The quiet vindication of Fukuyama

Liberal democracy works better than any alternative

Francis Fukuyama’s 1989 declaration of “the end of history” became one of the most ridiculed ideas in political thought. The suggestion that liberal democracy represents the final form of human government seems absurdly premature, especially after the rise of China, the resurgence of authoritarianism, and democracy’s visible struggles in some established Western nations. But when we look at the empirical evidence and think long-term, I think it’s more likely that he was right.

At least he’s right that robust capitalist economies paired with liberal democracy have all sorts of edges over the authoritarian alternative. It might take a long time for democracy to completely win out, materially, but that was precisely one of Fukuyama’s points.

I’m sure all of you are familiar with the main contents of his thesis, but because people love to straw-man, I guess I have to do some obligatory throat-clearing. Namely, by “the end of history,” Fukuyama didn’t mean events stop, or that nothing bad can ever happen, or that wars will disappear, or that everyone will come to ideologically love democracy, or that democracies can never materially backslide, or that anything is guaranteed at all. He meant that the long-run competition over workable political orders tends to favor liberal constitutional democracy paired with markets because it best solves recurring problems of information, accountability, peaceful succession, and overall material prosperity/human development. Liberal capitalist democracy is the system with the fewest internal contradictions, to adopt a Marxian term.

Okay, so how do we know that democracy works well? Well, John Gerring et al. (2022) surveyed more than 1000 statistical estimates from over 600 published papers on democracy’s correlates over the past 20 years. Across many outcomes involving social policy, economic policy, human rights, criminal justice, and overall government quality, most studies report statistically significant positive results. Here, positive is coded to correspond with a normatively desired outcome, such as reductions in mortality and corruption and increases in growth and human rights.

Democracy demonstrates quite a robust positive correlation with human rights, transparency, corruption reduction, human development/health, trade openness, low militarism, and quality of government. When vote-counting studies, authors find that around two-thirds or more of the papers (64%-77%) point in the normatively desired direction. (Though I’m not sure if “low military spending” is normatively desired on Fukuyama’s thesis. Because if democracies are militarily weak, they can get outcompeted by autocracies. This has been thrown into sharp relief over the past few years here in Europe.)

An unsurprising caveat is that there is evidence of publication bias in the literature, with studies more likely to be published if they show statistically significant results. This holds for both positive and negative results, but the effect is stronger on the positive side. In any case, when the authors correct for this, the overall positive and significant result persists. Additionally, democratic effects are more likely to be found for outcomes that are easily attained by government action than for those that lie beyond its reach. I don’t think there’s any bias here, as it actually makes great sense for this latter asymmetry to show up if the positive (causal) effect actually exists.

The Chinese model that many observers have held up as a viable alternative to democracy appears less compelling in this light. Because China has achieved remarkable economic growth, some people view its authoritarian capitalist model as an overall viable alternative for promoting good governance and human development. But this aggregate evidence suggests otherwise. When you look across the full range of governance dimensions rather than focusing narrowly on GDP growth, democracies systematically outperform autocracies.

What about economic growth?

However, another caveat worth noting is that there’s only mixed evidence for a positive correlation with a key variable that would be important for Fukuyama’s thesis: economic growth shows up as significant in only 55% of the studies. What should we make of this? In lieu of any other evidence, we should accept it and admit that democracy just isn’t clearly ahead, economically speaking. However, there is some additional evidence we can draw on.

First, has China, specifically, really achieved remarkable economic growth in comparison to its democratic East Asian peers? I’m not so sure. True, South Korea and Taiwan were also not democratic for the first few decades of their economic ascension, but that’s not the point. Following Acemoglu and Robinson, I’d argue that you can get fast growth as a closed authoritarian regime – if you’re exceptional, of course, and very poor – for a few decades. But, after that, perhaps you can keep continually growing for decades to come only if you transition to democracy. That’s precisely what happened to South Korea and Taiwan (and Japan at an even earlier period, although its development has stalled since the 1990s).

Here are two quick comparisons. The left panel shows economic development in the decades after each country started growing fast. The comparison between China and South Korea is the fairest here, because both started their fast-growth period at very similar levels of development. The right panel shows an overall fairer comparison between all four countries, because it shows their developmental trajectories from the point at which each reached a real per capita GDP (PPP) of $10k.

A fair accounting of China’s record would also tally up the tailwinds and headwinds that don’t show up in a single growth line. That is, the country had unusually favorable demographics through the 2000s, accession to a historically large trade-and-investment cycle, and massive investment-led expansion with diminishing returns.

Second, and more systematically, Carl Henrik Knutsen’s recent work suggests that focusing solely on average growth effects misses the crucial dimension of economic stability and risk mitigation. While some autocracies achieve spectacular short-term growth, they also experience catastrophic economic collapses at rates democracies simply never match. The variance in economic performance is dramatically higher under autocratic rule.

Knutsen (2021), “A business case for democracy: regime type, growth, and growth volatility,” Democratization.

As you can see, democracies (in red) are significantly less likely to experience negative growth in almost any given timeframe, and far less likely to experience severe economic crises with growth below -5%. More careful regressions show the same. When comparing volatility in growth rates over decades, autocracies show variance roughly 1.6 times as high as that of democracies, even when controlling for differences in income levels and other relevant factors. This pattern holds across more than two centuries of data and remains robust even when accounting for country-specific effects. In recent decades, the disparity in variance has gotten even bigger.

Another important consideration is that we have some evidence of growth mismeasurement because autocracies systematically inflate their GDP statistics. Look at the image below, which is based on the findings from a wonderful paper by Luis Martinez (more specifically, this). He uses satellite night-lights as an independent measure of real economic activity, then compares that to countries’ reported GDP. I’m not completely sure whether this is true, but I wouldn’t be surprised if autocracies misreport their development. (I’m also aware that if it turns out to be true, then the happy reports of the developing world economically outpacing the developed world over the past few decades might also be wrong or at least overestimated.)

So, okay, let me stop here. I don’t want to hand-wave away the relatively mixed evidence on democracy and growth. We need more studies to be sure.

But apart from growth, the data are relatively clear. There’s no better practicable political system than democracy. If by 2050, we routinely observe large, innovative autocracies delivering high living standards while suppressing rights, the rule of law, and open information, then I think Fukuyama’s thesis can be declared wrong. But we’re not seeing that right now, and to my mind there aren’t any strong indications of it changing over the coming decades.

(Unless, of course, the coming AI singularity kills, or supremely enriches, us all.)