Accounting for large fiscal government size

There is extensive evidence of the negative effects of the fiscal size of government – the spending, taxing, and ownership roles of the state – on economic performance (Bergh & Henrekson, 2011). But how do governments get big? As the size of government is measured as part of the Fraser Institute's Economic Freedom of the World index, we observe a wide variety of countries as having very large governments. We see the Nordic countries, which are among to be the most democratic in the world. We see a variety of autocracies. We see small countries like Lesotho, Eswatini, and Timor-Leste. When governments spend and own a large proportion of the economy, is there any ready explanation as to why?

The approach here is narrative-based, though informed strongly by data. It is likely that that a large number of variables influence the size of government, and many of them may influence different aspects of government size. Factors that cause a government to begin nationalising industries may be different from the factors that cause it to expand its pension system, for example. In the academic literature, there are many explanations that very plausibly tell us why some states are a bit bigger or a bit smaller, on the margin.¹ What we wish to do instead is to identify commonalities among groups of countries with very large governments, where the commonality plausibly has a direct pathway to the countries' size of governments. For simplicity's sake, we are considering the top quartile of countries with the largest governments, or 42 countries.

Again in contrast to the academic literature, we are not even making a claim to the ‘average treatment effect’ of the three variables we eventually identify. None of them is a mechanical law and each narrative has a counter-example. Rather, we wish to establish certain basic descriptive facts concerning the size of government. A great deal of confusion has arisen as a result of the lack of clarity regarding such basic descriptive facts; for instance the suggestion that the size of government should be removed from Economic Freedom of the World because of its lack of positive correlation with the rest of the index (Ott, 2018; cf. Murphy, 2022a). And although the issue is beginning to get rectified, there has long been confusion about the sustainability of Nordic ‘socialism’ (Lawson & Powell, 2019, pp. 5–14; Christensen et al., 2023). In this article, we hope to similarly facilitate improved discourse concerning the nature of the size of government.

Today, Economic Freedom of the World scores 165 jurisdictions. Of those, the top quartile of countries in terms of the size of government includes countries that are wealthy and countries that remain undeveloped. Jurisdictions rarely have uniformly low or high measures when it comes to the size of government as some components and subcomponents are negatively correlated with one another. Components 1A and 1B (government consumption and transfers and subsidies) tend to run together (R = 0.41). Elevated consumption and extensive transfers and subsidies among the wealthy nations of Western Europe generally mean these countries earn low scores in these components. Components 1C and 1E (government investment and government ownership of assets) also tend to run together (R = 0.46). Historically this might be where we would find countries dabbling in socialism, but now low scores in these components indicate that a country is dominated by natural resource extraction. The last remaining component of the size of government area (1D, the top marginal tax rate) is positively correlated with 1A (R = 0.08) and 1B (R = 0.041) but negatively correlated with 1C (R = −0.22) and 1E (R = −0.27).

Additionally, the Fund for Peace's Fragile States Index and the Economist Intelligence Unit's Democracy Index offer intuitive categorisations that allow us to characterise the state and political institutions of the countries in question. The Fragile States Index categorises countries, with additional gradations, on a spectrum of state fragility from ‘Sustainable’ to ‘Stable’, to ‘Warning’, and to ‘Alert’ (Turner et al., 2023). The Democracy Index assigns countries to the categories ‘Full democracy’, ‘Flawed democracy’, ‘Hybrid regime’, and ‘Authoritarian regime’ (EIU, 2023, p. 4). We can use these indices to provide characterisations of each country's institutions.

As will be shown, the two primary avenues for governments to grow very large are, predominantly, Wagner's law and the resource curse. Wagner's law (Karceski & Kiser, 2020; Lamartina & Zaghini, 2011; Murphy, 2022b) states that, as economic performance improves, government spending as a share of the economy increases. This is the ready explanation for the large welfare states in Europe. The natural resource curse can be stated a few different ways (e.g. Sachs & Warner, 2001), but in this instance we can see how large bounties from mother nature can lead governments (most famously in the Middle East) to fail to leave extraction in the hands of the private sector, so they can more easily grasp it themselves. In the words of the public intellectual Fareed Zakaria:

Countries associated with Wagner's law are wealthy and democratic with strong states. Their governments became large through government consumption (1A) and transfers and subsidies (1B). There is no relationship between Wagner's law and government investment (1C) or government ownership (1E), and a minimal relationship with the top marginal tax rate (1D) (Murphy, 2022b). Governments whose large states are associated with the resource curse tend to be autocracies, and they are all over the map in terms of their economic performance and the strength of their states. Their large states are associated with government investment and government ownership rather than government consumption and transfers and subsidies. The preponderance of countries subscribes to one of these patterns of the other. We will refer to 1A and 1B combined as ‘welfare state activities’ and 1C and 1E combined as ‘state control of the economy’.

We will consider the 42 countries in the top quartile of the Economic Freedom of the World index as having the largest governments (note that this corresponds to the bottom of the rankings in the ‘Size of Government’ area). Complete data for the different variables to be discussed below are tabulated in the Appendix. A total of 34 of the 42 countries is readily explained by either Wagner's law or the natural resource curse.

These two explanations for the size of government are well-worn in social sciences; Wagner's law was first observed in the nineteenth century. Simply looking at the list of the countries with the largest governments causes the petrostates and the European welfare states to jump off the page. The question was only for how big a chunk of cases would these two explanations be satisfactory.

First, consider natural resources. A total of 49 countries in the world have natural resources rents as a percentage of GDP in the double digits. If we use 10 per cent of GDP as a cut-off, then the resources curse explains 15 to 18 of the 42 cases of the world's largest governments. The 15 countries that are unambiguous resource curse countries are Algeria, Azerbaijan, Guyana, Iraq, Kuwait, Lesotho, Libya, Oman, Papua New Guinea, Qatar, Russia, Saudi Arabia, Timor-Leste, the United Arab Emirates, and Venezuela.

There are three cases of ambiguity. Arguably, both Wagner's law and the natural resource curse apply to Norway and Botswana. Both countries have large natural resource rents; they likely have larger governments than they would without these resources and are therefore stricken with some version of the resource curse, but they have otherwise escaped it and are functioning social democracies (though in the case of Botswana there is reason for caution: see Hammond, 2023). The third ambiguous case is Bolivia. The country's government is large probably because of the resource curse, but it just barely misses the definition with natural resource rents accounting for 9.5 per cent of GDP. (In addition, Tajikistan and Tanzania have non-trivial proportions of the GDP driven by natural resource rents; we return to them later.)

Aside from Botswana and Norway, the 16 clear cases of Wagner's law are Austria, Belgium, Croatia, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Italy, Japan, Luxembourg, Netherlands, Slovenia, and Sweden. If we set aside Japan, and pretend that the Nordic countries are one country, this set of countries is actually geographically contiguous. All 16 countries exhibit the mainline trajectory of development where well-functioning institutions, social democracy, wealth, and welfare state capitalism all arrive together. Figure 1 graphs all Wagner's law and resource curse countries by their natural resources and GDP per capita. Both indices are log scales (with one being added to resource rents so all countries can appear on the graph).

Figure 2 graphs all the above countries with their democratic institutions on the x-axis and state fragility on the y-axis. Of the Wagner's law countries, the ones to the left with weaker democracies are Croatia and Hungary, and the single country up and further to the right is Greece. Aside from Botswana and Norway, no resource curse country overlaps with the institutions of the social democracies, although four (Papua New Guinea, Guyana, Timor-Leste, and Lesotho) come somewhat close. The stable autocratic petrostates appear towards the bottom and to the left of the chart, and the petrostates that are coming apart a bit at the seams are at the upper-left edge of the chart.

The manner in which the welfare state capitalist democracies have large governments differs drastically from the manner in which the resource curse countries have large governments. We can subtract state control of the economy (1C and 1E) from welfare state activities (1A and 1B) from the for the two groups of countries. The resulting calculation is the ‘welfare state’-ness of each country's size of government, that is, how much of the size of government is focused on welfare state activities, if we set aside the overall size of government. The average of the 18 Wagner's law countries (including Botswana and Norway) is + 4.42. The average of the 15 resource curse countries (including Bolivia, excluding Botswana and Norway) is – 1.64.

Let us also work through the four more democratic of the resource curse countries which are visibly different in Figure 1. Guyana saw its oil production really surge only in 2019; although we may be optimistic about the country, we may also ominously say ‘wait’, as the resource curse simply hasn't had time to take hold yet (see Bahadur, 2024). Papua New Guinea and Timor-Leste both share a tiny corner of the globe and may persist as a pair of outliers. Timor-Leste is also one of the youngest countries in the world, not gaining formal sovereignty until 2002. Lesotho is a tiny enclave within South Africa. Of the four, Papua New Guinea is positive in welfare state-ness, though not strongly so (+ 1.15), while Timor-Leste (− 1.44), Lesotho (− 1.69) and Guyana (− 2.32) are more typically negative for resource curse states.

Finally, where does Botswana ultimately ‘belong’? For one, it is considerably poorer than any of the other welfare state capitalist democracies, although in terms of economic performance it is still right around the global average. But in this particular instance, it's probably worth stepping back from the narrative to note that the stylised facts expressed here are just that, and it is incorrect to ascribe single causes to any country. Moreover, Botswana has the 41st largest government in the world, and therefore just barely squeaks into the conversation. Its natural resources, and the fact that it appears to be a ‘normal’ country on the path to the end of history (Fukuyama, 1992), are sufficient in combination with one another to get it on this list. There may be no deeper interpretation to it than that.

Therefore, 34 of the 42 countries are accounted for by either Wagner's law or the natural resource curse. But why would a country have a very large size of government without either? The eight remaining countries to explain are China, Egypt, Eswatini, Guinea, Rwanda, Tajikistan, Tanzania, and Tunisia. The two final edge cases that could have been included as resource curse countries are Tajikistan (9 per cent of GDP) and Tanzania (6.7 per cent), though I will avoid ascribing the size of government to their natural resources. None of these countries is close to being Wagner's law welfare-state capitalist democracies; at best, according to The Economist, they are ‘hybrid regimes’ that are closer to autocracies than full democracies (EIU, 2023). The nature of political institutions, therefore, may play a role.

Most laypeople are likely unaware of Wagner's law or the resource curse as explanations of why some countries have large governments. But most people know that some ideologies, particularly both left-wing and right-wing versions of socialism, call for large governments. The Database of Political Institutions (Scartascini et al., 2021) contains assessments, by year, of the executive's ideology for each country. Unsurprisingly, all 46 years that the database has scored China's political institutions has it with a left-wing executive. But this is also true for all 29 years when Tajikistan received a score, all 46 when Tanzania received a score, and 37 of the 42 years Tunisia was scored (the other five years were ruled by the party Nidaa Tounes, which is coded as centrist but is arguably also left-wing). And what of the other four countries, Egypt, Eswatini, Guinea, Rwanda? Throughout the sample period, Egypt, Eswatini, and Rwanda were led not by conservative governments, but by governments with no politically coherent ideology, at least according to the Database of Political Institutions. Guinea was ruled for ten years by a left-wing executive; the rest of the period it was ruled by an executive with no coherent political ideology. In other words, of the countries with large governments that did not grow that way because of natural resources or because they were developing into democratic welfare states, all were ruled by autocratic governments with either leftist or an incoherent political ideology.

Since the 1980s trade barriers have fallen, many important dimensions of regulation have loosened up, and until very recently inflation has been tamed. We still do not really know how wealthy countries can reduce the size of their governments or, more specifically, ease up on their welfare state activities. Nevertheless, we have somewhat satisfying accounts, at least, for why countries with the largest governments got that way.

But even as I use the word ‘account’ I should be careful, as a fuller statistical account may show that the average effect of one of the three factors is smaller than implied, or that their effects on the size of government are not homogeneous. The resource curse itself is very heterogeneous in its effects. Giving sturdy financial foundations to a lasting authoritarian dictatorship is one way the resource curse plays out. Creating so much chaos that no persistent state can maintain a stable monopoly of violence is another, and you can't have a very large size of government if you don't have a government. To illustrate this point: one country with a very small government that would qualify as a resource curse state, under our definition, is actually Somalia.

The above analysis does not create a roadmap for having a smaller state, but it instead shows that the likely cause of very large governments in any particular case in the present day is rather explicable. The stylised fact we have gleaned is that one of three discernible things must be in place for a country to get such a large government. The problem, of course, is that one of those three things is becoming a wealthy, and failing to become wealthy in order to limit the size of your state does not seem like a very good deal.