Tax in Developing Countries: Increasing Resources for Development
Written evidence submitted by the London School of Economics and
Timothy Besley, LSE and Adnan Khan, International Growth Centre
Building fiscal capacity is central to state building. Historically, warfare and changes in political systems have played a central role in building effective states and increasing the size of tax extracted. The inability to tax is both a symptom and a cause of underdevelopment. Difficulties in collecting taxes are generally a symptom of a weak economy and an ineffective government. And governments without tax revenues must rely on external sources of support to provide public services.
This submission discusses how insights from recent research can be useful in thinking about these issues and in formulating policy responses.
Figure 1: Historical Data
We first lay out some basic macro-economic facts based on available data.  A good starting point is to look at the history of a sample of countries using data from Mitchell (2007).  Figure 1 illustrates how the average tax take has increased over time from around 10% in national income to around 25% in the sample as a whole. Equally striking is the increasing reliance on income taxation which made up about 5% in revenues in 1900 but about 50% by the end of the last century. The significant increase in the income tax share during two world wars is striking and appears to have been persistent, outlasting the wars.
In general, richer countries tend to raise more tax revenue as a share of national income than poorer countries. This is illustrated in Figure 2, where the left hand panel plots the overall tax take as a share of GDP from Baunsgaard and Keen (2005) against the log of income per capita (from the Penn World Tables), with both being measured in the year 2000 with observations coloured by income. The right hand panel looks at the same relationship using the time-series data from Mitchell (2007) to plot five-year averages of the tax share over the twentieth century against national income, distinguishing observations by time period. The cross-section and time-series patterns are strikingly similar. Higher-income countries today raise much higher taxes than poorer countries. Perhaps most striking is the fact that the tax share in national income of today's developing countries does not look very different than the tax take 100 years ago in the now developed countries.
Figure 2: Cross-country and Historical Tax Collection
To get a further indication of such differences across countries, it is interesting to look at the relative use of different tax bases. Arguably, trade taxes and income taxes are two polar opposites. To collect trade taxes only requires being able to observe trade flows at major ports. This is a much easier proposition than collecting income taxes which requires major investments in enforcement and compliance structures throughout the entire economy.  An interesting indication of fiscal-capacity investments is obtained by holding total tax revenue constant, and then asking how large a share of tax revenue is collected from trade taxes and income taxes, respectively.
These shares are plotted against each other in Figure 3.  Again, we report the cross-sectional pattern for the year 2000, based on data from Baunsgaard and Keen (2005), and the time-series pattern over the last 100 years based on historical data from Mitchell (2007). The income-tax share is displayed on the vertical axis, and the trade-tax share on the horizontal axis. We observe a clear negative correlation: countries with a higher reliance on income taxes tend to have less reliance on trade taxes. The left panel also shows a striking pattern by income: high-income countries depend more on income taxes, while middle- and -- in particular -- low-income countries depend more on trade taxes. The right panel of Figure 3 shows that the move from trade to income taxes has also been a feature of the historical development of tax systems. Again, the cross-sectional and time-series patterns look strikingly similar.
Figure 3: Income and Trade Taxes
Figure 4 looks at the personal income tax, plotting the relationship between the share of income taxes in total taxes and income per capita. The left hand panel separates the observations into three groups according to their share of taxes in national income: countries that raise more than 25% of taxes in GDP, countries that raise 15-25% of taxes in GDP, and countries that raise less than 15%. Those in the high-tax group again look distinctive, raising more of their tax revenues as income tax. The right hand panel again colours observations according to the time period. Once again, there is a pronounced similarity between the historical situation of older nations and the developing world today.
In contrast to the amount of taxes raised in GDP, statutory tax rates on goods and income are not particularly different between rich and poor countries. This reinforces the observation that the main difference between tax systems lies in the ability to enforce taxes – particularly extending the domain and outreach of income taxes and VAT. Hence, these are the key investments in fiscal capacity building that need to be studied to provide an evidence base for policy making. However, this requires studies that are tailored to the particular circumstances of individual countries and the specific challenges that they face. The DFID funded International Growth Centre has recently begun collaborations with tax authorities to build this evidence base and points to how an evidence-based policy agenda can be developed that it is useful for policy.
Figure 4: Income Taxes
Interpreting the evidence
It is useful to think about fiscal capacity building as purposeful and forward-looking decisions by governments, i.e. building an effective state as an investment problem. So an incumbent government must weigh the present costs of investing against future expected benefits. Benefits depend on how the proceeds of taxation will be used and hence reflect stability and institutions. The schematic representation below is from Besley and Persson (2011a, Chapter 1). In green, it depicts three factor which shape incentives to build fiscal capacity.
Figure 5: Building Fiscal Capacity
Political stability affects the time horizon over which governments view the returns to investment in fiscal capacity. Short-lived governments have little incentive to make changes that will benefit future incumbents.
Cohesive political institutions, those with strong legislative and judicial institutions, limit corruption and encourage public revenues to be spent on public services which benefit a broad segment of the population. This raises the demand for fiscal capacity. The item in blue on common versus redistributive interests emphasizes that the underlying interests and cleavages matter. Times of warfare traditionally created a strong common interest. In modern states there is a need to find an alternative to military spending as a means of creating common spending interests. This means working on factors which improve the efficiency and outreach of public service provision.
Even if institutions are cohesive, translating demand for spending into fiscal capacity is muted when there are natural resource rents or aid dependence. There is reliable evidence (Jensen (2010)) showing that investments in non-resource based taxation respond to oil price shocks in oil states. The evidence on cash aid dependence is less clear cut but the basic argument is similar.
Growing the economy has two main benefits in terms of tax collection. First, for most countries growth in tax take
Tim: I think you are referring in the first benefit to the link between tax take and formality, adn in the second to the link between growth and formality. But please check.
Tim: I think you are referring in the first benefit to the link between tax take and formality, adn in the second to the link between growth and formality. But please check.comes from extending the domain of the formal market economy. Taxes are typically levied on market transactions which must be measured for tax purposes. Moreover, there is a strong correlation between financial intermediation and the level of income – greater use of the formal financial sector also makes the government better able to observe and hence tax transactions. Second, growth is associated with greater formality and firms are at the heart of tax compliance in all developed economies. Even for income taxation, it is employers who bear the greatest burden of tax compliance for most citizens, as with PAYE in the U.K. This explains why VATs and the personal income tax grow in importance as sources of revenue.
There is a two way relationship between building a broad taxation and a market economy. Governments who achieve broad based taxation have strong incentives to create private-sector growth, the proceeds of which can be taxed and used to provide public services. Governments who lack broad-based taxation will use inefficient forms of taxation – particularly taxes on international trade, indirect forms of taxation such as licenses and user charges and the inflation tax.
Research and Policy Agenda
It is clear that building a more effective tax system is bound up with the standard array of policies which seek to promote development – both political and economic. But there are also specific issues around creating improved systems of revenue collection. Doing research on tax systems in developing countries is however often plagued by poor data availability. So evidence based intervention, especially where that evidence is drawn from studies specific to the country in question is frequently impossible. And poor levels of capacity for research and data management are an impediment also. Moreover, cultures of poor compliance are often allied to low levels of transparency and weak incentives to undertake projects which support the case for more evidence.
Recently, however, a new line of empirical research on tax enforcement is emerging which involves close collaboration between researchers and tax authorities in developing countries in designing and evaluating policy reform experiments, often randomized control trials. These studies have the potential to contribute to an evidence base that draws on ideas generated from frontier research and is grounded in the policy realities of developing countries. What helps gain local support for such studies is the value of working with researchers rather than consultants and donors. The motivation is to generate a local evidence base for effective reforms. For example, an ongoing study in Punjab, Pakistan by Ben Olken (MIT) and colleagues sponsored by the International Growth Centre aims to inform policy on revenue collection, corruption and the overall efficiency of tax collection system by designing and evaluating wage and incentive schemes for tax officers. Pakistan has one of the lowest, and declining, tax-to-GDP ratios in the world.
The government has fully approved the project, including randomization of tax collectors to treatments which are large: in the wage treatment, officials’ wages are almost tripled, and in incentive treatments, incentive payments equal an average of 30% of the increase in tax revenue. The study is also developing objective performance indicators for tax officials by utilizing the frequent movement of tax inspectors to different tax circles to estimate the extent to which individuals affect tax performance. It will examine the correlates between tax inspector characteristics and their fixed effects to understand the determinants of effective tax performance. This high degree of cooperation represents a unique example of intellectual and financial collaboration between researchers and policymakers-the interventions were jointly designed, the government is bearing the intervention and researchers its evaluation costs.
Raising wages to help reduce rent seeking is a cornerstone of many tax reform policies (e.g. the salaries of selected employees of the Federal Board of Revenue in Pakistan were doubled as a part of a reform). Despite the considerable cost of such pay raises, there is surprisingly little empirical evidence on the topic. From a policy perspective, since none of these reforms have been rigorously evaluated there is little evidence to formulate more effective policies. One can categorize incentive-based approaches to along two broad literatures. One suggests that public servants are underpaid, leading to low effort and a high likelihood of rent seeking. The other suggests that the problems stem from low-powered incentives and that reward and penalty mechanisms based on performance are needed. The Pakistan study contributes to both by examining both wages and incentives.
There are several theoretical reasons why higher wages could reduce rent seeking and enhance effort. In economics, the literature on the role of compensation for public servants in preventing corruption starts with Becker and Stigler (1974). In their model of efficiency wages, higher wages – coupled with monitoring – prevent corruption because they provide rents to the public servant that can be taken away in the case of malfeasance. Another commonly articulated argument is that there is a positive income elasticity of honesty: bureaucrats are corrupt "to feed their families," so bureaucrats with higher wages can consume the luxury good of honesty. A third story emphasizes selection: if wages are low, only dishonest people will bother to join the civil service (since only they can obtain a reasonable net wage), whereas with high wages, honest people may join as well. All three stories suggest that raising wages will reduce corruption and increase tax revenue, but they differ in the mechanism. With efficiency wages, the reduction comes by pairing higher wages with monitoring; with honesty as a luxury good, the reduction comes even in the absence of robust monitoring; with selection, the reduction comes over time as new, more honest people select into the civil service. This study is exploring the first two of these theories and is also exploring with government the feasibility of looking into selection effects by allowing new recruitment into vacant tax positions.
Another common approach is the use of output-based incentives in combination with audit mechanisms. Output-based incentives have been shown in many contexts to increase effort, though there is little evidence for civil servants. Output based incentives are more feasible for tax collection than other bureaucracies because tax revenue constitutes a clear output on which to base incentives. One concern with incentives for tax collectors is that they could lead to overzealousness and extortion. This study is therefore investigating not only whether incentives lead to an increase in revenue, but also whether they lead to extortion or over-assessments.
The pilot for the study with 11 tax inspectors has completed, while the l arge scale experiment with 151 inspectors is ongoing since July 2011 . While results from the main experiment are still awaited, the pilot results suggest a positive impact on total revenue but that it is sensitive to the credibility of reforms. Also, due to small sample size it is not yet possible to distinguish between the impacts of different treatments. From gov ernmen t perspective, the findings suggest a positive NPV and thus a cost-effective project.
Another International Growth Centre study currently underway in collaboration with the National Board of Revenue in Bangladesh is looking at the role of incentives on firms’ behaviour. Raj Chetty (Harvard) and colleagues are currently examining whether social rather than economic incentives can encourage tax compliance. In particular, they are exploring if concerns for social status can act as a ‘carrot’ to pay taxes. Henrik Kleven (LSE) is currently working with the Federal Board of Revenue of Pakistan through IGC to look into behavioural responses to their tax system (individual and corporate income tax, and VAT) with a view to reducing distortions and increasing tax revenue. This kind of work is in its infancy but shows how a proper collaboration between research expertise and a revenue authority can potentially pay dividends.
It is hard to divorce the creation of fiscal capacity from more general forces that shape state effectiveness and economic development. Countries that have the underlying preconditions for growth and state effectiveness will tend to build effective fiscal systems along the way. Giving aid and support to such countries may accelerate this process but the case for external intervention is weak. Countries such as India and China which are on favorable growth trajectories are fully aware of the need to develop their fiscal systems with or without external encouragement. On the other hand, African countries that are heavily dependent on natural resource rents and countries like Pakistan that have historically been heavily dependent on external inflows have little incentive to develop their fiscal systems which in turn negatively impacts their longer-term growth trajectory and their ability to develop accountable institutions.
More problematic are countries – particularly weak and fragile states – for whom the failure to build a fiscal system is simply a reflection of the state of the economy and polity. This only reinforces Schumpeter’s famous comment from nearly a century ago that "The fiscal history of a people is above all an essential part of its general history. An enormous influence on the fate of nations emanates from the economic bleeding which the needs of the state necessitates, and from the use to which the results are put." (Joseph Schumpeter, The Crisis of the Tax State, 1918). But where there is a will to embrace reform, there is also scope to encourage governments to undertake more ambitious and innovative schemes for tax collection in collaboration with researchers that are both carefully conceptualized and scientifically evaluated. Since greater revenue mobilization and improved public sector performance are among the top priorities of developing countries, the lessons from these studies are likely to broadly influence both the intellectual and policy debate on the topic.
Whether donors can incentivize governments to undertake such reforms raises some important issues. Aid conditionality rightly remains a contentious topic. And in the area of taxation, withdrawal of aid as governments increase domestic tax revenue take would generate perverse incentives. Thus donors need to think of means of rewarding governments for increasing their tax take, particularly when this comes from investments that reform the system and are likely to lead to long-run and sustainable progress. Providing support for better research, not just conventional technical assistance, is also an important consideration. Our experience in working with the Pakistan tax authority shows that there is potential when there is buy in from the authority in question and international expertise is on hand to participate in research design.
February 21st 2012
Baunsgaard, Thomas and Michael Keen , "Tax Revenue and (or?) Trade Liberalization", mimeo, IMF.
Becker, Gary S., and George J. Stigler. . "Law Enforcement, Malfeasance, and Compensation of Enforcers". The Journal of Legal Studies, 3 (January): 1–18.
Besley, Timothy and Torsten Persson [2011a], Pillars of Prosperity: The Political Economics of Development Clusters, Princeton: Princeton University Press.
Besley, Timothy and Torsten Persson [2011b], "Public Finance and Development" draft chapter for the Handbook of Public Economics.
Jensen, Anders , "Natural Resources and Ability to Tax", mimeo, LSE.
Kleven, Henrik, Martin Knudsen, Claus Thustrup Kreiner, Soren Pederson, and Emmaneul Saez. 2011. "Unwilling or Unable to Cheat? Evidence from a Tax Audit Experiment in Denmark". Econometrica 79, pp. 651-692.
Mitchell, Brian R. , "International historical statistics: Africa, Asia and Oceania 1750-2005;
the Americas, 1750-2005; Europe, 1750-2005", Palgrave Macmillan.
 The following charts and discussion are taken from Besley and Persson (2011b).
 This sample only includes a number of countries that existed already in 1900, where we are reasonably confident that the data are comparable across countries and time. The countries in this sample are Argentina, Australia, Brazil, Canada, Chile, Colombia, Denmark, Finland, Ireland, Japan, Mexico, Netherlands, New Zealand, Norway, Sweden, Switzerland, United Kingdom, and the United States.
 This implies that developing an informational environment in which third-party reporting makes it difficult to cheat can reduce the level of tax evasion. I n a n experiment conducted in collaboration with Danish tax authorities, Henrik Kleven (LSE) and co-authors (2011) find that announcing high probabilities of future audits generated substantial future tax revenue through behavioural responses to a higher perceived probability of detection.
 Other taxes not included in either trade or income taxes include indirect taxes such as VAT, property and corporate taxes.