The popularity of a religion is probably determined by various forces, one of which is – as with any activity involving humans – economic incentives. This post examines how certain practices of Islam helped historical Arabia become more prosperous, and how those same practices later limited economic growth in the region.
When Islam arose it provided an important solution to some problems, namely the constant raids of traders from richer oasis cities by poorer nomadic tribes in the desert. It did so by aligning economic incentives in such a way that it was better for the poor not to raid, and for the rich to redistribute some of their income to the poor to prevent raids.
This theory is put forward in a working paper by Michalopoulos, Naghavi and Praolo (2014). Let us start at the beginning. Consider 6th-7th century Arabia. Geography in Arabia is very unequal in terms of agricultural productivity: there are some rich cities around oases, and vast regions of arid deserts. These geographical inequality led to income inequality with cities being much richer.
The rich regions were involved in trade with foreigners, but nomadic tribes in the desert often robbed traders. There was constant feud. This created incentives for some kind of an income redistribution from rich to poor regions to ensure the safe passage of traders through the desert.
The theory argues that a simple redistribution system that may have arisen in the pre-Islamic world will not be effective enough to solve this problem. In such a system the poor tribes have an incentive to raid traders even after receiving income transfers. What is needed is a more complex, “dynamic” redistribution system, and this is what Islam effectively introduced.
Around the 7th century, various forces made the demand for such a redistribution system quite high. Firstly, the raiding Bedouin tribes were no longer policed at all (as some of the more powerful oasis kingdoms that did police them before collapsed in the 6th century). Secondly, the Byzantine-Persian wars made Arabian trade routes especially attractive. As it can be seen in the map below (source), Arabia provided an excellent opportunity for traders to avoid the two warring empires.
Consider first the inadequacy of the pre-Islamic redistribution system. Suppose the rich tribe proposes to redistribute some of its income (including revenues from trade) to the poor tribe in exchange for safe passage through the desert. Let us refer to this redistribution system as zakat.
The authors show such a situation is not stable. In essence, there is nothing to prevent the poor region from accepting the zakat system, getting “handouts”, and then still raiding the caravans. They can thus double-cross the rich region.
For those familiar with elementary game theory, this is exactly like the prisoner’s dilemma. The payoff matrix below shows what would arise in the model if we plug in some numerical values for all the variables/parameters (best responses in red).
We can clearly see that the Nash equilibrium is no zakat/raid, corresponding to defect/defect in the prisoner’s dilemma. Furthermore, we can also see that zakat/no raid would leave both parties strictly better off than this Nash equilibrium (this is like cooperate/cooperate in the PD). But such an outcome is unstable because both parties have an incentive to deviate. Especially, the poor can take the zakat payments and then deviate by raiding the rich, and this would give them a higher payoff.
Now, let us see what Islam does to remedy this situation. In the model, people’s utility is influenced by two things: their own consumption, and the gross income of their children. The gross income of their children can be increased in two ways: by direct bequests, and by investing their income in public goods (called waqfs). The latter one probably requires some clarification.
Waqfs are sort of donations to the community that richer people make in Islam. They usually take the form of hospitals, schools, mosques; i.e. public goods. These goods benefit the whole society, including the rich and poor regions. In the model, the presence of waqfs increases the human capital of the population. Human capital in turn increases wages. So if one invests in waqfs, the wages of one’s children will be higher. This will thus also deliver some utility.
But why would one invest in waqfs as opposed to directly bequeathing savings to the children? Savings have a return on them, but this return can be thought of as interest. Islam to a certain extent prohibits charging interest, or riba. If thus anti-riba laws are strong enough, then returns on savings will be too low. It will then be worth it more to transfer wealth to the next generation via waqfs.
Thus sufficiently strong anti-riba laws and the presence of waqfs creates an interesting situation. Indeed in this case, if zakat is introduced, no raid/zakat will be a stable outcome. The poor will not deviate anymore because the waqfs also benefit them.
Again, for those with some experience in game theory, the illustration below might be of help. The situation with Islam can be modeled as an extensive form game. We can see that the equilibrium is adopting Islam by the rich, and no raids by the poor.
We can see that the poor will never raid if the rich adopt Islam, they will always raid if they don’t adopt Islam. And since the rich are better off in the adopt/no raid scenario than in the not adopt/raid scenario, they will adopt Islam.
This result holds as long as there are sufficiently strong anti-riba laws, high enough returns from trade and a large enough desert region surrounding the oases. (A small desert region can make raids less problematic, because traders can quickly cross the desert and thus can easily avoid raiders.)
The institution of Islam is shown to be stable in the long run as well, if desert regions are large enough. The size of the arid region is actually very important for the results. The figure below summarizes this by denoting the size of this region by lambda.
We can first see that if the arid region is small there will be no raids and Islam will not be implemented. This is because a small arid region reduces returns to raiding, and raiding has fixed costs. So if the region is too small, the poor are better off not raiding at all. Then we enter a region where static redistribution might be introduced but it won’t be sustainable. This is the prisoner’s dilemma matrix discussed above. Beyond this region, Islam will be worth introducing. But for lower levels of lambda, it might not be sustainable in the long run. If the arid region is large enough, then Islam will be introduced and will be sustainable in the long run.
Finally, the authors show that in the long run the Islamic doctrine reduces the income of the rich by limiting wealth accumulation in the form of private capital. While in the short run, it is precisely this doctrine that allows the rich to maintain a high income, and to allow pre-industrial Islamic society to flourish; in the long run, this will cause problems. The lack of private capital will prevent the Islamic world from succeeding in an era where large-scale capital investments (e.g. shipping, industrialization) are important.
For those interested, there is also an empirical companion paper from the same authors that is nicely summarized here. It basically lends credence to the claims of this theory.
This paper also shows how institutions matter for development. And more importantly, how institutions that are beneficial in one situation may be disadvantageous in another. I blogged about another manifestation of this phenomenon: how persistent institutions affected the labor market on the Midwest.