Tuesday, July 7, 2015

Violence and the state

The always worth reading Prof. Gene Callahan posted -- citing Jared Diamond’s example of warfare among the Dani of New Guinea -- that violence is rooted in human nature, not the state. Prof. Callahan observes:
The problem isn't the State: the problem is human beings. And the problem with admitting that problem is you're not left with an easy slogan with which to get funding: "Hate the State" is catchy, but "Hate the human being" isn't going to get you many speaking engagements.
In a subsequent comment, he further cites Steven Pinker on the comparatively low rate of violence of the C20th. In Pinker's words:
If the wars of the twentieth century had killed the same proportion of the population that die in the wars of a typical tribal society, there would have been two billion deaths, not 100 million.
If one is going to cite the state as a net generator of violence, one has the dreadful problem that the evidence is against you. (Which does not stop people telling comforting stories about Somalia.) That agricultural (as in permanent-field) societies tend to be considerably less violent than foraging societies is easily explained by the incentive effects of large stationary asset (one’s land) increasing the risks from violence, and so reducing the willingness to engage in it, while increasing the pressure to evolve effective mediation or other constraints. Though the intermediate state of slash-and-burn horticulture appears to actually drive up the rate of violence, possibly due to increased competition for resources being greater than increased vulnerability of assets. (To put it another way, it does not create strong assets but it does create more production to fight over and more capacity to do so as population density goes up.)

A much bigger problem for the violent-state thesis is that the establishment of the modern state with its effective monopoly of organised violence has clearly led to a fall (pdf) in homicide rates starting in the C16th and C17th in North-Western Europe and spreading south and east in subsequent centuries. And the evidence Pinker adduces from a wide range of scholarly studies in Pp47-56 of his The Better Angels of Our Nature that even the most violent of state societies is far less violent than almost all non-state societies.

Control/violence trade-offs
Not that there really is anything particularly surprising about this. First, economic theory does rather predict that a monopoly generally results in less production of that which is monopolised – the monopolist restricts supply to drive up the return. While pre-modern states rarely achieved a monopoly of violence, they usually aspired to a dominance of organised violence: a dominance they certainly sought to drive up the rate of return.

Second, the state both has the standard producer interest in blocking competition (i.e. being the dominant, and preferably only, provider of organised violence in its territory) and, via the taxing authority, an interest in more taxable activity. Less private violence or risk of private violence leads to more production and more transactions. Rulerships have a basic interest in law and order, that’s why they provide it. As the state’s administrative capacities expand, the more it can act on both these incentives. So, other things being equal, stronger state means less violence. Hence the rise of post-medieval organised-violence-monopolising states has seen falling rates of violence and non-state societies being strikingly more violent than state societies.

The notion of the state as a net generator of violence seems to be based on two things – war, and confusing oppression with violence. Yes, of course states wage war but, as Diamond, Pinker and others point out, war predates the state (or even rulership). Indeed, amelioration of the dangers of war is an incentive to an effective state (as Somalia has recently discovered). Rulers have perennially boasted about their war-fighting prowess; both to intimidate rival rulers and encourage confidence to produce and transact. The larger the territory of the ruler, the further away raiders are likely to be. Borderlands might be less subject to the control of the ruler but they were also more violent.

It is also true that being the dominant provider of violence creates the capacity to oppress. The paradox of rulership is precisely that the ruler is both protector against predators and the most effective predator. Oppression may be based on the threat and capacity to engage in violence but it does not mean there will be more violence. Indeed, given the point is to extract a surplus from one’s subjects, the opposite will tend to be true. There is something of a control/level of violence trade-off here; one that C18th and C19th British opponents of an organised police force generally acknowledged, as do some modern American opponents of gun control.

Part of the appeal of gun control precisely being that society is made up of people of diverse motives, risk assessments and rationality. In the face of such diversity, state management of weaponry both economises on one's own efforts (purchasing weapons[s], learning how to use them, managing their possession and use) and potentially lowers the risks of such diversity. A judgement that depends on one's level of confidence in the state and fears about one's neighbours -- hence Steve Sailor's real estate theory of gun control (which also helps explain the rural-city gap in attitudes to gun control in Australia without the American slavery-and-race baggage).

The failures of drug prohibition illustrate this control-violence trade-off. The state withdraws its protection from particular transactions (sale and use of specified narcotics) and associated property. The result of the state’s withdrawal is increased violence. Of course, the presumptive claim is that the state has the capacity, via its bans, to stop such transactions. This turns out not to be true but its falsity exposes the control-violence trade-off.

Wars of righteousness
Where war and oppression not meaning that the state is a net generator of violencer becomes murkier is when states wage war against some section of their subjects. Most notoriously, the Nazi and Leninist wars of class, ethnic, religious, etc extermination; what political scientist R.J. Rummel calls democide and attempts to quantify.

In a grim sort of way, such wars of righteous extermination are examples of how greed is often preferable to other negative motives, as greed is, indeed, self-limiting. You cannot tax the efforts of the dead. But if your sense of righteousness entails that some group should not exist, then the state, as the most effective predator, is the most effective means for putting your sense of righteousness into exterminatory effect. (To the extent, for example, that the Nazi state actually harmed its war-fighting efforts in its drive to exterminate the Jews and other targeted groups.)

Monotheism started righteous extermination with queers and apostates but the process has since been secularised and expanded. (Righteousness meaning normative claims that trump morality: being enjoined to lead the community into stoning your own brother or sister to death for worshiping another deity counts as trumping morality.) The combination of expanding administrative and technological capacity for states with secularised righteousness has been a grim one, starting with the French Revolution (of course) and its brutal (as in over-the-top use of violence) suppression of the Vendee (the secular version of the Albigensian Crusade: the former being in revolt against the State of Virtue and the second in "revolt" against God) and moving on to the aforementioned Nazi and Leninist wars of social extermination.

Yet, even with all those slaughters added in, the C20th of the most powerful and capable states history has known was still less violent than the forager norm. The state is not a net generator of violence compared to the no-state alternative, though different forms of states have different propensities to violence. Which is to say, managing the paradox of rulership is the central problem of politics precisely because we cannot escape from it.

[An earlier version was posted at Skepticlawyer.]


  1. Of course there is one issue with your argumentation, you find yourself with a BIG endogeneity problem. There's a reason why Taleb is always annoyed by Pinker.
    States do no emerge randomly. They tend to appear in the most productive areas. These places are those where where high productivity allows taxation and thus the rise of the state. But they are also the places where productive activities are the most likely to be more rewarding than predatory activities.
    This implies that peace and the state are likely to emerge at the same place, for the same set of reasons, but not to be (initially) tied to one another.
    Inversely places where productivity is low, predatory behaviours remain relatively rewarding and taxes cannot be raised leading to the lasting presence of a tribal system of government.
    So is a place peaceful because it's rich or because it has a strong government? Is Somalia a tough place to live because it's poor or because it is run by warlords?
    Same thing for historical evolution within state societies. Do they become more peaceful because their state apparatus increases or do they become more peaceful at the same time as (but unrelated to) the growth of their wealth?

    1. Surely, the case of Somalia answers your question: there are lots of places as poor as Somalia, but with a lot less violence.

      And states emerged much later than farming. So, if stateless societies in the regions where states (eventually) emerged were more violent that the subsequent state societies, the question is also answered.

      States also generally do not increase the income of the mass of the population, which tends to remain remarkably static until the 1820s take off. Some states increase the breath and wealth of above-subsistence elites more than others.

      Moreover, there is a sharp decline in homicide rates precisely as the post-medieval state achieves a much higher monopoly of violence but without any equally sharp increase in economic activity (and certainly not in mass living standards). Pushing the elite away from use of violence towards more peaceful mechanisms is key, but again looks much more like a result of a shift in the state rather than an economic shift.

  2. Actually there are not "a lot" of countries as poor as Somalia, according to the UN, in 2013, it was the poorest country in the world. Then if you look at the countries that are with Somalia at the bottom of the ranking you find yourself with a number of places that no one would qualify as safe nor as directed by a strong state (Niger, Congo, Afghanistan, Nepal).

    One way to quick the endogeneity out of your argument would be to compare individual societies not only through time but one with another. For instance, 1500s Castile had a strong well established state, probably stronger than England's or that of the Spanish Netherlands, but people were noticeably poorer. Of course, the murder rate in the Iberian society was a lot higher than by the North Sea. The Brits and the Belgians were busy creating stuff while the Spaniards were killing one another over grave matters such as did your horse exchange an immodest glance with my mare?

    Many things happened in early modern Europe that could explain the historic fall of the murder rate, not least in several countries a noticeable rise of living conditions probably as early as the 1550s. Good luck enticing the elite away from dueling and honour killing if all you have to offer in compensation is poverty and boredom.

    And now if you look with each society, poorer regions tend to be the most violent ones. One of the best cases I know is that of Corsica, in France. They strictly speaking have the same state as the meek Parisians but they are noticeably poorer and killing one another is the national sport of the island.

    1. Violence disrupts social capital, social capital connects to income levels, so do low income levels generate violence, or does violence generate low income levels? And what breaks the cycle?

      As for the wealth of elites, it depends at what level. I doubt that the Spanish nobility was notably poorer than the English or Netherlands equivalents. But the latter societies began to develop a wider and more mercantile elite. But their states directed themselves much more to promoting commerce, while in Spain career paths were all about struggling over get into the state's flow of silver and positions.

      No one would claim that the state is the only factor behind levels of violence--African-Americans share as state with other Americans, but have startlingly higher homicide rates. (Of course, they may not quite experience the same state as other Americans, but even so.)

      The question is not does the state uniformly lower violence, but does the state lower violence more than not having one? And even the income counter-argument is problematic; even though states have very little effect on the income levels of the mass of population (which don't change that much until the 1820s Industrial Revolution take off), they are very important in the level and breadth of elite living standards.

  3. I may have gone too far in my criticism. Let me just concentrate on my initial qualm: the emergence of a state apparatus is not independent from variables that also affect the level of violence in a society. This means that you cannot answer satisfyingly the question you ask yourself: "does the state lower violence?" Specially not with highly speculative statistical evidence on the left hand of the equation and such a vaporous concept as "stateness" as the main explanatory variable. This is a methodological statement, no amount of observation can contradict it.

    The prudent answer to your question seems thus to be: I don't know because I can't know. Any statistical trend seemingly supporting that claim is tainted by the fact that the development of a state apparatus is endogenous. If you were to back up your claim you'd probably have to take a non-state society which was suddenly invaded by a state society and see what was the effect on the local murder rate. You need a natural experiment.

    PS: thanks for the answers

    PPS: your statement on the minimal improvement of living standards pre-1820s absolutely needs to be revised. Compared to the lows of c.1300, the average 1750 Englishman had a real income 5 to 10 times higher. Same thing for the future USA, Netherlands and several wealthy cities (Paris, Antwerp, Bordeaux, Geneva). See Robert C. Allen data on the matter. The regretted John Munro had convincingly showed that the upward trend can probably be dated from as far back as 1450-1500. True these improvements were later dwarfed by those observed over the past 200 years but they remain impressive.

    1. Yes, NW Europe seems to be the exception in living standards. How much, and when, the cliometricians are still arguing over. But you get post-medieval declines in violence in regions of Europe where overall living standards are stagnant or declining but states are still increasingly effective at imposing a monopoly of violence. Which is powerfully suggestive evidence for a state monopoly of violence reducing overall levels of violence.

      And the development of a state apparatus is not endogenous in the sense you seem to mean. That is, farming does not create a surplus which leads naturally to a state. States have to be built, and there appears to have been a lot of experimentation and false starts. The only reliable way to create a sustained social surplus is by expropriation, but the apparatus of expropriation requires a supporting surplus: a chicken-and-egg problem. Hence the experimentation and false starts.

      Sustained conflict, particularly across ecological frontiers (farmers v pastoralists) seems to produce the combination of increasing coercive specialisation and multi-generational authority required to produce a state.

      Which makes the emergence of states look at least somewhat independent of the variables that affect violence. Indeed, it is hard to see how you could sustain a state unless you did, at least somewhat, reduce the level of violence.

    2. The Yangste river valley and Tokugawa Japan may also be exceptions.