You are currently viewing What’s the right dose of democracy?

What’s the right dose of democracy?

If you haven’t done so already, take the reader survey here!

In medicine, there’s the occasional discussion of the dose response function of some drug. People wisely assume that it isn’t just a linear function where more medicine is always better. Rather, people accept there is an optimal dose. This model is called the hormesis model, it looks like this:

You are already familiar with this idea. Vitamins are good, but you don’t want to get too many, which will in fact kill you with a sufficient amount. It may seem strange, but even things we regard as poisons can show positive effects in low doses, the hormesis pattern. Here’s some data about bacterial growth in the face of anti-biotics:

Apparently, some bacteria benefit from a low dose of even penicillin! This kind of finding has lead many people to propose that low levels of ionizing radiation are also beneficial. It’s hard to know for sure, but the fact is rather important for the regulation of the nuclear energy. The various accidents usually lead to low levels of exposure for a large group of people. If such exposure is neutral (the threshold model) or even beneficial (the hormesis model), then clearly we have some very wrongheaded ideas about how to regulate this industry and when to evacuate people (Fukushima forced evacuations killed more people than were expected to save based even on the linear model).

In tax policy, there is the same idea, but from the perspective of the government: how high should taxes be in order to maximize government revenue? 0% tax, the government won’t earn anything obviously, 100% tax and the government will have a very hard time getting citizens to report any taxes, or do any work. Somewhere between these two ends lies the optimal value. That idea is called the Laffer curve:

Now, what about democracy? I know, in public life everybody operates on the falsehood that more democracy makes for better government. Certainly very low levels of democracy are bad for government and progress. So what is the right amount of democracy? Garett Jones’ book 10% less democracy: why you should trust elites a little more and the masses a little less makes the case that we have too much democracy.

So to spell it out:

The idea is not really new. After all, no country has direct democracy for all citizens. Most countries in fact have rather little direct democracy. Jones argues that this is a good thing. His model is simple enough: voters are short-sighted, ignorant, somewhat irrational compared to various government bureaucrats one could instead have running things. His evidence of this basically consists of summarizing a bunch of economics studies that looked into this. In Jones’ view, we should just appoint a bunch of economists to run most government things. There’s quite a number of paragraphs where Jones is very proud of his fellow economists:

Occasionally my fellow economists are derided as naive, out-of-touch scholars who know little and care less about how the real world works. But the central bank independence literature is just one example of how my fellow economists have kicked the tires and looked under the hood of their own findings. Part of the reason Cukierman and others did that is because economists know that other economists are a tough, skeptical audience. In our profession, we pass around stories of economists giving seminars on their research papers—usually at the University of Chicago, home to more Nobel laureates in economics than any other university—and never getting more than ten minutes into their slides because the other professors in the room tear the paper apart piece by piece for the next ninety minutes. The fear that such a prospect creates in the human heart spurs us to strengthen our arguments, find the data, check and see if we’re actually correct or if we’re just living in a dream world of our own creation.

So independent central banks are more likely to fight inflation, and part of the reason is probably because they delegate power to my team: the economists. But would it be asking too much of independent central banks to also cut the risk of a massive financial crisis? Apparently not. Countries with more independent central banks appear to have fewer financial crises, not more of them. Economists Jeroen Klomp of the University of Groningen and Jakob de Haan of Munich’s prestigious CESifo published a well-timed paper on the topic in 2009 just as the global financial crisis reached its peak.¹³ Looking at data from a global sample weighted toward middle- and upper-middle-income countries (where more of the financial crises happen), they looked at what happened between 1980 and 2005. They measured financial distress by examining a number of dismal symptoms: falling bank credit, falling bank share values, spikes in relevant interest rates, plus many other measures.

My George Mason colleague Bryan Caplan conclusively documented in his famed book The Myth of the Rational Voter that the least educated are the most likely to support policies that are largely rejected by academic economists: higher taxes on imported goods, rent control, and tougher government-enforced rules that make it harder for firms to fire workers, to name just a few. As Princeton’s Blinder puts it in Advice and Dissent, “A more economically literate public would be a blessing.”¹⁰

The studies

I think a nice way of reviewing a science book is to pick a bunch of random passages and see how a given study is described and how it actually is, and especially, whether it is decent or the usual p = .04 bullshit. This kind of approach has been used before e.g. Schimmack did it with some social psychologist book. I didn’t exactly do this because I am too lazy to read say 20 economist studies, but I did check out a few.

Judges and favoritism:

If there’s one thing you’d expect an elected judge to do, it would be to stick up for the people who put her or him into office. Amateur cynics think that means that elected judges will favor the big businesses that paid for their election campaign, but a professional cynic like me thinks that means that an elected judge will stick up for the voters. Let’s look at the evidence that elected judges pay attention to the voters.

The temptation of elected judges to favor local citizens over outsiders shows up in this blunt quote from an elected judge, Richard Neely of the West Virginia Supreme Court of Appeals, the highest court in West Virginia: “As long as I am allowed to redistribute wealth from out-of-state companies to injured in-state plaintiffs, I shall continue to do so. Not only is my sleep enhanced when I give someone else’s money away, but so is my job security, because the in-state plaintiffs, their families, and their friends will reelect me.”

Economists Eric Helland of Claremont McKenna and my George Mason colleague Alex Tabarrok found formal evidence that Neely wasn’t the only judge thinking that way. Across the United States, elected judges typically favor home-state citizens. Drawing on thousands of state court lawsuit trials—torts—from 1990 to 1995, they first found evidence for the fairly obvious fact that state judges give bigger awards when the defendant—the person being sued—was from out of state rather than in state. That could reflect a general bias against outsiders; it could also reflect the fact that in most states, the deepest pockets, the biggest corporations with plenty of money to pay out, are typically out of state. It’s likely a bit of both, but that anti-outsider tendency isn’t what’s of interest here. What we care about is whether the elected judges within a given state are more generous than the appointed judges in that same state when the person being sued is out of state. And indeed elected judges are more generous: the average award paid by an out-of-state defendant was about $140,000 higher when the judge was elected rather than appointed.

Seems a lot. It’s this study 20+ year old study: Alexander Tabarrok and Eric Helland, “Court Politics: The Political Economy of Tort Awards,” Journal of Law and Economics 42, no. 1 (1999): 157–188. Also see Eric Helland and Alexander Tabarrok, “The Effect of Electoral Institutions on Tort Awards, American Law and Economics Review 4, no. 2 (2002): 341–370. I looked and that value 140k, seems to be this:

Although awards against out-of-state firms are higher than average in both elected and nonelected states, the out-of-state penalty is larger in elected states. In elected states awards are $364,950 higher than average, while in appointed states they are only $219,980 higher than average. The difference in differences, what we will call the ‘‘elected effect,’’ is thus $144,970. The ‘‘elected effect’’ is highly statistically significant (F[1, 7637] = 5.1567 with P = 0.0220).


Their own regression table [table ] lists it as p < .01 but according to their own F statistics above, the value is .02. Using the table’s estimate and standard error, the t value is 2.27, which should be about p = .02.

Appointing judges:

Lawyers have one favorite way to judge who’s a good judge: citations. If a lot of judges refer approvingly to an old case judgment that you wrote, then you’re a good judge. The law tends to be conservative in the sense that a judge likes to be able to show that her judgment is rooted in precedent, in established, well-argued principles. In real life there may be a trade-off between an “established” principle and a “well-argued” principle, since some established principles of law can be based on horrible but politically popular reasoning. But for our purposes, what matters is that when a judge is casting about for a way to come to a decision that will stick, a judgment that’s less likely to be overturned by a higher court, she’ll look for well-written judgments written by others in her profession. And the legal opinion she’s citing doesn’t even have to be a winning judgment. Judicial dissents that you agree with are a great way to build a case for your own currently controversial view, a great way to lay the groundwork for judicial innovation. So whether you’re going along with the dominant paradigm or trying to subvert it, citing old judicial precedents is a path to success.

Whose judicial opinions are more likely to be cited? In the language of the field, which judgments are “higher quality”? Opinions written by appointed judges. This question has been investigated quite a few times by different scholars, and they reach the same conclusion, so let me turn to the most comprehensive study of the issue, by Elliott Ash and W. Bentley McLeod: “Judges selected by nonpartisan elections write higher-quality opinions than judges selected by partisan elections. Judges selected by technocratic merit commissions write higher-quality opinions than either partisan-elected judges or non-partisan-elected judges.”

It’s very circular. Maybe the technocrats just elect people based on who seem good at writing law papers that people cite, and then later, unsurprisingly, they have more citations. But let’s check out the study anyway: Elliott Ash and W. Bentley MacLeod, “The Performance of Elected Officials: Evidence from State Supreme Courts,” NBER working paper 22071 (2016):

It looks terrible. There’s only one real hit for p < .01, everything else is < .05, and a lot of stuff is < .10. Waste of time.

Next study:

It turns out that even if you know that a nation officially uses British common law, even if you know how rich it is, how far from the equator, and how much diversity it has, greater judicial independence is still a robust predictor of stronger private property rights. By my reading of the paper by La Porta and his colleagues, greater judicial independence is also a solid predictor of less government ownership of banks, less labor market regulation, and less red tape to start a new business. (That’s a personal interpretation. Interpreting statistical results sometimes is a bit like evaluating a painting, a matter of taste.)

It’s this one: Rafael La Porta, Florencio Lopez-de-Silanes, Cristian Pop-Eleches, and Andrei Shleifer, “Judicial Checks and Balances,” Journal of Political Economy 112, no. 2 (2004): 445–470.

Note the sneaky asterisks. *** is .01, not .005!. So with controls, the so-called robust results, we see p values of: 2x < .01, 3x .01 to .05, 2x .05 to .10. In other words, not that robust. 2 of 7 are robust and decent.


Some cities in California appoint their treasurers and others elect their treasurers. Cities can have elections to decide whether the city treasurer should be appointed by the city government; the default is that they’re elected. Whalley checks to see which kinds of cities have lower borrowing costs: ones with appointed treasurers or elected ones. The interest rate paid on a city’s debt is a useful index of how well the city is running its finances. Managing a city’s borrowing costs is complicated. Making your case to the financial system that your city is a good credit risk means focusing on a lot of details, and there are a lot of financial institutions that would love to make a California city pay high interest costs. If you can bring your city’s borrowing costs down by just a few tenths of a percent each year, you’re doing a great job. So Whalley’s overall question is this: Do cities with appointed treasurers pay lower interest rates on their debt?

Any particular California city might have lots of reasons for paying a different interest rate from the city next door. It might have a lot of citizens with lower education levels, it might have a lot of children enrolled in school who eat up revenue and whose parents aren’t paying much in taxes, or it might just be a small city that could be put at risk if one big employer moves out of town. A more obvious reason might just be that one city might start off with a lot more debt than another. Whalley shows that even if you know all of those facts about a California city, plus many more, the appointed treasurers are able to get lower interest rates on the city debt—about half a percentage point lower on average—than elected officials. In his basic tests, appointed treasurers win hands down.

But maybe there’s some other feature of the city that he didn’t quite catch—a statistical je ne sais quoi. He has two more tricks up his sleeve from the standard statistical tool kit. First, he treats every city as its own experiment and looks just at differences in interest rates before and after a city switches to having an appointed treasurer. Over the period Whalley examined, 1992 to 2008, forty-three cities held referenda to ask whether they should switch to appointed treasurers. He’s therefore able to look at the before-and-after differences of these elections in two ways, and the second is worth our attention: regression discontinuity design (RDD). That’s roughly equivalent to comparing interest rates in cities that just barely voted for an appointed treasurer (like a 51% vote) to interest rates in cities that just barely rejected an appointed treasurer (like a 49% vote). In a case like that, a city that just barely accepts is probably a lot like a city that just barely rejects, so this is as close to an experiment as we’re likely to get outside a petri dish.

This RDD method finds an even bigger benefit of appointed treasurers: seven-tenths of a percent lower interest rates every year. The average city in the sample has about $30 million in debt, so that comes out to a savings of $210,000 per year. It’s probably worth giving up some local voice to get those savings.

Study: Whalley, Alexander. “Elected Versus Appointed Policy Makers: Evidence from City Treasurers.” Journal of Law and Economics 56, no. 1 (2013): 39–81.

Table for fixed-effects model:

P values are both bad, between .01 and .05 and the second is between .05 and .10.

RDD table:

P values are both bad: between .01 and .05.


They looked at data from “the original 15 EU member states over seven years (1997–2003)” to see how independence affected a crucial element of competition: How much were big incumbent telecoms allowed to charge to upstart newcomers that wanted to interconnect through the incumbents’ networks?²³ So if a new France-based telecom wants to connect a call from France to Sweden by way of Germany, is the German telecom allowed to set a discriminatory high price just for newbies, perhaps as a ploy to curtail competition? Or does the regulator insist that the German telecom set a lower price, close to cost and close to the apparent prices that it charges other established competitors? This example is, of course, an illustration (details vary from year to year, industry to industry), but the question of whether new firms will be treated—let’s go ahead and use the word fairly—will shape whether your industry winds up with healthy competition between firms or with just a few high-priced regional monopolies.

Edwards and Waverman found a clear result. Yes, regulator independence predicted lower (fairer!) interconnect prices, but that was driven by one crucial feature: whether the regulator was confronting a government-owned telecom. Professors often worry about whether big business will distort market competition, but it turns out that the old bumper sticker has a lot going for it: “Don’t steal: the government hates the competition.” Independent regulators were apparently more likely to stand up against telecoms owned by big government.

Clear result? OK, let’s see. Study: Edwards, Geoff, and Leonard Waverman. “The Effects of Public Ownership and Regulatory Independence on Regulatory Outcomes. Journal of Regulatory Economics 29, no. 1 (2006): 23–67.

EURI-I is their independence variable, which has p < .01 (I calculate to be about .0009, t = 3.2). Looks fine! Their interaction, however, that Jones makes a big deal out of, is p < .05, and looks to be about .011. So that one is more doubtful. Model 3 has a better p value, I don’t know what the difference is with the two variants of the interaction. The last two models also have dubious p values.

Overall, then, my examination of the basis of his “bro, just listen to us economist experts, and give up democracy” is mostly a bunch of p-hacked studies. I wonder if economists don’t need a worse reputation than they already have as being extreme science cheaters, but I am not even sure they cheat more than the other social scientists. Certainly, economists have some overall more sound ideas about methods and policy, so all in all, yeah sure, we should indeed hire more economists into various technocrat roles if the alternative is sociologists or liberal arts people.

Jones doesn’t discuss at all the failings of economists or elites. He quotes favorably that survey they run of top economists:

When it comes to international trade, economists are about as unified in their views as any group of experts can be. The University of Chicago’s Initiative on Global Markets (IGM) regularly surveys both leading economists around the world, as well as a separate European-focused group of economists on a variety of policy questions. Regardless of how the questions have been phrased in the past decade, both the global IGM panel and the European IGM panel have overwhelmingly agreed that reducing trade barriers is good for the nation that lowers them. Letting foreigners compete in your markets is good for you, partly because you get cheaper goods and partly because the invisible hand will move your nation’s workers over to more productive uses.

Here’s a statement from 2016 that the IGM’s economists were asked to disagree or agree with: “Freer movement of goods and services across borders within Europe has made the average western European citizen better off since the 1980s.” Of the European economists surveyed, 68% strongly agreed and another 24% agreed—just about everybody. And since economists routinely consider how trade changes the income distribution, the IGM European panel was asked to disagree or agree with this statement: “Freer movement of goods and services across borders within Europe has made many low-skilled western European citizens worse off since the 1980s.” Here there was more disagreement: 20% agreed that lower-skilled citizens were often worse off, and 2% more strongly agreed, while 42% disagreed and 10% strongly disagreed. The uncertain were 18% of the sample. That means about one in five thought many lower-skilled workers were worse off, and another one in five weren’t sure. But there’s nothing close to a consensus that intra-Europe trade deals hurt lower-skilled Western European citizens. Pie-growing change often entails real costs, of course, and they’re not always the kind of costs you can measure with money. When the family moves to a big city so the parents can earn more, there’s probably one kid in the family who loses a treasured best friend, never to be regained.

A 2018 survey about European trade with China yielded similar answers: freer trade with China appeared great for Europeans, and while it certainly hurt European workers (as workers) in key industries, that alone wasn’t sufficient to argue for a policy like higher import taxes on imported Chinese steel.

Now, I’ve written at this one before. The same survey panel was also asked about Germany’s expected economic success with migrants, and it had this completely insane result:

How’s it going halfway into the period? (German original):

A new study published on Tuesday found that 49% of refugees who have come to Germany since 2013 were able to find steady employment within five years of arriving.

“This means that labor market integration is somewhat faster than for refugees from previous years,” said the Insitute for Labor Market and Vocational Research (IAB), who carried out the study.

IAB, which is part of Germany’s Federal Office for Migration and Refugees (BAMF), compared the numbers since 2013 to arrivals between the early 1990s and 2013, when the rate was slightly lower at 44%.

The Institute praised Germany’s efforts with integration and language learning for the progress: “In addition, significantly more has been invested in language and other integration programs for asylum seekers and recognized refugees since 2015.”

Employment rate and income are the main ways citizens contribute economically. This doesn’t sound like net benefits to me. The current German national unemployment rate is 5.3%, and that includes these people. Considering the welfare use and subsidized housing, language classes etc., this cohort is very likely to be a net negative for German citizens. And this was an easy forecast to make because there are ample such prior waves of immigrants.

To be fair, I think the main reason for this colossal failure of economists to predict a super easy case was that they were answering in public with their own names. If only economists would learn from social psychologists about how public responding leads to conformism bias, and sometimes a lot of it. Unfortunately, there doesn’t seem to be any anonymous surveys of economists about policy issues. If you can find any, post them in the comments! I’d be very interested in seeing what the differences are between public and private beliefs of economists.

My point with the above is that regular people were much more right than economists on the most important of matters: demographics. Jones is probably right that technocrats can do better for many complicated topics, but voters are usually more correct on matters of demographics. What does it matter if you can reap a 10% increase in government efficiency if your grandchildren cannot live in your capital anymore due to Muslim and African criminals roaming the streets? Demographics issues are very difficult to reverse, but government tax policies can be changed whenever one wants to.

That said, Jones is right with his general thesis. We probably have too much democracy. If voting was restricted to non-criminals, people not on welfare etc., then society would have much more sensible economic policies. Whenever it becomes possible to vote yourself to other people’s money, you will get an underclass that lives this way and it will keep expanding. For the last 100 years or so, we have also had to suffer an upper class that loves socialism. Maybe we should make it so that only the middle class can vote. Jones helpfully quotes Aristotle on the topic:

Aristotle, a student of Plato who in turn was a student of the ill-fated Socrates, emphasized that some sort of balance was likely to lead to better government:

If I was right when I said in the [Nicomachean] Ethics that . . . goodness consists in a mean, it follows that the best way of life is a mean, a mean which can be attained by everyone. These same criteria must be used to judge the excellence or otherwise of a constitution, which is, so to speak, the life of a state . . . Every state has three parts: the very rich, the very poor, and the middle class.¹

Aristotle spends some time elaborating on the merits of the middle class—the “mean” or average class. He says they tend to listen to reason (compare Chapter 5), they are enough like each other that friendliness and social equality are more likely to arise, and compared to the poor and rich, respectively, they’re more likely to pay taxes and less likely to covet government power. Aristotle concludes, “Therefore, of course, a state which rests upon the middle class is the best constituted in respect of those elements which, in our opinion, constitute a state. . . . All this goes to show that the best political society is one where power lies with the middle class.”²

But not every state can hand power solely to the middle class. In Aristotle’s day, that was partly because many states didn’t have a big enough middle class, and in our modern age, it’s partly because of the cultural norm of near-universal suffrage in democracies. When inequality is extreme, Aristotle sees little hope: “In those democracies which have no middle class and the poor far outnumber the rich, trouble ensues and the state soon goes to pieces.”³