Populism, State Capacity, and Why It Pays to Play by the Rules

 

Populism is on the rise around the world but the forms it takes can vary widely. Populists can align with the traditional left or right and can be liberal or conservative in their social values. What holds this diverse ideological family together is a sense that the will of the “true people” is being stymied by a corrupt elite — and that strong leadership is necessary to break that system. As tracked by a recent Ipsos poll, populist thought includes beliefs that “the system is broken,” “the economy is rigged to advantage the rich and powerful,” “traditional parties and politicians don’t care about people like me,” and “to fix our country, we need a strong leader willing to break the rules.”

This commentary focuses on the last of these beliefs, the need for a strong leader willing to break the rules. Ipsos found that 49 percent of respondents in 28 countries agreed with this sentiment. In the United States, the breakdown was 40 percent “agree” vs. 27 percent “disagree.” (See page 27 of the report.)

The Republican ticket in the U.S. presidential campaign of 2024 has made no secret of its willingness to break the rules. Donald Trump has blustered about “the termination” of certain rules, “even those found in the Constitution.” Vice presidential candidate J.D. Vance, meanwhile, has declared that as president, Trump should “fire every single midlevel bureaucrat, every civil servant in the administrative state, replace them with our people” . . . “and when the courts stop you, stand before the country, and say—” (invoking an apocryphal quote from Andrew Jackson) “the chief justice has made his ruling. Now let him enforce it.”

Of course, such statements are easy to shrug off as mere electioneering. But we should not dismiss them as purely cynical. In fact, there are hints that a sizeable segment of the American elite has come to believe that the country requires a leader empowered to weaken or outright violate deep-seated rules. The “tech right” that has coalesced around figures such as Vance, after all, draws inspiration from such thinkers as Curtis Yarvin, who advocates “absolute executive supremacy.”

Even the Supreme Court has come to worry that rules may constrain power too tightly. In its July ruling in Trump v. United States, far from warning against rule-breaking, the court found the president to be broadly immune from criminal prosecution for official acts. The decision declared that the nature of presidential power requires limits on accountability — the president should be free to take “bold and unhesitating action,” and exposure to criminal liability might make him “unduly cautious in the discharge of his official duties.”

The trouble is that populist voters and their elite supporters may be relying on a misleading notion of strength. We often hear some head of state described as a “strong leader” when the only indication of strength is the ability to hold onto power year after year. But the ability to stay in power and break rules is not the same as the ability to meet the aspirations of the broad populace. In fact, my research shows just the opposite. Governments are strong where leaders follow the rules. Where leaders break the rules, governments are too weak to fix much of anything. That holds across both democratic and authoritarian regimes. Let’s look at the evidence.

The ability to “fix things” as state capacity

The first step is to find a way to measure the ability of a government to “fix what is broken.” To me, that sounds a lot like what political theorists call state capacity. Broadly, that term refers to the ability of a government to accomplish its aims. Most writers treat state capacity as an ethically neutral concept, avoiding judgments as to whether a given regime’s aims are normatively good or evil. The specific measure that I use averages state capacity across three elements: administrative capacity (how well the bureaucracy works), security capacity (maintaining peace and fighting crime), and fiscal capacity (collecting taxes, managing the budget, and paying debts.)

Next, to determine whether a country’s leaders break the rules or follow them, I develop a measure of rule compliance. For this, I use an equally weighted index of compliance with four kinds of rules: Rule of law pertains to the honesty and efficiency of the judicial system. Government integrity means adherence to rules that restrain corruption. Executive constraints are rules that govern the transition of executive power, and those that impose legislative and executive checks on the power of incumbent chief executives. Finally, quality of market institutions is a matter of compliance with rules regarding property rights, investor protections, contracts, and other regulations of private businesses. This set of rules aims to capture features of governance that are relevant to the whole spectrum of regimes from democratic to autocratic and from liberal to oppressive. My concept of rule compliance thus covers only a subset of all laws and directives. It does not include “rules” that apply only to specific regime types, for example, requirements for accurate counting of votes in a democracy or standards for dress or facial hair in a theocracy.

The third important variable in my analysis is a measure of procedural democracy that focuses on a country’s commitment to holding fair elections with free and open participation. Readers who are familiar with some of my earlier research should take note that procedural democracy is a narrower concept than liberal democracy, which encompasses procedural democracy plus personal freedoms, free markets, and constitutional limits on government. “Democracy” and “democratic” as used in this commentary refer strictly to the procedural sense of the terms. I use the terms “authoritarian” and “autocratic” interchangeably as opposites of procedurally democratic.

The underlying data that I use to construct these three key variables come from the 2023 Legatum Prosperity Index. Keep in mind that there are lags in reporting. None of the individual indicators portray the situation earlier than 2022, and many reflect the situation two or more years in the past. I have converted the Legatum indicators to “Z-scores,” where a value of zero represents the mean for the 166 countries in the data set. The units are the number of standard deviations that a given observation falls from the global mean for that variable. For brevity, I often use the terms “democracy” and “autocracy” to mean countries with Z-scores for procedural democracy greater or less than zero, “strong” or “weak” for those with state capacity scores greater or less than zero, and “compliant” or “noncompliant” for those with rule-compliance scores above or below zero. Interested readers can find the full data with sources and definitions in an online supplement.

State capacity across the political spectrum

Figure 1, which shows how state capacity varies across the autocracy-democracy spectrum, sets the stage. Overall, democracies tend to have higher capacity than nondemocracies, but the fit is not strikingly close. As measured by statistic R2, procedural democracy accounts for only 41 percent of the variation in state capacity across the entire sample. Using an alternative set of procedural democracy scores compiled by Freedom House gives an even lower R2 of 30 percent.

The relationship between democracy and state capacity is not uniform across the spectrum, however. It is strong (R2 = 0.77) among the 88 democracies to the right of the central axis, but for the 78 autocratic regimes on the left, there is no statistically significant relationship. There are more autocracies with weak state capacity (59) than with strong state capacity (19), but it is not possible, statistically speaking, to attribute differences in state capacity within this group to whether a regime is egregiously undemocratic or only slightly below the mean on the democracy scale.

Given the modest influence of procedural democracy, then, it is worth asking what other factors account for differences in state capacity, especially on the authoritarian side of the distribution. Rule compliance turns out to play a major role.

The importance of playing by the rules

Figure 2 adds rule compliance to the data shown in Figure 1. Compliance scores are proportional to the area of the bubble representing each country. Shaded bubbles show positive scores, and empty bubbles show negative scores.  

It is not hard to see that rule compliance is positively associated with state capacity. Out of 71 countries with state capacity higher than the global mean, only 13 have negative scores for rule-compliance. Out of 69 countries with positive rule-compliance scores, only 11 have negative scores for state capacity. As measured by R2, 83 percent of the variation in state capacity among countries is explained by rule compliance.

Unlike the effect of democracy on state capacity, the effect of rule-compliance holds across the spectrum.

To the right of the central axis, Figure 2 is dominated by a group of 48 strong, rule-compliant democracies, shown in solid blue. Just four democracies, all small, have above-average state capacity combined with below-average rule compliance. Among the 36 weaker democracies, shown in green, just 10 have positive scores for rule compliance. For all democracies together, the R2 for state capacity and rule compliance is 0.84.

The surprise, perhaps, is how important rule compliance turns out to be for state capacity among autocracies. Ten authoritarian countries shown in red, led by Singapore, the United Arab Emirates, and Qatar, have positive scores for both rule compliance and state capacity. I call this group “institutional autocracies,” since solid institutional foundations are needed to maintain rule-compliance. (Despite their high ratings on some key aspects of good government, however, the institutional autocracies might not fit everyone’s idea of good places to live. All 10 of them have negative scores on a separate index of human rights and individual freedoms, not shown in the chart.)

For the authoritarian countries together, the R2 for state capacity and rule compliance is 0.72, not far below what it is for democracies. Leaving out the 10 institutional autocracies reduces the R2 to 0.53, lower but still statistically significant. So, even among the weak autocratic regimes, the difference between mediocre compliance and flat-out flouting of the rules accounts for more than half of the difference between slightly subpar state capacity and outright state failure.

Which matters more, democracy or rule compliance?

If both democracy and rule compliance are conducive to strong state capacity, which matters more? Multiple regression analysis provides some insights.

Across all countries, a regression of state capacity on procedural democracy and rule compliance finds that the two independent variables together account for 84 percent of the variation in state capacity. Essentially all that effect comes from rule compliance, with no statistically significant impact from democracy. However, as before, the pattern is very different among democracies and autocracies.

 Among democracies, the combined influence of the independent variables is slightly stronger (R2 = .86). In this group, however, democracy and rule compliance scores each make a significant positive contribution to state capacity. The effect of rule compliance is stronger, but not by much. Among authoritarian regimes, the combined effect of the independent variables is still significant, although not quite so strong (R2 = 0.72). The effect of rule compliance on state capacity accounts for all of that. The finding that among autocracies, democracy scores have no significant impact on state capacity thus carries over from the simple correlation tests to the multiple regression.

A brief note on China

For the most part, this commentary has looked at broad patterns rather than individual countries. It seems worthwhile, however, to make a few comments on the special case of China. China’s size alone justifies the attention. Beyond that, it stands out clearly in Figure 2 as having, by a good margin, the highest state capacity of any country with a negative score for rule compliance.

It is tempting to consider China as proof that a government whose leaders thumb their noses at conventional rules of good government can nonetheless build a powerful and effective state. But a closer look suggests a more complex story.

China’s great transformation began with a set of reforms initiated by Deng Xiaoping after the death of Mao in 1976. In several ways, those reforms steered China toward a more rule-based style of government. Rejecting the legal nihilism of the Maoist era, Deng showed a more positive attitude toward the rule of law. He introduced term limits for high officials. And although there was no big-bang transition to a market economy, small village enterprises and special economic zones pioneered at least partial steps toward property rights, enforceable contracts, and investor protections.

The Legatum data used in this study begin only to 2007, but a different index of state capacity, developed by Jonathan Hanson and Rachel Sigman, covers 1960 to 2015. That series, charted in Figure 3, shows that Deng’s reforms coincided with an impressive surge in state capacity.

What does the future hold? The small dip in the Hanson-Sigman state-capacity measure at the tail end of the chart, which coincides with Xi Jinping’s first years in power, cannot be considered more than a hint. Unfortunately, it is not easy to accurately splice my Legatum-based state capacity estimates onto to the Hanson-Sigman series. However, the Legatum data do suggest that China’s rule compliance peaked about 2018. The backsliding has been concentrated in the areas of rule of law and executive constraints. How lasting the trend is and how it will affect China’s state capacity going forward remain to be seen.

Taken as a whole, then, the arc of China’s development from Deng to Xi does not represent a clear exception to the story that rule compliance, not rule-breaking, is the path to a strong state.

Conclusion

The findings reported in this commentary are systematically inconsistent with the populist hope that a “strong leader who breaks the rules” can “fix what is wrong.”

• Among countries that are at least minimally democratic, increases in procedural democracy or rule compliance or both are associated with higher state capacity.
• Among authoritarian regimes, procedural democracy has no statistically significant effect on state capacity.
• Rule compliance has a strong positive effect on state capacity across the full spectrum of regime types. If anything, rule compliance is more important to maintaining strong state capacity where democracy is weak or absent than where democracy is well established.

Of course, these findings hold only to the extent that we identify the ability to “fix what is wrong” with a high score for state capacity. The academic notion of “state capacity” may seem remote from what voters have in mind when they say “the system is broken.” Porous borders, street crime, and an ever-widening gap between the rich and the poor are more likely to be what they are thinking of. But the most admired populist leader can’t do much to fix those things by sitting in an office giving orders. The orders will come to nothing without a competent civil service, well-trained police and soldiers, and a treasury that can steer a fine line between crushing taxes and crushing debt.

The fact is that lack of state capacity is a big part of the reason that populist leaders who achieve power often disappoint the hopes of their supporters. To pick just one example, think of Boris Yeltsin in 1991, boldly standing on a tank in defiance of Communist coup plotters. Yeltsin was subsequently twice elected President of Russia, but his promised reforms failed to produce an effective state.

Often, populists also disappoint the anti-elitist sentiments of those who bring them to power. Having failed to deliver good government, they cling to power by granting favors to a new elite. Yeltsin had reached that point already by the time of his second election in 1996.

When their one-time supporters see that the system is still broken and that the new elite is no better than the old, populist leaders or their successors all too often start breaking the rules of democracy itself. Once elected freely and fairly, they allow no more free and fair elections. Think of Vladimir Putin, to whom Yeltsin handed power on New Year’s Eve, 1999. 

We should ask: Could it happen here?

Adapted with permission from "The numbers show that governments that follow the rules are actually stronger," previously published by NiskanenCenter.org

Climate Inflation is Coming. How Should Central Banks Respond?

The impacts of climate change are visible everywhere — wildfires in California, preseason hurricanes in the Caribbean, insufficient water in the Panama Canal, populations on the move everywhere from North Africa to Central America. Inflationary shocks are another looming worry. Food prices will become increasingly volatile, while property insurance rates will escalate — or insurance simply won’t be available. Labor costs will rise as employers spend to shield workers from hotter weather or raise wages where workers balk at heat exposure.

It all points not only to more inflation, but also to greater variation and less predictability across sectors and regions. In fact, that’s already happening. The figure below uses data from the Atlanta Fed to divide prices into half that are “sticky” in the sense that they rarely change and half that are “flexible,” meaning they go up or down with every bump to supply or demand. Since the late 1990s, the volatility of flexible prices has exceeded that of the “great inflation” of the 1970s. Even sticky prices are showing some ominous wiggles.

In the years ahead, more volatile inflation will make it harder for the world’s central banks, America’s own Federal Reserve Bank included, to meet their commitments to stabilize prices. Will they be up to the job? Not without some changes in strategy.

How Central Banks Fight Inflation

The most common strategy for stabilizing prices among central banks today is “inflation targeting.” The central bank of New Zealand was the first to try it back in 1990. Now it is the official policy of the U.S. Fed, the European Central Bank and many others.

The basic concept is simple. A central bank watches forecasts of inflation for the coming year. If the expected rate of inflation is higher than acceptable, the bank raises short-term interest rates. Those feed through to longer-term rates, which, in turn, slow business investment and consumer spending on homes, cars and other durable goods. If the forecast inflation rate is “too low” — we’ll get to what that means in a moment — the central bank cuts rates.

But policymakers must get the details right to make inflation targeting work well. Here are three not-so-fine points that will loom in a world of climate-driven volatility. 

Setting the right target rate. When you first think about it, you might suppose that if inflation is bad, the best target would be zero percent — no inflation at all. In the early days, some highly respected central bankers including Paul Volcker and Alan Greenspan did advocate a zero target. But today, major central banks set a target higher than zero, typically 2 percent, or a target range centered on a rate higher than zero.

One reason for adopting a non-zero target is the need for room to cut interest rates in recessions. In the U.S., for example, the Fed estimates that to stimulate the economy, it has to push interest rates below a “neutral rate” estimated to be about 0.6 percent higher than the prevailing rate of inflation. If inflation is currently 2 percent, for example, effective stimulus would require an interest rate lower than 2.6 percent. 

Two percent is not sacred. Roger Douglas, who was finance minister of New Zealand when that country first tried inflation targeting,is reported to have said, “I just announced it was gonna be 2 percent, and it sort of stuck.” 

The lowest possible setting lies in a range from 0 to 0.25 percent. A cut in the policy interest rate from a neutral 2.6 percent to 0.25 percent would provide a quite a bit of stimulus. However, if the initial inflation rate were zero, putting the neutral rate at 0.6 percent, there would be almost no room for stimulus. Any recession would be likely to last longer and the recovery from it would be slower.

A second reason for a positive target rate is to give more room for relative prices and wages to adjust on their own. Even when the economy is cruising along at full employment, differing trends in productivity, consumer tastes and world trade are always pushing some prices and wages up and others down.

Some welcome those relative changes and others resist them. The strongest pushback comes from workers in lagging sectors, who resist wage cuts. If they are unionized, they may strike. Even nonunion employers may find that their best workers quit in the face of wage cuts, while the demoralized remainder become less productive. Cutting prices in the face of falling demand is not resisted as tenaciously as cutting wages, but many firms succumb only under the extreme pressure of tanking sales.

A positive inflation target provides a cushion against these downward wage and price rigidities. A 2 percent inflation target would permit wage and price increases of 3 or 4 percent in leading sectors while still allowing increases of 1 percent, or at least no outright cuts, in lagging sectors.

But 2 percent is not sacred. Roger Douglas, who was finance minister of New Zealand when that country first tried inflation targeting, is reported to have said, “I just announced it was gonna be 2 percent, and it sort of stuck.” Today, many economists think a target higher than 2 percent would be appropriate, in part because it would give central banks more room to cut rates in future recessions. The prospect that climate change will increase price volatility only makes the logic of a higher target that much more persuasive. 

Targeting the right index. The next devilish detail is which of many inflation indexes to use as the target benchmark. In the U.S., the one that makes the headlines is the consumer price index, or CPI. The headline rate for the EU, the “harmonized index of consumer prices,” is slightly different. The Fed sets its target in terms of a third measure, the personal consumption expenditures index (PCE). Inflation benchmarked by the PCE tends to run about half a percentage point below the CPI.

But the PCE index is not the only measure of inflation that the Fed uses. It also pays close attention to “core inflation” — that is, PCE inflation calculated excluding food and energy. The thinking is that this core index gives a good picture of underlying inflation trends. 

Adapting inflation targeting to climate trends is likely to require raising the target rate itself, modifying the indexes used for short-term policy decisions and lengthening the period over which inflation is brought back to the long-term target after temporary shocks. 

Monetary policy can control overall inflation in the long run, but it can do little to affect relative price movements, such as isolated spikes in, say, beef or gasoline prices. Overreacting to such transitory shocks could easily backfire in a way that destabilized the economy.

Note that climate change could make current measures of core inflation obsolete. Rather than moving randomly up and down from a stable average, future food prices are likely to develop an upward trend relative to less climate-sensitive goods. The same could be true of energy prices. In that case, excluding food and energy from the target index would no longer make sense.

One remedy would be to switch short-term targeting to a “trimmed mean” index. Instead of arbitrarily treating food and energy as the transitory factors, trimmed mean indexes simply clip off whatever prices are in fact the most volatile in any given period. The Cleveland Fed already publishes a trimmed mean version of the CPI and the Dallas Fed publishes a trimmed mean PCE. 

And did I mention the multivariate core trend inflation index from the New York Fed? Maybe I shouldn’t, or you’ll stop reading…

Lengthening the time horizon. Central banks do not try to hold the rate of inflation at its target month by month. In fact, trying to control inflation over a very short time horizon would risk harm to the real economy by pressing too hard against price and wage rigidities. Accordingly, without locking themselves into a strict calendar interval, central bankers use language like “medium term” or “longer term” to suggest a time horizon of more than a year but no more than a few years. If climate change intensifies the volatility of inflation and turns previously transitory movements of food prices into long-term trends, there would be a case for lengthening the time horizon over which inflation is averaged out.

Taking all these adjustments together, then, adapting inflation targeting to climate trends is likely to require a combination of raising the target rate itself, modifying the indexes used for short-term policy decisions and lengthening the period over which inflation is brought back to the long-term target after temporary shocks.

Beyond Inflation Targeting

Inflation targeting, although common, is not universal. It is worth looking briefly at the problems faced by the central banks of a different group of countries — those that either have no separate currency of their own or maintain fixed exchange rates pegged to the value of another currency. Examples of countries with shared currencies include Ecuador and El Salvador, which use the U.S. dollar, and Montenegro and Kosovo, which use the euro but are not members of the Eurozone. Countries with fixed exchange rates include Saudi Arabia and several other Middle Eastern oil producers, which peg to the dollar, and several African countries, which peg to the euro. Finally, there is the special case of Eurozone members themselves, which do not have separate currencies but do have a vote in the policies of the ECB, which manages the euro. 

Any external shock to an economy, even a favorable one such as an increase in the global market price of a major export, creates a cascade of effects with both winners and losers. 

Economists have long studied the question of whether two or more countries are better off with shared or independent currencies. They agree on one point: currency areas work best when member countries are likely to be subject to similar economic shocks. For example, other things equal, two energy exporters might share a currency, but sharing would make less sense for an energy importer and an exporter. In many ways the same principle applies to the choice between fixed or floating exchange rates for countries with independent currencies.

Any external shock to a country’s economy, even a favorable one such as an increase in the global market price of a major export, creates a cascade of effects with both winners and losers. An export boom can push up input prices for producers in other sectors in the same country. A rise in import prices harms consumers but benefits producers of home-grown substitutes for imports. And so on.

When a country has its own currency, its central bank can mitigate the effects of such shocks. It can apply stimulus to avoid the pain of deflation, or control inflation by raising interest rates. A country that uses another’s currency has no such option. The central bank of Ecuador can do nothing to affect the exchange rate of the dollar, interest rates on U.S. government bonds or the quantity of dollars in circulation within its own borders.

Members of the Eurozone are a little better off, since each of them has a voice in decisions of the European Central Bank. But the ECB can’t — and certainly doesn’t — please everyone all the time. The Greek debt crisis that followed the global financial meltdown of 2007-8 is a case in point. The perceived interests of Germany and Greece diverged so sharply during that episode that Greece came close to leaving the Eurozone altogether.

Even countries that have their own currencies can get into trouble when they tie their exchange rates to other currencies. If the central bank of a fixed-rate country tries to stimulate its economy by cutting interest rates, it risks creating a run on its reserves of foreign currency. If its treasury tries to finance government spending by borrowing abroad, it can undermine investor confidence and risk a default on its debt. 

Here, Argentina is the poster child. To stop a hyperinflation in the early 1990s, Argentina pegged its currency firmly to the U.S. dollar. As far as inflation went, the policy worked. But it left the Argentine central bank unable to cope with a severe recession a decade later. In 2000, the country abandoned its currency peg and defaulted on its debt, leading to a severe crisis.

It is often a close call whether a country is better off within a currency area or outside it, or whether it would best have a fixed or floating exchange rate. An increase in the frequency and severity of relative price shocks driven by climate change could well tip the decision, causing some countries to consider moving away from shared currencies and fixed exchange rates toward floating rates.

Even the Eurozone is not immune. A 2013 retrospective on the effects of the global financial crisis concluded flatly that the Eurozone “does not currently represent an optimum currency area,” implying that the impact of economic shocks varied widely among members. The bloc survived, but in view of its diversity of climates, products and consumption patterns, it seems likely that climate change will pose an ongoing challenge to its survival.

The bottom line

Trends since the turn of the century and projections for the future make it clear that climate change will leave no part of the economy untouched. Central banks, like the rest of us, will have to deal with this reality. In particular, they will have to cope with the possibility of persistent upward pressure on the relative prices of climate-sensitive products such as food, as well as larger and more frequent shocks that can strike any sector of the economy.

Fortunately, central bankers are not without options. Inflation-targeting countries will have to rethink the 2 percent inflation goal that most of them have adopted, sometimes without much reflection. Going forward, 2 percent may not give enough room for stimulus in times of trouble. By the same token, policymakers may also want to reconsider which measures of inflation they use as benchmarks along with the time horizon they select for bringing inflation under control.

Central banks that impose fixed exchange rates already face constraints on inflation policy that make adjustment to shocks more difficult. We may see fewer countries opt for fixed exchange-rate pegs going forward. Countries in currency unions (like the Eurozone) and those that have adopted another country’s currency unilaterally face even tighter constraints. The Eurozone has brought considerable benefits by reducing trade barriers and facilitating capital flows among member countries, but past shocks have already pushed it to the limit. It is worth thinking in advance about who is better in the euro area and who is not, rather than waiting for the next crisis to force a choice at an inopportune moment.

Unfortunately, even the wisest policies will not entirely shield economies from coming climate impacts. Every shock will produce both winners and losers. The best course will be to maintain enough policy flexibility in advance to manage climate effects in ways that spread the pain in as balanced a way as possible.

Originally published by Milken Institute Review. Reposted with permission.

 

Good data on good government: Reformers should take note as political philosophy meets statistics

Good government has been a focus of political philosophers for centuries, but times change. Today’s thinkers, unlike Plato or Ibn Khaldun, can draw on abundant data to test and refine their theories on the merits and drawbacks of various regimes. And data is not just for theorists. Good data can provide would-be reformers with practical insights in their day-to-day efforts to make government more effective and strengthen democracy. A profusion of indexes compete to measure quality of government, state capacity, state fragility, liberal democracy, electoral democracy, varieties of democracy, and more. Which one is the best? Or are they all so alike that the choice doesn’t matter? 

But where to start? This paper offers a map through the thicket of governance data. The first section deals with some key terms and concepts. The second develops a pair of benchmark indexes that are closely aligned with the theoretical categories of state capacity and liberal democracy. The third examines the similarities and differences among a sample of widely used quality-of-government measures, and their implications for the relationship between liberal democracy and state capacity. The fourth section looks at what we can learn by looking beyond statistical regularities to specific country outliers. Finally, the conclusion takes up the question of how best to integrate theoretical, practical, and quantitative approaches to the study of good government.

Terms and concepts

Before going further, we need to nail down some key terms and concepts, beginning with state capacity. I will follow the lead of the many writers who define state capacity in terms of a specific list of government competencies. The Niskanen Center State Capacity Project identifies those as the ability of government to effectively raise taxes, maintain order, and provide public goods. Jonathan Hanson and Rachel Sigman, creators of one of the indexes used below, call these extractive, coercive, and administrative capacities. I prefer the terms fiscal capacity and security capacity for the first two, along with administrative capacity for the third. I take all of these three-part definitions as essentially equivalent. 

It is worth noting, however, that some other writers use the term “state capacity” more loosely. For example, in their book The Narrow Corridor, Daron Acemoglu and James A. Robinson write simply that “the capacity of a state is its ability to achieve its objectives (p. 12).” Others instead use state capacity more narrowly to mean the quality of the state’s executive and bureaucratic functions, that is, its administrative capacity alone. 

Similarly, we can define liberal democracy as a set of institutions or practices. I find the list of “liberal rights practices” identified by Kevin Vallier to be especially useful. Vallier’s list includes democratic constitutionalism, markets and private property rights, freedom of association, electoral democracy, and social protection policies. In the next section I introduce a measure of liberal democracy based on a variation of that list. 

In contrast to the formal definitions of the first two terms, I will use the term quality of government (QoG) more loosely. The Quality of Government Institute at the University of Gothenberg defines it as the presence of “trustworthy, reliable, impartial, uncorrupted and competent government institutions.” I am inclined to stretch that list by adding “just” and “conducive to human flourishing.” In line with that broad definition, I treat state capacity and liberal democracy as specific aspects of generalized quality of government and refer to all the various indexes used in what follows as QoG indexes.

One further note: However they are defined, there is a tendency to see state capacity as a positive construct while liberal democracy takes on a more normative coloration. Writers that follow this distinction understand a high rating for state capacity to mean simply that a government is able to achieve its chosen objectives, without judging whether the ends are good or bad. A government might, for example, receive a high state capacity score by successfully suppressing all but one preferred religion or efficiently carrying out a program of ethnic cleansing that outsiders might regard with horror. In contrast, the defining aims and institutions of liberal democracy, such as free and fair elections, limited government, and freedom of speech, are often seen not just as tools of good government, but as ends in themselves. 

It will become apparent as we go on, however, that aspects of state capacity and liberal democracy can bleed into one another in a way that makes it hard to maintain the positive-normative distinction as more than a tendency. For example, a commitment to liberalism may limit the range of permissible tools that states can use to pursue their ends as well as the ends themselves — there may be such a thing as too much state capacity. There is also room for disagreement as to whether certain aspects of governance, such as government integrity (or its opposite, corruption) should be counted as an aspect of state capacity that affects the ability of a government to achieve its ends or a liberal democratic norm. In the end, whether liberal rights practices bolster or weaken state capacity is a question to be explored through historical, institutional, or (as in what follows) quantitative analysis.

Benchmarks

There is no shortage of indexes that measure one or another aspect of quality of government. This section introduces a revised version of two indexes that I have developed in previous work. One is designed to fit as closely as possible with the theoretical concept of state capacity and the other with the concept of liberal democracy, as those are defined in the previous section. Accordingly, the pair can serve as benchmarks for comparison with more widely used measures, some of which cover only one or the other of the two key concepts, or combine elements from each to form a blended index of quality of government. 

To maintain methodological consistency, I have drawn all of the underlying data for the benchmark indexes entirely from the Legatum Prosperity Index, 2023 edition. The full Legatum dataset and a detailed report on methodology is available for download from the organization’s website. Data for the benchmark indexes is organized hierarchically. Each index consists of a number of components, each of which has a number of elements, which are, in turn, formed from a number of individual indicators.

The benchmark state capacity index (StatCap). 

The benchmark state capacity index consists of three components:

• Administrative capacity corresponds to the Legatum element government effectiveness, which comprises indicators of perceived government quality and credibility, ability to set and maintain priorities, efficiency of spending and asset utilization, ability to implement chosen policies, ability to learn from mistakes, and coordination across areas of policy.
  
• Security capacity is the unweighted average of the Legatum elements for war and civil conflict, violent crime, and property crime. All of these are coded so that high scores mean less conflict and crime.
  
• Fiscal capacity is the unweighted average of the Legatum indicators for country credit rating and country risk premium. Country credit rating is an average of the credit ratings from S&P, Moody’s, Fitch, and DBRS. Each of these ratings, in turn, reflects a wide variety of underlying  economic and financial indicators such as government debt, GDP growth, balance of payments, past payment records and defaults, and so on. Country risk premiums are based on data provided by Aswath Damodaran of the Stern School of Business at NYU.

The benchmark liberal democracy index (LibDem). The benchmark liberal democracy index consists of four components:

• Limited government is the average of the Legatum elements for executive constraints, government integrity, and rule of law. These are closely related to what Vallier calls democratic constitutionalism.
  
• Quality of market institutions is the average of the Legatum elements for regulatory quality, property rights, investor protections, and contract enforcement.
  
• Personal freedom is the average of the Legatum elements agency, freedom of speech, freedom of assembly, absence of legal discrimination, and agency, which Legatum defines as “freedom to act independently and make one’s own free choices.”
  
• Procedural democracy corresponds to the Legatum element political accountability. It covers the mechanics of democracy, such as parties and the right to vote; the degree of public agreement on democracy as a goal; and the availability of effective complaint mechanisms outside the electoral process.

The primary consideration for selecting the elements and indicators for each of the components was their degree of fit with the theoretical concepts of state capacity and liberal democracy. After initial selection, the indicators and elements for each component were further screened according to their statistical properties. Interested readers can find complete data for the benchmark indexes in Online Supplement A. The FAQ tab of that supplement includes further information on the selection and screening processes. 

Additional quality of government measures

This section introduces six additional quality of government indexes that are well established in the quality of government literature. Two of the six focus primarily on state capacity, two on liberal democracy, and two have elements that overlap both concepts. I have selected them to represent a variety of approaches. The failure to include other indexes in no way reflects any judgment on my part that they are inferior.

The first pair of indexes are based on data from the World Bank’s Worldwide Governance Indicators. The WGI does not explicitly sort its indicators into liberal democracy and state capacity categories, but it is not hard to do so. For a measure of liberal democracy (WGI-L), I use the average of WGI scores for “rule of law,” “control of corruption,” “regulatory quality,” and “voice and accountability.” For a measure of state capacity (WGI-S), I use the average of the WGI scores for “government effectiveness” and “political stability and absence of violence and terrorism.”

Next comes the index of state capacity devised by Hanson and Sigman (HSI). As noted above, it provides measures for extractive, coercive, and administrative capacities and aggregates them into an overall score. Its authors adhere explicitly to the notion that an index of state capacity should be positive in nature. As they put it, their index “eschews normative beliefs about what constitutes legitimate exercise of state power.”

The fourth of the established indexes is the Liberal Democracy Index from the Varieties of Democracy Institute. That organization publishes a large database that includes several composite indicators covering various aspects of quality of government. The one used here, which I abbreviate as VDem-LDI, can be found in Table 3 of their 2023 Democracy Report. Unlike the benchmark LibDem index, VDem-LDI includes an egalitarian component that can be construed as a measure of Vallier’s social protection practices. 

The final two indexes include elements that overlap the theoretical constructs of liberal democracy and state capacity. One is the Democracy Index (EIU-Dem) from the Economist Intelligence Unit. The other is the Fragile States Index (FSI) sponsored by The Fund for Peace. The FSI index, as published, assigns highest scores to the most fragile states, which tend to be both low in capacity and illiberal. (The relatively few illiberal but high-capacity countries and those that are weak but liberal have FSI scores somewhere in the middle of the pack.) For purposes of this study, I invert the scale so that the most fragile states have low scores.

Table 1 provides an overall comparison of the two benchmarks and the six established indexes. Coverage differs from one index to another. The results reported below are based on 153 countries that are included in all of them. Data for all eight indexes, standardized with means equal to zero and standard deviations equal to 1.0, can be found in Online Supplement B. Details on sources and methodology, along with underlying data, are available from the original sources.

Are they all alike?

The eight indexes in Table 1 fall into three distinct groups according to their fit with the qualitative categories of liberal democracy and state capacity. This section shifts attention from the qualitative to a key quantitative question: Despite their theoretical differences, are the indexes so much alike statistically that it matters little which of them is used in any given research project, or are there critical differences beneath their superficial similarities? 

Andrea Vaccaro raises exactly that question in a recent paper titled “Measures of state capacity: so similar, yet so different.” This section follows Vaccaro’s lead by decomposing the question of similarity into two parts, convergence and interchangeability. Convergence is measured by general statistical properties. Interchangeability is a matter of whether competing measures lead to the same conclusion when used to test one or another specific hypothesis.

Convergence. The most basic measures of convergence among the QoG indexes are simple bilateral correlations. Table 2 shows the degree of similarity between each pair of indexes as measured by the coefficient of determination. That statistic, abbreviated as R2, is the square of the simple correlation coefficient R. R2 shows the proportion of the variation in the values of one member of the pair that can be explained by the other member. For example, the second cell in the top row of the table shows that 87 percent of the variation in the index LibDem can be explained by the index VDem-LDI. That suggests that the two indexes, while not identical, convey much of the same information. All of the R2 values in the table are statistically significant. The degree of convergence is particularly strong within the three liberal democracy indexes and the three state capacity indexes, but there is substantial similarity even between pairs of indexes from different categories.

Another statistic, called Cronbach’s alpha, provides an alternative measure of convergence. Cronbach’s alpha is useful in comparing several different ways of measuring a given concept, for example, different items of a questionnaire for assessing the effectiveness of teachers in a school. The value of alpha is 0 if the items are completely unrelated and 1 if they are identical. Results above 0.7 are commonly interpreted as showing that the items are consistent measures of the same underlying concept. Many statisticians consider that alphas above 0.9 indicate that the set of items includes some that are redundant, so that a smaller subset could be used without losing much if any information. Cronbach’s alpha for the three liberal democracy indexes, the three state capacity indexes, and even for all eight QoG indexes together fall in the range from 0.9 to 1.0. 

Finally, the impression of strong convergence is further reinforced by a third statistical test, principal component analysis. That approach is used to assess the degree to which a number of measures of some concept can be reduced to a smaller number of principal components, or “latent factors.” In this case, one might anticipate that the eight indexes would reveal two latent factors, one related to liberal democracy and the other to state capacity. Instead, it turns out that nearly all of the information contained in the set of eight indexes can be attributed to a single principal component.1 

Interchangeability. But convergence is not the whole story. Even a high degree of convergence does not mean that any two measures are interchangeable in the sense that they all give the same result when used to test a specific hypothesis. 

Consider, for example, the hypothesis that, other things being equal, the institutions and practices of liberal democracy are supportive of strong state capacity. Clearly, the results shown in Table 2 are supportive of that hypothesis. However, Vaccaro finds that things change when we examine the narrower hypothesis that the relationship between democracy and capacity is nonlinear. More specifically, the hypothesis is that adding just a bit of liberal democracy to an initially authoritarian polity produces little if any gain in state capacity, but as democracy takes root more deeply, each additional liberalization gives a greater marginal boost to capacity.

Rather than trying to replicate Vaccaro’s test of the nonlinearity hypothesis, I took a different approach. I looked at the nine possible pairings of the three state capacity indexes for the 153 countries included in this study with each of the three liberal democracy indexes. For each pairing, I fitted a curve using liberal democracy scores on the horizontal axis and state capacity scores on the vertical axis. A significant increase in slope moving from left to right would be consistent with the nonlinearity hypothesis, while a straight-line relationship would support its rejection.2 

Figure 1 shows results for three of the nine pairings. In each chart, the fitted curve is shown in red while a blue linear trendline provides a reference. The results do not support interchangeability. Figure 1a, which uses LibDem and StatCap, shows a statistically significant curvature. The curvature is even more pronounced in Figure 1b, which relates VDem-LDI to WGI-S. However, Figure 1c, which uses WGI-L as the measure of liberal democracy and HSI for state capacity, shows no significant curvature. The red fitted curve coincides almost precisely with the blue reference line.

An examination of all nine pairings shows that most of the differences result from the choice of liberal democracy index. For all three state capacity measures, the regressions show moderate but significant curvature when the independent variable is LibDem or VDem-Lib, but no significant curvature for WGI-L. That suggests that the state capacity indexes, while not identical, are much closer to being interchangeable than the liberal democracy indexes. Keep in mind that these results are only illustrative. If applied to other hypotheses or to other samples from the underlying data, the degree of interchangeability might differ.

So much for the regularities. What about the outliers?

Up to this point, we have focused on one dominant regularity in the data, namely, the fact that on average, a country’s level of liberal democracy is a useful predictor of its state capacity. However, even a casual examination of charts like those in Figure 1 shows that although most countries cluster around a positively sloped line, there are conspicuous outliers. This section looks at what can be learned from the outliers, particularly those that fall far above the trendlines. At least since the 1930s, authoritarian leaders around the world have insisted that Western liberalism is the enemy, not the friend, of effective government. Although high-capacity illiberal states are the exception, not the rule, it is worth asking which countries do fit that pattern, and what, if anything, they have in common.

To sort the outliers from the rest, it will help to introduce a new term. I will call the amount by which a country’s state capacity score exceeds the value that is predicted by its liberal democracy score its capacity-liberalism anomaly (CLA).3 A strong, positive CLA fits the pattern of trains-on-time authoritarianism. The CLA can be computed either one-by-one for the nine possible pairings of the three state capacity indexes with the three liberal democracy indexes, or it can be averaged across all nine pairings. 

Table 3 shows the top 10 countries by CLA among the 153 countries in the dataset. Column 2 shows the average CLA for each country across the nine pairings. The remaining columns show which countries made the top 10 for each individual pairing, with the numbers in parentheses showing their rank within that pairing. Only one country, Qatar, makes the top 10 for all nine pairings. In addition to the countries shown, 18 others made the top 10 list for at least one pairing.

Certain shared traits seem to predispose countries toward the high-capacity illiberal pattern. Five of the top 10 have communist or post-communist political systems and five are major oil exporters. (Kazakhstan falls in both groups and Singapore in neither.) Three of the top four — Singapore, the United Arab Emirates, and Qatar — have state capacity scores in the top quartile among all 153 countries, ranking 9th, 27th, and 28th respectively. China’s state capacity, in contrast, ranks a much less impressive 46th out of 153, behind such competitors as Botswana, Malaysia, and Bulgaria. China’s high CLA is due less to exemplary state capacity than to a dismal ranking of 119 on liberal democracy, which places it right between Djibouti and Russia. Singapore and Hungary are the only countries among the top 10 by CLA that have positive average liberal democracy scores.4

It is worth noting that the agreement on CLA rankings is far from complete. Even Qatar, the only country that makes the top 10 in all nine pairings, has rankings for the individual pairings that vary all the way from first to tenth. The inconsistency of top 10 rankings provides additional evidence of less than complete convergence and interchangeability among the QoG indexes.

More can be learned about the CLA outliers by looking beneath the broad QoG indexes to their constituent elements. For example, the benchmark LibDem index includes components representing limited government, quality of market institutions, personal freedoms, and procedural democracy. All of the CLA top 10, even relatively liberal Singapore, have negative scores for personal freedoms and procedural democracy, that is, scores below the global mean of 0. However, on limited government and market institutions, their performance is more varied. Singapore and the UAE have very strong scores for both of those elements. Qatar and Saudi Arabia are a bit behind, but still positive. China, in contrast, scores below the global average for limits on executive power, and although its score for quality of market institutions is positive, its market institutions are weaker than any of the five petrostates in the top 10. Vietnam, Belarus, and Oman score below the world average on all four LibDem components.

Where to next?

This paper has aimed to encourage both academic researchers and would-be reformers to make greater use of quantitative data on quality of government. What have we learned that is of practical use? What are the questions that need further exploration? Here are some key points, as I see them:

First, not every discussion of state capacity or liberal democracy need take a deeply quantitative approach. Nevertheless, cross-country comparisons and within-country trends can provide context for even the most abstractly theoretical discourse or the most pragmatic policy analysis. As Vallier notes in his book on political and social trust, “we now have access to data on how social stability and cooperation are sustained, data that earlier social contract theorists did not have, and so we can speak more directly to the problems we currently face.”5

Second, some thoughts on how to choose among the many available measures of quality of government. Although broad correlations reveal substantial similarities among widely-used indexes, they are not fully interchangeable. Writers who choose to use QoG data, even casually, should be careful about the alignment between the available numbers and the meanings they attach to terms like state capacity and liberal democracy. For example, some writers on state capacity are primarily concerned with the ability of a government to execute the policies it chooses – what I have called administrative capacity. If so, it might not make sense to choose a measure of state capacity, such as StatCap, HSI, or WGI-S, that includes substantial weights for security capacity (war, crime, terrorism) and fiscal capacity (taxation, credit ratings, debt). Instead, it might be better for the purpose at hand to extract a single element or group of elements from those broader measures of state capacity. The same goes for liberal democracy. For some, democracy is simply a matter of holding free and fair elections. For others, procedural democracy is only one element of a broader liberal order that includes limited government, free markets, and personal freedoms. Finding the right data requires more than grabbing the first index that comes along with “democracy” in the title.

Third, although there is a strong convergence among the most popular indexes of quality of government, they are not fully interchangeable. Even if one index appears to be the best conceptual fit for the purpose at hand, it is worth double-checking to see if any hypotheses it appears to support hold up when tested against alternatives.

Finally, nothing said here is intended to denigrate nonquantitative analysis. Lord Kelvin overstated the case when he famously wrote that if you can’t measure what you are talking about, “your knowledge is of a meagre and unsatisfactory kind.” Even so, there are great opportunities for cooperation among theorists, hands-on analysts, and data junkies in advancing the frontiers of knowledge about democracy and state capacity. 

With these points in mind, there is no end to the variety of topics waiting to be explored. How do specific factors such as women’s agency or political trust affect state capacity and liberal democracy? What can we learn from time-series data as compared to the cross-national approach taken in this paper? How can newly available data contribute to better understanding of long-running controversies such as the resource curse? Imagination is the only limit.

                                                                ***

1. For full results on all of these convergence tests, see Online Supplement B. ↩︎
2. The curves were fitted using multiple regressions in the functional form y = a + bx + cx2, with x as the liberal democracy score and y as the state capacity score. For this set of regressions, the variables were normalized to a scale of 0 to 100. A positive and statistically significant value for the c coefficient was interpreted as consistent with the nonlinearity hypothesis. The regression as a whole showed a statistically significant relationship (p < .01) for all nine pairings but the curvature was significant only for pairings with LibDem and VDem-LDI as the independent variable. (For full results, see Online Supplement B.) ↩︎
3. More specifically, a country’s CLA is the residual from a linear regression of state capacity on liberal democracy. The CLA scores in Table 3 are standardized so that the mean is zero and the standard deviation is 1.0. Any positive score thus shows unexpectedly high state capacity. ↩︎
4. Full CLA rankings for the 153 countries can be found in Online Supplement B. For those who are curious, the United States falls close to the trendline, with a slightly positive CLA of 0.15. It ranks 21st of 153 in terms of state capacity and 18th in terms of liberal democracy. ↩︎
5. Vallier, p. 16. ↩︎