**Part 3: Resources to End Extreme Poverty**

## **Mobilizing Resources for the Poor** †

#### **Kathleen Beegle and Alejandro de la Fuente**

† This book chapter builds on a previously published work: Beegle, Kathleen, and Alejandro de la Fuente. 2019. Mobilizing Resources for the Poor. In *Accelerating Poverty Reduction in Africa*. Edited by Kathleen Beegle and Luc Christiaensen. Washington, DC: The World Bank Group.

#### **1. Africa's Poverty Financing Gap Remains Large**

Beyond shifting development priorities and policies, the agenda to accelerate poverty reduction in Africa requires harnessing more resources. The message about spending more and spending better to address the critical needs for the poor is essential to meet SDG goals. Assessing a country's poverty financing gap requires a sense of the needs of the country's poor, as well as of the country's capacity to mobilize the resources to meet them. This is challenging, conceptually and in terms of data. One metric regularly used to gauge needs is the aggregate poverty gap (APG). It is the monetary value of the gap between the income of the poor and the international poverty line aggregated across the poor population. It gives an estimate of the amount necessary to mechanically lift all the poor out of poverty through redistribution. As such, it provides a first (and imperfect) benchmark.<sup>1</sup>

In 17 out of 45 countries with data, who have over one-third of the poor in Africa, at least 10 percent of GDP (in 2016 prices) would be needed to fill the aggregate poverty gap. All but two (Lesotho and Zambia) of these are low-income countries. For Burundi, the Central African Republic, the Democratic Republic of Congo, Madagascar, Malawi, and Mozambique, the gap requires over 50 percent of the country's GDP. By way of comparison, government tax revenues were only 9 percent on average in Africa's low-income countries. Filling the poverty income gap

<sup>1</sup> One downside is that the APG does not provide a direct estimate of the amount of public investments and support needed to strengthen the earning capacity of the poor today, and of their children in the future (through human capital investment today), nor an estimate of the amount needed to prevent those around the poverty line from falling back. Still, it is a frequently used starting point for considering a country's poverty financing needs and whether it has, in principle, the domestic means to meet them. For applications of this method, see, for example, Chandy et al. (2016), Olinto et al. (2013), Ravallion (2009), and Sumner (2012).

would leave nothing for public good provision, so clearly not a realistic option. Not surprisingly, the APG is 3 percent or less of GDP (in 2016) for most middle-income countries (17 out of 20 countries), with Lesotho, Nigeria, and Zambia being exceptions. In most of the non-low-income countries, the challenge is not so much the amount of resources required by the poor to reach the poverty line, but the decision and effort to redirect resources to the poor to raise incomes.

Using a different, but related metric, closing the poverty gap would also imply an infeasibly high tax rate on the non-poor in many countries (Figure 1). In 22 (mainly middle-income and resource-rich) countries out of the 43 for which there are data, it would imply a rate of less than 10 percent on the income of the non-poor above the poverty line. Due to the depth of poverty and the number of poor, even redistributing the income from a country's billionaires would only have a modest impact on poverty (Chandy et al. 2016). Despite rapid growth in natural resource revenue, for most countries in Africa, natural resource revenue is also not sufficiently large to address the poverty gap, even in theory (Figure 2). Only in five African countries (Angola, Botswana, Gabon, Mauritania and the Republic of Congo) would a direct transfer of 7 percent (or less) of resource revenues fill the poverty gap.

These numbers are indicative that, particularly, Africa's low-income countries are unlikely to have the financial capacity to overcome poverty, and that international financial assistance will continue to be required. Other direct estimates of the cost of making some core social services available or so-called financing of the SDGs confirm the large gap (Development Finance International and Oxfam 2015; Greenhill et al. 2015; Schmidt-Traub 2015).

**Figure 1.** High levels of poverty imply high tax rates on the non-poor to cover need. Source: Authors' calculation.

#### **Poverty rate (US\$ 1.90)**

**Figure 2.** Resource revenues are not sufficient to eliminate the poverty gap. Source: Authors' elaboration with inputs from Nga Thi Viet Nguyen and Rose Mungai. Subset of 23 countries out of 48 with resource revenues and complete data on the level of resources.

#### **2. Fiscal Systems in Africa**

#### *2.1. Revenue and Spending Space*

States get tax revenues directly (e.g., personal and corporate income tax) and indirectly (e.g., value added tax (VAT), excise taxes, and customs duties). Some governments obtain further revenues through grants from donors and international organizations and natural resources, when available. These different revenue sources as well as the ability of governments to manage arrears and borrow, and to draw in private capital for public–private partnerships determine the fiscal space for African governments to spend. There are huge challenges to both raising revenues domestically and increasing other sources of revenue, including from international aid, which is in fact decreasing, or international financial markets, given rising debt levels.

In most of Africa's low-income countries, the domestic revenue imperative remains stark. Most have tax revenues relative to GDP under 13 percent (that is revenues net of grants) the 'tipping point' below which executing basic state functions and sustaining one's development becomes problematic (Figure 3) (Gaspar et al. 2016). For Africa's low-income countries, the average 2013 tax revenue share of GDP was in fact only 9 percent. It was slightly larger for lower–middle-income countries (19 percent). The Organisation for Economic Co-operation and Development (OECD) average in 2015 was 34.3 percent (OECD Organisation for Economic Co-operation and Development).

**Figure 3.** Most African countries have a domestic revenues deficit. Source: de la Fuente et al. (2018) based on the International Centre for Tax and Development (ICTD)/United Nations University—World Institute for Development Economics Research (UNU-WIDER), Government Revenue Dataset, June 2016.<sup>2</sup>

However, a country's level of economic development does not fully predetermine its capacity to raise revenues. Government revenue as a percentage of GDP was more than 20 percent in Mozambique and Zimbabwe, both low-income countries. Lately, there has also been an improvement in domestic revenue collection across Africa. The region experienced the largest increase in tax revenue across the globe since the turn of the century (IMF 2015). As already stated, however, this improvement is beginning from a low level, and, disconcertingly, projections find that the countries with the lowest domestic resource mobilization levels are also expected to grow these revenues at lower rates, further widening the gap (Development Initiatives 2015).

<sup>2</sup> https://www.wider.unu.edu/project/government-revenue-dataset (accessed on 22 September 2021).

Most African countries rely heavily on indirect taxes levied on the sale of goods and services. This includes VAT, trade taxes paid at the port, and excise taxes (such as fuel taxes). VAT, in particular, has led the way to a raise in domestic revenues. Indirect taxes are often also invisible to consumers, and, if kept simple, easier to administer. This makes them a preferred tax instrument in many lower-income countries, where administrative capacity is limited. In addition, informal businesses are widespread in low-income countries; they are generally cash based and hard to tax. Therefore, lower-income countries rely more on indirect taxes than middle-income countries, but this has pernicious consequences on welfare, as Section 2.3 shows.

Direct taxes are the second main source of revenues for African countries. Yet, total revenues from personal income taxes amount to only 2 percent of GDP in sub-Saharan African countries (excluding South Africa). The main direct taxes are personal and corporate income tax. Their contribution as a share of GDP has not been improving either because governments discouraged marginal increases in corporate and personal income taxes. Or simply because income earners avoided complying. Property taxation contributes very little (recorded at 0.1–0.2 percent of GDP, for those countries where reliable information exists) (Moore and Prichard 2017).

Some countries in Africa also generate substantial revenues from natural resources. Out of 37 countries for which data are available, 22 are considered resource rich: from oil-rich countries like Chad and the Republic of Congo to diamonds in Botswana and minerals in Niger or Mauritania. In these countries, revenues from natural resources make up between 10–20 percent of GDP (Figure 3). Tax revenues in developing countries with substantial natural resources tend to be higher than for countries at the same income level that lack such resources. So, in principle, resource revenues can enhance spending on pro-poor sectors such as the social sectors (for example, health and education), agricultural and rural development, as well as social protection programs including cash transfer schemes strengthening the poor's risk management capacity. However, often revenues go directly from extracting companies to governments, without citizen involvement. This weakens the ability of citizens to scrutinize government expenditures. As a result, poverty reduction is slower and multiple human development indicators are worse in resource-rich countries in Africa than in other countries at the same income level (Beegle et al. 2016; de la Brière et al. 2017).

Taken together, the low base on which to tax, the limited capacity to tax more, and the political inability to channel national income from natural resources to pro-poor spending result in a large poverty financing gap. Low-income countries face the greatest needs, have the lowest taxable base and are least efficient in

mobilizing revenues. Financing from foreign donors or international organizations will remain a critical source of funding for many of the poorest African countries in the foreseeable future.

While domestic resources are the largest resource available to African countries in aggregate, aid makes up more than 8 percent of gross national income (GNI) for half the low-income countries in Africa (Figure 4).<sup>3</sup> It is often geared towards pro-poor sectors such as health, agriculture, and education. For example, aid finances three quarters of public health spending in Rwanda (Development Initiatives 2015) and donor funds finance 90 percent of public agricultural spending in Burundi (Pernechele et al. 2018). The sectors of education, health and financial support to the poorest through safet nets account for around a third of all donor aid.

**Figure 4.** ODA is a large share of GDP in low-income countries. Source: OECD database (2017).

Unfortunately, while global ODA has been increasing, reaching an all-time high of USD 140 billion in 2016 (at current prices)—ODA to African countries increased marginally in nominal terms from USD 45.8 billion in 2013 to USD 46.3 billion in 2017 (after a dip to USD 42.5 billion in 2016). In per capita terms, though, it has declined in nominal terms from USD 48.3 to 42.6 given population growth. The decline is, at

<sup>3</sup> We lack estimates of aid inflows from international charities, international nongovernmental organizations (NGOs), and private donations.

least in part, because donor countries were spending more in their own countries on refugees and asylum seekers. Such spending more than doubled in three years, from less than 4 percent of total donor spending before 2013 to 11 percent in 2016. Germany and Italy spent more on in-country costs than they gave in aid to Africa; Norway and Switzerland had increases in in-donor refugee costs and decreases in aid flowing to developing countries. Four Development Assistance Committee (DAC) donors—Greece, Italy, Austria, and Hungary—allocated more than 50 percent of their bilateral assistance in 2016 to in-donor refugee costs. When in-donor refugee costs are excluded, only three countries—Norway, Luxembourg, and Sweden—out of the 29 DAC donor countries reached the United Nations target of 0.7 percent of ODA/GNI in 2016 (Sebany 2017).

The combined resources from domestic revenue and ODA at current levels do not suffice to reach the SDGs goals related to universal education, universal health, and scaled up safety nets in developing countries; billions more are needed (Greenhill et al. 2015; Manuel et al. 2018). The costs for education, health and financial support needed for the poorest in Sub-Saharan Africa totals USD 262 billion (in 2017 prices) (Manuel et al. 2018). Some target getting an extra trillion (Development Finance International and Oxfam 2015). In light of the shortfalls, ODA is increasingly also being used to catalyze private sector investment in developing countries, though the jobs and poverty impact of blended finance needs to be better understood (Sebany 2017). Donors should recommit to the original ODA target of spending 0.7 percent of their national income on development aid overseas and reverse the trend of a declining share of ODA to Africa. In 2015, DAC countries spent 0.3 percent of ODA/GNI globally and 0.1 percent in Africa. If donors met aid targets (0.7 percent of GNI), the financing gap in low- and low–middle-income countries would be met (Greenhill et al. 2015).

Governments could in principle also borrow, domestically and internationally. Yet, many will find it difficult. Lenders may be unfamiliar with small countries who do not normally borrow. Countries that do borrow may have large existing debts and may not be able to raise additional sums.<sup>4</sup> Standard & Poor's has downgraded four African countries since the start of 2017, namely, Gabon, Namibia, the Republic of Congo, and South Africa. Additionally, for those with an International Monetary

<sup>4</sup> As such, between 2010 and 2017, seven countries accounted for over three-fourths of the total African bond debt issued: Angola, Côte d'Ivoire, Ghana, Kenya, Nigeria, South Africa, and Zambia (World Bank 2017).

Fund (IMF) program, there may be additional restrictions related to taking on debt.<sup>5</sup> A few countries are facing repayment problems, for example, Mozambique and the Republic of Congo. Additionally, even those with low debts may find it difficult to borrow when they most need to, due to the move to normalization of the monetary policy in advanced countries, a decrease in other sources of funding, and rising sovereign risks in the region.

#### *2.2. A Mixed Record on Spending on Pro-Poor Sectors in Africa*

Many measures to tackle poverty are embedded in the provision of basic services and direct transfers (for example, schools, clinics, or cash transfers that help to build human capital and manage risks) as well as in the sectoral allocation of public spending towards sectors that are more likely to benefit the poor, such as agriculture. As such, tracking pro-poor spending is usually sectorally focused even though, importantly, within-sector spending choices can also have quite different effects on poverty (Owori 2017).

Five key points emerge. First, while a number of countries are close to meeting or exceeding global targets for pro-poor sectoral spending as a share of GDP or government expenditures, absolute (per capita) spending levels are very low, often with room for expansion through reallocation, for example through a reduction in energy subsidies. Second, within-sector spending is often ill-targeted to the needs of the poor and, implementation is inefficient. Third, as a result of both these factors, many poor still pay for access to basic services critical for human development; out-of-pocket expenditures are high, or lack the public goods needed to increase their earnings (e.g., agricultural innovation and rural infrastructure). Fourth, resource-rich countries spend less on education and health than other African countries of similar income level, and spending is less efficient. Finally, in health and education, as well as agriculture and risk management (humanitarian aid), a large share of funding in many countries comes from donors, questioning government commitment and independence as well as the sustainability of pro-poor spending.<sup>6</sup>

<sup>5</sup> Seventeen countries have an IMF Extended Credit Facility and/or Extended Fund Facility (Benin, Burkina Faso, Cameroon, the Central African Republic, Chad, Cote d'Ivoire, Gabon, Ghana, Guinea, Guinea-Bissau, Madagascar, Mali, Mauritania, Niger, Sao Tome and Principe, Sierra Leone, and Togo). Two countries have IMF Stand-By Agreement and/or Stand-By Credit Facility (Kenya and Rwanda).

<sup>6</sup> For a discussion on exploring how, when, and why poverty can be a priority in the national budget, see Foster et al. (2003) which summarizes five African country case studies that explore this.

Among the social sectors, governments consistently spend more on education (4.3 percent of GDP on average across low- and middle-income countries in Africa), typically followed by health (1.8 percent of GDP), and social safety nets (1.4 percent of GDP) (Figure 5). On average spending is in the target range for education (4–6 percent of GDP per capita or at least 15% to 20% of public expenditure to education) under the Education for All (EFA) initiative. However, spending is below the target for health spending (about 4 percent of GDP per capita since the Abuja Declaration target is 15% of public expenditure to health). Spending on social safety nets is lowest, but much lower in most countries, given there is a concentration of social safety net spending in Southern Africa. This is also well below the share spent on energy subsidies (3.8 percent of GDP on average). Agriculutral spending as a share of GDP is 1.4 on average. Given low levels of GDP per capita, the absolute levels of pro-poor spending per person can be strikingly low, especially in low-income countries. Additionally, there is important heterogeneity across country groupings and sectors. Resource-rich countries, for example, spend less on critical social services (education, health, social safety nets) as a share of their GDP (i.e., given their income level) than their non-resource rich counterparts.

Spending is not usually tracked subnationally, although one could make the case that this should be carried out in many sectors. Some evidence suggests that the poorest places are not getting equal, let alone greater, spending. Recent work, using geo-tagged aid data and data sources as a proxy for poverty (night lights, other remoteness measures, and health outcome estimates) finds that aid specifically is disproportionately going to richer areas (Briggs 2018). Country-level studies often show disparities in public spending suggesting the same. Government health expenditure in the Democratic Republic of Congo were 1.8–3.5 times higher in Kinshasa than in provinces with lower poverty rates, and, though not adjusted for price-level differences, this disparity is reflected in starkly unequal access to service and health outcomes (Barroy et al. 2014). In Ghana, government spending per pupil is higher in regions with lower poverty rates (Abdulai et al. 2018). Even when spending data are not readily available, since the bulk of health and education spending is salaries, disparities in staffing per capita between poor and less-poor areas (which is well documented in many studies) reflects, in large part, overall unequal spending. Unequal investments in social sectors partly explains why geography is one of the strongest predictors of within-country inequality (Beegle et al. 2016).

**Figure 5.** There is diversity in spending, but education dominates. Source: Country average spending as a percent of GDP among low- and middle-income countries in Africa; education, public health, and mililtary from WDI; energy subsidies from IMF (2015); social safety nets from Beegle et al. (2018); agriculture from the SPEED database.

#### *2.3. Are Africa's Fiscal Systems Impoverishing?*

Fiscal systems can have an impact on poverty and inequality, both through the government's overall fiscal situation and through the distributional implications of tax policy and public spending. Many policies can enhance equity. Governments can use taxes and transfers to redistribute income ex post and they can use public spending—through the provision of public goods and services—to reshape the distribution of 'opportunities' and foster mobility within and across generations (Bastagli 2016; Inchauste and Lustig 2017; Lustig 2018).<sup>7</sup>

<sup>7</sup> The provision of quality public goods and services can help individuals increase their stock of assets—for example, in terms of human capital such as education, health, or skills; their financial capital; or their physical capital such as land or machinery, thereby equalizing opportunities. Promoting an environment of investment and innovation can expand access to opportunities as individuals use their capital and labor to generate income—for example, utilizing their skills to participate in the labor market or using their land for agricultural production. Social protection systems—including safety nets, subsidies, and transfers—also act as a mechanism for equity, redistributing resources to the most vulnerable.

One increasingly used tool to assess who bears the burden and benefit from the different instruments upon which domestic resource mobilization and government spending depend, is Fiscal Incidence Analysis (FIA). A summary and expansion of the FIA tool applied to 11 African countries through Commitment to Equity (CEQ) Assessments shows that many fiscal systems in the region are at best neutral in terms of poverty impacts or, at worst, sometimes, poverty increasing (de la Fuente et al. 2018). South Africa and Namibia are exceptions, as the fiscal systems of these two countries deliver significant additions to income through direct transfer spending (Figure 6). Yet, even when the poverty rate is unchanged or has fallen like in Namibia and South Africa, African fiscal systems may still create burdens for some poor and vulnerable households. That is, some poor and vulnerable individuals may end up paying more in taxes than they receive in transfers—a phenomenon known as 'fiscal impoverishment' (FI) (Higgins and Lustig 2016).<sup>8</sup> The FI index summarizes the number of poor<sup>9</sup> individuals who are estimated to have experienced net losses from fiscal policy (i.e., they have paid more into the fiscal system in taxes than they are estimated to have received from it as benefits). The FI index is expressed as a rate among either the overall population or the poor population. When FI is stated in terms of the latter, it demonstrates how well the fiscal system did at protecting poor and vulnerable households from experiencing losses. The proportion of poor households who are disadvantaged by the fiscal system can exceed 80 percent in countries that deliver very few cash benefits directly like Comoros, Ghana, Mali, Togo, Uganda, and Zambia (Figure 7). This does not correct, however, for the proportion of poor households that are net beneficiaries of the fiscal system and escape poverty as a result.

<sup>8</sup> Note that this holds in the aggregate, as those who benefit and those who pay may not be the same poor or vulnerable individuals.

<sup>9</sup> The FI index estimates the net losses experienced by those who are "post-fisc" poor, or those would be classified as poor given their CEQ Consumable Income levels. The Fiscal Gains to the Poor (FGP) index, meanwhile, estimates the net gains experienced by those who are "pre-fisc" poor, or those who would be classified as poor given their CEQ Market Income levels.

0% 10% 20% 30%

Ghana (2013)

Mali (2014)

Namibia (2009/2010)

Senegal (2011)

South Africa (2010)

Togo (2015)

Uganda (2013)

40% 50% 60% 70% 80% 90% 100%

Zambia (2015)

**FI Index**

**Figure 6.** Fiscal Policy in Africa Frequently Increases Poverty. Source: de la Fuente et al. (2018).

Ethiopia (2011)

5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55%

Comoros (2014)

Ghana (2013)

Mali (2014)

Namibia (2009/2010)

Senegal (2011) – per...

South Africa (2010)

Tanzania (2011)

Togo (2015)

Uganda (2013)

Zambia (2015)

**Figure 7.** Fiscal systems create net losses for the poor even when incidences of poverty are reduced. Source: de la Fuente et al. (2018).

Underpinning these patterns are three proximate causes or drivers of this FI in Africa. First, there is heavy reliance on consumption taxes like VAT to raise revenues and compensate the low levels of taxes collected from other sources, including corporate, income, and property taxes. Second, some governments spend large amounts on energy subsidies which fail to reach most poor households and agricultural subsidies which have low returns compared to other agricultural investments. Third, social protection systems provide only limited targeted, direct transfers to the poor, either because few households are covered, or transfer amounts are relatively low or both. For these reasons, it is further anticipated that the group of poor people who escape poverty by virtue of being net fiscal receivers is also small.

Note also that the FI index and the discussion directly below refers to reductions in the cash-based financial position or in the purchasing power of individuals. The FI index does not attempt to include the benefits provided by the provision of in-kind benefits like education, health, or infrastructure services as in-kind benefits cannot be "eaten"; i.e., they neither increase nor decrease purchasing power over other goods and services.

Direct taxes create very small burdens for the bottom 40 percent while indirect taxes paid by the bottom 40 percent often represent 10 percent or more of pre-fiscal income (Figure 8). Subsidies—even when they are extensive—provide little benefit to poor and vulnerable households who often do not access the subsidized services as much as the wealthy (such as electricity and transport fuel). Direct transfers provide compensation (for the bottom 40 percent) equal or greater than taxes paid only in South Africa and Namibia.

In the aggregate, the total cash benefit transferred to the poorest 40 percent of the population through subsidies and direct transfer programs is smaller in absolute magnitude than the burden created (for the same population) by direct and indirect tax instruments. In other words, most individuals in the bottom 40—including most poor individuals—can expect to be net payers instead of net recipients.<sup>10</sup>

Even if the fiscal system makes a portion of the poor net payers, one could argue that this would be fine as long as this may be the only way to finance strongly progressive and extensive public expenditure on sectors that benefit the poor such as education and health. However, is this the case for Africa? It is not clear that the

<sup>10</sup> To repeat: we refer here to the cash-based financial position purchasing power of individuals and we are not including the value of in-kind benefits like education, health, or infrastructure services.

poor benefit from in-kind spending in education and health as much as they could, given the problems with the quality of the services received.

It is important to note that a limitation of the fiscal incidence analyses reported here is that it does not account for infrastructure spending which in some countries may benefit the poor in terms of higher quality of life and/or more access to markets.

**Figure 8.** Indirect taxes paid outweigh subsidy and transfer benefits for the bottom 40 percent of national populations. Note: Direct transfers and subsidies represent 104 percent and 7.0 percent (respectively) of market income plus pensions in the bottom 40 percent of South Africans (in 2010). The vertical axis in this figure is truncated at 40 percent so that fiscal systems in the non-South African countries are more clearly comparable. Source: de la Fuente et al. (2018).

#### **3. Mobilizing More and Less Harming Revenues**

#### *3.1. Heavy Reliance on Indirect Taxes and Unreliable Direct Taxes*

As the previous section demonstrated, how taxes are raised matters to poverty as much as the amount raised, with the Bottom 40 often significantly affected by indirect taxation. VAT is preferable for efficiency and effectiveness reasons, but it can hurt the poor. Tax exemptions on goods and services primarily consumed by the poor provide a way to mitigate the negative effects. Yet, such goods and services are

few and far between. Additionally, often the goods and services exempted from VAT are consumed in significant quantities by the non-poor, implying important foregone tax revenues. Furthermore, the revenues raised through VAT and other indirect taxes will need to be properly channeled to the poor or vulnerable so that they become net receivers of the fiscal system. Targeted cash transfers provide are an alternative way to compensate the poor. However, the amount of resources that are dedicated to cash transfers is often insufficient (both because of insufficient coverage and low levels of transfers) and needs to be weighed against other competing needs (spending on education, health, WASH, infrastructure, security, and so on). Section 4 sketches some ideas to inform those decisions.

Direct taxes, on the other hand, tend to be progressive because richer people more often have formal jobs. However, economists are quick to diagnose that direct taxes can affect efficiency and long-run growth—by disincentivizing investment, human capital acquisition, and innovation. Yet, the evidence suggests that for low-income countries, shifting away from consumption taxes (e.g., VAT) in favor of income taxes appears to have no negative effect on growth (McNabb and LeMay-Boucher 2014).

More importantly, the small formal sector in many African countries limits the scope for collecting more revenues through direct taxation. Personal income tax is generally limited in economies with large informal sectors as there are few formal employers. Nevertheless, there is room for direct taxation of a wider base of taxpayers, including from the informal sector. Inducing tax compliance also fosters good governance more widely; it comes along with a demand for state institutions that are more responsive, accountable, and competent.

Taxpayer noncompliance is a continual and growing global problem, but studies suggest that developing countries, many of them in Africa, are the hardest hit (Cobham 2005; Fuest and Riedel 2009). Part of the reason is that it often does not seem to pay to taxes. If taxpayers perceive that they do not obtain corresponding benefits from government collectors, tax compliance decrease (Junquera-Varela et al. 2017; Ali et al. 2014; Mawejje and Okumu 2016).

#### *3.2. Taxing the Rich*

In many African countries, the numbers of wealthy are growing fast (McCluskey 2016), as are the prices of real estate—one of the major assets held by the rich. Yet, many rich people pay relatively low taxes on their assets and incomes/enterprises. In Ghana, income tax revenue could have been higher by 22 percent (equivalent to 0.5 percent of GDP) if everyone who filed income tax in 2014 had paid full amounts of income tax due (Asiedu et al. 2017). Wealthy individuals often have significant

investments in local land and property and underdeclare their income from such activities. Out of 71 high-ranking Ugandan government officials owing large domestic business assets (like hotels and schools), only one had ever paid personal income taxes between 2011 and 2016 (Kangave et al. 2016).

The barrier to collecting more property taxes may be largely political, but some technical measures can also widen the base for these taxes. Recent experiences in Sierra Leone point to at least three options for improvement.<sup>11</sup> First, simplified valuation methods that rely primarily on observable features of properties (as opposed to sophisticated, often imported, information technology systems). Second, transferring the responsibility for valuation and property tax collection away from central tax agencies through hands-on and continuous training of local staff (instead of high cost, but short term, training programs). Additionally, third, long-term partnership at the local level is needed and should include continuous support to, and pressure on, political leaders when they have inevitably confronted political resistance (Jibao and Prichard 2016). Relatedly, concentrating the responsibility of collecting property taxes into those with stronger incentives to collect revenue can yield great results. In Lagos, Nigeria the local government undertook an overhaul of governance and property taxation since the early 2000s with the determination of Lagos' leaders to realize their 'mega-city ambitions', in part to attract increased investment (Goodfellow and Owen 2018).

#### *3.3. Corporations and Cross-Country Competition*

Without overlooking domestic policies and revenue sources, additional revenues could further be raised from multinationals. A large portion of the tax bill of multinationals is domestic (through levies, payroll taxes and import taxes). However, multinational companies can minimize their tax bill on profits through transfer mispricing. Simply put this takes place when a company can appear to lose money—or to make very little profit—in the country it is operating in, while making money in secrecy jurisdictions—trading with a subsidiary—where there is no real production and sales activity going on, and remarkably low-tax or no tax applied.<sup>12</sup> Trading goods that are mispriced to avoid tariffs is not illegal, but there

<sup>11</sup> Property tax collection increased at least threefold from 2007 to 2011 in nominal terms, or at least doubling in real terms in the four mid-sized city councils of Bo, Kenema, Makeni, and Koidu-New Sembehun.

<sup>12</sup> First, a corporation working in a developing country sets up a subsidiary in a tax haven. Second, they sell their product at an artificially low price to this subsidiary—enabling them to declare minimal

is widespread agreement that multinationals should refrain from this type of tax minimizing behavior.

However, evidence shows that multinational companies do give in to this temptation. A recent study using confidential tax return data of South African firms with connections to tax havens with no corporate tax, report 47 percent lower profits and have a 7 percent higher likelihood of reporting a loss (Reynolds and Wier 2016). The size of these responses is roughly twice as large as what have been observed in developed countries. This supports the commonly held view that multinational firms operating in developing countries are more aggressive in their tax planning.

When adding it all together—aggressive tax planning by multinationals, high reliance on the corporate tax and increasingly lower corporate tax rates, increased exposure to multinational activity, and increased complexity in multinational corporate activity—the future does look dire for African corporate tax revenues.

#### *3.4. Tapping Mining Income*

For some countries, a major cause of revenue losses is related to revenues generated in extractive industries. Natural resources as a prominent source of government revenues remains relevant despite recent downturns, given the prospects of new mineral resource discoveries and the eventual bounce back of falling commodity prices (Roe and Dodd 2017). At the same time, there is now a shared consensus that government revenues from extractive industries are far too small.

According to the IMF, the effective tax rate in mining is typically 45–65 percent of export value (cited in Africa Progress Panel 2013). In 2010–11, Sierra Leone, Ghana, and Zambia received only between 2 percent and 12 percent from natural resource taxation and royalties (Christian Aid and Tax Justice Network Africa 2014). A conservative estimate of the losses in concession trading in the Democratic Republic of Congo in copper and cobalt mining found USD 1.36 billion in losses for 2010–2012, compared to the budget in health and education of USD 698 million (Africa Progress Panel 2013).

The failure of African countries to capture income from the extractives sector is driven by a mix of factors. These include overly generous tax incentives and tax

profits and consequently pay very little tax to the government of the developing country. Thirdly, their subsidiary in the tax haven sells the product at the market price—for comparatively huge profits coupled with a low tax rate (or none at all). In other words, corporations are manipulating prices to pay minimal taxes. See blog post of 7 March 2014, on the "Africa at LSE" blog (http: //blogs.lse.ac.uk/africaatlse/2014/03/07/tax-evasion-the-main-cause-of-global-poverty/) (accessed on 22 September 2021).

dodging, as well as weak tax revenue authorities and the corruption of elites. In some cases, governments give generous tax concessions to extractive companies which undercut its own revenue code and the government lacks the capacity or will to properly track what the industries should be paying (as an example see the discussion on Liberia in Sustainable Development Institute 2014). African state companies in the extractive sector lack transparence and the problem is compounded by the 'global governance deficit' in some international extractive companies that are major investors in Africa (Africa Progress Panel 2013). Levying appropriate royalty payments and corporate taxes from private companies has helped countries like Ghana and Zambia to raise more revenues in the recent past.<sup>13</sup>

#### **4. Towards Better Spending for the Poor**

The fiscal agenda to reducing poverty in Africa is not only about greater revenues and spending more. Improving the efficiency and equity of that spending, to be more impactful for poor and vulnerable households is equally critical. This means getting more for each dollar spent, but also spending more in the sectors and sub-sectors as well as the places that improve the lives of the poor more effectively within the given budget.<sup>14</sup>

#### *4.1. Overspending on Subsidies*

Consumer price subsidies are one way to 'pay back' consumers some of their taxes. They are almost always regressive: those with assets or services to subsidize are generally better off than the poorer segments that often pay indirect taxes that pay for the subsidies. For instance, less than 15 percent of kerosene subsidies in the region are received by the bottom 20 percent—the fuel type most used by the poor (3 percent in the case of liquified petroleum gas and gasoline). For African countries, on average, providing USD 1 to the poorest 40 percent of households through untargeted gasoline subsidies is accompanied by spending USD 23 to the top 60 percent of households (Coady et al. 2015). Two-thirds of global poverty in 2012 based on USD 2.50 per day would have been covered with redistribution of national fossil-fuel subsidies to the poor (Sumner 2016). Subsidies are, hence, a very

<sup>13</sup> See Natural Resource Governance Institute (2014) for more discussion on taxing and revenue collection from natural resources.

<sup>14</sup> This also relates to finding the right sources of financing, including crowding in private sector finance and public-private partnerships (often in infrastructure), to enable governments to allocate more resources to pro-poor investments.

inefficient way of increasing the consumption of the poorest households. Replacing energy subsidies with a basic income guarantee could both save money and have health and environmental benefits (Coady et al. 2017; IMF 2017).

Within agriculture, farm input subsidies were almost phased out in the 1990s, during a period of structural adjustment in Africa, but they have made a strong comeback due partly to residual support for subsidies among African leaders and partly to the uncertainties about food supply during the 2007/2008 global food and fertilizer price instability. Ten African governments spend roughly USD 1.2 billion annually on input subsidies alone, primarily on fertilizers (Goyal and Nash 2017). In principle, farm input subsidies could make a dent on poverty by making available key inputs to a large population of poor farmers and potentially raise their productivity thereby promoting household and national food security and enhance rural incomes.<sup>15</sup> However, have farming input subsidies delivered?

The existing body of research shows modest impact of fertilizer subsidy programs on yields and overall production;<sup>16</sup> this, in turn, attenuates the subsidy programs' contribution on retail food prices or poverty reduction (On poverty: see Ricker-Gilbert 2016 for a review in Malawi; Mason and Smale 2013, Mason and Tembo 2015 in Zambia; and Jayne et al. 2016 for Africa). This lack of impact of input subsidies on productivity and poverty gets magnified because countries in Africa do not spend much on agriculture. Farm input subsidy programs have crowded out other complementary public investments that have proven more efficient drivers of agricultural productivity growth. Take the cases of Malawi and Zambia—two of the largest spenders on agriculture in the region: in 2014, the budget allocation to fertilizer and seed subsidies was over 40 percent of the total budget to the Ministry of Agriculture (Goyal and Nash 2017).

<sup>15</sup> Farm input subsidies, particularly on inorganic fertilizer, have been justified on the basis that soil nutrients, particularly nitrogen, are essential for maize production, and that most smallholders lack the cash resources or access to credit that would enable them to purchase inorganic fertilizer at commercial market prices.

<sup>16</sup> There is no one-size-fits-all rule for deciding what is an optimal response rate; but in Malawi and Zambia—the two countries in Africa where input subsidies are the largest relative to agricultural spending—estimates suggest modest returns to fertilizer use at best. Burke et al. (2012) found that, on average, the response of maize is 2.7 kilograms (kg) of grain per kg of subsidized fertilizer acquired by households, which is only 50 percent of the Government's expected maize-fertilizer response rate of 5 kilograms. In Zambia, participation raises maize production by 1.88 kg of maize per kg of fertilizer, which is considerably smaller than similar application in other countries, like Kenya, where participation in a similar scheme NAAIAP raises maize production by 361 kg on average, other factors constant (Mason et al. 2016).

Removing subsidies and shifting that spending to public goods and services could improve efficiency and possibly equity. Such reform creates winners and losers and thus brings political pressures to the government. Vested interests and populist pressures exist in all countries. Transport leaders, mining companies, and politically connected firms will want to hold on to energy subsidies, for example, to maintain the preferential treatment in their business as well as to raise barriers to entry for newcomers. The political economy of agricultural subsidies is no less real. Political influence concentration is associated with more subsidies (Figure 9). Nonetheless, some countries have managed to remove subsidies (Inchauste and Victor 2017). To address the politics of reform, it may be necessary to compensate affected groups to preempt opposition. Such compensations may not be cost-efficient, but failing to compensate them (for instance, in the Dominican Republic, transporters and middle classes for removing the fuel and electricity subsidies) could have stopped the reform from passing altogether. Secondly, consumers need to see what they get in exchange for rising prices if the process is to be sustained. Strong communication on the need for price liberalization and trust in the ability of government to handle competing interests is important to sustain price increases.

**Figure 9.** Greater concentration of political influence can result in more subsidies. Source: Bolch et al. (2017). Note: The index is measured by an index of how many individuals at the bottom of the income distribution (the potential winners from more redistributive policies starting from the poorest) would need to come together to outweigh the opposition from the top of the income distribution by accounting for the wealth owned by those individuals.

When, and if, subsidies are scaled back, it needs to happen with a scaling-up of social protection systems. Redistribution has been shown to significantly increase the odds that reforms will succeed. A review of reforms in the Middle East and North Africa classifies all reforms that are combined with cash and in-kind transfers as successful, as opposed to only 17 percent of those without such transfers (Sdralevich et al. 2014). However, greater revenues for government do not 'automatically' lead to higher allocations for safety net programs as Ministries of Finance come under many competing demands to reallocate the savings. A concerted effort from civil society or from external financiers to ensure that as part of the subsidy reform, safety nets are funded adequately, is vital. In recent years, IMF has suggested introducing or expanding social protection programs to compensate vulnerable households during price subsidy reforms (Feltenstein 2017). Equally useful, politicians could earmark part of those savings to build credible commitments to carry out the reform as intended.

#### *4.2. Boosting Pro-Poor Spending within Sectors*

Certainly, increased government spending on sectors that are critical for the poor—such as agriculture, WASH, education, health, and safety net systems is part of the solution. However, at the same time, current spending could be made more impactful for the poor. In two dimensions, the spending in these sectors underperforms for the poor: in terms of within-sector allocations and in terms of the productivity of spending.

Within-sector spending is not neutral with regard to the poor and non-poor. For example, in education inequality in public sector spending in Africa is common and means that children from wealthier households benefit more from public resources allocated to education. This results from two channels. First, children from poor households are less likely to attend post-primary schools for which per pupil spending is higher (Darvas et al. 2017). Second, within school levels, more public resources go to schools in wealthier areas (often urban) (Bashir et al. 2018). This is, in some cases, due to horizontal imbalances in funding resulting from decentralization of service delivery. Partly, this reflects the fact that teacher salaries are by far the largest category of public expenditures on schooling. The distribution of teachers, especially trained and experienced, is biased toward urban schools leaving rural schools with higher pupil/teacher ratios. Additionally, urban public schools have better infrastructure and learning materials.

In health, government expenditures are skewed toward tertiary services. In the Democratic Republic of Congo, 87 percent of government health expenditure were

focused on hospitals, used disproportionately by the wealthy (Barroy et al. 2014). The unequitable spending relates to both staffing and non-staff costs. Again, in the Democratic Republic of Congo, the modest operating budget almost entirely goes to hospitals. Though hospitals can presumably help people avoid large health costs and income shocks, evidence suggests this spending is off target from a poverty perspective. Capital investments in both education and health services need to be rebalanced toward primary education and care, which are usually more cost-effective. Public investments in curative care are especially regressive, driven by the lower use of such services by the poor (Castro-Leal et al. 2000). Lower usage is attributed to several factors, including the perceptions of poor households about illness as well as low access and quality of services for poor househoilds.

Spending more on services that are needed and utilized more by the poor, does not necessarily imply it is effective. The effectiveness of spending is as important as its magnitude; but the quality of public schooling, health care, and other service provision is generally low, even when adjusted for spending levels. A handful of African countries are relatively efficient with respect to early grade education and are also managing to improve their efficiency (Figure 10 upper-right quadrant). The bottom-left quadrant of the figure shows these countries have a current high level of inefficiency and the index has deteriorated over time. Beyond primary, there are large inefficiencies in spending in secondary education in Africa. These are largest in low-income countries where the consequences are arguably greatest in terms of poverty reduction (Grigoli 2015). Globally, health care systems in Africa are the least efficient and this is also the region with the neediest people (Sun et al. 2017).

In agriculture, ample evidence shows that rebalancing the composition of public agricultural spending in Africa could reap massive payoffs for reduicng poverty and increasing agricultural productivity. While studies often show low returns to spending in the sector, specific types of spending (such as investments in core public goods related to R&D, technology generation and diffusion, and market linkages) yield high returns for productivity. The inevitable conclusion is that choices about how to allocate public agricultural spending matter significantly (see the detailed discussion in Goyal and Nash 2017).

There is no single solution to mis-targeted resources and poor quality of services. A number of approaches can be identified. Improved financial accountability is one avenue in health (CMI 2006) and in education (Hubbard 2007). There are a range of other avenues to improve pro-poor investments, such as better financial management, results-based financing approaches, private provision, decentralization, better inputs and support to civil servants, and information/social accountability. Many of these

have been detailed in other reports (for example, see the discussion in de la Brière et al. 2017). Technology can serve an important role (see Technology Spread).

How best to improve efficiency in spending remains an exigent space for further experimentation and learning.

**Figure 10.** Internal Efficiency in Education Remains a Challenge. Source: Bashir et al. (2018). Note: African countries in red. Figure plots the current value of the internal inefficiency index and its improvement over a 35-year period. Positive values indicate less inefficiency.

**Author Contributions:** All authors have contributed equally to the manuscript. All authors have read and agreed to the published version of the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **Development Cooperation, Growth and Poverty Reduction: A Survey of the Evidence**

**Rainer Thiele**

#### **1. Introduction**

Over the first decades of its existence, the modern system of development cooperation, which was established in the 1960s under the auspices of the Organization for Economic Cooperation and Development (OECD) Development Assistance Committee (DAC), was mainly concerned with achieving higher economic growth in developing countries. The intellectual underpinning came from the well-known two-gap model (Chenery and Strout 1966), which assigned to donors the role of providing external resources in order to overcome developing countries' savings and foreign exchange gaps. The structural adjustment programs that dominated the development discourse of the 1980s also aimed at putting recipient countries on a higher growth path by implementing macroeconomic and structural reforms.

Only in the 1990s did a consensus emerge that poverty reduction should be regarded as the ultimate objective of development cooperation. While this shift was associated with an increased use of instruments directly targeting the poor, such as cash transfers, it still left ample room for growth-promoting activities, because growth has been shown to lead to lower poverty in most instances. The literature on pro-poor growth (e.g., (Ravallion and Chen 2003)) added considerations of inequality to the agenda, either stating that the reduction of excessive inequality is a goal in itself or stressing that lower inequality raises the poverty-reducing potential of economic growth. Finally, with the adoption of the Millennium Development Goals (MDGs) in September 2000, helping improve non-monetary poverty indicators such as child and maternal mortality became a new priority among donors, prompting a significant shift in aid allocations from production sectors and physical infrastructure to social infrastructure.

The present paper provides an overview of the empirical evidence regarding the impact of international development cooperation on economic growth, (monetary and non-monetary) poverty and inequality in order to assess whether donors have directly or indirectly contributed to achieving internationally agreed poverty reduction targets. This might also give an indication of what to expect from development cooperation when it comes to moving towards the very ambitious Sustainable Development Goal (SDG) of completely eliminating absolute poverty.

#### **2. Aid and Growth**

This section first looks at the relationship between development cooperation and economic growth in recipient countries, which is arguably the most controversially debated topic of the aid effectiveness literature. It then highlights donors' recent efforts to help recipients raise exports and inflows of foreign direct investment (FDI), which can be regarded as one particular mechanism through which foreign aid might spur recipient growth.

#### *2.1. Direct Growth E*ff*ects*

The vast existing empirical literature on the effectiveness of foreign aid in raising economic growth and development in recipient countries has so far yielded ambiguous results (e.g., (Qian 2015)). In an influential set of meta-analyses surveying the aid effectiveness literature, Doucouliagos and Paldam (2009) conclude that aid has failed to significantly improve economic growth. In contrast, reviewing a range of empirical aid–growth estimates published since 2008, Arndt et al. (2016) conclude that the large majority of studies have found positive impacts, particularly when effects are assessed over longer time periods. An important strand of the aid–growth literature argues that foreign aid can only be expected to be growth-enhancing under specific conditions. Most notably, Burnside and Dollar (2000) suggested—and confirmed empirically—that donors could contribute to economic growth in developing countries only if they focused their engagement on recipients with reasonable levels of governance. This finding shaped policies by the World Bank and other donors for quite some time, even though Easterly et al. (2004) showed early on that Burnside and Dollar's estimates are not robust in the presence of minor changes such as the use of an updated and enlarged dataset. Even studies that find a positive growth effect of foreign aid generally point to moderate magnitudes. Clemens et al. (2012), for example, estimate that raising economic growth by one percentage point per year in the average recipient country would require aid in the order of 10 percent of GDP.

The observation of results varying strongly across different studies can be due to a number of reasons. One key difficulty is that much of the existing literature examines aggregate foreign aid, which comprises a set of very heterogeneous components. While some components such as emergency assistance are not at all expected to

affect economic growth rates, others may do so through widely differing mechanisms (Qian 2015). Cash transfers to the poor, for instance, are likely to have immediate and direct income effects, whereas the process that eventually leads to income effects of projects supporting women empowerment is much more complex and time-consuming. The aggregation of aid also increases the difficulty of developing credible strategies for the identification of causal effects. It has been almost impossible to find an exogenous source of variation that fulfils the exclusion restriction, i.e., does not affect growth through channels other than aid.

The few studies that carefully address the endogeneity of aid have not been able to resolve the empirical ambiguity. Dreher and Langlotz (2020), for example, instrument foreign assistance with a shift-share instrument along the lines of Nunn and Qian (2014), interacting donor-government fractionalization and the probability of each recipient country to receive aid from a particular donor. Their results suggest that the effect of foreign aid on economic growth is insignificant. Galiani et al. (2017) apply an approach that resembles the quasi-experimental regression discontinuity design. They exploit the fact that some of the poor countries receiving aid from the World Bank's International Development Association (IDA) crossed the income threshold over the period under consideration and thus became ineligible for IDA grants. Their identifying assumption of this approach is that countries above and below the threshold only differ in that those above receive less aid. Consequently, the authors employ as an instrument for aid, whether a country is above or below the threshold. They find that foreign aid increases growth. Specifically, according to Galiani et al. (2017), "a one percentage point increase in the aid to GNI ratio from the sample mean is shown to raise annual real per capita growth in gross domestic product by approximately 0.35 percentage points", which is a sizeable but still moderate effect.

#### *2.2. Impact on Foreign Trade*

Trade is one specific area in which donors have pledged to commit additional resources, especially since the aid-for-trade initiative was launched at the Hong Kong Ministerial Meeting of the World Trade Organisation in 2005. The reason this area has received particular attention is that trade and trade liberalisation can make a substantial contribution to economic growth and poverty reduction (e.g., (Winters et al. 2004)). However, a range of factors may prevent low-income countries in particular from taking advantage of trade opportunities. Among the obstacles are trade restrictions adopted by industrialised countries and the developing countries themselves as well as structural weaknesses on the supply side such as low levels of human capital and an insufficiently developed infrastructure.

By tackling such supply-side bottlenecks, aid for trade holds the potential to foster exports from developing countries. This is not to deny, however, that the donors may provide foreign aid mainly to support their own exports to aid-recipient countries (e.g., (Hoeffler and Outram 2011; Nowak-Lehmann et al. 2009)). Hühne et al. (2014) integrate the recipient and donor perspectives in a nested gravity model where they test for differences in the effects of aid for trade on the trade flows in opposite directions. According to their empirical estimations, aid for trade promotes trade in both directions, with moderate quantitative impacts: A doubling of aid for trade means that exports from recipient to donor countries increase by about five percent, while imports by recipients from donors increase by about three percent. Hence, the results do not support the skeptical view that donors grant aid for trade primarily to promote their own export interests.

The results obtained by Hühne et al. (2014) point to "important limitations in the effectiveness of aid for trade. Strikingly, the significantly positive effects on recipient exports do not hold for the low-income group of recipient countries. Aid for trade rather [appears to promote] the exports of middle-income countries, most of which are probably less dependent on aid to overcome supply constraints." Likewise, as Hühne et al. (2014) show, aid for trade turns out to be "more effective in promoting the exports of countries in East Asia and Latin America than the exports of countries in Sub-Saharan Africa, even though the need for support appears to be most pressing in large parts of Sub-Saharan Africa".

Several studies (e.g., (Calì and te Velde 2011; Helble et al. 2012; Hühne et al. 2014) show that the impact on trade is positive for all three component parts of aid for trade, namely aid for 'economic infrastructure', aid for 'productive capacity', and assistance in 'trade policy and regulations'. As one might expect, the third and most directly trade-related component of aid for trade exhibits the strongest trade-enhancing effect. For policy makers, it could thus pay off to put a stronger emphasis on support in the area of trade policy and regulations. This subcategory is so far fairly small and includes, for example, assistance in trade negotiations and technical support for meeting sanitary standards, which could turn out to be particularly beneficial for poor countries with weak administrative capacities.

#### *2.3. Impact on Foreign Direct Investment*

FDI has the potential to transfer technology, provide well paid employment opportunities, and promote economic growth and reduce poverty in developing countries, but it has remained strongly concentrated in a few large and relatively advanced emerging economies (e.g., (Nunnenkamp and Thiele 2013)). Hence, the challenge of spreading the potential benefits of FDI across developing countries as envisioned in the Monterrey Consensus of the United Nations persists.

There are several channels through which foreign aid might foster FDI in developing countries. Aid can, for example, be expected to increase the productivity of private investment if it improves complementary factors such as economic and social infrastructure (Selaya and Sunesen 2012) or the regulatory environment. The small existing empirical literature on the relationship between aggregate aid and FDI is inconclusive. According to the pioneering cross-country study by Harms and Lutz (2006), the effect of aggregate foreign aid on foreign direct and portfolio investment was not significantly different from zero in the 1990s. Donaubauer et al. (2020) report a significantly positive impact of aid on FDI for lower-middle-income countries, but an insignificant one for low-income countries. Asiedu et al. (2009) find even negative effects of aid on FDI in low-income recipient countries. Kimura and Todo (2010) use bilateral data in a gravity-type setting to investigate the relationship between aid and FDI. Estimating gravity equations for the top five donor countries (France, Germany, Japan, the United Kingdom, and the United States) in the period 1990–2002, they find that foreign aid, in general, does not have any significant effect on FDI. The only exception is a positive impact of Japanese aid on Japanese investment in recipient countries, which the authors call a "vanguard effect".

A clearer and more positive picture emerges when considering specific aid categories rather than the very heterogeneous aggregate aid figures. Selaya and Sunesen (2012) confirm their hypothesis that donor support for complementary factors such as human capital formation is associated with more FDI, whereas aid invested in physical capital comes at the expense of lower private investment. In substantive terms, the regression results suggest that one aid dollar invested in complementary factors draws in around two dollars of FDI in the long run. Donaubauer et al. (2016) provide evidence that aid for physical infrastructure has promoted FDI over the period 1990–2010. Lee and Ries (2016) explore whether aid for trade has promoted greenfield investment by lowering operating costs. According to their estimates for the period 2003–2013, this has indeed been the case at least in the more advanced recipient countries. Examining the sector-specific transmission mechanisms in a structural gravity framework, Donaubauer et al. (2020) find that aid for physical infrastructure, post-primary education, and governance as well as aid for trade remove investment barriers and thereby increase FDI stocks.

Accordingly, donors could become more effective in supporting FDI by putting a stronger focus on the aid categories that help improve the provision of inputs complementing private investment.

#### **3. Aid, Poverty, and Inequality**

Foreign aid could help mitigate poverty and inequality within recipient countries if two critical conditions were met. Donors would have to allocate aid in line with their rhetoric on pro-poor growth, by targeting the most disadvantaged population groups. At the same time, the authorities in the recipient countries would have to ensure that aid actually reaches the poor. Both conditions are likely to be violated to at least some extent. From the literature on aid allocation across recipient countries, it is well known that donors pursue a mix of motives, being motivated partly by developmental concerns and partly by commercial and political self-interest (e.g., (Hoeffler and Outram 2011)). Commercial donor interests may have as a consequence that foreign aid, e.g., in the area of physical infrastructure, is concentrated in industrial clusters rather than remote areas where the poorest people are living. Likewise, using aid as a means to buy political support by the local elite implies that it favors the rich and influential rather than the poor within a particular country. On the recipient side, aid may be used to provide goods and services that benefit the poor, but it has also been shown to induce rent-seeking and elite capture (e.g., (Reinikka and Svensson 2004; Andersen et al. 2020)).<sup>1</sup>

#### *3.1. Impact on Poverty*

#### 3.1.1. Monetary Poverty

The first MDG, which simply states that the share of people living in absolute income poverty should be cut by half between 1990 and 2015, has been reached globally if not in all developing countries. Whether aggregate foreign aid has contributed to the decline in poverty is hard to assess due to the same reasons discussed above for the case of economic growth. Still, the consensus is that, on average, foreign aid has been associated with falling monetary poverty. The soundest empirical studies so far, by Hirano and Otsubo (2014) and Arndt et al. (2015), confirm this view using per capita income of the poorest quintile and the poverty headcount (at \$1.25 and \$2 per

<sup>1</sup> This paragraph closely follows Box 3.5 in Stephan Klasen et al. (2018).

day) as poverty indicators, respectively. According to Arndt et al. (2015) estimates, a doubling of aid reduces poverty by around 15 percentage points.

Kaya et al. (2013) go beyond the aggregate perspective and look at the poverty impact of one specific category, aid for agriculture. Employing a fixed effects panel approach they estimate that a one percent increase in agricultural aid reduces the headcount poverty ratio by 0.2 percent in the aid recipient countries. The study also found that the growth elasticity of the headcount poverty ratio ranges from 1.7 to 3.5 across different specifications, which leads to the conclusion that agricultural aid is effective in poverty reduction directly and indirectly through growth. In a similar vein, Hirano and Otsubo (2014) show that social aid significantly and directly benefits the poorest quintile in society, while economic aid increases the incomes of the poor through growth.

#### 3.1.2. Non-Monetary Poverty

While the objective of cutting monetary poverty by half may have attracted the strongest public attention, OECD Development Assistance Committee (DAC) donors appear to have put even more weight on achieving the non-monetary MDGs if one judges them by their aid allocation decisions: the average share of overall aid budgets devoted to supporting social infrastructure investments rose substantially from about 20 percent in the early 1990s to over one third throughout the 2000s.

The existing empirical evidence tends to show that aid targeted at the social sector has helped improve various MDG-related indicators. According to cross-country studies at the macro level, more aid for education has been associated with increased primary school enrollment, less repetition and higher completion rates (D'Aiglepierre and Wagner 2013; Dreher et al. 2008). Likewise, aid for health has been shown to lead to lower infant mortality (Bendavid 2014; Mishra and Newhouse 2009).

More recent studies for several Sub-Saharan African countries (De and Becker 2015; Odokonyero et al. 2018; Kotsadam et al. 2018), which are based on geocoded data at the sub-national level and thereby mitigate the methodological problems that arise in particular from the unobserved heterogeneity prevalent in cross-country settings, corroborate the previous findings.

For the case of Malawi, De and Becker (2015) estimate significant, positive effects of education aid on raising school enrolment, of health aid on decreasing disease severity and of water aid on decreasing diarrhea incidence based on a combined propensity-score matching and difference-in-differences approach. The estimated effects are modest but non-negligible: an average health project, for example, leads to close to one fewer work day lost due to illness, per person per year.

Furthermore, employing a difference-in-differences estimator, Odokonyero et al. (2018) find that aid allocated to Uganda's health sector had a fairly "strong effect on reducing the productivity burden of disease indicated by days of productivity lost due to illness but was less effective in reducing disease prevalence".

Kotsadam et al. (2018) match geographic aid data with available georeferenced survey data from five Nigerian Demographic and Health Surveys. Their difference-indifferences estimates suggest "that children born to mothers who live in locations close to one or more aid projects have a lower risk of dying before the age of 12 months". In substantive terms, aid is estimated to lower the infant mortality rate by about one percentage point, or more directly by 10 children per 1000 born, which is again a non-negligible effect. The mortality-reducing potential of foreign aid seems to be particularly strong for less privileged groups like children of Muslim women, and children living in rural areas.

#### *3.2. Impact on Inequality*<sup>2</sup>

As in the case of poverty, the expected impact of foreign aid on inequality depends on the extent to which funds are well-targeted and aid capture is avoided. The latter tends to be inequality-increasing as rents are typically captured by local elites endowed with a disproportionate share of a country's economic and political power (Angeles and Neanidis 2009).

Given these counteracting factors, the question of whether foreign aid has reduced within-country inequality is an empirical one. The evidence so far—all obtained using the Gini coefficient as the indicator of inequality—is limited and ambiguous. Herzer and Nunnenkamp (2012) find that foreign aid exerts an inequality-increasing effect on income distribution. According to Chong et al. (2009) as well as Arndt et al. (2015), there is no robust association between aid and inequality. Shafiullah (2011) as well as Hirano and Otsubo (2014) conclude that aid reduces income inequality.

The mixed results of these studies may partly be due to differences in country samples and time periods as well as differences in the panel data techniques employed. Yet, the most recent analysis by Hirano and Otsubo (2014) also points to a more substantive explanation. The authors detect a considerable heterogeneity of the estimated impacts across aid sectors. Specifically, aid given to the social sector, which increased disproportionately over the period covered by their study, is shown

<sup>2</sup> This subsection closely follows Box 3.5 in Stephan Klasen et al. (2018).

to have the strongest and most robust inequality-reducing effect. This is in accordance with the evidence of effective support for social infrastructure presented in the previous section. Yet, even for social sector aid the impact on inequality as measured by changes in the Gini coefficient is not significant in quantitative terms.

Overall, the part of foreign aid dedicated to the social sector appears to be effective in improving social indicators that matter for the poorest segments of the populations in recipient countries. This is even though the targeting of social sector aid towards primary services—while having improved—still leaves much to be desired. The share of educational aid budgets allocated to post-secondary education, for instance, is still roughly equal to the share primary education receives (Lanati and Thiele 2020). Further improvements in targeting may be seen as a realistic next step towards increasing the poverty- and inequality-reducing potential of foreign aid.

#### **4. Concluding Remarks**

The evidence presented in this paper suggests that development cooperation can help achieve growth and poverty reduction in partner countries, even though the effects are likely to be modest in many cases. Most confidence can be put into the finding that, in accordance with the MDGs, aid for social infrastructure has contributed to achieving non-monetary goals such as higher school enrollment and lower infant mortality. It remains to be seen whether donors will continue to play a positive role when it comes to reaching more ambitious SDGs such as ensuring quality education for all. The evolution of educational quality has so far clearly fallen short of the improvements realized in quantitative indicators (e.g., (Bold et al. 2017; The World Bank 2018)). For instance, many pupils leave primary school without being able to read simple sentences and solve basic mathematical problems (The World Bank 2018, p. 2). To help overcome the quality problems in social infrastructure and come closer to meeting the respective SDGs, donors need to shift their strategy. While building schools or health facilities was mainly a matter of providing resources, quality improvements will require much more nuanced interventions. These include technical support in curricula development as well as training programs for school teachers and health personnel. Governance reforms such as performance-enhancing incentives for teachers will also play an increasing role.

Foreign aid specifically targeted at facilitating integration into international trade and attracting FDI has also shown to be effective in most existing empirical studies, even though its positive effects tend to largely miss low-income countries. In contrast, it is inherently difficult to empirically identify income effects of foreign

aid at the macro level, which the long-standing and still unresolved debate about the aid–growth relationship illustrates in a forceful manner.

In any case, if the aim is to bring poverty close to zero in accordance with SDG1, the potential of measures that mainly work through "trickle down" growth is likely to be more limited than in the past. This is because many of today's poor live in a fragile state and/or face multiple development obstacles such as belonging to a discriminated minority, lacking assets or having limited access to markets and public services. Any donor response to this kind of structural poverty has to be multi-dimensional, including support for building state capacity and targeted interventions such as land-tenure reforms and the establishment of road connections to remote areas.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


Shafiullah, Muhammad. 2011. Foreign Aid and its Impact on Income Inequality. *International Review of Business Research Papers* 7: 91–105.

Winters, L. Alan, Neill Mc Culloch, and Andrew McKay. 2004. Trade Liberalization and Poverty: The Evidence so Far. *Journal of Economic Literature* 42: 72–115. [CrossRef]

The World Bank. 2018. *World Development Report 2018: Learing to Realize Education's Promise*. Washington DC: The World Bank.

© 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **A Safety Net for You, a Safety Net for Me? Donor Promotion of Social Protection Schemes Faces Policy Coherence Issues**

**Fritz Brugger**

#### **1. Introduction**

Social protection schemes are increasingly recognized to be pivotal for preventing and reducing poverty, as they provide a safety net for the most vulnerable (Fiszbein et al. 2013). During crises, they are even more important to prevent people from falling into poverty, as the COVID-19 pandemic has demonstrated. While social protection was not even mentioned in the Millennium Development Goals (MDGs), it has gained prominence in the Agenda 2030 discourse to the point that "achiev[ing] substantial coverage of the poor and the vulnerable" with social protection schemes was established as one target under the Sustainable Development Goal on ending poverty in all forms (SDG 1.3). In the Addis Ababa Action Agenda, governments committed to a "social compact", pledging to deliver social protection to their citizens, including social protection floors (UN 2015).

New protection schemes are beginning to emerge. For example, Tanzania, Kenya, Ethiopia, and Senegal are starting to create welfare systems and distribute cash to poor households. Receiving USD 43 every three months in Senegal and USD 13 per month in Tanzania is not a fortune, but it is a start. Today, 90% of the bill of Tanzania's program is picked up by donors such as the World Bank and the British and Swedish governments (Economist 2019). This may work for kick-starting protection schemes but relying on donor money is not a sustainable solution.

To finance social protection schemes over the long-term, the availability of sufficient government income through domestic tax collection is indispensable (Adaba 2016; Fiszbein et al. 2013). In 2016, 55% of the world population—as many as 4 billion people—were not covered by any social protection cash benefits, with large variations across regions: from 87% without coverage in sub-Saharan Africa to 14% in Europe and North America. Rolling out social protection across sub-Saharan Africa is an even bigger challenge, with 41% of people subsisting on less than USD 1.90 a day (UN 2019).

The approach of donors to support domestic resource mobilization is inconsistent. On one side, donors have pledged to provide strong international support for social protection initiatives and to explore "coherent funding modalities to mobilize additional resources, building on country-led experiences" (UN 2015, p. 6). Donors have also signaled an appetite to support capacity building for tax authorities more vigorously and have stepped up their commitments (ATI 2019).

On the other side, donor countries are defending international tax rules that benefit their business communities, and which, to a degree, work against the ability of developing countries to obtain a fair share of taxes from multinational corporations operating in their jurisdictions. Foreign direct investment into developing countries increased twentyfold between 1990 and 2015, twice the growth rate of global foreign direct investment (FDI) (UNCTAD 2019). This is not least because donor countries through the 'beyond aid' agenda of the Organization for Economic Co-operation and Development (OECD) have long promoted FDI as an effective way to create jobs and generate tax revenues in low income countries.

Balancing the promotion of tax revenues in developing countries while at the same time protecting domestic economic interests is a classical policy-coherence-for-development dilemma. The basic idea behind policy coherence for development is to align non-aid policies with development goals. In the best case, policies across government are mutually reinforcing (positive coherence). In the minimum case, policy coherence for development avoids that public policies have an adverse impact on developing countries.

Better coherence between donors' aid and non-aid policies is expected to have a greater impact than—in this case—simply increasing aid budgets for capacity building of tax authorities and for supporting social safety nets (Brown 2015; Siitonen 2016). The OECD declared in its 2012 Strategy on Development that "enhancing policy coherence for development is one of the primary objectives" while the SDGs (Goal 17, target 17.13 and 17.14) recognize greater policy coherence as an issue of systemic relevance to the successful implementation of the Agenda 2030 (OECD 2016, 2012a; Spencer 2012; Verschaeve et al. 2016).

The remainder of the chapter analyzes how donor countries deal with the policy coherence dilemma. It first reviews the evolution of domestic resource mobilization on the donor agenda over the last two decades and particularly since the financial crisis of 2008. Next, we turn to the reform of the international corporate taxation agenda. More precisely, we review the reform of the transfer pricing rules in the Base Erosion and Profit Shifting (BEPS) reform process that was initiated in 2012. Then, we consider whether the institutional platform on which international

tax rules are negotiated—the OECD—affects policy coherence outcomes. The last section concludes.

#### **2. The (Re-)Discovery of Domestic Resource Mobilization Programs**

Tax revenues in low- and middle-income countries are significantly lower compared to the OECD average measured as share of GDP. While the OECD countries collect taxes at the level of 34% of GDP, the respective average value for Africa is only at around 18% (OECD 2018a). According to the International Monetary Fund, countries collecting less than 15 percent of GDP in taxes are assumed to be below a tipping point to make a state viable and put it on a path to growth. As of 2015, 35 of the world's 75 poorest countries are below this threshold. Domestic public resources are lowest where extreme poverty is highest (Gaspar et al. 2016).

Although the collection of tax requires substantial administrative capacity that is missing in most low-income countries, the topic received relatively little attention from development agencies until a decade ago. Domestic resource mobilization was mentioned for the first time in the Financing for Development (FfD) conference in Monterrey in 2002 but it was particularly after the financial crisis in 2007/8 that domestic resource mobilization was discovered as a previously "neglected factor in development strategy" (Culpeper and Bhushan 2008). Aid skepticism and budgets under pressure sparked interest in increasing tax capacity as an exit strategy to the point that the 2008 FfD conference in Doha put domestic resource mobilization center stage (UN 2008). Later, it became one of the SDG targets (SDG 17.1) and the first out of seven action areas adopted at the third FfD conference in Addis Ababa in 2015 (UN 2015).

The high-profile support for tax capacity building is not (yet) met by financial commitment of bilateral donors. Tax activities were estimated to amount to some 0.15 per cent of total development assistance only in 2015 (PCT 2016). Starting from this low baseline of USD 224 million of effective disbursements in 2015, donors pledged to double spending for tax projects by 2020 in the framework of the Addis Tax Initiative (ATI 2019).

In light of the rapidly growing tax-related activities (for a mapping of donor activities, see e.g., Avis 2017; PCT 2016, p. 13f), an effort was made to improve the coordination of actors and programs and to "better frame technical advice to developing countries". The Platform for Collaboration on Tax, established in 2016, brings together the four largest multilateral organizations active in tax matters: International Monetary Fund (IMF), United Nations (UN), World Bank Group (WBG), and Organization for Economic Co-operation and Development (OECD).

Particularly noteworthy is the presence of the OECD, which signals a change that the OECD has undergone related to developing countries. The OECD is a membership-based organization of currently 37 high-income countries and at the same time, the key institution for the negotiation and dissemination of international tax rules. Now, the OECD has declared to become a key actor in the domestic resource mobilization agenda and "to mainstream development across all of our work" (OECD 2018b, p. 4). In contrast, the UN, IMF and World Bank are not promoters of the international corporate taxation soft law but do advise developing countries on tax. Hence, their mandate makes them more concerned with protecting source taxation, which is not necessarily aligned with the position of the OECD (Picciotto et al. 2017, p. 10). For example, the IMF pointed out in its influential paper on spillovers in international corporate taxation that tax treaties, which are typically based on the OECD brokered model convention, significantly restrict the rights of countries to tax activities where they take place ("at source"), reducing the corporate tax base of capital-importing states, which are mostly developing countries (IMF 2014). The IMF concludes that developing countries should sign double taxation treaties "only with considerable caution" and expresses the need to "protect and expand their corporate tax bases in the face of challenges in applying the [arm's length principle]" (IMF 2014, p. 10, 34). From this perspective, the Platform for Collaboration on Tax can be seen as an attempt to increase policy coherence among multilateral organizations towards developing countries. Less clear is on whose terms this coherence will settle.

A more technical attempt to facilitate the coordination of external support for reform is the Tax Administration Diagnostic Assessment Tool (TADAT) to vet the health of a country's tax administration and evaluate the progress of tax policy reform and capacity building initiatives by way of subsequent repeat assessments. Unfortunately, of the 94 national performance assessments carried out by mid-2020, only 19 are publicly available, making it hard to assess the effectiveness of technical assistance for domestic resource mobilization.

The most recent international effort is the Tax Inspectors Without Borders initiative by OECD and the United Nations Development Programme (UNDP). The program takes a "learning by doing" approach whereby experts from OECD countries work directly with local tax officials on audits and audit-related issues concerning international tax matters such as abusive tax avoidance by multinational enterprises (MNE), including transfer pricing and thin capitalization. The roughly USD 500 million in tax revenue recovered in four years and 59 deployments in Africa, Latin America, Eastern Europe, and Asia (TIWB 2019) stand against the USD 100–240 billion in lost revenue annually that developing countries suffer due

to avoidance (Crivelli et al. 2015). Corporate tax planning contributes to this loss with an estimated USD 30–52 billion in trade misinvoicing, USD 5–50 billion through transfer pricing manipulation and USD 3–9 billion via treaty shopping arrangements (UNCTAD 2020).

All efforts of the international community to support domestic resource mobilization at the technical level are—sooner or later—confronted with the fact that tax is deeply political. The collection of tax not only requires substantial administrative capacity, but more importantly, requires a state to be considered legitimate since the vast majority of tax is collected when there is a high level of voluntary compliance (Di John 2011, 2010). Changes in a country's tax policy and administration are largely driven by domestic economics, politics, and institutions (Di John 2010; Fjeldstad and Heggstad 2011; Fjeldstad and Moore 2008). This means that technical support for tax authorities is confronted with the capacity of elites to influence tax policy formulation and administration, as well as the involvement of tax collectors and public servants themselves in rent-taking (Forstater 2018, p. 10). Property taxation is a case in point. Land and property are among the most visible indicators of personal wealth and property tax is widely recognized as efficient, administratively straightforward, and a progressive way to collect revenue (Ali et al. 2017; Ramírez et al. 2017). Nevertheless, property taxes are rarely used in low-income countries, they amount to 0.1–0.2 per cent of GDP, while in most OECD countries, property taxes account for 1–2 per cent of GDP. Property taxes tend to target the economic and political elites who have the power to prevent tax policies being enacted or enforced (Moore and Prichard 2017, p. 16). While challenges to domestic resource mobilization posed by domestic politics are important, this chapter focuses on the policy coherence dimension that donor countries face.

#### **3. Reform of the Cross-Border Taxation Rules**

The global financial crisis has not only triggered donor interest in domestic resource mobilization. The financial crisis has equally triggered the OECD/G20-led BEPS project, claimed to bring about the "most fundamental changes to international tax rules in almost a century" (OECD 2015). Such reform processes are prime opportunities to increase policy coherence.

Based on the findings of a recent research project (Brugger and Engebretsen 2020), this section summarizes the outcome of—and political economy behind—the deliberation over the reform of the OECD transfer pricing rules. What sounds—and is—very technical guides not only transactions between related entities but, in essence, defines the scope for shifting profits from high to low tax jurisdictions.

#### *3.1. Why Low-Resource Countries Struggle to Tax Multinational Enterprises*

The allocation of taxing rights between capital importing and capital exporting countries is based on the "separate entity" concept. This means that each permanent establishment of an MNE in a country is taxed as a separate entity. Furthermore, transactions between related entities (i.e., controlled by the same MNE) should be priced "at arm's length", i.e., as if they were market-based transactions between non-related entities (Langbein and Fuss 2016; Perry 2017).

In tax jargon, this is called "transfer pricing". A transfer price is the price charged by a company for goods, services, or intangible property to a subsidiary or other related company. Transfer pricing therefore allocates the tax base generated by the profits of MNEs among the national jurisdictions within which those enterprises operate.

Whilst transfer pricing is a legitimate feature of the commercial activities of MNEs, abusive transfer pricing occurs when income and expenses are improperly applied, distorting the allocation of profit among the countries in which multinationals operate for the purpose of reducing overall taxable income of the enterprise (CITCAR 2020; De Mooij and Liu 2018; OECD 2020a). Transfer pricing is one of the most complex global tax planning tools employed by MNEs.

Transfer price manipulation not only reduces a country's tax base, but also provides a substantial advantage to MNEs in comparison with single-country firms because only the former can use this type of international tax planning strategy. In fact, it is comparable to a subsidy which MNEs receive but domestic enterprises do not (Baistrocchi 2013; Cooper et al. 2017; Rixen 2011). Drawing on evidence from different industrial sectors for a group of OECD countries, Bartelsman and Beetsma (2003) estimate that at the margin, more than 65% of the additional revenue resulting from a unilateral tax increase is lost because of income shifting towards lower tax jurisdictions. The bulk of profit shifting seems to be done by the largest companies (Wier and Reynolds 2018). In addition to denying a country's essential tax revenue and putting local businesses at a competitive disadvantage, transfer pricing contributes to harmful competition at a global level among tax jurisdictions, as countries attempt to lower their tax rates to attract MNEs to their own jurisdictions in a race to the bottom dynamic (Baistrocchi 2013).

Developed countries, through the OECD, have become the authority on transfer pricing over the preceding decades. The OECD issues the Transfer Pricing Guidelines (TPG) for Multinational Enterprises and Tax Administrations that define acceptable methodologies to implement the arm's length principle (ALP), i.e., to establish market prices for cross-border transactions of MNEs (OECD 1995, pp. 1–6).

Although the OECD Guidelines on Transfer Pricing are not legally binding, they have attained "canonical" status (Picciotto 2018, p. 19) due to the unique combination of international "soft law" regulations (TPGs and the OECD Model Tax Convention endorsing the ALP) and their translation into national "hard law" through bilateral tax treaties, building on the OECD Model Convention (Rixen 2008a).

The crux of the matter is how the arm's length price is best established. The OECD guidelines accept five standard methods which all rely on an assessment of facts and circumstances of the individual transaction to determine the final price that should correspond to comparable market transaction. Conducting such a comparability analysis requires considerable technical capacity and expertise, something that is often lacking in resource-constrained low-income countries (Solilová and Nerudová 2015). Even where this capacity is available and sophisticated techniques for adjusting data from other regions for use as comparators are deployed, it produces a wide range of putative comparables at best (e.g., Gonnet et al. 2014). The Platform for Collaboration on Tax concludes that "a comparability analysis provides only an approximate answer and that some flexibility is needed to determine a principled answer" (PCT 2017, p. 66). Individual adjustments on a case-by-case basis offer considerable discretion to both taxpayers as well as the officials in charge of establishing transfer pricing. This makes the exercise vulnerable to manipulation, increases the risk of legal disputes, or translates tax assessment into a negotiation with the taxpayer "to achieve a sensible, arm's length result" (Ibid., p. 67).

The ALP is evidently complex and difficult to apply for resource-constrained countries, and simpler alternatives exist. Why do the OECD guidelines not recognize and promote simpler alternatives?

The most far-reaching alternative concept to the ALP is the idea of taxing MNEs as single firms instead of treating each subsidiary as separate entity. The corporation's total worldwide profit (or loss) is then attributed to each jurisdiction, based on factors such as the proportion of sales, assets, or payroll in that jurisdiction. Such formulary apportionment—also known as unitary taxation—would be a systemic change to how international corporate taxation works (Clausing and Avi-Yonah 2007).

The middle ground is occupied by so-called "simplified methods" to establish the price of a transaction. Simplified methods still rely on separate entity accounting, but instead of assessing the circumstances of each transaction, they define rules to establish the price for a transaction with no or only very limited adaptation possibilities. Such simplification looks like an attractive option to resource-constrained tax administrations in the Global South. An example is Argentina's so-called sixth method, which applies benchmark prices to commodity trading and which was later

adopted by a few other Latin American commodity producers and Zambia for its copper sector (Durst 2018; Grondona 2018). More comprehensive is the Brazilian fixed margin method which specifies the profit margins to be applied to each type of transaction (Rocha 2017). While Brazil's fixed margin method clearly opposes the OECD recommendation of considering the specific facts and circumstances of each transaction, it achieves administrative simplicity as well as a low level of tax disputes (Picciotto 2018, p. 33), a fact that is acknowledged by MNEs such as Shell operating in Brazil: "We find that fixed margins provide a level of certainty to the results of an inspection, which simply assessing risks and functions would not. In essence, fixed margins are easier to follow, regulate and inspect" (Gasper 2016).

In 2013, when the BEPS reform was launched after the global financial crisis and the subsequent public outcry over exposés detailing the aggressive international tax planning schemes employed by some of the world's biggest MNEs, the OECD Secretary General together with the Head of Tax declared that it was "essential to simplify and strengthen the transfer pricing rules for the benefit of both developed and developing economies" and to "alleviate the compliance burden for both tax authorities and taxpayers" (OECD 2012b).

Yet in 2019, after the reform, the global tax system continues to rely on the ALP, notwithstanding its limitations (Okauru 2018). Unitary taxation has so far failed to challenge the dominance of the ALP (De Robertis 2018). Even less far-reaching proposals to move from the discretionary-based ALP approach towards a more rule-based settlement of transfer prices as the Brazilian or Argentinian methods have been met with fierce resistance. Facing the pressure from low-income countries, the OECD included (but not endorsed) in the 2017 revision of the transfer pricing guidelines the so-called sixth method in addition to the five recommended methods. At a closer look, however, it comes with the condition that a detailed comparability analysis was to be conducted between the economically relevant characteristics of the controlled transaction and the specification of the quoted price (Collier and Andrus 2017, p. 249)—a requirement that essentially removes whatever merits the sixth method has in ease of administration (Picciotto 2018, p. 25).

Even this "uneasy compromise" (Büttner and Thiemann 2017; Perry 2017), which largely satisfied OECD member countries and business community demands, was criticized by tax practitioners and the big accounting firms in particular (Ernst and Young 2015; PWC 2015). Developing countries, academics, and civil society deplored the missed opportunity for reform (ICRICT 2019a) and former senior OECD tax officials noted that, after all, the transfer pricing guidelines have become not simpler but even more complex to implement (Collier and Andrus 2017; Andrus

and Oosterhuis 2017). What explains this outcome? A discourse network analysis covering the period from the mid-1990s to the end of 2018 sheds light on the deliberative dynamics during the BEPS reform (Brugger and Engebretsen 2020).

#### *3.2. The Perseverance of the Arm's Length Principle*

It is well documented in the academic literature that the driving force behind the transfer pricing guidelines has been an epistemic community centering around the OECD tax network and consisting of the organization's staff, experts from member countries' tax administrations, and the private tax law and accounting community since the mid-1990s (Büttner and Thiemann 2017; Seabrooke and Wigan 2016). As corporate taxation has become an increasingly politicized topic after the financial crisis, devising rules for transfer pricing could no longer take place in closed groups of like-minded specialists. In response, the OECD started to invite public consultations on reform proposals. In these consultations, the epistemic community defended its policy project by reaffirming the accuracy of the ALP and its positive effect on FDI and international cooperation. Even more importantly, those defending the status quo managed to activate their constituency to the degree that accounting companies, tax lawyers, and MNEs firmly dominated the public debate as well as public consultations around the 2017 reform of the transfer pricing guidelines. As Figure 1 illustrates, their share of the debate grew from 27% right after the financial crisis to over 56% between 2014 and 2018. Over time, the statements of the "pro ALP camp" also grew more hostile towards simplified methods, with an increasing share of the statements demanding modifications to simplified methods, claiming that they would cause double taxation or putting into question the legitimacy of alternatives to the hegemonic transfer pricing methodologies altogether.

In contrast, the "reform camp" that advocated for simplified TP methods, mainly civil society organizations and think tanks, was clearly on the defense (Figure 1). They were not able to effectively break the epistemic community's ranks nor build a broader coalition in support of reform, resulting in their share of the debate becoming marginalized over time; it always remained below 15% (Figure 1) (Brugger et al. 2019; Brugger and Engebretsen 2020).

**Figure 1.** Intensification of the debate on simplified methods (bars represent the overall number of statements in a time period, right *y*-axis) and changing actor dynamics over time (icons represent the share of statements per actor group, left *y*-axis). Source: (Brugger and Engebretsen 2020).

#### **4. Legitimacy Deficits in International Corporate Taxation Governance**

In addition to the discursive struggle over the technicalities of transfer pricing, negotiations over international corporate taxation also have an institutional dimension. For the last few decades, the OECD established itself as the platform that largely defined international tax governance. The institution has derived its global legitimacy from providing solutions that are accepted because of their problem-solving capacity, typically called output legitimacy. The OECD's Achilles' heel is its lack of input legitimacy. As a membership-based organization, it represents only a small number of high-income economies and is perceived as "the rich countries' club" (Rixen 2008b, p. 148), as the statement of an Indian representative to the UN illustrates:

[The] OECDModel Tax Convention and the OECD Transfer Pricing Guidelines have been developed on the basis of consensus arrived at by the Government of 34 countries (all developed countries). These guidelines only protect the interests of the OECD countries which are partial to such Convention. Since the Governments of developing countries are not party to the OECD Guidelines, it is improper to suggest that they represent internationally agreed guidance knowing fully well that concerns of developing countries have not been taken care of in the OECD Model Convention and OECD Transfer Pricing Guidelines. (Spencer 2012, pp. 25–26)

The more representative UN has not (yet) been able to establish itself as a meaningful counterweight to the OECD in international tax matters (Hearson 2017; Mosquera Valderrama et al. 2018; Rixen 2008a; Zagaris 2005). Its "Committee of Experts on International Cooperation in Tax Matters" (UNTC) comprises 25 members (15 from developing and 10 from developed countries) acting "in their expert capacity" rather than directly representing their government's interests; they meet twice per year for four days (UN ECOSOC n.d.). In contrast, the OECD Centre for Tax Policy and Administration (CTPA) has a 155-person strong secretariat supporting the various OECD committees, work programs, task forces, and dialogue platforms (OECD 2018b).

The UN committee has published its own UN Practical Manual on Transfer Pricing for Developing Countries. It is largely consistent with the OECD approach but has a more user-friendly focus (Collier and Andrus 2017; Soong Johnston 2017). There is also a dominance of OECD representatives in the UN tax body observable. At one stage, 48% of the members of the UNTC were from OECD member countries, including its chairman, and some of them serve both on the OECD and UN tax committee (Spencer 2012). While it may be an overstatement that the "OECD has had in effect operational control of the UN Tax Committee" (Spencer 2012, p. 23), a notable overlap in personnel is apparent.

Since the beginning of the century, repeated efforts to upgrade the UN Tax Committee of Experts to an International Tax Organization have failed. Initially, a more representative and legitimate global tax body under the auspices of the UN has been suggested by the UN High-level Panel on Financing for Development and outlined in the 2001 Zedillo Report in preparation of the FfD conference in Monterrey (UN 2001). The International Tax Organization was supposed to "take a lead role in restraining tax competition designed to attract multinationals with excessive and unwise incentives . . . , sponsor a mechanism for multilateral sharing of tax information . . . [and] most ambitious of all, an International Tax Organization might in due course seek to develop and secure international agreement on a formula for the unitary taxation of multinationals" (UN 2001, pp. 28, 65).

However, global tax issues were not even been discussed in Monterrey (2002) or in the follow-up conference in Doha (2008) (Lesage et al. 2010). The only result was a slight upgrading of the UN tax Committee from an Ad Hoc Group of Experts to a Committee of Experts and the increase from one to two meetings per year.

In 2015, the international community agreed at the 3rd FfD conference in Addis that "efforts in international tax cooperation should be universal in approach and scope" (UN 2015, p. 13) and that the representation "of developing countries in decision making in global international economic and financial institutions" must be enhanced "in order to deliver more effective, credible, accountable and legitimate institutions" (SDG 10.6).

This consensus did not translate into support for an International Tax Organization. A respective proposal was promoted by the G77, the group developing countries at the UN. The proposal was frustrated by the strong opposition of OECD member countries and the EU in Addis and a year later at the UN Conference on Trade and Development in Nairobi in 2016 (Deen 2016; G77 2015, 2017; Picciotto et al. 2017, p. 8ff).

In parallel, the OECD took steps to bolster its own input legitimacy by opening up and rendering the policy-making processes more inclusive (Christensen and Hearson 2019). From 2012 to 2015 the OECD engaged in a "dialogue-by invitation only" through the Global Forum on Transfer Pricing. Criteria on who was invited and governance mechanisms of the forum remained classified (Christians and Apeldoorn 2018). In 2016, the mandate to work on the "development of guidance on transfer pricing" moved to the newly created "Inclusive Framework" (IF) (OECD 2017a, p. 13). The IF is open to all jurisdictions that commit to the BEPS reform package and its consistent implementation, and pay a membership fee. BEPS Associates "participate on an equal footing" but what that exactly means remains opaque as agreements with "Associates", governance structures of the IF, and procedural information are not public.

The core mandate of the IF is monitoring the implementation of tax reform measures that have already been agreed on by the G20 and OECD members (Hearson 2017). Hence, the process of broad-based implementation of the BEPS minimum standards will roll out the governance mechanism largely devised by the OECD at the global level (Fung 2017; ICRICT 2019b). As the number of countries willing to implement policies originating in the OECD advances (137 as of January 2020), so does the organization's standing and legitimacy as a forum for technical problem-solving and diffusion of international tax standards (Christians 2010; Sharman 2012; Vega 2012).

There has been an opening up of the OECD over the last decade, which demonstrates institutional learning about the need for inclusivity in establishing a broadly accepted global tax policy mandate. However, several facts raise doubts about the sincerity of this move to create a level playing field, particularly with respect to decision making power. First, the IF was created after the substantial discussion on the BEPS actions was already over and the focus shifted to implementation. Second, the Council delegated to the Committee for Fiscal Affairs (CFA)—in which IF members

are supposed to participate "on equal footing"—the authority to approve future amendments to the TPGs in 2017, but only after the 2017 revision of the TPGs was approved by the Council (OECD 2017a, 2017b). Third, the same Council can at any time unilaterally take back the competencies delegated to the CFA. Fourth, as long as the full terms of the partnership engagement, governance structures, and procedural information of the IF are not publicly accessible, it is not possible to empirically analyze and evaluate the stake of developing countries in the IF (Christians and Apeldoorn 2018; OECD 2003). Finally, the OECD Council has retained a gatekeeper role by approving the formal invitation to new BEPS Associates (OECD 2017a, p. 18).

After all, the IF is operating in the shadow of the OECD hierarchy and only full membership in the OECD signifies actual decision-making power. The OECD is vigilant that the policies of potential new members do not stray too far away from the organization's principles and guidelines so as to unbalance the existing settlement. Brazil, as the only major economy that does not follow the OECD TPGs, is a case in point. The country officially submitted its bid for membership in 2017 (Hearson and Prichard 2018). Angel Gurría, the OECD Secretary General, made clear that the candidate's alignment with the OECD TPGs is a condictio sine qua non for accessing the OECD: "Transfer pricing is one of the key areas where alignment with the OECD's internationally accepted standard is necessary. This constitutes a core principle and a benchmark that needs to be met by any new Member wishing to join the OECD" (Gurría 2018; OECD and Federal 2019).

#### **5. Conclusions**

Stronger domestic resource mobilization is recognized as essential to build social safety nets and fight poverty. While the development discourse tends to focus on technical support to tax administrations, this article highlighted donor countries' policy coherence gap in the area of tax and development. With FDI promotion being a pillar of the development agenda, the question of cross-border taxation cannot be excluded.

The topic is of broader relevance because domestic revenue management, FDI, and policy coherence for development are all part of the "beyond aid" agenda, a term that signifies the changing role of development cooperation against the backdrop of a growing critique against development aid, a changing donor landscape, stalling official development assistance (ODA) budgets, and the importance of global governance (Janus et al. 2015).

Zooming in on the reform of transfer pricing technicalities, the article finds that the announced simplification that would benefit low-income countries has not

materialized so far. A strong coalition behind the ALP managed to fend off calls for simpler and more rule-based methodologies that narrow the room for discretion and profit shifting. What appears to be primarily technical OECD policy generation is highly political. Interest politics trump policy coherence (OECD 2016, p. 40).

This finding renders obsolete the often-voiced fear that an International Tax Organization—or any other more inclusive forum to deliberate over international corporate tax governance—would unduly politicize international taxation. Yet, a more inclusive global tax governance architecture would certainly alter the balance between OECD members and developing countries.

SDG 10.6 calls for the "enhanced representation and voice for developing countries in decision-making in global international economic and financial institutions in order to deliver more effective, credible, accountable and legitimate institutions". Yet, a closer reading of the indicator monitoring SDG 10.6 reveals that this target does not include global tax governance. It considers developing countries' membership and voting rights in the Bretton Woods institutions, the African, Inter-American and Asian Development Bank, UN, WTO, and the Financial Stability Board. Inclusiveness in international tax governance is not considered, and tax policy-making is kept at the discretion of the OECD.

The aim to maintain its role as a pivotal actor in global tax policy provision might lead to more organizational reform of the OECD in the future, when new events shake the existing settlement. The next such impetus for reform is the digitalization of the economy. The fact that the "platform economy" does not need a "permanent establishment" to conduct business in a country removes a key pillar of the current international corporate taxation. Moreover, with most of the platform economy operating out of the US, their interest is by and large opposed to that of the EU and other OECD member countries. This rift among OECD countries might impede even the so far prevalent lowest common denominator agreements among its leading member countries that are home to large multinationals. This opens the possibilities for unprecedented alliances between jurisdictions across income groups that stand to lose out on the digital economy (Fung 2017).

The IF is mandated with the work on the digitalization of the economy (OECD 2018c). Most of the substantive work is undertaken by the Task Force on the Digital Economy, a body under the OECD Committee of Fiscal Affairs. The first proposals submitted for public consultation in 2019 break new ground: they consider solutions that go beyond the arm's length principle, including a minimum tax rate and some formulary components (OECD 2019a, 2019b, 2019c). However, in January 2020, the program of the IF was replaced by the more moderate reform proposals authored by the OECD Center for Tax Policy and Administration (OECD 2020b). The final agreement on the response to the tax challenges arising from the digitalization of the economy will bring more clarity regarding the inclusiveness of the IF and the extent to which low-income countries can exert influence on the coordination of cross-border taxation. If this is not the case, calls for an International Tax Organization are unlikely to disappear.

**Funding:** This article is largely based on research that was funded by the Swiss Programme for Research on Global Issues for Development, grant number 169564.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

MDPI Books Editorial Office E-mail: books@mdpi.com www.mdpi.com/books

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18