smaller group of jurisdictions is involved in the conflict, and between to -0.4% to -1.2% if a larger number is involved, compared to a scenario without tax- and trade-disputes. These results represent a substantial reduction in global GDP, stemming from efficiency losses due to the tariffs compounded with reductions in wages and capital returns as well as a slow-down in investment driven by declining exports. Household income would be impacted to the same extent as GDP, indicating that households would bear most of the costs of the shrinking economy. World trade, measured as real global import volumes, falls by about twice the rate of real GDP.

305. In contrast, an approximation of the impacts of investment cost increases under the consensus scenario suggests a reduction of less than 0.1% of global GDP, compared to a hypothetical scenario where failure to secure consensus does not lead to any future tax- and trade-disputes.

4.1.6. Overall Assessment against the background of COVID-19

306. Under both scenarios, global investment and output are being severely affected by the COVID-19 pandemic. The COVID-19 crisis is affecting firms, economies and governments in ways that could modify the expected impact of the reform. For example, the economic crisis will have strong negative effects on the profitability of most MNEs in the short- and possibly medium-term. Under the consensus scenario, loss carry-forward provisions could imply that any impacts of Pillar One and Pillar Two on investment costs, and therefore global investment, are likely to be less than estimated and will take effect over a much longer timeframe. Given the structure of certain highly-digitalised markets, increased demand for digital services could translate into higher profitability for certain large MNE groups, implying that a larger share of profits will likely be subject to Pillar One over time, in line with profitability increases.

307. Against the background of the COVID-19 crisis, the negative consequences of a failure to secure consensus are likely to be more severe. As the global economy is very fragile, aggravated tax- and trade disputes could compound the effect of the COVID crisis or hinder the post-crisis recovery. DSTs could lead to double taxation and asymmetric corporate taxation, inducing firms to forego investment opportunities that would be economically viable in the absence of DSTs. Political pressure on governments will increase, as the tax challenges from digitalisation remain unresolved at a time when many highly-digitalised MNEs are experiencing higher sales and rising valuations.

308. Securing a consensus-based solution to the tax challenges arising from digitalisation will support global investment compared to alternative counterfactual scenarios, which are likely to exacerbate an already negative economic outlook. While increases in investment costs could slightly weigh on global investment under the consensus scenario, this negative effect would be quantitatively small compared to alternative scenarios capturing the potential costs of the tax- and trade-related conflicts that may arise if a consensus-based solution cannot be secured. In addition, the reform would enhance tax coherence and tax certainty, and the efficiency of capital allocation compared to the no-consensus scenario, which would also support investment and growth.

4.5.1. Methodological approach

323. Consistent with the MNE group level perspective, the empirical approach considers a stylised organisational structure that is held constant throughout the analysis and calibrated based on jurisdiction-specific data on profit and asset location. For the purposes of the empirical calibration, it is assumed that the ultimate parent entity of the MNE group carries out an investment in the jurisdiction where it is located, conducts operations through its subsidiaries located in other jurisdictions included in the dataset and produces a final consumer good that is sold to a global consumer base. Further assumptions are that the firm is a large MNE in a profit position and that investment is financed by retained earnings; the treatment of loss-making firms is, therefore, not considered. The investment is constructed as an unweighted average across three broad asset categories, non-residential structures, tangibles assets and acquired intangibles; statutory tax rates, depreciation rules and macroeconomic parameters are taken from OECD Corporate Tax Statistics (OECD, 2020[15]).7 Personal income taxation and other taxes at the international level, e.g., withholding taxes, are not considered.

324. Effective marginal tax rates (EMTRs) and effective average tax rates8 (EATRs) are calculated at the MNE group level with respect to an additional investment in the jurisdiction of the ultimate parent entity9, assuming that a share of its profits are shifted to other group subsidiaries located in jurisdictions with lower tax rates. The extent to which profits are shifted is approximated based on the consensus estimate from the profit shifting literature (Johansson et al., 2017[16]), (Heckemeyer and Overesch, 2017[17]), (Beer, de Mooij and Liu, 2020[18]). The location of shifted profits as well as the revenue structure are empirically calibrated, consistent with the data matrices constructed for the development of the tax revenue estimates (see Chapter 5), and shifted profits are taxed at the respective statutory CIT rates (see Hanappi and González Cabral (2020[12]) for a sensitivity analysis). This approach ensures that cross-country heterogeneity in the activities of MNEs, and thus in the potential impact of the proposals, is reflected in the analysis.

325. To capture the impact of the proposals, ETRs are first calculated10 for a baseline case with profit shifting, building on data from 2019 on corporate tax rates and bases as published in OECD Corporate Tax Statistics (OECD, 2020[19]). Holding organisational structure and profit location constant, the baseline case is then compared to the post-implementation case to determine ETR changes at the MNE group level. Pillars One and Two include several design features that determine the extent to which the MNE group level ETRs will be affected. While the specific parameters are yet to be determined, the analysis relies on a set of assumptions, on a without prejudice basis, to model the expected effects of both pillars.11 The specification underlying the analysis discussed below assumes12 that Amount A13 under Pillar One entails a 10% profitability threshold based on profit before tax over turnover, a 20% reallocation percentage to market jurisdictions and a restriction to ADS and CFB. In considering Pillar Two, a 12.5% rate with jurisdiction blending is assumed; the carve-out on depreciation expenses is assumed to be 10% and approximated using the value and location of tangible assets while the carve-out on payroll cannot be covered in this chapter due to data limitations. A revenue threshold of EUR 750 million is modelled for both Pillars. Using these parameters, the share of firms and profits that are in scope of Pillar One and Pillar Two are calibrated using jurisdiction-specific, firm-level data.14 Additional results for a range of design options and parameter values are presented in Hanappi and González Cabral (2020[12]).

4.5.2. Empirical results

326. Consistent with the assumptions about stylised organisational structures, the results in Figure 4.2 identify MNE group level changes in ETRs associated with real investments in the jurisdiction of the ultimate parent entity.15 Changes in profit shifting due to the implementation of the proposals are not captured in the calculation of the MNE group level ETRs. As discussed in Chapter 3, the new provisions are expected to reduce the amount of profits shifted to low-tax jurisdictions; this means that overall effective taxation would increase, on average, while tax liabilities under Pillar Two would fall if changes in profit shifting are accounted for. Insofar as that is the case, the direct impact of minimum effective taxation on the group level ETR would be lower. Following the approach taken with respect to the revenue results, the jurisdiction-specific results are aggregated into three groups based on GDP per capita: high income jurisdictions, low and middle income jurisdictions, and investment hubs.16 GDP-weighted aggregated results are shown, for each of the three groups as well as the global average, in Figure 4.2.

327. Several initial insights emerge from the analysis of MNE group level ETRs. First, the share of MNEs that will be unaffected by the new rules is significant due to the revenue threshold and various scope restrictions.17 Second, the results suggest small effects of Pillars One and Two on MNE group level effective average tax rates (EATRs) and effective marginal tax rates (EMTRs), based on an average across all MNEs. The global GDP-weighted average change in the EATRs from Pillar One and Two is estimated to be just over 0.3 percentage points (Panel A), representing a small impact compared to 24%, i.e., the weighted average EATR in the sample, or the 6 percentage point reduction in the EATR observed over 1999-2017 (Section 4.7.3). The corresponding change for the EMTR is around 1.4 percentage points, again representing a small change compared to the weighted average EMTR of around 25%. Third, much of this increase is driven by Pillar Two. While the global weighted-average change in the EATR due to Pillar One is close to zero (0.01 percentage points), the change due to Pillar Two is estimated to be around 0.3 percentage points. The same pattern is observed for the EMTR (Panel B in Figure 4.2). Fourth, the effect of the new rules on ETRs realised on investments in investment hubs is larger compared to investments in other jurisdictions.

Figure 4.2. Changes in Effective Tax Rates due to Pillars One and Two

ETRs are calculated at the MNE group level assuming an investment in the jurisdiction of the ultimate parent entity; the vertical axis shows GDP-weighted average changes in ETRs in percentage points by income groups.

Note: Pillar One (Amount A only) considers a 10% profitability threshold on Profit/Turnover, 20% reallocation percentage to market and a scope that is restricted to ADS and CFB. Pillar Two considers a 12.5% rate with jurisdiction blending and a 10% carve-out on depreciation expenses (approximated using the value of tangible assets). In addition, it is assumed that an MNE group that claims the benefit of the carve-out should be required to make a corresponding and proportional adjustment to the covered taxes for the calculation of the ETR. The alternative option (i.e. not making a corresponding and proportional adjustment to the covered taxes) would be difficult to model with the available data. See Chapter 3 for more details. As described in endnote 12, real investments in jurisdictions without a full-fledged CIT system are not considered; those jurisdictions are thus excluded from the Figure, but profits can still be shifted to these jurisdictions. A revenue threshold of EUR 750 million is assumed, on a without prejudice basis, for the modelling of both Pillars. The combined effect does not include interaction effects of both Pillars. The results in the Figure reflect the average effects across all MNEs. The number of jurisdictions is restricted to those available in the OECD’s Corporate Tax Statistics (OECD, 2020[19]), with the exception of Estonia and Latvia where profits are only taxed upon distribution. Jurisdiction groups are based on the World Bank classification of countries by income group. Investment hubs are defined as jurisdictions with a total inward FDI position above 150% of GDP, based on raw FDI data.

Source: Hanappi and González Cabral (2020[12]).

328. A range of further issues should be considered when interpreting these results in terms of their impacts on investment and global output. Although the quantitative analysis discussed above focuses on group level ETRs, some additional insights on entity level ETRs emerge as a corollary; in particular, both Pillars narrow the dispersion of entity and group level EATRs and EMTRs across jurisdictions. The reduced dispersion in tax rates reduces the tax gains from profit shifting, thus likely resulting in a corresponding reduction in the amount of shifted profits. As the marginal gain from profit shifting falls, corporate tax bases may become less elastic and, as a consequence, a given tax rate increase may be associated with lower costs in terms of tax base lost due to relocation – governments may thus be able to raise funds more efficiently.

329. In addition, a reduction in ETR differentials across jurisdictions is likely to improve the efficiency of the international allocation of capital (Englisch and Becker, 2019[20]). Firms make investment decisions based on post-tax rather than pre-tax returns. If profits are taxed at source and foreign income is tax-exempt, as is increasingly the case,18 ETR differentials can be substantial, potentially inducing firms to choose locations where investments yield lower pre-tax, but higher post-tax returns. As entity level ETR differentials decrease due to the new tax rules, there is less scope for tax-induced distortions and global output19 could thus increase as capital allocation across jurisdictions becomes more efficient.20

4.5.3. Tax Incidence

330. The tax incidence of the reform can impact economic agents and their investment behaviour and should thus be considered in the evaluation. As opposed to statutory incidence (i.e., the legal obligation to remit a tax), the economic incidence associated with a new corporate tax liability may be borne by shareholders or may be shifted onto other economic agents, e.g., workers or consumers, through price adjustments. The extent to which these different economic agents bear the incidence of the reform could affect the impact of cost increases on firms’ investment, therefore having important implications for the assessment of the new rules in terms of efficiency.21 For example, if an increase in investment costs would fall mostly on workers or consumers, it could lead to a smaller reduction in firms’ investment compared to the case where a larger share is borne by shareholders.

331. In general, the extent to which the economic incidence can be shifted onto others depends on a range of factors including supply and demand elasticities as well as market structure (Fullerton and Metcalf, 2002[21]) (Auerbach, 2006[22]). In the context of corporate taxation, it is typically assumed that in open economies (with competitive product markets), where capital is relatively mobile across jurisdictions while labour tends to be less mobile, firms will be able to shift a comparatively large share of the burden onto workers through wage decreases. However, only a few empirical papers directly investigate these theoretical insights and subsequent research has questioned the underlying assumptions, suggesting that, based on empirical estimates, shareholders bear a larger share of the economic incidence than previously assumed (see Gravelle (2013[23]), Clausing (2012[24]) (2013[25]) and references therein).

332. While empirical research on tax incidence is not conclusive due to the scarcity of disaggregated firm-level data, recent studies suggest that price adjustments to tax changes often follow more complex patterns, specifically when standard theoretical assumptions are not fully satisfied. For example, if firms earn more than the normal return to capital, e.g., due to their dominant position in a given market, corporate tax would fall, at least in part, on economic rents; as a consequence, capital allocation would be less distorted and downward pressure on wages would be lower (Auerbach, 2006[22]). Furthermore, Clausing (2013[25]) argues that profit-shifting allows multinational firms to decouple the location where profits are reported from the location of economic activities, thus partially insulating workers in high-tax jurisdictions from adverse wage effects associated with a given corporate tax increase. In addition, recent studies based on disaggregated data suggest that firm-level heterogeneity and rent-sharing between workers and shareholders could have implications for tax incidence (Arulampalam et al. (2012[26]); Serrato and Zirdar (2016[27]); Fuest, Peichl and Siegloch (2018[28])). Furthermore, the evidence discussed in Section 4.6 suggests that investment of entities in more profitable MNE groups is less sensitive to taxation, a result that could also be driven by the existence of economic rents at the MNE group level.

333. The literature on corporate tax incidence often assumes that product markets are competitive in which case it is impossible for firms to shift incidence onto consumers. Although this assumption may be warranted in more general settings, it is less likely to be the case in the context of highly-digitalised firms operating in markets characterised by winner-takes-all (or winner-takes-most) dynamics. In particular, the presence of multi-sided markets could have implications for tax incidence analysis, thereby potentially affecting investment decisions. Theoretical research suggests that in multisided markets optimal prices do not necessarily correspond to marginal costs (Rochet and Tirole (2003[29]), (2006[30])). If this is indeed the case, it is possible that tax changes induce an adjustment in the optimal price structure, for example, if the digital platform has some degree of market power and is sharing economic rents with different groups of end users. However, the academic literature on this particular topic is still limited and recent contributions suggest that there is a considerable degree of uncertainty with respect to tax incidence in multisided markets (Kind, Koethenbuerger and Schjelderup (2008[31]), (2010[32]), Belleflamme and Toulemonde (2016[33]), Kind and Koethenbuerger (2017[34]), Tremblay and Tremblay (2017[35]), Bourreau, Caillaud and De Nijs (2018[36]), Cui (2019[37]), Bibler, Teltser and Tremblay (2019[38])).

334. Compounding these considerations, the new rules are likely to elicit additional behavioural changes by MNEs and governments that could lead to further indirect effects on efficiency and global output relating, as discussed in the Section 4.7.

4.7.1. Fiscal space

349. Having already accounted for the corresponding direct effects on investment in the Section 4.5, the overall assessment of the new rules also has to account for the expected increases in fiscal space. As discussed in the revenue analyses, the new tax rules are expected to directly increase tax revenues for most governments (see Chapter 2 and Chapter 3). While the optimal use of additional fiscal space depends on the country-specific context, an increase in tax revenues, having already accounted for the change in investment costs, is likely to have positive effects on the economy, as it can be used, e.g., to support public or private investment or to reduce public debt levels.

350. Furthermore, the reduction in ETR differentials noted in the Section 4.5 affects not only real investment, but it also reduces the potential gains from profit shifting. As a result, MNEs would be expected to engage in less tax evasion and corporate tax bases may become less elastic on average, e.g., if the sensitivity of profits to tax is increasing non-linearly with tax rate differentials. A given tax rate increase would, therefore, be associated with lower costs in terms of tax base lost due to relocation and governments may thus be able to raise funds more efficiently; e.g., tax administrations could raise the same revenue with lower tax enforcement levels. Through this channel, the new tax rules would thus reduce the marginal cost of public funds (Slemrod and Wilson, 2009[45]), thereby further strengthening fiscal space across all jurisdictions.

351. The positive impacts of increased fiscal space are likely to be particularly important for developing countries. The marginal cost of public funds may be higher in developing countries due to capacity constraints in terms of tax enforcement and administration (Dabla-Norris et al., 2019[46]; Keen et al., 2015[47]). Insofar as compliance and administration functions related to Pillar One and Pillar Two are carried out by entities in ultimate parent jurisdictions, additional tax revenues could be raised at a lower cost per unit of public funds. Moreover, the need for domestic resource mobilisation to support public investments tends to be greater in developing countries, suggesting that public investments could yield stronger positive effects in those countries compared to developed countries where the return on additional public investment is likely to be lower (Besley and Persson, 2014[48]).

4.7.2. Compliance and administration costs

352. Introducing new tax provisions to implement Pillars One and Two will increase filing requirements, and compliance costs, for those MNEs that are within scope. As discussed in Chapter 2 and Chapter 3, the policy design includes several scope restrictions implying that only large and relatively profitable MNEs will be in scope. This means that increases in compliance costs will be borne mostly by those firms that currently have the lowest compliance costs measured as a proportion of turnover (see Box 4.2). In addition to this, some tax administrations may experience increases in administration costs as a result of implementing the new rules. Simplified administrative processes currently under discussion for various components of both Pillars would help to limit these costs.

Figure 4.5. Trends in Multilateral Agreement Procedure (MAP) cases for OECD Countries

Source: OECD calculations based on MAP Statistics at: https://www.oecd.org/ctp/dispute/

To ensure a balanced panel, data are provided only for the following countries: Australia, Austria, Belgium, Canada, Chile, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Korea, Luxembourg, Mexico, the Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey, the United Kingdom, and the United States.

Figure 4.6. Time to closure of cases under Multilateral Agreement Procedure

Source: OECD calculations based on MAP Statistics at: https://www.oecd.org/ctp/dispute/

353. Several indirect effects on firm behaviour could, at least partially, attenuate the expected increase in compliance costs. First, tax planning represents a larger proportion of the administrative costs of large MNEs relative to smaller firms (see Box 4.2). It is expected that the new rules will reduce the incentives for MNEs to engage in profit shifting. To the extent that firms reduce or abandon the use of complex tax planning structures, in the medium term this could reduce compliance costs for firms and administration costs for tax authorities.

354. Second, there has been an expansion of Mutual Agreement Procedure (MAP) cases and an increasing volume of outstanding MAP cases in the international tax system (Figure 4.5). These cases are already creating significant compliance costs for MNEs. It is noteworthy that a large share of these MAP cases (51% in 2018) are transfer-pricing cases, which typically take much longer to close (Figure 4.6). While time-to-closure remains high for transfer pricing cases, it has recently been reduced for other, non transfer-pricing-related cases. Some of the transfer pricing cases refer to related party distributors, especially limited-risk distributors. By standardising the remuneration of related party distributors, Amount B of Pillar One would reduce the likelihood of these cases ending up in MAP. 26

355. Third, the reforms considered under Pillars One and Two are expected to include an expansion and improvement of dispute prevention procedures, suggesting that the additional compliance burden of implementing the proposals may be somewhat reduced. MNEs would need to allocate fewer resources to appeals, litigation and audits than would be the case in the absence of these additional procedures. These aspects, coupled with a potential reduction in MAP disputes due to simplifications under Amount B of Pillar One, may attenuate any increases in compliance costs for MNEs that result from the implementation of the new rules.

356. As the implementation of the proposals creates additional compliance costs for MNEs that are within scope, they may also lead to additional administration costs for tax authorities. As is the case for MNEs, a reduction in disputes due to expanded dispute prevention and resolution procedures may reduce administration costs. Simplification measures (such as Amount B) may also obviate transfer pricing disputes in certain areas. However, the overall level of administration costs may nonetheless rise as a result of the implementation of the proposals. Additional administration costs may be larger for those administrations that act as the lead tax administration as part of the proposed centralised and simplified administration system, and so may not be equal across jurisdictions.

4.7.3. International tax competition

357. Empirical evidence generally supports the view that jurisdictions use their CIT system, including statutory rates as well as the definition of corporate tax bases, to compete for MNE investment and the tax revenue it generates over time (see Box 4.3). First, statutory CIT rates have been declining since the mid-1980s, not only in OECD countries but also in many developing and emerging economies (Devereux et al., 2002[63]) (Klemm and Van Parys, 2012[64]). In particular, the unweighted average statutory CIT rate in OECD countries has declined from above 32% in 2000 to around 23% in 2020, and from around 28% to just below 21% in a sample covering more than 90 developing and developed countries over the same time horizon (see Figure 4.7 based on OECD Corporate Tax Statistics (2020[19])). Second, available data on forward-looking effective average tax rates27 (EATRs), capturing not only the statutory rates but also standard components of the corporate tax base, show a decline in the unweighted effective average tax rate from 29% to around 23% in a balanced panel of OECD and G20 countries over the period 1999 to 2017. An alternative data source covering mostly European countries shows a similar decline in the unweighted EATR from 24% to 20% over the years 2005 to 2018 (Figure 4.8).

358. In addition, empirical research has also identified strategic interactions in tax policy design across jurisdictions, implying that the policy choices of governments regarding tax rates and bases are affected by other governments’ tax policy choices (see Annex 4.A). Interactions can arise in many different areas, going beyond statutory CIT rates and bases; for example, policy decisions on preferential regimes, CFC rules or double tax agreements can have implications for policy design in other jurisdictions. Ultimately, these strategic responses to tax competition produce negative fiscal externalities (or spillovers) that could drive corporate effective tax rates below their optimal level and lead to inefficiently high taxation of less mobile tax bases such as labour or consumption. In this case, unilateral (or uncoordinated) policy design is likely to produce a globally inefficient outcome, i.e., a reduction in global economic output compared to the case with multilateral (or coordinated) policy design (Gupta et al., 2014[65]).

359. However, the theoretical literature reviewed in Annex 4.A also provides some arguments for the potentially efficiency-enhancing effects of tax competition. For example, competitive pressure could have positive efficiency effects if capital becomes partially immobile once the location decision has been taken and governments cannot credibly commit to keeping effective tax rates stable in the absence of tax competition. Alternatively, positive efficiency effects could also materialise if competitive pressure limits public sector growth that is driven by rent-seeking or self-interested behaviour by political leaders and government officials.

Figure 4.7. Combined statutory CIT rates 2000 to 2020, by Region (unweighted)

Note: The data series represent unweighted average combined CIT rates across all countries covered in the second edition of OECD Corporate Tax Statistics; zero-tax jurisdictions are included; in the case of progressive tax systems the highest applicable rate is used.

Source: OECD Corporate Tax Statistics (2020[19])).

360. The new tax rules introduced under Pillar One and Pillar Two will affect the strategic interactions in tax policy design across jurisdictions, potentially reducing the extent to which governments engage in tax competition. Under Pillar One, the share of the corporate tax base within the scope of Amount A is determined by reference to profitability at the MNE group level; a fraction of this share is then allocated to market jurisdictions28 and taxed there at prevailing rates. This approach is likely to reduce competitive pressure on governments through two different channels. First, consumers and users can generally be considered as relatively immobile; the allocation of a share of the corporate tax base to market jurisdictions thus directly reduces the tax base that can be attracted through lower effective taxation. Second, assessing corporate tax bases at the MNE group level, rather than on a separate basis for each of the entities of the MNE group, is likely to reduce the sensitivity of MNE investment with respect to tax rates in specific locations. As the amount of investment that can be attracted through a given reduction in the effective tax rate will be lower, governments will have some scope to retain higher levels (or resist pressure to implement lower levels) of effective taxation compared to the case without an MNE group-level assessment.

361. Under Pillar Two, minimum taxation is implemented by introducing four new rules (see Chapter 1), leading to significant implications for governments’ optimal tax policy choices. The objective of the new provisions is to ensure that the effective tax rate on MNE profit that would otherwise be taxed below an agreed minimum rate is brought up to this minimum rate, which has to be decided by the Inclusive Framework. The income inclusion rule (IIR) will allow jurisdictions where ultimate parent entities are located to top-up taxes on low-taxed profits earned by subsidiaries in other jurisdictions such that effective tax rates are brought up to a given minimum rate. In the absence of complementary rules, MNEs could have incentives to invert, i.e., to change their residence for tax purposes, in order to avoid minimum taxation if some jurisdictions do not adopt the IIR (see Annex 4.C). To avoid creating such an incentive, the undertaxed payments rule (UTPR) would allocate top-up tax proportionately among entities applying UTPR in a co-ordinated way first to those entities making direct payments to the low-tax MNE entity and then amongst all entities in the group that have net intra-group expenditure (Englisch and Becker, 2019[20]). Furthermore, two additional components, the switch-over rule and the subject to tax rule, complement these rules, contributing to the objective of ensuring minimum effective taxation.

362. Pillar Two introduces a lower bound on the effective tax rates that governments can offer in order to attract foreign investment, thereby limiting the extent to which competition can take place through the corporate tax system. However, it is possible that competitive pressure persists and some governments respond by providing other forms of public support. In some cases (e.g., grants or subsidies), this support will be more transparent than if it was delivered through the tax system. In addition, the existence of a lower bound does not necessarily imply that effective tax rates will remain stable (or even increase) across all participating jurisdictions. As reviewed in Annex 4.A, some theoretical models29 suggest that a reduction in the effective tax rate in jurisdictions with relatively high rates could, under certain assumptions, become more attractive with minimum effective taxation because it is common knowledge that low-tax jurisdictions cannot lower their rates below a certain minimum. However, it is unclear whether this theoretical result is empirically relevant, given that recent evidence at the German subnational level points in the opposite direction (Schwerin and Buettner, 2016[66]).

Figure 4.8. Effective Average Tax Rates (EATRs), by Data Source (unweighted)

Time series data on EATRs is currently available from two different sources, the Oxford Centre for Business Taxation (CBT) and the Leibniz Centre for European Economic Research (ZEW); the unweighted averages shown in this graph are computed for balanced panels covering 38 and 35 countries respectively.

Note: The ZEW time series covers 35 mostly European countries over 2005 to 2018: Austria, Belgium, Bulgaria, Canada, Switzerland, Cyprus30, Czech Republic, Germany, Denmark, Estonia, Spain, Finland, France, UK, Greece, Croatia, Hungary, Ireland, Italy, Japan, Lithuania, Luxembourg, Latvia, Macedonia, Malta, Netherlands, Norway, Poland, Portugal, Romania, Sweden, Slovenia, Slovak Republic, Turkey and the USA. The time series from the Oxford University Centre for Business Taxation covers 38 countries over 1999 to 2017: Argentina, Austria, Brazil, Belgium, Bulgaria, Canada, Switzerland, Chile, Germany, Denmark, Estonia, Spain, Finland, France, UK, Greece, Hungary, Indonesia, Ireland, Israel, India, Iceland, Italy, Japan, Republic of Korea, Luxembourg, Mexico, Netherlands, Norway, New Zealand, Poland, Portugal, Russia, Sweden, Slovenia, Turkey, USA and South Africa.

Source: ZEW rates are based on the European Union (EU) Commission project TAXUD/2018/DE/307; Oxford CBT rates have been sourced from the website.

363. A different strand of the theoretical literature investigates the role of zero-tax jurisdictions specialised in attracting accounting profits that are shifted outwards from other jurisdictions. On the one hand, the existence of such jurisdictions increases the marginal cost of public funds faced by other jurisdictions, thus implying a net loss in global economic output (Slemrod and Wilson, 2009[45]). On the other hand, it has also been argued that allowing for a certain amount of profits to be shifted outwards could be an efficient strategy in a context where governments compete for mobile and immobile tax bases but tax policy cannot discriminate between the two (Dharmapala, 2020[67]).

364. Taken together, these theoretical and empirical insights suggest that the new rules will not put an end to tax competition, but they are likely to dampen the downward pressure on effective tax rates across many jurisdictions, mostly due to the reduction in ETR differentials. While the provisions envisaged under Pillar One clearly reduce negative fiscal externalities, scope restrictions will limit the overall impacts of these changes in the incentive structure at the global level. Minimum effective taxation under Pillar Two introduces a lower bound to tax competition; however, the resulting changes in the incentive structure are complex and optimal policy responses will depend on a range of other factors such as the size and structure of the economy, regional interactions or the general tax structure.

4.7.4. Tax incentives to support innovation

365. Innovation is a key driver of productivity, investment and growth at the firm level as well as for entire economies. However, due to the intrinsic nature of innovation, there are reasons to believe that individual entrepreneurs and businesses are likely to generate less innovation than would be socially optimal, in the absence of government intervention (see Box 4.4).

366. To support innovation and innovation activities31 such as research and development (R&D), governments have a wide array of policies at their disposal. Targeted policies can take the form of direct support via grants or public procurement of R&D services, or indirect support via the tax system, i.e., R&D tax incentives. While governments often opt for a mix of direct and indirect measures, the use of indirect support instruments has generally increased across the OECD in recent years. In some cases, governments combine tax incentives that provide preferential treatment to the inputs to innovation, e.g., R&D expenditure such as the wage of researchers; with tax incentives providing preferential treatment to the output of innovation, e.g., the income from a patent (Appelt, Galindo-Rueda and González Cabral, 2019[83]) (see Box 4.4). However, despite the increased use of tax incentives in policy, evidence on their effectiveness is mixed, appearing more positive for input-based incentives than for output-based incentives (see Annex 4.Bs). 32 Moreover, providing preferential treatment to the income from a highly mobile asset has been found to lead to the strategic location of IP assets, a key profit shifting channel identified in the literature (Beer, de Mooij and Liu, 2020[18]).

367. Notwithstanding the impact that the broader tax environment has on innovation, a key change that the new rules under Pillars One and Two introduce is the potential limitation of the use of targeted measures such as expenditure- or income-based tax incentives to foster innovation. In particular, minimum effective taxation, as proposed under Pillar Two, will limit the potential tax benefits that governments can provide through these instruments. The extent of the tax benefits offered will depend both on the design of the incentive and the general characteristics of the tax system (see Annex 4.B). Therefore, this potential reduction in generosity will only come to bear in situations where tax incentives drive effective taxation at the jurisdiction33 level, i.e., taking into account taxation on other non-qualifying income, below the minimum threshold. In cases where the minimum threshold limits the benefits of tax incentives under Pillar Two, only a subset of large MNEs would be affected while domestic and potentially also smaller multinational firms would still be able to benefit from tax incentives as before. As the final design parameters under Pillar Two are still under discussion, the introduction of carve-outs would also reduce the number of firms affected and limit the reduction in tax benefits from Pillar Two to an even smaller number of cases.

368. The provisions envisaged under Pillar One will have much less direct effect on the use of tax incentives to support innovation. Both types of tax incentives, expenditure- and income-based, decrease the effective taxation of innovative projects, thus increasing their profitability. To the extent that the reallocation of a percentage of residual profits under Pillar One leads to an increase in the tax rates applicable to the income derived from innovative projects, the new rules could lessen the reduction in effective taxation intended by tax incentives. This effect would be more likely in the case of income-based incentives.34 However, given the size of the reallocation percentage currently under discussion, the generosity and effectiveness of expenditure- and income-based tax incentives are unlikely to be significantly altered by this potential interaction with a reallocation of tax bases under Pillar One. Overall, only a relatively narrow subset of large and highly profitable MNE groups will be subject to these effects as determined by the scope of Pillar One.

Figure 4.9. Business R&D expenditures, by type of expenditure

Note: The definitions used in the graph are based on the Frascati Manual. Current expenditures are composed of labour costs of R&D personnel and other current costs used in R&D. Other current expenditures are defined in the Frascati Manual 4.23 as those non-capital purchases of materials, supplies, equipment and services to support R&D, including external R&D personnel. Country coverage for this indicator is limited to those available in the OECD R&D Statistics. The underlying data refers to the latest year available, mostly to the year 2017. The data in the graph refers to 2018 for Chinese Taipei, Japan and Korea, 2016 for Argentina, Chile, Estonia, Latvia, Poland and United Kingdom; to 2013 France and New Zealand; to 2011 for China; and to 2009 for the Netherlands.

Source: OECD, R&D Statistics, September 2020.

369. The set of available innovation policies is broad enough to enable governments to respond flexibly and adapt their policy mix to these structural changes in the international tax environment if there is a concern about the level of innovation in their economies.35 After all, incentives through the tax system are only one instrument in the policy toolbox of governments. For example, direct support measures could still be employed.36 However, even within the tax system, tax incentives delivered through the CIT are not the only instruments available to governments to support innovation. First, expenditure-based tax incentives that provide relief to payroll taxes or social security contributions will, in principle, be unaffected, as they do not depend on corporate income taxes. These types of incentives provide an immediate subsidy to labour costs, a key component for R&D making up around 60% in R&D statistics, as shown in Figure 4.9.

Figure 4.10. R&D active innovative firms as a percentage of innovation-active firms, 2016

Note: The indicator refers to product or process innovative firms. Data covers OECD and partner economies. Data on Israel, Mexico, Ireland and Luxembourg are not available. The statistics for Colombia and Argentina refer to the manufacturing sector.

Source: OECD, based on the 2019 OECD survey of national innovation statistics and the Eurostat’s Community Innovation Survey (CIS-2016), http://oe.cd/inno-stats, January 2020. Data and notes available here: https://www.oecd.org/sti/inno/innovation-indicators-2019.zip

370. Second, there are other tax instruments beyond targeted R&D tax incentives that can be used to promote innovation activities. In fact, the type of innovation activities firms conduct varies by country, across industries and firm types (Castellacci, 2008[93]; Galindo-Rueda and Verger, 2016[94]; Galindo-Rueda, Verger and Ouellet, 2020[95]), and thus the optimal policy mix. Among innovation–active firms, the share engaging in R&D averages 47% for OECD countries for which data is available, with great variability across jurisdictions, see Figure 4.10.37 This only highlights the role of innovation activities beyond R&D. The acquisition of capital assets and knowledge typically represent a significant share of total innovation expenditures, reaching around 50% of innovation expenditures among EU jurisdictions for which this breakdown is available, with significant variation across jurisdictions (Eurostat, 2020[96]). Accelerated depreciation schemes that seek to reduce the cost of acquiring new equipment, such as hardware or software, and that appear to be effective tools in promoting investment (Ohrn, 2019[97]; Zwick and Mahon, 2017[39]) could be considered as part of the policy-mix.38

371. Taxation is ultimately one among many factors affecting innovation decisions. Well-functioning product, labour and capital markets, with regulations that do not overly penalise failure may all act to increase the returns to innovation (Andrews, Criscuolo and Menon, 2014[98]). A skilled workforce, an intensive innovative environment, stable macroeconomic and regulatory conditions, competition and openness to trade and policies that help surmount innovation barriers (e.g., regulatory barriers to competition or insufficient human capital) are all key elements in the development of a sound innovation environment (Bloom, Van Reenen and Williams, 2019[99]; OECD, 2015[100]; OECD, 2010[101]). Notwithstanding the limitations that the Pillar One and Pillar Two reforms will impose upon the use of CIT-related tax incentives, governments will continue to be able to draw upon a wider range of instruments to rebalance the innovation policy-mix to mitigate any impact on the support that firms affected by the new rules are receiving, if deemed necessary.

372. Although the new rules, especially minimum effective taxation under Pillar Two, could limit the generosity of some tax incentives for R&D and innovation under specific conditions, they will also curtail the potential tax gains from profit shifting through the strategic location of IP assets. This will dampen the incentives for governments to engage in tax competition over mobile tax bases. To the extent to which the location of R&D and innovation-related activities was previously driven by tax considerations, rather than other commercial considerations, the new rules would, therefore, be expected to support an efficiency-increasing relocation of these activities. This effect further strengthens and reinforces the impacts of the reduction in ETR differentials on the efficiency of the international allocation of capital, as discussed in Section 4.4. Taken together, these changes in the incentive structure faced by MNEs and governments are thus likely to lead to a more efficient location of innovation activities, and investments more generally, which would be more strongly affected by non-tax considerations related to wider business and economic conditions.

4.7.5. Tax incentives to support economic development

373. Foreign Direct Investment (FDI) can play an important role in promoting economic development due to its potential to generate positive external effects in the local economy, e.g., through a spillover in knowledge and expertise to local workers and businesses. Governments use investment tax incentives to influence the size as well as the sectoral or regional location of FDI, with the aim39 of, e.g., creating jobs and boosting exports in their jurisdiction (see Box 4.5).40

374. Jurisdictions differ with respect to their general tax structures and the type of incentives used. Low income jurisdictions (or developing countries), in particular, collect much less tax revenue as a share of GDP compared to higher income jurisdictions, an empirical finding that is often related to a lack of administrative capacity (Besley and Persson, 2014[48]). Given prevailing differences in public infrastructure, human capital and the general business environment, their comparative advantage are typically cost-related, e.g., larger labour cost differentials or lower corporate taxation (Vukšić, 2013[102]). Against this background, high income jurisdictions tend to rely more on R&D-related tax incentives, which are typically narrower in scope and more reliant on administrative capacity, while low income jurisdictions often resort to more generous tax incentives such as, e.g., tax holidays or exemptions, which are broader in scope but require less capacity to administer (IMF-OECD-UN-World Bank, 2015[103]). In some cases, tax incentives are combined with other measures, e.g., policies aimed at creating a favourable business and legal environment, in a Special Economic Zone (SEZ), an approach that is also more prevalent in low- and middle-income jurisdictions (Figure 4.11).

Figure 4.11. Special economic zones (SEZs) by income group

Source: UNCTAD (2019).

Note: Special economic zones does not include logistic hubs.

375. Under certain conditions the new tax rules proposed could serve to limit the ability of government to provide generous tax incentives, including investment tax incentives. In particular, tax provisions under Pillar Two are aimed at ensuring a minimum level of effective taxation. This implies that the maximum level of tax benefits that governments can provide to foreign investors will be limited because other jurisdictions (e.g., in the case of the application of the IIR, the ultimate parent jurisdiction) will be able to apply a top-up tax to the low-taxed profits in order to bring effective tax rates at the jurisdictional level up to the minimum rate. While the minimum tax rate could also become binding in the case of very generous credits and allowances, this effect will be most significant for the most generous tax incentives, such as corporate tax holidays and exemptions, which often reduce CIT liabilities to zero, either indefinitely or for a defined period. Tax provisions introduced under Pillar One are less likely to limit the tax benefits obtained through investment tax incentives since a potential reallocation of tax bases will be determined at the MNE group or segment level and entail the reallocation of only a share of residual profit. However, even after the implementation of Pillar One and Pillar Two, governments will still be able to use tax incentives as before to attract FDI as long as effective tax rates do not fall below the minimum rate at the jurisdictional level.

376. Although the potential limitation in the generosity of tax incentives could reduce the scope of policy instruments available to governments to attract foreign investment, it is unlikely to produce efficiency losses at the jurisdiction or global level. First, the effect of tax incentives on cost differentials across jurisdictions could be smaller than anticipated if tax competition induces an increasing number of jurisdictions to offer incentives, leading to a level effect without creating a comparative advantage for any particular jurisdiction. Second, tax incentives may induce distortions across different investments and may result in windfall gains for projects that would already have taken place in absence of the incentives. Third, competitive pressure is likely to be stronger in developing countries due to the resource requirements associated with the administration, auditing and evaluation of targeted regimes based on a narrower scope, thus further limiting potential resource mobilisation. Fourth, reductions in domestic resources available to governments in developing countries are unlikely to be compensated by increases at the global level if investment is merely relocated to other jurisdictions with similar economic characteristics. There is therefore a risk that tax incentives fail to achieve the stated policy objectives at the jurisdictional level while at the same time leading to a globally inefficient allocation of capital.

377. In line with these arguments, the literature finds only limited evidence of the effectiveness of tax incentives in terms of attracting additional FDI. On the one hand, the existing literature typically does not investigate whether investment tax incentives obtained in a specific jurisdiction lead to an increase in global investment of a given MNE group, e.g., driven by a reduction in the effective tax rate at the group level. On the other hand, even at the jurisdiction level, there is only mixed evidence of their effectiveness in attracting foreign investment (Box 4.5 and Hajkova et al. (2006[4])).

378. However, even if they succeed in attracting investment, tax incentives are not necessarily the most cost-efficient use of public funds dedicated to investment promotion. The net benefits of investment tax incentives, taking into account forgone tax revenue, are often not well understood, due to capacity constraints and/or lack of thorough evaluation. These constraints are of particular relevance for developing countries, where reliable information on revenue forgone is often not available. As tax revenues are a key source of finance to deliver often lacking public goods and services in developing countries, such as education and skills development, health and infrastructure, it is possible that alternative policies41 would deliver superior results in terms of investment, growth and sustainable development.

Figure 4.12. Distribution of statutory CIT rates by income group

Note: For each of the income groups, the boxplot shows the 25th percentile (lower edge of the box), the median (straight line in the box) and the 75th percentile (upper edge of the box); the mean is indicated by a cross and each circle corresponds to an observation.

Source: OECD Corporate Tax Statistics.

379. Irrespective of the cost-efficiency and effectiveness of tax incentives in general, developing countries will have substantial policy space to support foreign investment through their corporate tax systems, even after the introduction of the proposed rules. Figure 4.12 shows that, based on the most recent available data, statutory CIT rates are highest, on average, in lower and lower middle income countries; forward-looking effective average tax rates, taking into account standard components of corporate tax bases, also tend to be higher in developing countries (OECD, 2020[15]). On average, in developing countries both types of rates are generally well above the minimum effective tax thresholds that are currently being debated (Figure 4.12). Developing countries therefore retain considerable policy space for the use of tax incentives, even after the implementation of Pillar One and Two, especially compared to emerging or advanced economies. As with R&D tax incentives discussed above, governments will continue to be able to rely upon investment tax incentives to incentivise the location of FDI in their jurisdiction as long as the resulting effective tax rates do not fall below the minimum threshold. Given that in a number of developing countries current incentive regimes are very generous, e.g. relying mainly on tax holidays and exemptions, an adaptation of investment promotion strategies towards a cost-efficient use of tax incentives could thus be a key priority for developing countries aiming to promote economic development in the context of the evolving international tax system.

380. Taken together, the available evidence suggests that, in the absence of otherwise attractive economic conditions, the cost-efficiency and effectiveness of tax incentives is limited. Moreover, tax incentives could increase tax competition and undermine domestic revenue mobilisation efforts, especially in developing countries. Against this background, the introduction of the new tax rules under Pillar One and Pillar Two could strengthen the bargaining position of developing countries wanting to move away from certain, particularly costly, policy choices that deliver unclear net benefits. A reduced reliance on tax incentives would imply, in turn, that MNE location decisions are more affected by commercial considerations, instead of tax considerations, thereby potentially reinforcing positive effects on the efficiency of international capital allocation, as discussed in the context of R&D tax incentives and, more generally, in the Section 4.4.

4.7.6. Firm competition and investment

381. Recent research suggests that structural economic changes related to digitalisation and globalisation have important effects on firm concentration and price-cost mark-ups at market and industry level. While the debate on these effects is not settled, evidence from the last several decades suggests that, in certain countries and industries, fewer firms account for a higher share of revenues, output and/or employment (see Box 4.6). In particular, several empirical studies suggest that these developments are mostly driven by the highly digitalised sectors of the economy, as network effects give rise to winner-takes-all (or winner-takes-most) dynamics allowing some firms to gain competitive advantages (Bauer and Latzer, 2016[127]) (OECD, 2018[128]) (OECD, 2019[129]). To the extent that these findings are associated with a decrease in the intensity of competition among firms,42 they could imply a reduction in consumer welfare, as well as business investment, innovation and productivity growth (Gutiérrez and Philippon, 2017[130]).

382. The Pillar One and Pillar Two proposals are not targeted to influence competition dynamics among firms, as this is primarily the focus of competition and regulatory policies rather than tax policy. Nevertheless, from a tax perspective, Pillar One and Pillar Two would contribute to a more even playing field among MNEs, and between MNEs and non-MNE firms. In turn, this could contribute to reducing, or at least slowing down the trend towards greater concentration and other potentially adverse effects associated with such concentration.

383. The impact of the proposals differs across firms due to the thresholds and scope restrictions embedded in the new provisions. The design of Pillar One is targeted towards MNEs based on their size, through the global revenue threshold, and their profitability, through the profitability threshold (see Pillar One Blueprint report (OECD, 2020[13])) in addition, its application is potentially limited to businesses providing automated digitalised services (ADS) and consumer facing businesses (CFB). The new provisions under Pillar One, therefore, target the largest and most profitable MNEs - these are the firms that have been most strongly associated with the secular trends towards firm concentration and price-cost mark-ups (Calligaris, Criscuolo and Marcolin (2018[2]); Bailin Rivares et al. (2019[131])). Under Pillar Two, no sector-specific scope restrictions are considered. However, the new provisions introduced to ensure minimum effective taxation will imply that those MNEs engaging in profit shifting to low-tax jurisdictions will experience larger increases in their tax liabilities under Pillar Two, in line with the policy intention to address remaining concerns about base erosion and profit shifting (BEPS). Since larger MNEs tend to engage more in tax planning and related tax savings could be used to crowd out other firms (Sorbe and Johansson, 2017[132]), the expected reduction in profit shifting due to the new tax rules, and Pillar Two in particular, could further counteract the observed trends in concentration.

384. The impacts of Pillars One and Two reinforce each other, possibly contributing to a dampening of the trend towards greater concentration at market and industry level. As large and profitable MNEs, especially those who are engaging in profit shifting, will be affected the most, the new rules are expected to produce a more level playing field among MNEs, and vis-à-vis their smaller and domestic competitors, as well as those firms not engaging in profit shifting. If less concentrated markets and industries are indeed associated with more business investment and innovation, the reform could have additional positive effects on growth through this channel. However, tax policy is only one among many factors affecting concentration dynamics; its quantitative impact also depends on the wider competition and regulatory policy context.

4.8.1. Global economic implications if consensus is reached

386. Taking stock of the findings presented in Sections 4.3 to 4.7, several main insights emerge in relation to the expected investment effects at the group and entity level, as summarised in Figure 4.13. To begin with, it is worth highlighting that the proposals operate at the MNE group level, determining tax bases based on the global performance of the MNE group as a whole rather than on a separate entity basis. Depending on the final policy design, the inclusion of a number of scope restrictions would imply that the effects of Pillar One and Pillar Two on investment costs would be limited to a subset of MNEs. It is currently proposed that the scope of Pillar One is limited to MNE groups that perform ADS and CFB activities, and by a profitability threshold that targets the impact of the rules to more profitable MNEs, and by a revenue threshold that targets larger MNE groups. Pillar Two, on the other hand, is focussed on large MNEs achieving low effective tax rates at the jurisdictional level. Particularly, where a formulaic substance-based carve-out is provided, it is likely that these rules will mainly affect firms engaging in profit shifting.

Government responses and other indirect effects on investment

388. As the discussion in Section 4.7 shows, potential indirect effects on investment are typically context-specific. Furthermore, indirect effects could interact, positively or negatively, with each other, thus either reinforcing or offsetting overall effects on investment at the jurisdiction level; the overall direction of these effects at the jurisdiction level therefore depends mostly on government responses to the evolving international tax system. For example, increased fiscal space could be used to increase support for business innovation or to expand tax administrations with the aim of lowering compliance costs; a more level playing field for domestic and multinational firms could be combined with additional support for young, innovative firms; lower competitive pressure on effective tax rates could facilitate the development of a different investment promotion strategy. A successful adaptation of domestic policies to the changing incentive structure could, therefore, lead to additional positive effects on global investment and economic output.

389. In particular, the analysis highlights several policy areas where the response of individual governments to the changes in the international tax system could have important effects on investment at the jurisdiction and, ultimately, the global level. To start with, investment costs would increase in jurisdictions where firms were subject to ETRs below the minimum threshold prior to the reform, either due to minimum effective taxation under Pillar Two or because low-tax jurisdictions increase their domestic ETRs. As a result, ETR differentials across jurisdictions would decrease, implying lower tax gains from relocation and thus higher relevance of non-tax factors for location decisions. In addition, lower profit shifting incentives would reduce the cost of public funds to the extent that tax administrations need to spend less on enforcement activities. Responding to this structural change, low-tax jurisdictions could decide to adapt their broader policy strategies to improve their attractiveness for foreign investment, over and above pure cost considerations. As discussed in Section 4.7, a similar response could be adopted by countries, including developing countries, where they have previously chosen to grant generous, and often inefficient, tax incentives to attract FDI, but can strengthen domestic resource mobilisation and open up more fiscal space for public investments.

390. Moreover, specific consideration should be given to policy areas where investments yield positive externalities, as tax policies are often used to address concerns about underinvestment in such contexts. While spillover effects from FDI could also yield positive effects on local economies, the main policy area where external effects are expected to be quantitatively relevant is innovation and, in particular, business R&D. Section 4.7 has presented in-depth discussions of various tax and non-tax policies aimed at supporting innovation and their effectiveness based on empirical evidence. This analysis suggests that, while policy strategies may need to be adapted in certain jurisdictions, e.g., to bring preferential regimes in line with recommendations of the OECD/G20 Base Erosion and Profit Shifting project or to shift public support from indirect to direct support measures, a wide array of policy instruments remain available to governments after the implementation of Pillar One and Pillar Two in order to achieve their intended policy objectives.

391. Although the changing incentive structure could thus induce behavioural reactions by MNEs and governments, leading to a more efficient allocation of capital across jurisdictions, and therefore, an increase in global economic output, as summarised in Table 4.2, it is beyond the scope of this overall assessment to quantify these effects. However, in the event that some positive efficiency effects are realised at the global level due to the implementation of Pillar One and Pillar Two, these effects could be expected to, at least partially, offset any negative effects associated with higher investment costs at the MNE group level.

Figure 4.13. Investment effects at MNE group and entity level

MNE responses at the group level and their impacts on global GDP

392. Turning to the analysis of investment costs at the MNE group level, the empirical results show that, on average, there are only small effects of Pillars One and Two on EATRs and EMTRs.45 The global GDP-weighted average change in the EATRs is estimated to be just around 0.35 percentage points; the corresponding change for the EMTR is around 1.40 percentage points, with Pillar Two driving most of this increase. Moreover, the effect on investment costs is larger for projects located in investment hubs, most of which have comparatively low effective tax rates. These results rely on the following illustrative assumptions on the design of Pillar One and Pillar Two. In Pillar One, only Amount A is modelled, assuming a 10% profitability threshold based on profit before tax over turnover, a 20% reallocation percentage to market jurisdictions and a scope restriction to ADS and CFB sectors. In modelling Pillar Two, a 12.5% rate with jurisdictional blending is assumed, including a 10% carve-out on depreciation expense (approximated using the value and location of tangible assets).

393. These results provide a starting point to explore the extent to which an eventual implementation of the Pillar One and Pillar Two proposal could lead to a reduction in the global level of MNE investment. In particular, a quantitative approximation of the overall impact on global output in the medium to long run (i.e. 10 years following implementation) can be derived by combining the empirical results from Sections 4.5 and 4.6 with additional assumptions and modelling techniques. To provide a quantitative approximation of the investment effects that is consistent with the OECD model for long-run projections,46 an estimate of the change in the total business investment rate can be derived - this calculation is illustrated in Figure 4.14.

Figure 4.14. Estimating the change in the total business investment rate

Source: OECD illustration.

394. The calculations are based on the following aggregate figures.

• Global average EMTR change. Assuming a 10% profitability threshold based on profit before tax over turnover, a 20% reallocation percentage to market jurisdictions and a restriction to ADS and CFB for Amount A under Pillar One and a 12.5% rate with jurisdiction blending, including a 10% carve-out on depreciation expense (approximated using the value and location of tangible assets), for Pillar Two, the combined effect on the global GDP-weighted average EMTR,47 measured at the MNE group level, is about 1.4 percentage points. The approximation uses the EMTR, rather than the EATR, to better align with the empirical literature on the tax sensitivity of investment.

• Tax elasticity of investment. The tax sensitivity of investment is based on the industry-level estimates developed by Sorbe and Johansson (2017[7]), as their empirical approach is aligned with the estimates discussed in Section 4.6 while also allowing for a consistent approximation of the long-term effects on investment. Building on this approach, the short-term sensitivity, -0.05, is translated into a long-term sensitivity of -0.12, and adjusted downwards to -0.08 to account for the empirical finding that tax sensitivities are on average around 35% lower among MNE groups with profitability above 10%.48

• Share of MNE investment. The empirical results from Sections 4.5 and 4.6 refer to investment costs and elasticities of MNEs; however, not all business investment comes from MNEs. Since the MNE share in global business investment is not directly observed in the data, it is approximated49 based on the global average MNE share in gross value added (32%), including domestic and foreign MNEs, using data from the Analytical AMNE50 database.

395. Bringing the aggregate figures together, as shown in Figure 4.14, suggests that increasing investment costs would lead to a decline in the total business investment rate of around 0.04 percentage points (1.4*-0.08*0.32=-0.04). Although this approximation accounts for lower tax sensitivities among the MNE groups targeted by the new rules, it corresponds to an upper bound, given that the tax sensitivity does not take into account that group level responses are likely to be smaller due to relocation of investment within the group. Further, this tentative approximation is subject to a number of caveats that have been highlighted in relation to the underlying empirical work discussed throughout this chapter (Millot et al. (2020[44]), Hanappi and González Cabral (2020[12]).

396. Once the change in the total business investment rate has been estimated, it can be used to approximate the GDP change, using a stylised framework consistent with the OECD model for long-run economic projections (see Box 4.7 for a brief description of the modelling approach). Based on this approach the change in the investment rate, -0.04, thus translates into a reduction in GDP of around 0.07% over a medium- to long-term horizon.

397. While the above calculation represents only a simple approximation of the likely effects on GDP, a sensitivity analysis can be conducted to understand how the quantitative result would change with respect to some of the underlying assumptions. Specifically, the same calculation can be done using the average tax sensitivity of investment, -0.12, without adjusting downwards to account for the lower tax sensitivity of highly profitable MNEs. In this case, the change in the investment rate is estimated to be -0.05 (1.4*-0.12*0.32=-0.05), leading to a slightly higher reduction in GDP, i.e., 0.11% over the same horizon. Alternatively, the change in the EATR can be used, instead of the EMTR, to approximate the impacts of the proposals on backward-looking ETRs. Using the change in the global GDP-weighted average EATR, 0.35 percentage points, as a basis for the approximation would result in a significantly less negative effect on GDP, i.e., -0.02% over the same horizon.

Box 4.7. Assumptions to model long-run economic impacts

The impacts of a reduction in the investment rate on GDP can be assessed based on the following framework that can be seen as a stylised and simplified version of the OECD model for long-run economic projections. Production in the economy is assumed to be a Cobb-Douglas function of capital $K$ and labour $L$, with labour share, $1-\alpha$.

$Y_t=A{K}_t^{\alpha }{L}^{1-\alpha }$

In this equation, $A$ represents total factor productivity (TFP). Capital depreciates at a constant rate, $\delta$, so that its evolution is determined by the accumulation equation.

$K_{t+1}=\left(1-\delta \right)K_t+I_{t+1}$

$I_{t+1}$represents investment, which is assumed to be a constant share, $\gamma$, of GDP, $I_{t+1}=\gamma Y_t$. Given that in the steady state the investment rate is fixed to the level of the depreciation rate, $\frac{I}{K}=\delta$, the fall in investment is implemented as a permanent reduction in the investment-to-GDP ratio, $\gamma$. To align the parameters with the OECD long-term model, it is assumed that the capital-to-output ratio equals 1.5, $\frac{K}{Y}=1.5$, implying that a shock of $ϵ$ on the investment rate, $\frac{I}{K}$, as derived in the econometric estimations, can be translated as a permanent reduction in $\gamma$ of $1.5\cdot ϵ$ . The shock is implemented starting from a situation where the economy is in its steady state, with the initial parameters set at $\alpha =0.24$, $\delta =5\%$ (per year) and $\gamma =7.5\%$before the shock.

398. Taken together, the two main channels through which the proposals affect global investment and output are, first, an increase in investment costs at the MNE group level, reducing global investment levels and, second, a reduction in ETR differentials, reducing profit shifting and improving the global capital allocation across jurisdictions. The first effect is negative and a tentative quantification suggests that it could lead to a reduction of global GDP by less than 0.1%. However, there could be a second, countervailing effect on global GDP if capital allocation across jurisdictions becomes more efficient; however, the extent to which this would be the case depends on individual governments’ policy responses to the evolving international tax system as well as on MNE responses along other decision margins, e.g., with respect to their organisational structure. Since these effects are difficult to quantify, only the effect of the first channel has been modelled in this quantitative assessment.

Table 4.1. Consensus scenario: Summary
Direct Effects on Investment	Indirect Effects	Summary
Scope restrictions imply that many MNEs will be unaffected and increases in investment costs are targeted to highly-profitable MNE groups and those engaged in profit shifting	Fiscal space increases for most jurisdictions and lower profit shifting incentives could make it easier to raise additional public funds	Increases in investment costs could induce a small reduction in global investment
Increases in investment costs are expected to be quantitatively small, mostly driven by Pillar Two	Compliance and administration costs could increase, but this potential increase would likely be lower than in a situation without agreement	This could lead to a reduction in global GDP by less than 0.1%, which may be partly or fully offset by positive indirect effects
ETR increases at the entity level may lead to some relocation away from low-tax jurisdictions	Indirect effects are often context-specific and adaptation of policy mixes may be required in certain jurisdictions	Policy adaptations may be needed in certain jurisdictions in response to changes in the international tax system
Entities in more profitable MNE groups are less sensitive to tax increases	Many policy instruments remain available after implementation of Pillar One and Pillar Two to achieve intended policy objectives
Reduction in ETR differentials across jurisdictions is likely to reduce the incentives for profit shifting and could improve the efficiency of the international allocation of capital	Economic efficiency could increase, e.g., if firm competition is strengthened	A consensus-based multilateral solution would lead to a more favourable environment for investment and growth than would likely be the case in the absence of an agreement
Source: OECD illustration.

4.8.2. Global economic implications if no consensus is reached

399. The tax challenges arising from digitalisation were described in detail in the BEPS Action 1 Report (OECD, 2015[139]). Although the report considered several options to address these challenges, no agreement was reached at the time. Subsequently, the Tax Challenges Arising from Digitalisation – Interim Report 2018 noted that changes to the international tax rules were required and that a coherent and concurrent review of two key aspects of the existing framework, the nexus and profit allocation rules, should be undertaken with the view towards a global consensus in 2020 (OECD, 2018[128]).

400. As stated in the Interim Report, some jurisdictions adopted the position that, in the absence of a consensus-based global solution, interim measures to address concerns about the tax challenges arising from digitalisation would likely be implemented on a unilateral basis. Notwithstanding the fact that diverging positions existed on both the merits of and the need for interim measures such as transaction-specific digital services taxes (DSTs), an increasing number of jurisdictions have implemented or are considering the implementation of DSTs. In addition, a number of jurisdictions have introduced other unilateral measures including alternative applications of the permanent establishment (PE) threshold, withholding taxes or specific regimes targeting large MNEs, e.g., diverted profit taxes. While these additional unilateral measures are not included, Table 4.2 presents an overview of jurisdictions that have either legislated or are considering the implementation of a DST or similar measure.

Modelling the economic implications of tax and trade disputes

409. The counterfactual scenarios aim to consider the potential state of the global economy assuming the failure to secure an agreement on international taxation. As mentioned above, such an agreement is assumed to entail a consensus on the design of the proposals as well as a commitment to withdraw any existing DSTs and to refrain from implementing such measures in the future. In contrast, the counterfactual scenarios not only capture the economic impacts of unilateral measures on global investment and output, but also the potential trade-related ramifications of an increasing fragmentation of the international tax system and, in particular, their feedback effects on economic outcomes. To this end, the following analysis builds on a cooperation with the OECD’s Trade and Agriculture Directorate using the OECD METRO Trade Model,55 a computable general equilibrium (CGE) model, capable of simulating economic impacts of taxes as well as tariff retaliation in a consistent framework.

410. The OECD METRO Trade Model is routinely used to study changes in trade policy and other factors (see Annex 4.D and the model documentation (OECD, 2020[147])). It covers 65 countries and regional aggregates and 65 commodities across the global economy, based on external and internal data sources such as the Trade in Value Added56 (TiVA) database, to analyse how production and trade are linked in global value chains and how resources such as labour, capital and natural resources are allocated across economic activities. It allows for the simulation of the economic impacts of a wide range of trade policies, including border tariff rates, non-tariff measures, export restrictions, domestic taxes and support programs.

Box 4.8. Digital services taxes and retaliatory tariffs in the OECD METRO Trade Model

The DST is implemented as an ad valorem tax on intermediate sales from the business services sector to all buyers, assuming that the revenues from digitally delivered services, such as revenues from advertising on a digital platform, are located exclusively in that sector.1

While the DST is not the focus of the analysis, its implementation by the respective country grouping in each scenario anchors the size of the tariff retaliation and the counter-retaliation. To calibrate the DST rate, the French case serves as a benchmark based on the observation that the 3% DST in France is expected to generate revenues of around EUR 450 million in its first year. With that information, and taking the value of intermediate sales of business services in France as the benchmark, a corresponding DST rate on the business services sector can be approximated for either a 3% or a 5%2 rate for each country, however, it is not possible to model revenue or income thresholds, as the model does not distinguish individual firms.

Given this approach, the French DST rate of 3% translates into a 0.12% tax on the entire business services sector – i.e., the DST-equivalent sector-level tax rate. Hence, the implicit assumption is that roughly 1/30 of the intermediate sales of business services correspond to digitally delivered services under the scope of the DST. This DST-equivalent tax rate is applied to intermediate sales of business services in all countries that are simulated to implement a DST, yielding the targeted tax revenues as shown in Annex Table 4.D.1 and Annex 4.D.

The retaliating region, i.e., the United States, implements retaliatory tariffs on targeted sectors to match the DST revenues in countries that implement a DST, times a retaliation factor of either one or five. The size of the (ad valorem) retaliatory tariff rate is thus a function of the DST revenues in the implementing country, the imports into the United States of the targeted sectors in the baseline data and the size of the retaliation factor. Tariffs are simulated on trade in non-service sectors that are known to be typically targeted for tariff action and are symmetrically applied to the same products, which makes the results comparable between countries.

Based on these assumptions, the retaliatory tariffs increase with the DST rate and the retaliation factor, leading to a set of calibrated tariff rates. Counter-retaliatory tariffs are assumed to match US retaliatory tariffs exactly in terms of revenues, without an additional counter-retaliatory mark up (see OECD (2018[148]) and Annex 4.D). All counterfactual tax and tariff rates are assumed to be permanent and calculated on baseline data, i.e., without taking reactions and adjustments into account that will occur once a tax or tariff is put in place.

← 1. It should be noted that the statistical classification of digital services is still subject to considerable discussion, and the choice made in this study follows closely the insights in the Handbook on Measuring Digital Trade, Version 1, OECD, WTO and IMF (2020[204]) and mapping into the GTAP v10 sector aggregation which is used as input in the METRO database. More information is available at https://www.oecd.org/sdd/its/Handbook-on-Measuring-Digital-Trade.htm.

← 2. The 5% rate is also modelled on the basis that a number of jurisdictions have proposed DSTs with a higher rate than France (see Table 4.1 above).

411. Although the DST is not included in the set of standard trade policy measures, it can be incorporated in a stylised way as an ad valorem tax on the share of digitally delivered intermediate sales from the business services sector to other sectors. As described in Box 4.8, its design is calibrated using French DST provisions and revenue estimates to produce an approximation of the DST base. In line with the variation in DSTs across jurisdictions, the DST rate is assumed to be either 3% or 5% (Table 4.2). Retaliatory tariffs are a standard feature of trade modelling that is captured within the existing framework. In the calibration used for this analysis, a range of plausible retaliation factors is considered. At the higher end, a worst-case scenario with a retaliation factor of five is considered, reflecting the broad magnitude of the tariff retaliation that was initially announced in the dispute between France and the United States. At the lower end, a proportional retaliation scenario is considered, broadly in line with the tariff response announced by the USTR on 10 July 2020 (see Annex 4.D for sensitivity analyses).

Figure 4.15. Trade retaliation: sequence of events

Simulation of the different cases follow a three-step sequence of events.

Source: In each of the cases, the number of countries implementing a DST or similar measure varies; the group of countries counter retaliating varies accordingly. The country groups are described in Table 4.3 and Table 4.4.

412. Figure 4.15 outlines the sequence of DST implementation, tariff retaliation and counter-retaliation underlying the counterfactual simulations presented in this section (see Annex 4.D for a description of additional simulation results). To capture a broad range of possible trade developments, as discussed above, two cases are distinguished, each corresponding to a different group of countries implementing a DST. In the case with narrow DST implementation, only those jurisdictions that have already legislated a DST or are subject to USTR Section 301 investigations are assumed to adopt such a policy in the event that the multilateral process does not reach a consensus, i.e., groups 1, 2 and 3 in Table 4.3. In the case57 with broad DST implementation, all jurisdictions in the model except for the United States, the People’s Republic of China and Hong Kong (China) implement a DST or similar measure, i.e., groups 1, 2, 3 and 6 in Table 4.3.58

Table 4.3. Country groupings
The following country groups are used to construct the policy simulations in Table 4.4.
Group	Label	Countries	Share in global GDP
1	DST legislated and/or effective	France	4%
2	Section 301 investigation underway - EU	Austria, Czech Republic, Hungary, Italy, Spain, Belgium, Bulgaria, Croatia , Cyprus59, Denmark, Estonia, Finland, Germany, Greece, Ireland, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Sweden	16%
3	Section 301 investigation underway – non-EU	Brazil, India, Indonesia, Turkey, United Kingdom	12%
4	Retaliating	United States	22%
5	No DST, no retaliation	China, Hong Kong (China)	14%
6	Other (no DST except in the case with broad implementation)	Rest of the World	33%
Note: In the case with narrow DST implementation, only those jurisdictions that have already legislated a DST or are subject to USTR Section 301 investigations are assumed to adopt such a policy in the event that the multilateral process does not reach a consensus, i.e., groups 1, 2 and 3. In the case with broad DST implementation, all jurisdictions in the model except for the United States, the People’s Republic of China and Hong Kong (China) implement a DST or similar measure, i.e., groups 1, 2, 3 and 6. Group 6 includes the following countries: Australia, Canada, Israel, New Zealand, Norway, Philippines, Tunisia, Russia, Kenya, Japan, Korea, Brunei Darussalam, Cambodia, Philippines, Singapore, Viet Nam, Argentina, Chile, Colombia, Peru, Costa Rica, Switzerland, Kazakhstan, Saudi Arabia, Morocco, South Africa, Malaysia, Mexico, Thailand as well as an aggregate category covering the remainder of the global economy.
Source: OECD.

413. In each of the two cases, it is assumed that the unilateral implementation of DSTs or similar measures is followed by the United States imposing retaliatory tariffs on specific goods and sectors in the countries that have implemented DSTs. The targeted sectors are selected based on official announcements and experience from past trade disputes (see Box 4.8 and Annex 4.D). Finally, the countries that have implemented DSTs in the first step of the sequence are assumed to counter-retaliate against the United States in equal measure (OECD, 2018[148]). This modelling assumption is also based on public statements made by French and European Union officials. Again, the targeted sectors are selected based on official announcements and past experience.

Table 4.4. Overview of the simulated cases
Different DST rates and retaliation factors are simulated for both cases. Group numbers refer to country groups in Table 4.3.
Case	DST rate	Tariff Retaliation Factor	Simulations
Narrow DST implementation	3%, 5%	1, 5	DST implemented: group 1,2,3 Tariff retaliation: group 4 Tariff counter-retaliation: group 1,2,3
Broad DST implementation	3%, 5%	1, 5	DST implemented: group 1,2,3,6 Tariff retaliation: group 4 Tariff counter-retaliation: group 1,2,3,6
Note: As sensitivity analysis, additional scenarios have been simulated and are described in Annex 4.D.
Source: OECD