In general as a production process gets more complex and requires many specialised and non-substitutable inputs, it is more vulnerable to disruption. This is proposed as a cause of collapse for many sophisticated empires throughout history. The economic fallout from the Iceland volcano fiasco got me wondering: does trade follow the rule that complexity leads to fragility, and if so how can we reduce that?

Short run effects

There is a compensating benefit to trade which increases stability which complicates the situation: unexpectedly low production in one place can be made up by unexpectedly high production in another. To give us some context I’ll consider robustness of food supply, one of the most important production and distribution processes for a society’s stability. For simplicity I’ll imagine three kinds of disruption: local crop failure (a local 30% output shortfall), global crop failure (a randomly distributed 30% output shortfall globally) and a halving of trade (from war, protectionism or natural disaster for example).

Self-sufficiency but no trade: Imagine a world where there is no trade in food between regions. Each region has to aim to produce enough food to feed itself. Each region has to worry about a bad season and regional output can be very variable and if they don’t produce enough food, they will starve. Given this they will aim to produce more food than they need and stockpile lots of grain in order to make a famine very unlikely. They are nonetheless more vulnerable to local shocks than if they could buy food from elsewhere in these emergency scenarios. A global shock has the same impact. However such a situation is not at all vulnerable to trade disruption as nobody relies on trade.

Self-sufficiency and trade: Imagine that in this world trade was suddenly opened up so that all regions could trade with one another. Initially production patterns do not change, so each region is still dedicating the same resources to food production and has the same distribution of expected outputs. In the case of a global supply shock, trade will help a little if a region can get by for a while on a low food supply; those regions with an especially bad crop failure can buy from those regions with a more mild failure. But local crop failure can now be covered with imports from other regions which had bumper crops that year. Vulnerability to trade disruption is no greater now because no region relies on trade except when there is a local crop failure and previously they would have been ruined in that situation anyway. This situation is much more robust than the previous one

No self-sufficiency but trade: Now over time people adapt to this new trade and some regions start producing a lot of food and others start producing less food. In a market system this concentration and specialisation will be based on comparative advantage. Local crop failure is only an issue when trade also fails. Assessing vulnerability to global failure is hard. The total global expected output and stockpiles of food would probably go down a bit over this time. The ‘law of large numbers’ means less excess capacity or stockpiled food is necessary to keep the same risk of famine as existing in scenario 1 because total global food production is less variable than local food production. However, specialization between regions makes food cheaper making stockpiling cheaper and allows us to produce most food in the regions with the least variable output. Compared to self sufficiency and trade we are more vulnerable to trade disruption.

Local crop failure

Global crop failure

Trade disruption

Self-sufficiency but no trade



Not vulnerable

Self-sufficiency and trade

Not vulnerable

Vulnerable (but less?)

Not vulnerable

No self-sufficiency but trade

Not vulnerable

Vulnerable (?)


As we can see, self sufficiency and trade is unambiguously the most robust of the three options, but it is ambiguous which of ‘self-sufficiency but no trade’ and ‘no self-sufficiency but trade’ is the worst. That will depend on how frequently local crop failures, global crop failures and trade disruptions occur and how harmful each is.

What can we do?

If we can have both lots of trade and some back-up systems appropriate for a non-trade world we will be especially robust.

A subsidy for local production to make you less reliant on trade, but open trade when local output falls short is a possibility, though it would be costly and useless against global failures as other countries will proportionally reduce their production. Robin Hanson proposes among other things that previously agreed prices be allowed to rise in emergencies, which gives private parties a reason to maintain excess and robust local production capacity if a supply shock seems likely. The more lucrative opportunities to price gouge during disasters a society commits to having, the more robust it will be.  The subsidy of stockpiles which are only released in a disaster would increase robustness to all kinds of failure, even those where law and order break down. To avoid crowding out other stockpiles or more robust production methods, government stockpiles should commit to sell at high prices rather than give the stockpile away.

In my next post I’ll consider long run effects of trade and complexity on robustness and explain why restrictions on trade or any other policy which reduces productivity growth would be counterproductive.