Democracy for Realists, Part III

There are two ways of thinking of the role of politicians in a representative democracy. The first, which was previously addressed, is a vision of the politician as an avatar: they are to ‘represent’ the majority views in their constituency, and put aside any personal views that conflict with that. This is the operative view of the folk theory of democracy, which the authors of Democracy for Realists have provided ample evidence against.

For another quick example of the flaws in this view, take the recent polls showing that Obamacare, now that it’s on the chopping block, is more popular than ever. Republicans believe they were elected with a mandate to repeal it (set aside the problem that more people actually voted for Democrats). The only thing that has changed about the ACA over the last year has been who controls political power. If we were to take the folk theory seriously, it would imply that a non trivial percent of the population has simultaneously carefully considered the evidence and come to new conclusions regarding their preferred health care system. This strains credulity. Instead, it stands as testament to the fickleness of our opinions. Which beliefs should politicians represent? The one the people held when they elected them, or the ones they have a few months later?

In the accountability model, the primary way democracy operates is by holding politicians accountable for their past performance. In contrast to the folk theory, politicians are viewed not as avatars but as as trustees of districts. Voters have the power to remove politicians if they don’t like their performance, but otherwise there is no active mandate.

The accountability model, then, views voters as exercising indirect control of policy without sophisticated knowledge by assessing politicians’ performances, rewarding success with reelection, and failure with ouster. Would I need an understanding of the mechanisms of monetary policy to know that tight money leads to recessions? No, because I need only see that I lost my job because the tight money policies of the party in power led to a shrinking economy.

Through the threat of diselection, politicians are incentivized to act in the interest of the voters – if a policy were good for a politicians but bad for voters, the accountability model would predict that politicians would not implement that policy lest they are kicked out at the next election. Elections serve as a solution the principal-agent problem between constituents and their representatives.

The accountability models faces two challenges: 1) It requires voters to accurately calculate changes in their welfare between elections, and 2) Separate out the things which the government is responsible for from those that are not. If these conditions are not met, the accountability model falls apart; when a politician’s chance of reelection is no longer tied to their performance, they are free to act in their own interest without fear of assured consequences. As you might suspect, Achen and Bartels provide considerable evidence that this is the case.

To begin, it is very well established that the state of the economy (particularly income growth) has a measurable effect on election outcomes. This makes sense; the economy is central to our lives and is responsible for a great deal of our well being. Politicians that manage an economy poorly are most certainly not acting in the general interest, and should be appropriately removed from office.

However, there’s a catch: studies show that it’s only the performance of the economy in the six months before an election that matters to voters. Our memories are too short; we overweight recent performance and underweight the past. So, an economy could be in shambles for 3+ years, but if personal income begins to grow substantially in the summer before a presidential election, the incumbent party’s chance of reelection is much higher than it ‘should’ be.

The potential consequences of this fact are worse than simple unfairness. It incentivizes incumbents to intervene and juice the economy just before an election. I’m a bit skeptical as to the role this plays in American elections, though the authors do show that personal income growth is statistically significantly higher in US presidential election years, over and above the growth rate of national GDP.

That being said, the degree of control the president has in controlling of the economy is often popularly overrated. Perhaps there is tinkering around the edges, but the timing of major economic events, like recessions, is essentially random. It follows from this that the results of elections will tend to be random also. Take the 1992 election for example. It was preceded by a mild recession in 1990-91 that still lingered into the summer of ‘92 – unemployment was still rising as late as that June. If that recession had happened just one year earlier, the effects would have largely been gone by election time and George H.W. Bush would have certainly won reelection. In this instance, the policy consequences were probably not that great; Bill Clinton’s New Democrats were not that far away from the Republicans of that era. But today, the divide between the parties is wider and growing. In an era of polarization and a powerful executive, random control of the presidency seems disastrous.

Of course, there is more to the story than the economy. Achen and Bartels also cover popular punishment for natural disasters and other events that government has no plausible control over. The main example they use are droughts. They provide convincing evidence that drought leads to more anti-incumbent votes in areas that are reliant on farming and ranching. Why is this problematic for the accountability model? If voters were punishing poor responses to such events, by definition half of the time the response will be above average. But the authors show this makes no difference in people’s voting patterns: above average, below average, incumbents are punished just the same. Again, this removes the operating mechanism through which the accountability model works, namely, actual accountability.

To close, an interesting concept the authors address is the ‘social construction of blame.’ After all, it’s not as though we hold government responsible for any bad thing that happens to us; if we get into a car wreck on the way to the polls, we aren’t going to change our vote. (Though there is some evidence that events like important sports championships may influence voting by affecting mood; the Cubs did win the World Series just before Donald Trump was elected president…).

What is necessary is a popular story linking blame to the incumbents, whether accurate or not. Two examples to illustrate: Repeated plagues of locusts in the 1870s seemingly had no effect on incumbents in sparsely populated Nebraska, despite devastating effects on farmers; without clear communication and organization, there was no shared interpretation of the plague that placed blame on the incumbent Republicans. And during the 1918 Spanish Flu epidemic, which killed about half a million people in the United States, there was also no punishment effect. At that time, there was no plausible popular story linking government to a public health issue.

Again, another theory of workings of democracy is shown to be inadequate. People are simply not ‘rational’ enough in aggregate for the accountability model to be plausible. So if voters do not vote in accordance with ideology, nor do they consistent elect competent managers, then why do people vote the way they do? And can the idea of democracy as producing responsive government be saved? We’ll see next time as Achen and Bartels begin to construct a realist theory of democracy.


Hypocrisy and Nihilism

That is the subject of an op-ed in the LA Times from a couple weeks back by, again, Jacob T. Levy, which has forced a reconsideration on my part.

I’ve long considered hypocrisy one of the most unforgivable sins in all of political discourse. To see pundits rail for years against policy X while party A is in power, only to suddenly discover its merits in the second week of a November when party B takes control is unbearable, as is witnessing policy Y be justified by an appeal to the very values it abrogates. War is peace, freedom is slavery, we have always been at war with Eastasia, etc etc. 

To be sure, I’m not innocent here either; we’re all guilty to some extent of mood affiliated beliefs. What upsets me is the shamelessness of some political actors when it comes to hypocrisy, and I’ve reached a place where I tend to dismiss any sort of argument coming from a blatant hypocrite. But this is wrong and short sighted.

One example: I have long had a particular loathing of Thomas Jefferson, a man who could talk about the rights of man and rail against tyranny while also owning human beings. (Please, no arguments about how he hated slavery and wished it would disappear, but was trapped by the times. As is told in Daniel Walker Howe’s brilliant history What Hath God Wrought, as the slavery debate heated up in the early 1800s, Jefferson repeatedly wrote in support of the South’s right to slavery.) 

But what this article reminds me is that, as distasteful as hypocrisy is, it’s far preferable to nihilism. It is better to imperfectly aspire to some values than to aspire to nothing at all.

And that seems to be the situation we are currently in when it comes to our political leadership. The particular example in the article is the Trump interview with Bill O’Reilly, in which the President dismisses arguments against Vladimir Putin by insisting that we aren’t so innocent either. 

Of course, it’s true that America has a problematic history of clandestine operations and bloody interventions (when it served our interest). But we don’t gun down journalists in the street. We don’t poison political rivals with radioactivity. To equivocate the US with that type of behavior is disturbing.

I was never a fan of George W. Bush’s freedom rhetoric; I thought it was overplayed and often silly. But now we have a President who doesn’t talk about these kinds of values at all. Trump’s main themes are safety and jobs, and he rarely speaks of broader principles. These are the kind of justifications we are used to hearing from illiberal strongmen: I will keep you safe, I will keep you fed, I will give you jobs.

One of the themes of Francis Fukuyama’s work is that ideas have power. The invention of the liberal democracy was not the inevitable result of the forward march of history. It required the intersection of several independent factors, one of which was a particular idea about the legitimacy of government requiring the consent of the governed.

Imagine that the Declaration of Independence made no mention of inalienable rights or freedom; or imagine there never was any such document at all in the course of the War of Independence. After all, to some extent the document was propaganda; many of the Founders were well enough off and only wanted to increase their relative status. Would our history have played out the same?

I think the answer is no. To be sure, the values laid out in the Declaration didn’t prevent a shameful program of Indian extermination in the following decades, nor another century of slavery. But without such rhetoric, would the franchise have been extended so early in spite of the Founders skepticism of democracy? Would the Second Founding have begun in the aftermath of the Civil War?  Would the postwar liberal international order even exist if the War of Independence had been fought to the cry of ‘America First’?

It’s not enough to point to the hypocrisy of political leaders and walk away in disgust. In doing so, we run the risk of falling prey to political nihilism, which is incompatible with liberal democracy and a belief in any kind of progress. If this nihilism becomes the political mainstream, we really will be no different from Russia or any other illiberal polity.

Immigration and Integration

I recently read this interesting Lyman Stone post on immigration via a related post on Noahpinion; both are worth a read. Mr. Stone’s article is on the longer side, and does an excellent job of laying out all the arguments surrounding immigration. While I don’t know that I agree 100% with his analysis, I appreciate that he takes a nuanced, data informed position (and has lots of graphs!). The gist of both is that, contra immigration hardliners, the claim that a broad immigration ‘pause’ is necessary is not supported by evidence.

Immigration skeptics, like the President and his advisors, often rely on claims of danger to support their position. Trump actively campaigned with tales of Mexican criminals flowing across the border. Given that there is zero evidence that immigrants, illegal or otherwise, commit more crimes than the average American, I suspect this is often a cover, intentional or not, for something else. What that ‘else’ is, is a fear that immigrants either 1) won’t assimilate to American culture or 2) rather than us assimilating them, they assimilate us.

This fear is, on its face, not wholly unreasonable, especially if you believe that there is truth to the idea of American exceptionalism. Perhaps those values that have made America unique would be diluted and lost if too many ‘different’ people came here. What the posts I linked at the beginning argue is that, even if this were true, that is not the situation we are currently in.

To be clear, another thing immigration skeptics tend to be shy about expressing (but not Trump!) is that most of their fears are directed at Mexican immigration. And that makes sense, because in the last few decades Mexicans have far outnumbered any other group of immigrants; no other single group comes close. But even bigger than this surge was that of Germans and Irish during the mid 19th century (both were close to 5% of the total US population at their max, versus about 4% of foreign born Mexicans today)- and we all know the end of that story. So there’s historical reasons to be optimistic: peoples who were once viewed with suspicion and anxiety were eventually so thoroughly absorbed into American society that nobody would question the Americaness of people of those heritages. It happened before, and it will happen again.

An argument against this might be that the German and Irish were Europeans, and American has a European heritage; what we’re dealing with today is fundamentally different. I just don’t buy that. It’s easy for us today to forget just how vehement the dislike and distrust of Catholics was in 19th century America, even into the 20th; it was still a big deal only 57 years ago for a Catholic to run for president. The Know Nothing party of the 1850s held that German and Irish Catholics were controlled by the Pope and were fundamentally incompatible with republican government. Now we look back on that and think it’s silly. I suspect we’ll look back on this period in much the same way.

Now, Mr. Stone makes the argument that given how difficult that earlier period of integration was, there is a case for limiting Mexican immigration while still encouraging overall immigration. Integration is harder when there is such a large population of your fellow immigrants that you can get by without really interacting with the rest of American society. While I think a deeper look at the exact mechanisms of integration is necessary, it’s an argument that makes sense and could lead to a compromise solution among actors of good faith on both sides of the issue.

One final important point Mr. Stone makes is that nativist backlash against immigration is often spurred not by high levels of immigrants, but high growth rates in the recent past. By this token, since the net rate of Mexican immigration to the United States has been close to zero since the recession, the backlash may  dissipate soon. I can only hope it does so before the current administration puts any extreme immigration policies in place.




Terrorism, Part II

This is a continuation of an exploration of the Global Terrorism Database. Last time I mentioned the GTD codes for whether an attack was ideologically international in scope. What follows are three charts breaking down global terror attacks by this variable:

In the last few years there has been an uptick in ideologically based terrorism, alongside the overall rise documented last time. Is there a meaningful difference in the average number of deaths depending on the type of attack?

A simple t-test of the number of deaths from clearly domestic vs clearly international terrorist attacks is not significant, suggesting there is no meaningful difference. However, performing the same test against the unknown attacks does show a meaningful difference for both domestic and international attacks.  This suggests to me, assuming this is not due to problems with the data, that attacks that remain unknown are more likely to be the work of individuals or amateurs rather than professional, organized groups, and hence less deadly.

Have attacks gotten more deadly over time? Breaking the sample into pre-2001 and post-2001, there does seem to be a statistically significant difference in deaths/per attack, with attacks having grown deadlier, though not by a huge amount (I ran an analysis excluding 2001 in case 9/11 was having an outsize effect; there was essentially no difference in result):

data: subset(dat, Year < 2001)$nkill and subset(dat, Year >= 2001)$nkill
t = -4.4096, df = 147780, p-value = 1.036e-05
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
-0.3717482 -0.1429688
sample estimates:
mean of x mean of y
2.215667 2.473026

Now for the United States. First, the raw number of deaths from terrorist attacks. I had to manually cut off the graph, because otherwise the deaths from 9/11 would totally swamp all the other data.  We can see that there has been an uptick in recent years, though it doesn’t appear much different from what was occurring in the 70s.


What kinds of attacks are these? (Note that again I limited the y-axis).


The difference is that the attacks in the 70s were definitely domestic (think the Weather Underground), whereas the ones of late are of uncertain origin. Since 9/11, however, there have been no deaths from an outside based terrorist group. (One aside, it’s strange to me that the Oklahoma City bombing is coded as an unknown ideology). Also, while there has been a rise in the last few years, it must be noted that this number is dwarfed by the number of people killed in car accidents, accidental gun deaths, or drug overdoses.

Now these days, when people think of terrorism, they overwhelmingly associate it with jihadism or other kinds of Islam-linked extremism. (Indeed, the new Trump administration has elected to focus their domestic counter-terrorism efforts solely on Islamic extremism). Is this fair? That is, how much terrorism is actually associated with Islam?

Of course, this is a very difficult question to answer. It requires untangling complex aspects of identity, such as religion, ethnicity, and culture. Was the Irish Republican Army a ‘Catholic’ terrorist group? The Ulster Defence Association a ‘Protestant’ group? Both groups were composed of members of those respective religions, but it seems hard to say there were particularly ‘Catholic’ or ‘Protestant’ aspects to their terrorism. As such, groups like the mujahideen, who fought the Soviet occupation of Afghanistan, would not seem to be fundamentally ‘Islamic’, though they were composed of Muslims.

That being said, there is obviously a stain of terrorism, like that pioneered by Al-Qaida, that is rooted in a radical vision of the meaning of Islam. I’ve a cursory attempt at separating out such groups for analysis. The GTD codes for the perpetrating group, if known. I created two groupings: the first consists of attacks made by groups with a name matching any of several keywords associated with jihadism, such as ‘Al-Qaida’, ‘Islamic State’, ‘Al-Shabaab’, etc. I did not include Palestinian based groups, because generally the dominant goal of such groups is politically related. The other grouping is everything else, with the additional exclusion of unknown and unaffiliated groups.

Assuming my groupings were close to correct, we see that organized jihadist terrorism didn’t become a significant player until the early 90s, though it was still dwarfed by other kinds. But since about the mid 2000s, the two have moved in tandem, and are currently equal. However, in the second figure you can see this is very sensitive to the inclusion of the most terror-ridden nations: removing Iraq and Syria, the level of jihadist terrorism goes down significantly; removing Afghanistan lowers it even further. The situation in these countries, having been destabilized by civil war, may be distortive in understanding global trends.

Global Terrorism since 1970

I’m working on a larger project at the moment, but want to use this space to post some of the work in progress. I’m looking at data from the Global Terrorism Database, which tracks a staggering amount of detail on terrorist attacks going back to 1970.

To begin, how has the number of terrorist attacks changed over time?


There is a striking increase in attacks after 2011, with the current rate about triple what it was before.  Where are these events occurring?


Clearly these events are not evenly distributed throughout the world. The spike of the last six or so year is driven almost entirely by the Middle East & North Africa and South Asia, with a much smaller spike in Sub-Saharan Africa.

These have just been the sheer number of events; what about fatalities? First, a look at the percent of events having at least one recorded fatality:


There seems to have been a slight trend towards increased deadliness. Now, the total number of recorded fatalities by region:


The pattern is broadly similar to the raw number of attacks. Finally, a look at the average number of people killed per attack, again by region:


Beyond an obvious spike for 9/11 and a handful of other isolated events, there doesn’t appear to be any clear pattern over time. If terrorism were becoming more coordinated or tactics were becoming more advanced over time, we might to expect ‘better’ terrorists to kill more people per attack; that isn’t happening.

Looking at this data, it’s striking how concentrated terrorism is in just a few parts of the world. It’s also interesting to see terrorism fade from Central and South America after a moderately strong presence in the 80s and 90s.

The database also attempts to keep track of whether a terrorist attack was ideologically motivated or not, though there isn’t information on the particular kind of ideology. What I hope to do next is see if there is any pattern in the frequency of ideologically motivated attacks over time, and whether those are deadlier than non-ideological attacks.

Democracy for Realists, Part II

<Part I>

Some might argue that the disconnect between voting patterns and policy outcomes is a result of the political system. The argument might go something like, ‘It’s not the voters’ fault that politicians don’t reflect the will of the people; the corruption of political institutions is the culprit. If we could just put more power in the hands of the people and out of the hands of politicians, things would be better.’ This strain of thinking, best summed up as ‘the cure for the ills of democracy is more democracy,’ was a motivating force behind the Populist and Progressive movements in late 19th and early 20th century America and is fundamentally based on the folk theory of democracy. It still clings to the notion that voters have coherent and independent policy preferences.

But as the Achen and Bartels point out, the available evidence doesn’t look great here either. They look at two prominent examples of ‘more’ democracy: the progressive era reforms of the referendum, recall, and initiative; and the opening of the party primary system. We’ll begin with the latter.

Continue reading “Democracy for Realists, Part II”

Link Roundup

There have simply been too many good pieces published lately to keep up with. Here are some that I think are worth reading, with some select quotes and commentary.

First is David Frum’s worrying cover story in the Atlantic, ‘How to Build an Autocracy’, written even before the events of the past weekend. This is essential reading. Frum, the former George W. Bush speechwriter who coined the phrase ‘axis of evil’, is certainly no left wing shill, and people of all political persuasions should take him quite seriously. A short excerpt:

Yet the American system is also perforated by vulnerabilities no less dangerous for being so familiar. Supreme among those vulnerabilities is reliance on the personal qualities of the man or woman who wields the awesome powers of the presidency. A British prime minister can lose power in minutes if he or she forfeits the confidence of the majority in Parliament. The president of the United States, on the other hand, is restrained first and foremost by his own ethics and public spirit. What happens if somebody comes to the high office lacking those qualities?

Over the past generation, we have seen ominous indicators of a breakdown of the American political system: the willingness of congressional Republicans to push the United States to the brink of a default on its national obligations in 2013 in order to score a point in budget negotiations; Barack Obama’s assertion of a unilateral executive power to confer legal status upon millions of people illegally present in the United States—despite his own prior acknowledgment that no such power existed.

Donald Trump, however, represents something much more radical. A president who plausibly owes his office at least in part to a clandestine intervention by a hostile foreign intelligence service? Who uses the bully pulpit to target individual critics? Who creates blind trusts that are not blind, invites his children to commingle private and public business, and somehow gets the unhappy members of his own political party either to endorse his choices or shrug them off? If this were happening in Honduras, we’d know what to call it. It’s happening here instead, and so we are baffled.

The entire piece must be read; Frum soberly lays out all the ways in which a Trump administration could (and probably will) be corrosive to the already very weakened structure of our democracy. But Frum reminds us, this will only come to be if we allow it. This piece should galvanize any concerned citizen into action. Whether you are conservative or liberal, the stakes here are higher than any policy preference:

Those citizens who fantasize about defying tyranny from within fortified compounds have never understood how liberty is actually threatened in a modern bureaucratic state: not by diktat and violence, but by the slow, demoralizing process of corruption and deceit. And the way that liberty must be defended is not with amateur firearms, but with an unwearying insistence upon the honesty, integrity, and professionalism of American institutions and those who lead them. We are living through the most dangerous challenge to the free government of the United States that anyone alive has encountered. What happens next is up to you and me. Don’t be afraid. This moment of danger can also be your finest hour as a citizen and an American.

Next is an excellent piece by Benjamin Wittes from the Lawfare blog, a site that is a well regarded resource on national security law, examining in detail this past weekend’s executive order. (For my own quick take, see here). Spoiler: it doesn’t come out looking good.

The malevolence of President Trump’s Executive Order on visas and refugees is mitigated chiefly—and perhaps only—by the astonishing incompetence of its drafting and construction.

I don’t use the word “malevolence” here lightly. As readers of my work know, I believe in strong counterterrorism powers. I defend non-criminal detention. I’ve got no problem with drone strikes. I’m positively enthusiastic about American surveillance policies. I was much less offended than others were by the CIA’s interrogations in the years after September 11. I have defended military commissions.

Some of these policies were effective; some were not. Some worked out better than others. And I don’t mean to relitigate any of those questions here. My sole point is that all of these policies were conceptualized and designed and implemented by people who were earnestly trying to protect the country from very real threats. And the policies were, to a one, proximately related to important goals in the effort. While some of these policies proved tragically misguided and caused great harm to innocent people, none of them was designed or intended to be cruel to vulnerable, concededly innocent people. Even the CIA’s interrogation program, after all, was deployed against people the agency believed (mostly correctly) to be senior terrorists of the most dangerous sort and to garner information from them that would prevent attacks.

I actually cannot say that about Trump’s new executive order—and neither can anyone else.

Now something shorter from Tyler Cowen at Bloomberg View on bursting the filter bubble. In these times, not only does it seem increasingly difficult but also even more essential, to understand the views of those we disagree with. When many voices in media simply cater to their viewership and have no interest in actually informing the public or engaging in a productive dialogue, how can we burst the filter bubble? Cowen recommends engaging in a version of the Ideological Turing Test. The original Turing Test is a benchmark for artificial intelligence: if an AI can, through conversation, fool a human into thinking that the AI itself is human, then it has passed the Turing Test. Similarly, the Ideological Turing Test asks us to be able to explain an opposing view in such terms that we could fool others into thinking we held that view. It’s sound advice, and I anticipate using this space in the future to experiment with such pieces.

Finally, the Niskanen Center, a libertarian think tank, has been publishing a fantastic series of pieces on liberalism. Jacob T. Levy, whose work I wrote about earlier, in particular has been very good. While nearly everything so far is worth reading, these pieces in particular are essential:

The Future of Liberalism and the Politicization of Everything: A good primer on liberalism.

A Tale of Two Moralities, Part One: Regional Inequality and Moral Polarization: Expands on many of the issues of inequality and sorting I discussed in Our Kids Are Coming Apart.

The Party Declines: A defense of party politics. Novice political observers tend to be very cynical about political parties, and imagine that a universe without party politics would be a better one. Levy mounts a convincing case for strengthening America’s political parties, which are actually very weak, and this weakness is one source of our current political dysfunction.

The Defense of Liberty Can’t Do Without Identity Politics: In the wake the 2016 election, many voices called for the left to abandon the identity politics they felt had come to dominate Clinton’s campaign and subsequently lead to her defeat. Levy assures us that identity politics are instead essential to a functioning democracy. This Vox piece is a useful companion.

The Free Society is an Open Society: In which Levy argues that a war against immigration is at its roots a war against a free society.

That’s all for now!