Bringing in the supply side

I’ve never been entirely happy about the mainstream treatment of the supply-side in economics. Too often, this is either ignored in favour of a Keynesian obsession with aggregate demand, or ideologized (“all you need is free markets”), or just assumed to be a basic maths question (“assume a Cobb-Douglas production function“). I’m pleased, then, to see this new paper (pdf) from David Levine:

More advanced technologies demand higher degrees of specialization – and longer chains of production connecting raw inputs to final outputs. Longer production chains are subject to a “weakest link” effect: they are more fragile and more prone to failure…There is a kind of reverse “Keynesian multiplier” that magnifies the effect of real shocks. Consequently, more advanced economies should have higher unemployment rates and be more prone to crisis.

This paper is in the spirit of this one from Charles Jones, which also stresses the importance of complements in the production process. 
Both are a corrective to the standard view, which is that if one supplier fails, a firm can just switch to another. But, says Levine, this mightn’t be possible:

If I am a car manufacturer and my tire supplier fails, I just go to another supplier to get tires. But in general equilibrium, the other supplier was supplying tires to someone – and if I get those tires, some other automobile manufacturer does not. However: if another automobile making chain has a piston maker who has failed, then the tire manufacturer in that chain is unemployed and happy to provide me with tires. Hence when there are failures, some reduction in output must be accepted, but it is best to try to concentrate the failures all in the same chain.

This is, of course, no mere possibility. You can read the 2008-09 recession in these terms. When firms were faced with an inability to get bank finance, they could not substitute to either other banks or to other sources of finances. The result was that many production chains - all of which required bank finance - failed simultaneously.
One counter-argument here - which is often used against real business cycle theory - is that, in ordinary times such simultaneous failures are unlikely, simply because there are millions of production chains and  shocks are unlikely to be sufficiently correlated across these to generate macroeconomically significant fluctuations.
However, this recent paper by Xavier Gabaix questions this. Quite simply, large firms are sufficiently large that fluctuations in their fortunes do translate into macro fluctuations.
Put Gabaix and Levine together and we have a theory of economic fluctuations (I hate the word “cycle”) which is neither purely a demand-side story, nor one of pure equilibrium. Which must be an improvement.
And maybe an important one. A key fact for the UK economy recently has been that inflation has stayed higher than expected despite the recession. Which might be a sign that the recession was, perhaps in part, a supply shock rather than a demand one.

Comments

[Luis Enrique]

I'd like to see some more examples of long production chains failing because of shocks to one link. Where are these automobile manufacturers that are fragile and prone to failure in this fashion? They've got other problems. Meanwhile, I can think of short-chain production functions that are fragile. Subsistence farming, hit by a drought.

The story you tell about banking isn't a long chain story. "reliance on bank finance" != "long production chain". There are very simple farmers that need to borrow money each year to pay for seed.

I'm not sure banking is an example of a link that's hard to substitute for either. I'd have thought it was pretty easy to switch banks. What's an example of an easy-to-substitute input? Say a free house pub that can easily switch between brewers. Well if all the brewers simultaneously go bust, the pub is screwed. That does not tell us brewers are an example of a hard to substitute input.

[Luis Enrique]

I really don't think there's a "standard view" that inputs are always easily substituted. Do you buy the conclusion "more advanced economies should have higher unemployment rates and be more prone to crisis"? I don't. I'd have thought all lengths of chain suffer equally from weakest link problems.

NB if you like looking at the supply side in more detail, you should like The Macroeconomics of Specificity

http://ideas.repec.org/p/nbr/nberwo/5757.html

Cabellero has a book on it.

[wufnik]

Barry Lynn wrote a whole book about the fragility of the industrial supply system back in 2005--End of the Line.

[Alex]

Very interesting Chris, thank you.

[Unity]

Luis:

If you want a very literal example of a 'shock' to a single link impact on a long production chain, you should dig out the industry data on memory chip prices and cross-reference the data against significant geological activity affecting Taiwan.

Chip fabrication plants are notoriously sensitive to earthquake shocks, so even a relatively minor quake can knock out production for several days, if not a couple of weeks, while the equipment is reset and tolerances checked, causing significant volatility in the markets.

Not, perhaps, quite the kind of shocks you had in mind, but shocks nonetheless ;0)

[Alex]

"I'm not sure banking is an example of a link that's hard to substitute for either. I'd have thought it was pretty easy to switch banks. "

That wasn't the problem, though; it wasn't that particular banks stopped lending, it was that banks in general stopped lending.

[Luis Enrique]

Unity,

yes that's an example, thanks. The global electronics industry did not fail when that link did.

Alex,

well, quite.

[mat]

Luis;

The gabix paper shows that shocks to large firms can create macrocrises. The Levine paper shows that failures at firms reduce output as production chains get longer. That's micro roots for thinking off the banking crises as a supply shock; wether or not you think the banks are, themselves, part of a levine-chain.

Easy to test to. Compare output falls in long-production chain industries and the rest. If long production chains create large falls in output, then that should predate any demand side effects, which would take a while to work through, right?

[charlieman]

Luis:

A couple of years ago, a company supplying electrolyte to capacitor manufacturers used the wrong recipe (allegedly, there was an industrial espionage background to this problem). Consequently, a high proportion of capacitors from a variety of manufacturers were prone to premature failure.

The problem extended to those who use capacitors in their products, making this a long chain problem. Once the problem was identified, manufacturers of capacitors could work out which batches were at fault. But it was impossible for them (or intermediary sellers) to identify visually whether a particular capacitor was good or bad (there is not enough space on a capacitor to print a batch code). Nor was there a practical way to test a capacitor in situ in a finished product or on the shelf in a warehouse.

Consequently, manufacturers knowingly supplied goods (capacitors and finished products) that were faulty and would fail prematurely. Some goods may have failed early outside the manufacturer warranty period, creating a cost to the consumer. Others may have failed during the warranty period, creating a cost to the manufacturer.

The global electronics industry did not collapse, but many companies did. Owing to the fragility of the industry, assigning company failure to bad capacitors would be tricky. However, a quick internet search will throw up the names of many companies that suffered reputation damage (sometimes self inflicted) owing to a single supplier problem.

[Leigh Caldwell]

A related argument is that labour becomes more specialised as more specialist (and lucrative) opportunities arise. This is likely to make firms keener to hang onto employees when they have them, but also makes search costs higher for unemployed people.

In this model, unemployment would rise more slowly at the start of a recession and fall more slowly at the end. And large supply shocks would have a more persistent effect on unemployment.

This argument has slightly different implications to the supply chain story described above. At a glance, the specialised labour story seems to better describe the UK's (and maybe northern Europe's) recession while the supply chain version better fits the US.