Let’s suppose there are no natural rights, no right or wrong but what the law says – and the basic purpose of law is to preserve your own life. Let’s further suppose that chaos and constant warfare necessarily mark the absence of government, and that the more government there is, the less private crime there will be. In fact, some of these assumptions are probably wrong, but let’s stipulate them for the moment, for one could hardly come up with premises more likely to yield antilibertarian conclusions. After all, Thomas Hobbes built his entire justification of an absolute, authoritarian state on these assumptions.
Does an unlimited state protect the average person’s life better than a limited state? This is an empirical question that we need to refer to the evidence.Theorize the rate of killing in a society as a function of the authoritarian nature of the government. When the government is overly solicitous of criminal defendants, civil liberties, privacy, etc., the rate of killing is high, let us suppose. (We ignore for the moment the contra-Hobbesian evidence as to how government interference increases violence, for instance through the drug war.) As government interference into citizens’ lives increases, the rate of killing falls. However, at a certain point government becomes so powerful and irresistible that the greatest threat to a citizen’s life comes not from other private citizens, but from the government itself: the function rises. The figure below charts the U-shaped relationship that we might expect from Hobbes’ assumptions. If we are purely concerned with self-preservation (and no other goods or rights), we should choose the government that minimizes this function.
Now, if Hobbes is right, then the minimum to this function will be found in an absolutist state with essentially unlimited power. (The only right the sovereign lacks in Hobbes’ theory is the right to command you to kill yourself.)
The empirical evidence will consist in a comparison of murder rates by private citizens to rates of death by government. In Europe around 1900 murder rates were generally between 1 and 2 per 100,000 and rose slightly by the end of the century.[1][2] In the U.S., rates were about 1 per 100,000 in 1900 and rose to 6 by the end of the century (and spent periods around 10). In recent years, the number of world murders has been around 500,000. If we assume that this annual number of murders was about the average for the entire century (probably a very significant overestimate, since populations were much lower at the start of the century, and the murder rate was lower), then there were about 50 million murders in the 20th century.
By contrast, governments killed 262 million of their own citizens in the 20th century.[3] This total does not include inter-state wars, that is, the number of people in other countries that governments killed. It’s quite clear that in the 20th century governments killed several times as many of their own citizens as private murderers did. Furthermore, the governments that did the most killing were the unlimited, absolutist ones: Mao’s China, Stalin’s USSR, Hitler’s Germany, European colonial governments in Africa and Asia, Pol Pot’s Cambodia, Imperial Japan, etc. The average size of government in the 20th century must have been to the right of the minimum in the curve above.
Therefore, Hobbes was wrong that unlimited government is the best solution to the problem of self-preservation in anarchy. Limited governments have a far better record. On average, governments were too big in the 20th century.
Of course, the notion that liberal democracies are better at preserving citizens’ rights than authoritarian regimes is hardly surprising in this age. We might apply this same methodology, however, to distinguishing among regimes on how well they protect property rights. Assume that every liberal democracy has reached a sufficiently low level of death by killing, and that having our self-preservation assured, we are next most interested in minimizing the expected amount of coercive property transfer. Assume again, in a manner perhaps overly favorable to the case for large government, that up to a certain point increasing government size reduces the overall rate of coercive property transfer, both public and private, but beyond that minimum the amount of coercive property transfer rises due to a growing burden of taxation with little marginal benefit in terms of reducing theft and fraud. Assume further, in a manner again perhaps overly favorable to the case for large government, that the just distribution of property is whatever distribution actually obtains in a society. In other words, taxes that are paid for transfers or public investment will not be considered “coerced property transfer.” Coerced property transfer will only include private theft, vandalism, fraud, etc. and taxation paid to maintain the ongoing operations of government, that is, government consumption. The figure below illustrates the new theory:
Brand and Price (PDF) did a detailed study of the various costs of crime in the United Kingdom, taking into account security expenditure, insurance administration, the value of property stolen or damaged, emotional and physical impact on victims, lost output, health services, and cost of policing. That last category is excluded from my calculations, since it is part of the other side of the ledger. Considering only property crimes (robbery, burglary, theft, fraud, vandalism), the total annual cost came to 30 billion in 1999 pounds. UK GDP in 1999 was £891.1 billion at current prices.[4] Thus, the economic cost of property crime was about 3.4% of GDP. In the U.S. property crime rates are slightly higher (not nearly as different as homicide rates are), and the total cost of property crime might be slightly higher as well.
What is government consumption in both countries? In 1999, government consumption in the UK was 18.4% according to World Development Indicators (World Bank) and 18.3% according to the OECD. In 2006 (the last year for which cross-nationally comparable data are available), government consumption was 15.8% of GDP in the U.S.
Clearly, the burden of government in today’s democratic welfare states is several times greater than the total burden of all property crimes, even including the stress and economic loss associated with “random banditry” as opposed to “stationary banditry.”[5] The average size of the democratic welfare state is well to the right of the minimum of the curve above – we experience far more coerced property transfer than we need to, even should every transfer program undertaken by the government be just.
If we place ourselves behind a “veil of ignorance” in choosing our rules of justice, one plausible rule for a society would be that society should be organized so as to minimize the amount of coerced property transfer we can expect to suffer, subject to an acceptably small risk of bodily assault or enslavement. The evidence shows that, if this rule is indeed the one for judging the optimality of our society, government is far larger than optimal in the U.S. and other advanced welfare states.
PILEUS 
The government is not just a taker. It’s also a provider of services. The notion underlying democratic welfare states is that the state takes from all and provides services to all, for the good of all. This feeds back into your concerns by giving everybody solid incentive not to steal or kill.
This depends on good governance, which you in the states don’t have. You’d be better of focusing on “death by *bad* governance”
(From Australia, we we get free(ish) education and healthcare in a moderately advanced welfare state, all for less government spending as %GDP that you do. )
Yes, government does provide services – but that’s exactly the point of this exercise: to determine whether we get value for money. The state protects our lives and property – either directly by catching and punishing criminals or indirectly (so the argument goes) by alleviating poverty, providing education, etc. But it looks as if the effort to stamp out a relatively small amount of private crime has yielded a very large amount of “government crime” if you will (coerced transfer).
But you can’t estimate value for money by only counting cost. You also have to measure what you get. If all I get is protection against crime, then your measure is accurate. But I also get actual services as well, if only as a by-product. How do you propose to measure the value of them?
Sure, there are other services government provides… But then the government also forces out alternative providers of those services in most cases. So it’s not clear whether we’re getting a net benefit from those services or not.
I suppose another exercise would be to assume that government provision of non-protection services is perfectly efficient and then to subtract from government consumption that portion that is expended on providing generally available services. That would be the most generous possible set of concessions to the pro-big-government case. I suspect the results would still hold up, although probably not as dramatically.
Not that it takes too much away from your point, but I think you’ve mischaracterised Hobbes’ position. His notion of an absolute, authoritarian was much more formal than substantive, not least because the technologies of modern totalitarian states were unknown to him.
His notion of an authoritarian state boils down to a determinant unitary decision rule. This is something that all modern liberal democracies have. Various administrative, legislative and judicial mechanisms work there way to producing a determinate outcome about what you shall be compelled to do, whether generally or in a particular instance. We are nearly all subject to state authority in that sense.
What Hobbes was most concerned about were people going around proclaiming various levels of authority and disagreeing about what the right thing to do was. Basically, people impersonating sovereign power. It is very confusing and leads to violence. In some sense, we accept that today even under limited government. I am sure it must be VERY illegal to impersonate a policeman or a judge.
Hobbes may have missed out a bit on the uses of things like the separation of powers. It often seems like Hobbes preferred a monarch because that was the most limited form of Government of his time: only one pair of hands to meddle, so most ordinary people get left alone. But that amounts to details in Hobbes’ system. It can absorb most notions of limited government liberalism/conservatism and even recognise its benefits.
I would respectfully disagree with that interpretation of Hobbes. I agree that having a “final decider” is his major concern, but check out his chapter on “Things Tending to the Dissolution of a Commonwealth.” Hobbes explicitly opposes freedom of speech, freedom of religion, any conception of property rights that would exclude the sovereign’s will, & any notion of separation of powers or the rule of law (the sovereign subject to the law just like the citizens). Kant’s On the Common Saying seems much more congenial to the kind of political philosophy you’re envisioning – he explicitly rejects Hobbes’ authoritarianism and legal positivism but also endorses a unitary sovereign and a duty of all citizens to obey even an unjust sovereign.
This provides a good framework for limited government types to think about how limited they should be (probably smaller than first considered). I like how the life, liberty, and property concerns are neatly tied together.
But per the Australian’s comments, if the belief is that the purpose of government _is_ coerced property transfer, property rights arguments aren’t going to be convincing (but that doesn’t seem to be the point of this work).
I don’t think anyone thinks “coerced property transfer” is anything but a bad thing, even communists. After all, if you could achieve the welfare state without threatening to throw anybody in jail, who would oppose that? No one enjoys coercion for its own sake, except psychopaths.