1 Why are tax havens more harmful to developing countries than to other countries?¹

Memorandum written for the Commission to the Government Commission on Tax Havens

Date: 15th May 2009 by Ragnar Torvik, Department of Economics, NTNU, ragnar.torvik@svt.ntnu.no

1.1 Introduction – The effect of tax havens

Tax havens provide an economic opportunity to enrich oneself at the expense of society. When this opportunity is used, it lowers the public income of the countries where taxable income or wealth is withheld. In jurisdictions that are themselves tax havens, however, the effect will possibly be opposite – here the tax revenue could increase since some of the income or wealth will be taxed here. It is, however, obvious that the net effect of tax havens is a reduction in the total amount of tax paid – if this were not the case, the taxpayer would have no incentive to conceal income or wealth in tax havens.

This appendix studies the effects of the existence of tax havens on countries that are not themselves tax havens. A different question, which is not studied here, is the effects of being a tax haven. There is reason to believe that these effects are positive – it is in the interest of the Cayman Islands to be a tax haven. However, that is of little interest to the problems studied in this appendix – an analogy clarifies that the policy conclusion is not to stimulate tax havens: If a thief becomes richer by stealing this is no argument to allow theft. We must distinguish between created income and income derived by taking from others.

For countries that are not themselves tax havens, the effect is that tax income is lower than it would otherwise have been. The sign, then, is clearly negative. The magnitude is an empirical question. As the discussion in the Commission’s Chapter 5 shows, it seems that the sums withheld from taxation are substantial. For industrialised countries, there is reason to believe that the funds withheld from taxation are mainly private income. In developing countries, there are many examples of the great extent to which public funds, too, are concealed in tax havens, to enrich corrupt bureaucrats and politicians. DRC (Democratic Republic of Congo, Zaire), where Mobutu Sese Seko held power from 1965 to 1997, is a well-known example. Tax havens helped Mobutu conceal the great wealth his political position enabled him to steal. Acemoglu, Robinson and Verdier (2006, page 169) assert: “There is no doubt that the aim of Mobutu was to use the state as enrichment for himself and his family. He was a true kleptocrat.” The authors show that the consequences were catastrophic. The country’s natural resources and economy were plundered by the political elite. Income per capita in 1992 was half of what it was at independence in 1960.

The example of Mobutu shows that tax havens give the political elite in developing countries an instrument for concealing the wealth they plunder. The example also shows the key role played by a county’s resource wealth. For corrupt politicians, natural resources are a possibility for personal enrichment, and tax havens represent a means to achieve this end. The example also shows that the types of institutions that serve the interests of politicians are not necessarily those that serve the interests of the general population. Weak public institutions, corrupt bureaucracy, and little democracy were preconditions that made it possible to conceal such substantial sums in tax havens. The institutions that served Mobutu’s interests had different characteristics from institutions that would have served the interests of the general population – and he changed the institutions and undermined democracy to serve his own narrow self-interest. The halving of domestic income per capita cannot be explained by the direct effect of Mobutu’s plundering alone: “In the 1970s, 15-20% of the operating budget of the state went directly to Mobutu.” (Acemoglu, Robinson and Verdier 2006, side 169). In order to explain why per capita income was halved, we have to take into account the indirect damage the regime inflicted on the economy. By all accounts, this indirect damage was far greater than the sums that went directly to the private enrichment of the regime and its supporters. To understand the effect of tax havens on developing countries we should not confine ourselves to a traditional analysis that assesses damage only in terms of what is withheld or lost in public tax income.

In the following, least space will be devoted to the traditional effects of tax havens and most space to the effects specific to developing countries, effects that have been less analysed in the past. This is not because of a conviction that the traditional analyses are less important. The background for these priorities is rather that the effects that are common to industrialised countries and developing countries are so well documented in previous literature that we are here on firmer ground. In this case, there is less reason to enter into details. The effects that are specific to developing countries have been less analysed. Here, then, we will enter more into details, and contribute some new, and hopefully original, analysis. At the same time, this will mean that chapter 3 of this appendix may seem more speculative than the survey of well-established knowledge.

In 1.1 and 1.2, we briefly treat the effect of tax havens on public and private income. In chapter 2, we first discuss theory for the effect of tax havens on developing countries. We show why and how the increased opportunities for private income represented by tax havens can in reality serve to reduce private income. It is a challenge that it is difficult to find data for an empirical investigation of whether tax havens contribute to a reduction in private income. Tax havens and secrecy go hand in hand. Instead, the effect of other income that can give rise to similar mechanisms as tax havens are studied. Central here is the study of the paradox of plenty – the fact that countries rich in natural resources seem to have had lower growth rates than other countries during the past 40 years. The paradox of plenty is relevant for the study of tax havens because the effects of resource income can give rise to many of the same mechanisms as tax havens – in particular, they give rise to economic adaptations whose motive is the reallocation of existing income in favour of oneself, rather than the creation of new income. There is, however, reason to believe that the income opportunities provided by natural resources have a more benign effect than do tax havens – the extraction of natural resources also contributes to value creation. Tax havens give rise to adaptations that increase personal income, not by increasing the total value creation, but by changing the allocation of resources with the aim of paying less tax. Drawing conclusions on the effects of tax havens based on the effects of resource income will give skewed results – the effects of tax havens will by all accounts be less benign.

The paradox of plenty is central to the study of tax havens for another reason as well: Tax havens help corrupt politicians and destructive entrepreneurs to plunder a country. Recent research shows that countries with bad institutions and political systems are hardest hit by the paradox of plenty. Private entrepreneurs and politicians in such countries are faced with private incentives that are not compatible with the best interests of society as a whole: The combination of bad institutions and tax havens gives corrupt politicians and destructive entrepreneurs good possibilities to stash away the resource income they plunder.

Chapter 3 treats the effect of tax havens on institutions and democracy. While the general population may have an interest in stable democracy and good public institutions, tax havens will contribute to giving politicians an interest in the opposite. There are many examples that institutions aimed at stemming illegal flows of money are deliberately destroyed by the authorities, that people working in such institutions are pressured not to do their jobs, or even killed.

Incentives to change institutions so they correspond better to personal motives can also cast a light on important changes to the political systems in many developing countries. One such change is the increased incidence of presidential rule in many developing countries, particularly in Africa. Most African countries started off with a parliamentary system at independence. Today, nearly all the countries have changed their constitutions to a presidential system. This massive change may seem counter-intuitive, as research indicates that presidential rule in Africa gives policies that are less responsive to the interests of the general population than a parliamentary system does. Why has it been so tempting for the political elite in Africa to introduce presidential rule? One possible hypothesis is that in some countries presidential rule is established not because it serves the general population, but because it concentrates power with an economic and political elite.

Further in chapter 3, there is a discussion of how tax havens may influence the degree of democracy. Much recent research indicates that important economic motives may lie behind conflicts – it may be in the narrow economic self-interest of opportunistic agents to weaken democracy. Again, tax havens play a major part – they provide an opportunity to stash away the income plundered through the undermining of democracy, and make such a strategy more tempting.

If tax havens contribute to making a country’s institutions worse and to destabilizing democracy, the effects for the general population may be disastrous. Not only do tax havens contribute to lower incomes – they also contribute to undermining the basis of future growth and welfare.

While the contents of chapter 2 are based mainly on previously published research, part of what is presented in chapter 3 is based on research which has not yet found its way into scholarly journals, and which therefore exists only as unpublished notes. This means that what is presented here is to a lesser extent established knowledge, and must to a greater extent be regarded as speculation. However, the fact that the field of research is new is no argument for omitting it. The reader should nonetheless keep in mind that much of what is presented in chapter 3 is yet unpublished, and has therefore not undergone the quality control that underpins published research.

1.1.1 Tax havens – Reduced Public Income

When private agents take advantage of tax havens, tax revenue will be reduced. Lower public income will consequently lead to a combination of reduced public activity, and higher taxes on other activity. Compared to more developed countries, however, developing countries will find it more difficult to levy alternative taxes. A low level of economic activity means that there are few alternative tax bases. Deficiencies in institutional capacity and large informal sectors make tax collecting difficult. There is therefore a limited potential for alternative tax revenue. The loss of income caused by tax havens must mainly be compensated through a reduction in public economic activity.

At the same time, the need for public services and public investment is great in developing countries. Poverty, disease, and a limited level of education produce strong pressures and a great need for public services. Poor infrastructure makes public investment very necessary. Tax havens therefore contribute to reducing public activity in the countries where it is most needed. Tax havens reduce public income both in industrialised countries and in developing countries – but the negative effects of lower public income are greater for developing countries.

Tax havens make it relatively more profitable to apply talent to increasing the profitability of undertakings through tax evasion, than to apply talent to increasing profitability through operating more efficiently. Such skewing in the application of talent in the private economic sphere does not profit society, because society’s economic calculations must take into account that saved tax for the private entrepreneur is equal to a reduction in public income. Again the harm may be greater in developing countries, because entrepreneurs here are more of a scarce resource than in industrialised countries, and the misuse of such talent therefore has greater consequences.

In many developing countries, the quality of the legal system is lower, corruption is higher, and the public bureaucracy is less competent. This makes the probability of getting away with tax evasion high, and the cost of not getting away with it low. Since the use of tax havens is a greater temptation and carries a lower cost in developing countries, this, too, will ensure that the negative effects will be greater than in countries with a better legal system, less corruption and a more competent public bureaucracy.

1.1.2 Tax havens – Increased opportunities for private income

One potential argument against the assumption that there are costs connected with tax havens is that when the profitability of economic activity increases, economic activity will increase. One positive side effect of tax havens is thus that they can give rise to the development of industry and thus contribute to economic growth. Note, however, that this reasoning does not hold water. If tax havens lead to economic resources being used differently than would otherwise have been the case, the reason for this would be to save tax. Otherwise, resources would have been moved into this alternative activity even if there were no tax savings. Tax motivated movement of resources will therefore typically lead to an increase in private income smaller than the amount saved in taxes. Consequently, moving resources into other activity as a consequence of the existence of tax havens will reduce total value creation, not increase it: Total value creation is the sum of public and private income – and this has gone down.

1.2 Tax havens and developing countries

So far, we have seen that tax havens influence developing countries and well developed countries through the same mechanisms, but that the negative consequences are stronger for developing countries. In developing countries, however, mechanisms other than those relevant to more developed countries also make themselves felt. These mechanisms have their basis in, and are made worse by, the poor quality of public institutions and the weakness of the political systems of these countries. In their turn, tax havens will make the institutions even worse and the political systems even weaker. In this way, institutions and political systems are, on the one hand, a cause of the strong negative effects of tax havens on developing countries, and, on the other hand, tax havens are a cause for the declining quality of institutions and political systems. Such a negative spiral has grave consequences for economic and political development.

1.2.1 Economic opportunities and economic outcomes

Although the argument that tax havens give increased possibilities for income has the wrong sign, it may be the case that the greater income opportunities provided by tax havens can dampen their negative effect. If the increased opportunities for private income represented by tax havens result in higher private income, their effect will be less dramatic than the impression given by measuring the total loss in tax income. The economically relevant measure for society is the loss of tax income minus the increase in private income. If tax havens give increased private income, they would represent a cost, but this cost would be lower than one might first believe.

1.2.2 Theory: Tax havens in developing countries

Let us assume a country with a high crime rate, widespread corruption, a poor quality of public bureaucracy and a weak political system. In such a society, it will be relatively attractive for people to engage in destructive banditry, corruption, rent seeking, tax evasion etc., rather than to establish and operate productive enterprises. For such an economy, a decisive point is how tax havens influence the incentives for talented entrepreneurs, and what effect this will have on the economy as a whole. The theory presented in the following is based on Torvik (2002).

It is clear that the higher the number of entrepreneurs who choose to engage in productive activity, the higher the income of each entrepreneur will be. There are many reasons for this. More entrepreneurs engaged in productive activity means fewer in destructive activity, and thus less crime and corruption. In its turn, a reduction in crime and corruption makes it more profitable to engage in honest economic activity. A higher number of entrepreneurs in productive activity gives higher production, income and thus greater demand. Greater demand in its turn increases sales and profitability. A higher number of entrepreneurs in productive activity gives higher tax income, greater public income, and thus better public services and infrastructure. Good public services and infrastructure in their turn increase the profitability of private industrial activity. Figure 1.1 shows this relation. We measure the number of entrepreneurs in productive activity along the horizontal axis and the income of each entrepreneur along the vertical axis. The rising curve shows that the income for each entrepreneur increases with the increase in the number of other entrepreneurs engaged in productive activity.

Figure 1.2 shows how the number of entrepreneurs who choose to engage in unproductive or destructive activity influences the income for each and every one of them. We measure the number of entrepreneurs in such activity from left to right, and the income for each of them on the vertical axis. An increase in the number of entrepreneurs in unproductive activity, i.e., a movement from right to left in the figure, implies a lower income for each and every one of the unproductive entrepreneurs. There are several reasons for this. A large number of criminal competitors reduces the income of each single criminal. A great number of entrepreneurs in destructive activities implies that there will be fewer entrepreneurs in productive activity. Then, there will be less production, less to steal and lower income for each criminal.

Figure 1.3 shows how the country’s entrepreneurs are distributed between productive and unproductive activity. Where the curves intersect, the income for entrepreneurs in productive activity and in unproductive activity is equal, and no entrepreneur has an incentive to switch between productive and unproductive activities. If the distribution of entrepreneurs is not at the intersection in the figure, but to left if it, income from productive activity will be higher than income from unproductive activity. In that case, entrepreneurs will have an incentive to switch from unproductive to productive activity. The number of entrepreneurs in productive activity will then rise, and the number in unproductive activity will decline, until we are back at the intersection. If we are in a situation to the right of the intersection, we will see the opposite movement. Here, the income of the unproductive entrepreneur is higher than the income of the productive one, and entrepreneurs will switch from productive to unproductive activity until we are back in the intersection, where the income of an entrepreneur is equal in unproductive and in productive activity. The intersection and the distribution of entrepreneurs between productive and unproductive it implies, therefore represents a situation of stable equilibrium.

The question is now how tax havens influence this equilibrium in the economy. This is shown in Figure 1.4. Tax havens represent improved income opportunities for those who are engaged in destructive banditry, corruption, rent seeking, tax evasion etc. When the possibilities for evading taxes improve, it becomes relatively more attractive to be a rent seeker. The curve that represents income opportunities in rent seeking thus shifts upward in the figure, as shown with the stippled curve in Figure 1.4. The conclusion is surprising: Better income opportunities for every entrepreneur cause the income of every entrepreneur to fall!

The intuition behind this result is as follows: Tax havens make it more profitable to be a rent seeker, and thus more people choose rent seeking and fewer choose productive activity. A greater number of rent-seekers and a smaller number of productive entrepreneurs reduce the income of every productive entrepreneur. However, lower income from productive activity makes it relatively even more attractive to engage in rent seeking. This leads to a further fall in the number of productive entrepreneurs, a further fall in the income for those who are engaged in productive activity, an even greater reduction in the number of productive people and an increase in the number of rent seekers, and so on. Tax havens thus provoke a multiplicative process – but the multiplicator is negative. The reason this process will not continue indefinitely is that the income of rent seekers, too, falls when there are more rent-seekers and fewer productive agents.

In the new economic equilibrium it must be the case that what at first seemed like a better income opportunity for private entrepreneurs is a disadvantage for each and every one of them: Tax havens make the income of every entrepreneur go down – not up, as one might believe at first glance.

In developing countries, then, tax havens not only lead to lower public income – there is also reason to believe that they lead to a fall in private incomes. There is no conflict between private and public sector – tax havens are a disadvantage for both.

1.2.3 Theory: Tax havens and poverty traps

In the figures studied above, the income curve for rent seeking was steeper than the income curve for production. Now assume that the opposite is the case, as shown in Figure 1.5. Here the advantages for other entrepreneurs that there are many productive entrepreneurs are relatively great, while the disadvantages for rent-seekers of many other rent-seekers are relatively small.

At first sight one could believe that this figure gave the opposite conclusion of Figure 1.4: If we now make rent-seeking more advantageous and shift the curve upwards, Figure 1.6 shows that the curves will intersect at a higher level of income. Tax havens thus seem to give higher private income.

However, this is a logical fallacy. To se this, study Figure 1.5 again. If, in this figure, there are more productive entrepreneurs and fewer rent-seekers than at the intersection – i.e., if we are to the right of V* – then the income of productive entrepreneurs is higher than the income of rent seekers. This leads to a switch from rent seeking to productive entrepreneurship. If we start out to the right of V*, we will over time move further to the right in the figure – until we are in a situation where all entrepreneurs in the economy are engaged in productive activity. If we start out to the left of V*, the income of productive entrepreneurs is lower than the income of rent seekers. Since it is more profitable to be a rent seeker than to be productive, the number of entrepreneurs will decrease, and the number of rent seekers will increase. If we start to the left of V* in the figure, we will over time move further to the left in the figure – until we are in a situation where no entrepreneurs find it profitable to engage in productive activity:

V* is not a stable equilibrium point – but a tipping point. There are two stable equilibrium points in the figure – either all are rent seekers, or all are productive entrepreneurs. The equilibrium where all are rent seekers is a poverty trap. Precisely because all are rent-seeker and no one a productive entrepreneur, it will be more profitable for each individual to be a rent-seeker rather than a productive entrepreneur. The situation is therefore stable – but not inevitable. If all chose to be productive entrepreneurs, then every individual would be better off as an entrepreneur than as a rent seeker. In this equilibrium, everybody’s income would have been greater.

It is therefore clear that what Figure 1.6 really shows is that the critical mass of entrepreneurs needed for the outcome to be the good equilibrium increases from V* to V**. Tax havens make it more difficult to reach the good equilibrium, and make it easer to end in the poverty trap. If tax havens contribute to turning an economy that would otherwise have ended in a good equilibrium into a poverty trap, the growth effects of tax havens are worse – and not better – than indicated in the analysis above, where we had a unique equilibrium.

1.2.4 Empirical study: Do increased income opportunities reduce income?

We have seen that economic theory suggests that in developing countries the increased income opportunities represented by tax havens reduce rather than increase income. This may seem counter-intuitive – it would seem more reasonable if the profitable economic opportunities available to agents in a society materialized as good economic outcomes. However, the “income opportunities” tax havens represent are in reality income acquired by agents without their creating any value. There is reason to believe that the effect of this type of income opportunity in developing countries is negative rather than positive. However, it is difficult to conduct empirical research on the effect of tax havens – secrecy lies in their nature. Because of this, it is difficult to find out whether countries where the population makes widespread use of tax havens have a lower income than other countries. Even if the investigation of the negative influence of tax havens on income is difficult, however, some light can be shed on the problem indirectly in a different way: One may find other income opportunities that represent income appropriated by agents in the economy without value-creation – and investigate whether these lead to reductions or increases in income.

One such income opportunity is natural resources, like for instance oil and diamonds. This type of income accrues largely because of the accident that countries have natural deposits – and to a lesser degree because the agents of these countries are particular good at conducting industrial activity. This does not mean that it is not an advantage to have entrepreneurs who are good at converting such income opportunities to actual income. On the contrary – such countries may draw greater benefits from naturally given income opportunities than other countries.

Income from natural resources represents a greater value for society than “the income” derived from tax havens – this last income come in its totality from a reduction in public income. Consequently, there is reason to believe that income opportunities from natural resources will have a more beneficial effect than income from tax havens. If we use income from natural resources to investigate the effect of increased income opportunities, we will therefore underestimate possible negative effect of tax havens.

There is also another reason that the following discussion on how resource income affects an economy will be fairly detailed: Tax havens are part of the explanation why income from natural resources can be harmful to a country.

1.2.5 The paradox of plenty

The last decade has seen the appearance of an extensive international research literature that argues that countries with much income from natural resources end up at a lower income level. This phenomenon is often called ”The resource curse” or ”The paradox of plenty”. 2

It is important to note that this literature is relevant to the effect of tax havens in two ways. Firstly, this literature will, as argued above, represent an indirect test on how income opportunities that give rise to rent seeking can affect an economy. Secondly and maybe more importantly: Tax havens are a natural part of the paradox of plenty. As we will see, it is precisely when economic and political agents have the opportunity to conceal income from natural resources that natural resources are harmful. In the opposite case, it is not harmful for an economy to gain income from natural resources.

Figure 1.7 illustrates the paradox of plenty. Each point in the figure represents a country – in total 87 countries are represented, and they are all the countries for which we have the necessary data. All data used in this empirical analysis are taken from Mehlum, Moene and Torvik (2006a), and the data for all the countries are reproduced in the Appendix at the end of this memorandum.

The horizontal axis in Figure 1.7 shows the countries’ exports of natural resources as a share of total production (GDP) in 1970. Countries far to the right in the figure, then, largely specialize in the export of natural resources, whereas countries far to the left in the figure have little export of natural resources.

The vertical axis shows average annual economic growth since 1965. Countries high up in the figure thus have high growth, whereas countries low down in the figure have low growth.

Source Data source: Mehlum, Moene and Torvik (2006a).

The solid line in the figure is a regression line that shows the connection that best represents the relation between exports of natural resources and growth – countries with much export of natural resources have lower growth than countries with little export of natural resources. Natural wealth is accompanied by poor economic growth – and poverty from nature’s part is accompanied by good economic growth.

1.2.5.1 Correlation and causality

The correlation shown in Figure 1.7 has proven quite robust – can it then be said that exports of natural resources cause lower growth? No, and as we shall see, there are several reasons for this. The figure nonetheless represents a useful point of departure. However, to uncover the possible effect of the export of natural resources we must dig deeper. We must check whether the correlation holds water when we take into account that there may be other factors that affect both the export of natural resources and growth.

Assume that countries with substantial natural resources protect their markets from foreign competition in order to build up their own industry. Assume also that such protection does not in reality contribute to growth, but to inefficiency. In that case, we cannot based on Figure 1.7 argue that the export of natural resources leads to lower growth. Natural resources per se are not the problem for growth – but the fact that countries with substantial natural resources pursue policies that give inefficiency and low growth. Had these policies not been pursued, growth might have been as high as in countries with few natural resources. In other words, we lay the blame for low growth on the abundance of natural resources, when the blame should have been laid on the policies.

Another example: Some countries are poor and others are rich – in poor countries, production is low and in rich countries, production is high. If rich and poor countries have an equal amount of natural resources, it will represent a large share of production in poor countries and a small share of production in rich countries. When we measure natural resource exports as a share of total production, poor countries will, all other things being equal, present themselves as rich in natural resources – and rich countries as poor in natural resources. If poor countries grow more slowly than rich countries, Figure 1.7 only shows that poverty begets poverty – whereas wealth begets wealth. Export of natural resources may be completely irrelevant as an explanation of differences in economic growth. Figure 1.7 confuses the effect of natural resources with the effect of low income.

A third example: In poor countries the establishment of industrial activity will often give little profit – the only activity that gives economic yield is the cutting down of forests, the excavation of diamonds, or the pumping up of oil. In this case, Figure 1.7 does not show that the export of natural resources gives poverty – but that poverty gives specialization in the export of natural resources.

The examples illustrate a general point – there is a difference between correlation and causality. From Figure 1.7 we do not know whether it is low growth that leads to high exports of natural resources, or whether it is high exports of natural resources that lead to low growth, or whether there is a third factor (for instance politics) that co-varies with both exports of natural resources and growth. To investigate the effect of natural resources on economic growth it is not enough to look at these two variables in isolation – we must also include other factors.

Table 1.1 seeks to take this into account by including more variables. The table shows different variables that affect economic growth in the 87 countries with available data for the variables used. When a variable in the table has a positive sign, it means that a higher value of this variable increases growth in the countries, whereas a negative value means that the variable reduces growth. When a variable is marked with *, it means that it is reasonably probable that the variable’s co-variation with growth is not caused by pure coincidence (in the sense that the estimate is significant on a 5% level). Adjusted R2 shows how great a part of the variation in countries’ growth rates is explained by the variables in the analysis.

If the analysis in Figure 1.7 confuses the effects of being poor with the effects of high exports of natural resources, we can control for this by including the level of income in our analysis. In Regression 1 in Table 1.1, this is done by including a measure for income level in the various countries at the beginning of the period under investigation. We see that this variable has a negative sign – the countries that had a high income in 1965 have had a lower average growth after 1965 than the countries that had a low income. Thus, it seems not to be the case that rich countries on average grow more quickly than poor countries – rather, it is opposite – poor countries grow faster than rich countries.

In the light of the discussion above, we must also control for whether the countries have pursued economic policies that have largely sheltered them from foreign competition. In Regression 1 in Table 1.1, this is done by including a variable for countries’ freedom of trade in the period. We see that this variable has a positive sign – the countries that have largely pursued free-trade policies have, on average, grown faster than the countries that have to a greater extent sheltered themselves from foreign competition. Openness has contributed to faster growth. Countries that have shut themselves off from the outside world have had lower growth.

The most interesting point of Regression 1 in Table 1.1 is, however, that the tendency from Figure 1.7 is still present – countries with high exports of natural resources have grown more slowly than countries with low exports of natural resources. If a country increases its exports of natural resources as a share of GDP by 10 percentage points, the estimate indicates that its annual growth will decrease by 0.62 percentage points. This is a powerful effect – the difference between a growth of one percent and a growth of two percent is not one percent, but one hundred percent. If the effect is relevant for Norway, it indicates for instance that the export of oil and gas of 26% of GDP in 2008 reduces growth by 1.6 percentage points. That is more than half of a “normal” Norwegian growth rate. 3

Even if we have controlled for initial income and trade policies, the estimate may still not reflect how the export of natural resources influences growth. It is reasonable to assume that many of the countries that export natural resources have a weak protection of private property rights, much corruption, and a low-quality government bureaucracy. If we do not control for this, the weak economic development will be ascribed to the export of natural resources, while the real problem is poor public institutions. Regression 2 in Table 1.1 shows the effect of controlling for institutional quality. As we see, an improvement in institutional quality has a positive effect on economic development. For example, the analysis predicts that a country like Mexico – which has institutions of medium quality, with a score of 0.54 on a scale where 1 is the highest, would have had an annual economic growth 0.9 percent higher if its institutions were as good as Norway’s, with a score of 0.96 of 1. This accounts for the entire difference in growth between Mexico (2.2 percent annually) and Norway (3.1 percent annually) in the decades after 1965.

But again – the effect of an abundance of natural resources is present even if we control for institutional quality – from Regression 2 in Table 1.1 we see that it is significant and approximately as strong as in Regression 1. Even when we take into account the possibility that countries rich in natural resources may have institutions of poorer quality, growth is still lower in these countries.

Another possibility is that that countries rich in natural resources have a poorer investment climate than other countries – an abundance of natural resources can for example give rise to a feeling that it is not so important to stimulate investment because “we have enough to live on anyway”. If we not control for investment climate, weak growth could be ascribed to natural resources – even though the investment climate is the real problem. Regression 3 in Table 1.1 shows the effect of controlling for investment as a share of GDP. As expected, there is a close correspondence between investment and growth – higher investment gives higher growth. If Turkey, which had investments equivalent to 22,5 % of GDP in the period, had invested as great a part of its income as Norway, which invested 32,5 % of GDP, the analysis indicates that annual growth in Turkey would have increased from 2,9 percent to 4,4 percent.

However, we see in Regression 3 in Table 1.1 that the export of natural resources still has a negative effect on growth – the effect is significant and somewhat stronger than in the two previous Regressions. Even when we control for possible differences in investment climate, countries with abundant natural resources grow more slowly than countries with scarce natural resources.

It is also thinkable that there are other variables associated with an abundance of natural resources that affect growth – and which should therefore be controlled for. The reader who wants to study these effects more in depth is referred to Sachs and Warner (1995,1997) and Mehlum, Moene and Torvik (2006a,b). These analyses study the effect of controlling for variables like for instance level of education, income distribution, ethnic fractionalization, unstable terms of trade, and the size of the agricultural sector. These studies conclude that the negative effect of natural resources on growth is robust also when these factors are controlled for. 4

We see, then, that on average an abundance of natural resources leads to lower economic growth in a country. Income that agents in the economy to a large extent appropriate although they do not create much of value can have unfortunate effects on the economy. Domestic income may be reduced and not increased. Tax havens, too, give agents in the economy income opportunities even though they do not create anything. Taken in isolation, then, there is reason to believe that the net effect of tax havens are negative, particularly if we take into account that the income opportunities represented by tax havens in all probability have a less benign effect than income from natural resources.

Nonetheless what may be the most important question remains: Are there systematic differences between those countries where unproductive income opportunities are particularly damaging and those countries where they are less damaging? The empirical literature on the paradox of plenty has until recently focussed on the average effects of resource income on growth. This is interesting enough in itself. But for every Nigeria or Venezuela there is a Botswana or Norway – what is most interesting is not that natural resources on average can give lower growth – but that the differences between different resource-rich countries are so great. In order to understand the problem of tax havens and design policies we must study the variance rather than the average effects. When do unproductive income opportunities lead to economic failure and when do they not?

1.2.5.2 Institutional quality

Mehlum, Moene and Torvik (2006a) argue that natural resources will give different incentives in different countries depending on the quality of their public institutions. In countries where government effectively supports property rights, and where there is little corruption in the public bureaucracy, natural resources will contribute positively to growth. Natural resources in such countries will give private agents incentives to invest in exploiting this wealth – more natural resources therefore stimulate value creation and growth. However, in countries where property rights are poorly protected and there is much corruption, more natural resources will make it more tempting for entrepreneurs to purloin resource income rather than build up enterprises. Here, an abundance of natural resources will undermine investment, value creation and growth.

How can one then study whether natural resources have opposite effects on growth in countries with good and poor public institutions? The regressions in Table 1.1 estimate only the average effect of resources – to find out whether resources have different effects in different countries depending on institutional quality, one must include an interaction term, i.e., a term where resource wealth is multiplied by institutional quality:

(Resource exports • Institutional quality)

Regression 4 in Table 1.2 is the regression from Table 1.1 expanded by such an interaction term. The effect of an increase in resource abundance is now from Table 1.2 given by:

– 14,34 + 15.40 • (Institutional Quality)

This result strongly supports that the effect of resource abundance is opposite in countries with good and poor institutions. In countries with the worst imaginable institutions, the index for institutional quality has a value of zero – in that case the cross term is eliminated, and the effect of natural resources on growth is given by –14.34. In such countries, then, unproductive income opportunities are very detrimental to growth. In countries with the best imaginable institutions, the index for institutional quality has a value of 1 – in this case, the effect of more natural resources on growth is given by –14.34 + 15.40 = 1.06. In such countries, then, natural resources have a positive effect on growth. The growth effect of natural resources has opposite signs in countries with good and poor institutions. In countries with good institutions the paradox of plenty does not exist – here resource abundance becomes an advantage for economic growth.

From Table 1.2 we can also find how good the institutions need to be in order for the paradox of plenty no longer to be relevant. The positive and negative effects of natural resources on growth even out when:

– 14,34 + 15.40 • (Institutional Quality) = 0

This means that for countries whose institutional quality is greater than 14.34/15.40 = 0.93, natural resources do not have a negative influence on growth. Of the 87 countries included in the analysis, 15 have a value higher than 0.93 on the index for institutional quality. For this fifth of the countries with best institutional quality – among them Norway – resource wealth does not give lower growth.

A potential problem with regression analyses that include many countries is that there are other differences between Norway and Sierra Leone than those connected with institutional quality. More generally, it can be argued that the paradox of plenty may be relevant for Africa as the least developed continent, but not for the rest of the world. Regression 5 in Table 1.2 therefore excludes Africa from the analysis. However, we still see that the main results come through also if we study only countries outside Africa – the paradox of plenty is not a specifically African phenomenon.

Since one of the most marked characteristics of developing countries is low institutional quality, it is precisely in such countries that the income opportunities represented by tax havens inflict the greatest economic damage. The analysis above therefore supports the view that tax havens are far more damaging to developing countries than to industrialised countries.

As discussed above, the question of whether natural resources underestimate the problem of tax havens can still be raised. An abundance of natural resources has more positive effects on the economy than there is reason to believe that tax havens have. It is difficult to argue that agricultural production represents the same type of unproductive income opportunities as tax havens do. In this respect, the income provided by tax havens has more in common with diamonds and oil than with grain and vegetables. However, the data for resource wealth used in the analyses lump all natural resources together to a general measure of resource wealth. It should therefore be investigated whether this has any significance for the results – and more importantly – is it the case that those natural resources that can most be associated with unproductive income opportunities have the most detrimental effects on economic growth and development in developing countries?

Regression 6 in Table 1.2 gives an alternative measure for resource abundance – it only includes minerals and oil as a share of GDP. There are two interesting results to note. Firstly, the term that shows the negative effect of natural resources when institutional quality is low is stronger than before (-17.71 in column 3 against -14.34 in column 1). Secondly, the positive interaction term between natural resources and institutional quality is also stronger than before (29.43 in column 3 against 15.40 in column 1). These results imply that relative to natural resources in general, minerals/oil have a stronger negative effect on growth in countries with poor institutions and a stronger positive effect on growth in countries with good institutions. In oil economies, the difference between failure and success is greater than in countries that base their economy on other natural resources.

Boschini, Pettersson and Roine (2007) is the most thorough empirical analysis to date of how different types of natural resources influence growth – and how this depends on institutional quality. They use four different measures of resources, and show that the decisive factor for the effect on growth is the combination of institutional quality and the ease with which various natural resources can be seized. The worst thinkable effect of natural resources we see with diamonds in countries with poor institutions.

On this background it is reasonable to assume that the effects of tax havens on growth in countries with weak institutions are substantially worse than what one would believe by comparing with the effects of unproductive income opportunities in general. The types of incentive to which tax havens give rise are particularly damaging to growth precisely in those countries that most need growth.

1.2.5.3 Presidential rule versus parliamentary rule

In a new and very interesting contribution, Andersen and Aslaksen (2008) show the following: The paradox of plenty is relevant for democracies with presidential rule – but not for those with parliamentary rule. A simple way to show the result is Figure 1.8 and 1.9, taken from Andersen and Aslaksen (2008).

Figure 1.8 shows the link between resource wealth and economic growth for countries classified as democracies with presidential regimes. Here we see that there is a clear connection like the one in Figure 1.7 – countries with high exports of natural resources have low growth. The paradox of plenty is relevant for countries with presidential rule.

Source Andersen and Aslaksen (2008)

Figure 1.9 shows the link between exports of natural resources and growth for democratic countries with parliamentary systems. We see that among these countries there is no link between the export of resources and growth. The paradox of plenty does not apply to democratic countries with parliamentary systems.

Source Andersen and Aslaksen (2008)

This again indicates that there is a close link between politics and the paradox of plenty. However, we still have only a limited knowledge of why resource wealth is more damaging with presidential rule than with parliamentary rule. One hypothesis is that under parliamentary rule the government will to a greater extent reflect the wider interests of the people than under presidential rule. With a parliamentary system, the government depends on continuous support in parliament. Continual support of this kind in its turn requires policies that benefit broad strata of the people, and not just a limited power elite, or a narrow section of the population. Presidential rule, particularly of the type we see in many developing countries, is more characterised by the president’s strong personal power. Politicians come to depend on the president rather than the president on continual support from a broad stratum of politicians. In such a system, the president has great scope for adapting policies to the private interests of the power elite, rather than to what is in the best interests of the population.

1.2.6 Tax havens and private income – conclusion

In countries with weak institutions and with presidential rule, then, there is a tendency that what one would at first think should increase income on the contrary reduces it. Income opportunities that appear not because something is created, but rather because agents adapt to reallocate income in favour of them, have so strong damaging effects that they reduce total value creation and income.

The “income opportunities” represented by tax havens will for most developing countries contribute to a reduction and not an increase in domestic income. Countries with weak institutions will not be able to exploit such income opportunities productively – rather they will give rise to unproductive or even destructive activity.

In addition, the great majority of developing countries – practically all of Latin America and Africa – have a form of presidential rule where most power is concentrated in the hands of the president. Such political systems seem very vulnerable to the damaging effects of the type of income opportunities represented by tax havens.

In sum, the argument that tax havens also represent income opportunities does not seem valid for developing countries. The “income opportunities” represented by tax havens contribute to exacerbate rather than alleviate their problems.

1.3 Tax havens and changes in institutions and political systems

In most countries, the negative effects of tax havens will set in motion institutional and political changes adapted to limiting the problems. In well-functioning countries, this will be a natural response, which will reduce the damaging effects of tax havens. Again, however, it may be the case that developing countries have the opposite response – responses that contribute to exacerbating the damaging effects of tax havens.

For politicians in countries with strong institutions and political systems tax havens represent a problem – they hurt the economy and reduce public income. Institutional and political changes can, however, limit this damage. For politicians in countries with weak institutions and political systems, on the other hand, tax havens do not necessarily represent a problem – they can also represent a solution. Tax havens provide opportunities for concealing income derived from corruption and illegal activity, or income politicians have dishonestly appropriated from development aid, natural resources and public budgets. In short, tax havens improve the feasibility of regarding the country’s economy as one’s personal purse. In this way, the appearance of tax havens also gives rise to political incentives to dismantle rather than build institutions, and to weaken rather than strengthen the political system. There are many examples that institutions designed to work against corruption and tax evasion are deliberately weakened. It seems obvious that the reason for this is that certain agents see their self-interest served by making such institutions weak.

Ross (2001a) shows that in countries like the Philippines, Indonesia, and Malaysia the existence of rich tracts of rain forest contributed to the deliberate dismantling of state institutions by politicians. The rain forests provided the basis for opportunities to pocket large sums of money through ruthless exploitation of the forest – but for this to be possible, the state institutions established to counteract abuses and overexploitation had to be undermined. Politicians had incentives for dismantling institutions rather than building them – and the reason was the abundance of natural resources.

Ross (2001b) finds that countries with large oil deposits become less democratic. In such countries, democracy can represent a cost for politicians because it hinders them in using the large public income as they please. Large income from resources can therefore give political incentives for weakening democracy. In the same way, income opportunities provided by tax havens give politicians weaker incentives to enact democratic reforms, or can even give them stronger incentives to reduce the democratic control on those in power.

Collier and Hoeffler (2009) show how “checks and balances” – institutional rules that limit the political abuse of power and balance political power – enhance growth. However, they find that particularly in countries where such rules are important – for example because the country has substantial public income from natural resources – the rules are undermined by politicians. Again, the analogy to tax havens is obvious. It becomes in the interest of politicians to invest in a model of society where secrecy and opportunities for personal abuse of power are tolerated.

1.3.1 Presidential rule versus parliamentary rule

We have seen that there is a tendency that the political systems in countries with presidential rule are less able to use increased economic opportunities to achieve better political outcomes. There is reason to believe that tax havens do greater damage in such countries. This raises the question of why some countries have presidential rule whereas other countries have parliamentary rule. Until recently, the consensus was that this is largely determined by history, and is stable over time. Latin American countries have presidential rule. The most common understanding is that this was a political choice made when these countries became independent about 200 years ago, and that the political systems have since endured. In the same way, the fact that the majority of European countries are parliamentary is explained in terms of historical choices that have endured. Although this understanding may have its merits, it also raises some questions. For example – given that certain forms of presidential rule are economically and politically damaging – why do not countries with this form of government switch to another form? One hypothesis could be that in some of the countries with presidential rule this serves the members of the political and economic elite, consequently they do not see their self-interest served in a switch to another form of government, even though a switch would have been advantageous for the country as a whole.

The explanation that the political system is formed by historical choices and is therefore stable over time is inadequate. This can easily be seen if one concentrates on African countries. Here we find a remarkable pattern. When African countries gained independence, most of them had parliamentary constitutions. If we look at African countries south of the Sahara, we see from Table 1.3 that at independence, there were four times as many countries with parliamentary constitutions as with presidential rule. However, in country after country, constitutions were changed – and today only three of the 21 the countries that started out with parliamentary systems retain them. None of the countries that started out with presidential rule has changed to a parliamentary system. It is also remarkable that two of the three countries that still have parliamentary systems – Botswana and Mauritius – are the countries that have done best after independence.

Why have these countries chosen forms of government that seem to give worse economic outcomes? Why did Mobutu want to change his role from prime minister to president in Zaire in 1967? The same question can be asked of Mugabe in Zimbabwe in 1987, Stevens in Sierra Leone in 1978, Banda in Malawi in 1966 and Nkrumah in Ghana in 1960. There is little doubt that the transition to presidential rule in Africa represents a transition to a less democratic system, and to a system where the president personally has acquired substantial political power at the expense both of parliament and of the general population. In short, the transition in Africa is a transition that gives a narrower power elite greater political power – a power that can extensively be misused for personal gain. For opportunistic politicians it is therefore not difficult to understand that such a transition can serve their personal self-interest – even though it does not serve the interests of the country. The discrepancy between what serves the individual politician and what serves society is determined by the degree to which politicians are able to pursue policies that are not subjected to controls by the population. For opportunistic politicians tax havens represent a useful tool – they make it easier to adapt policies towards personal gain. The gains from switching to a system where the power of the general population is reduced and that of the political elite is increased is therefore greater for politicians the better their access to tax havens. Presidential rule with its concentration of power goes hand in hand with tax havens. The political system is not determined by the interests of the county – but by the interests of its political power elite.

The rise of tax havens increases the possibilities for personal enrichment through a political career, and at the same time makes it more tempting for opportunistic politicians to change institutions and political systems. In its turn, this can have an effect on what kind of people choose to be active in the political system. Not only do tax havens make it easier and more tempting for corrupt and dishonest politicians to further their own interests at the expense of the interests of society – tax havens may also have the side effect that the proportion of corrupt and dishonest politicians will be greater than it would otherwise have been.

1.3.2 Tax havens, natural resources and conflict

In the last few years, it has become clear that many civil wars can largely be explained by economic motives. Collier and Hoeffler (2004) find that higher income in the form of more natural resources increases the danger of civil war, whereas higher income in the form if higher productivity reduces the danger of conflict. One weakness of the first empirical studies of conflict is nonetheless that the causality is unclear: Do more natural resources lead to more conflict – or are countries with more conflicts typically in a situation where the only viable economic activity is the extraction of natural resources, whereas other industrial activity will not develop? Does poverty lead to conflict – or does conflict lead to poverty?

The more recent empirical literature seeks to clarify these causal relations. The most path-breaking contribution is Miguel, Satyanath and Sergenti (2004). How can one know whether it is poverty that leads to conflict or conflict that leads to poverty? If one can find a variation in income that is certainly not caused by conflict, then one may next find the causal link from poverty to conflict. Miguel, Satyanath and Sergenti (2004) use meteorological data for this. In Africa it is clear that the amount of rain has a strong influence on the income from year to year – at the same time, it is clear that civil wars cannot have an effect on the amount of rainfall! The variations in income caused by changes in rainfall can therefore be used to establish how changes in income affect conflict. The result is unambiguous: Poverty engenders conflict.

Besley and Persson (2009) study the causality between income from natural resources and conflict by exploiting the fact that all countries are too small to influence prices on the world market. Thus, changes in these prices cannot be caused by changes in conflicts within single countries. At the same time, Besley and Persson (2009) use time series data that enable them to control for any unobserved characteristics for each country. Again, the result is unequivocal: Wealth in natural resources engenders conflict.

But why does income in the form of natural resources give more conflict – while income due to higher productivity gives less conflict? Tax havens can partly explain this relation. Aslaksen and Torvik (2006) develop theory for this. Natural resources are beneficial to politicians whether there is democracy or not. In a democracy, public income from natural resources will, wholly or partially, be transferred to the population in the form of income transfers or expanded public services. Since politicians are accountable to the electorate, they will have relatively limited opportunities for using much of the resource income to enrich themselves, or for projects preferred by politicians, but not by the electorate. Nonetheless, it is reasonable for politicians to regard resource income as beneficial – either because they care about what is beneficial to the population or because they, even in under democracy, will to a certain extent be free to use some of the resource income at their discretion. Figure 1.10 shows this connection. The greater the resource income, the greater will be the utility to politicians in a democracy.

One alternative to democracy is to decide the struggle for political power through armed conflict. Armed conflict is expensive – an army must be established, equipment must be acquired, soldiers must be paid, and at the same time, a conflict will damage the means of production and the infrastructure. The cost of conflict is higher the higher the productivity of a country – high productivity gives high costs – and high costs makes the establishment of an army expensive. High productivity raises the cost of destruction – there is more to destroy. On the other side – if productivity is low, pay is low, and it is cheap to establish an army. At the same time, there is not much to destroy – so destruction represents less of a cost than when productivity is high.

However, for the kind of politician who is more concerned about his own good than that of the general population, there can also be advantages in winning power through armed conflict rather than through democratic elections. Where in a democracy you are accountable to the voters, and this limits the political freedom of action, with no democracy, you will to a greater extent be able to control the nation’s resources at will. For this reason, the utility of conflict for such politicians is higher the greater the resource income in relation to productivity. Moreover, since politicians have fewer barriers on their use of state funds than in a democracy, the utility to politicians of resource income grows faster under a non-democratic regime. The curve for the utility to politicians in Figure 1.10 therefore has two properties. Firstly, the expected utility of conflict will be negative if you have nothing to win – the conflict curve in 1.10 therefore begins below the utility to politicians under democracy. Secondly, the slope of the curve for utility to politicians under conflict will be steeper than the slope of the curve under democracy – in Figure 1.10 the conflict curve is steeper.

Figure 1.10 now clarifies that for the type of politician who primarily has personal economic motives, the choice between democracy and conflict will not be determined by how natural resources affect profitability under conflict, as many seem to argue. The central question is how profitability under conflict develops relative to profitability under democracy. Figure 1.10 shows the point at which politicians will see it in their interest to have democratic institutions – and when they will not see it in their interest, and instead resort to armed conflict. If natural resources in relation to productivity is lower than X*, then it is in the self-interest of politicians to support democratic institutions. There will therefore be a tendency for democracies to be more stable in countries with little resource income and high productivity. If natural resources in relation to productivity is higher than X* then it is in the self-interest of politicians to undermine democratic institutions and resort to armed conflict. In such cases, both parties in a conflict may have an interest in keeping the conflict alive!

A central factor in economically motivated conflict is that the wealth stolen by politicians can be safely channelled out of the country, so the funds are not lost if one loses political power. Easier access to tax havens will therefore shift the curve for the utility under conflict upwards, as is drawn with the stippled line in Figure 1.11.

We see from Figure 1.11 that tax havens make it relatively more attractive to undermine democracy and choose conflict instead – the critical level of resources relative to productivity needed for stable democracy goes down from X* to X**. Tax havens contribute to destabilizing democracies because the type of political agents who do not think of the best interests of the general population find it relatively more profitable to engage in conflict.

1.3.3 Theory: Tax havens, institutional change and income

Tax havens undermine the quality of countries’ institutions and political systems. This in its turn has a negative effect on income, an effect over and above the negative effects of tax havens on income studied above. This is shown in Figure 1.12. For those entrepreneurs who are engaged in productive industrial activity, weaker institutions and political systems represent a cost. More corruption, a weaker legal system, contracts awarded on the basis of political friendships rather than profitability, poor government bureaucracy, and so on, make economic activity less profitable than it would otherwise have been. The curve that indicates the income of entrepreneurs in productive activity in Figure 1.12 therefore shifts downward to the stippled curve.

It is not surprising that the deterioration in opportunities for productive entrepreneurs reduces the income of all entrepreneurs. It is more surprising that the income is reduced by more than the initial deterioration in income opportunities would indicate: The fall in income is greater than the vertical shift in the curve for the income of entrepreneurs in productive activity.

The intuition is that weaker institutions and political systems spark off a multiplicative process that exacerbates the initial problems. When the profitability of industrial activity is weakened by deteriorating institutions, this leads to fewer entrepreneurs engaged in productive activity, and more engaged in unproductive activity. This, in its turn, reduces the profitability of productive activity even further, the fall in productive activity accelerates, income falls even more, and so on. Income falls, then, not only as a direct result of institutional change – but also as an indirect result of the negative multiplicative processes set in motion by the institutional changes. The effect of poorer institutions and weaker political systems on income is therefore greater than one would expect based only on the reduced profitability of economic activity caused by weakened institutions.

1.3.4 Empirical studies: Tax havens, institutional change, and income

To all appearances, stopping such a process will have a great effect on economic growth. In the last decade, it has become clear that institutional quality may be the most important driving force of economic prosperity and growth. Acemoglu, Johnson and Robinson (2001) is the best-known study of the effect of institutions on domestic income. They estimate that if a country whose institutional quality at the outset lies on the 25th percentile could improve its institutional quality to place it on the 75th percentile, domestic income would increase 7-fold. Few factors have as strong an effect on growth as improved institutions. Precisely for this reason, the damaging effects of tax havens are so disastrous for developing countries – tax havens contribute not only to conserving poor institutions – but also to making them worse.

If tax havens not only affect institutional quality, but also have an effect on the choices political agents make between conflict and democracy, the growth effects become still more dramatic. Nothing is as damaging for development and growth as war.

1.4 Conclusion

The negative effects of tax havens are greater for developing countries than for other countries. There are many reasons for this. Reduced government income will have a greater social cost for developing countries than for industrialised countries. In addition, other mechanisms make themselves felt in countries with weak institutions and political systems. In such countries, “income opportunities” represented by tax havens for the private sector in reality contribute to the reduction of private income. Tax havens are central to the explanation of the paradox of plenty – and give resources that normally contribute to economic growth and development the opposite effect. The damage is particularly great in countries with weak public institutions, in countries with presidential rule, and in countries with unstable democracies. At the same time, institutions cannot be regarded as natural givens. Tax havens give the agents in the economy incentives to change institutions – but for the worse rather than for the better. Political agents are given incentives to weaken public institutions, to establish a type of presidential rule where much power is concentrated in the president’s hands, and to undermine democracy. For developing countries, the growth effects of putting a stop to the use of tax havens are great.

Explanation of variables:

IQ – Institutional Quality. This is an index based on data from Political Risk Services. The index consists of an unweighted average of five part-indices: (i) the degree to which the population of a country accept its law-enforcement institutions, (ii) bureaucratic quality, (iii) corruption in government, (iv) risk of expropriation and (v) probability that government will honour contractual obligations. The index goes from zero to one, with zero as den worst institutional quality and one as the best.

SXP – Resource wealth: Primary exports as share of domestic income in 1970.

GROWTH: A country’s average annual growth rate in GDP per capita in the period 1965 – 1990.

GDP65 – Initial level of income: Logarithm of GDP per capita in 1965.

OPEN – Openness for trade: An index that measures the proportion of years in the period when a country is characterised by openness for international trade.

I NV – Investments: Average percentage share of real investments of GDP.

AFR – Africa: Coded as 1 if a country lies in Africa and zero otherwise.

PRES – Democratic countries with presidential rule: Coded as 1 if the country is classified as et democracy in the period and if its political system is presidential, 0 otherwise.

PARL – Democratic countries with parliamentary regime: Coded as 1 if the country is classified as a democracy in the period and if it has a parliamentary political system, 0 otherwise.

Sources for data: Sachs and Warner (1995, 1997), Mehlum, Moene and Torvik (2006a), Andersen and Aslaksen (2008). For some of the countries, these works do not include data for all the countries. In such cases, the data sources referred to in these works are used to complete the data sets. For Hong Kong, there are no data for the political variable. Hong Kong is therefore classified neither as a democratic presidential regime nor as a democratic parliamentary regime.