Friday, December 26, 2008

”No Buffer, No Resiliency”? Maybe. But What Does That Have to Do with the Current Economic Crisis?

On his blog for The Atlantic, James Fallows appears to be supporting “the argument that today's lean, hyper-efficient, ‘just in time’ economy was magnifying the effects of today's economic collapse.” Sounds plausible, but how does it stack up against statements from sages such as Alan Greenspan—okay, bad example—suggesting that lean production practices had the effect of smoothing out economic cycles? Let’s give the blame-lean-production theory a closer look, using the auto industry as an example.

Let’s say that a financial crisis leads to a credit crunch that deprives a cash-starved, lean-production auto company of adequate financing to maintain its production, distribution and sales operations. This process is repeated up and down the supply chain, causing overall auto production to plummet. The idea behind Fallows’ argument I suppose is that if this auto company (and its suppliers as well as its distribution chain) had maintained stocks of parts and components, it could continue producing autos even without financing until… until something happened to reduce the growing piles of unshipped autos on the company yard and docks and unsold autos on the dealer lots? And how does allowing that to happen ameliorate, rather than compound, the difficulties for the auto company? The current financial crisis, which has nothing to do with lean production, precipitated an economic downturn, which has nothing to do with lean production, that has lead to a dramatic fall in auto demand, which has nothing to do with lean production. The financial crisis exacerbates the precarious financial situation of the Detroit Three, which is connected to lean production only to the extent that they may have lagged behind their foreign competitors in adopting the practice.

The potential flaw in lean production lies in its logistics, not economics. Supply chains can indeed be disrupted, sometimes with dire consequences. That is why, for example, states and energy producers hold large, expensive, but immediately accessible reserve stocks. That is why manufacturers maintain multiple sources. Both these needs, of course, are balanced against costs. Auto companies are in less of a spot than providers of products whose supplies must not allowed to be disrupted, though they did encounter serious problems when the Niigata Earthquake damaged the production facilities of the major supplier for a vital component. Even then, the industry as a whole coped remarkably well, with what wound up being a mere blip in production—nobody noticed anything unusual at the retail level.

On a final note, the assertion that “’Just-In-Time,’ is based upon ... a wholly unjustified wager that the economy and its supporting systems will always remain stable and never experience disruption” and that it has anything meaningful to do with the current economic difficulties just doesn’t hold water; the whole thing is a straw man that was set up to justify an unfounded argument. Nobody is making such assumptions save the author of the essay for which Fallows uncharacteristically fell without bothering to match it up against reality.

ADD: I urge those of you who usually don’t read the comments to click to see what Janne has to say. I hope that my response is as interesting.


Janne Morén said...

There is an argument to be had, though, looking at demand disruption in the other direction.

With just-in-time, the moment car sales slow, so do the parts orders for supply companies, and in turn so do the materials purchasing, and the purchase of transient labour - temporary workers getting laid off at a moments notice.

The economic pain gets distributed widely at a moments notice, hits economic indicators and public perception alike, and feeds back into even lower sales through dimming economic confidence.

With buffers, on the other hand, demand disruptions get distributed more slowly. Ups and downs alike get smoothed out, and even with a major crisis like now the lower rate of change at every level will result in less of a drop in confidence or transient indicators even if the end result is the same. You're basically avoiding aliasing effects that can create spurious transients that disrupt the system.

Not sure how well the above argument actually holds, but it doesn't strike me as completely implausible. Of course, a buffered system is generally non-linear and the devil is in the details; some buffer configurations may actually be less stable than non-buffered systems, while others may be very stable with only modest buffer sizes.

Jun Okumura said...

You’re right, Janne, the devil is in the details. I wasn’t quite sure about my argument myself, and had left it on my hard disk with half a mind to rewrite it in an more open-ended manner, to invite comments. Generally speaking, the more highly buffered a system is, the more capable it is to smooth out small drops in final demand but more likely to resulting in a major disruption when a major drop in demand forces a corresponding inventory adjustment unavoidable. Having said that, I don’t see much difference between the two in the face of a global financial meltdown whose economic effects will not run its course until global consumption patterns have readjusted themselves in a more sustainable manner. If demand is down and you can’t finance your running costs, then a buffer system or lack thereof should make little difference to the outcome.

By the way, the Chrome apparently insists on posting my comments as “Jun” while Firefox continues to go with “Jun Okumura”. Just thought you’d like to know.

Janne Morén said...

I think the argument I mostly want to make is that it's not the level, but the rate of change that is important - for business and macroeconomic activity just as for population, social mores and the cleanliness of my workroom.

If the change is slow enough that actors can adapt then the impact will be very small. If the change is too fast, though, it will have large consequences. That goes whether the change is positive or negative, by the way. My workroom has spent the best part of a year slowly degrading into a disorganized dump, but I never quite noticed and I've never suffered for it. Once I clean it up this weekend, however, the impact will be profound as I will no longer be able to find anything.

Similarly, even if the end result is the same, serious buffers will mean that economic actors will get time to adapt over time to a changing economic environment. They can plan earlier for a downturn, seek new lines of business, plan for maintenance during the coming downturn and avoid a few ill-conceived hires that would end up losing their job again six months later. The effects would be dissipated. In fact, I think it's very instructive to think of the economy as an unstable dynamic system, and see regulation as a damping factor. From that point of view, the details of regulation almost doesn't matter at all as long as it has a damping effect on the rate of change of the system. That's for a different post, though.

And I would also argue that the end result in fact would not be the same - a downturn is part real, but part perception, and a slower rate of change would alter the perception of a downturn, and quite possibly shorten it substantially.

Oh, and Chrome is from California; it may just be a more laid-back, informal kind of application and prefers the easy familiarity of "Jun" over the stuffier "Jun Okumura". If it starts addressing you with "whoa, dude" and offer you home-rolled smokes it may be time to back off on using it for a while.

Seriously, try clearing your cookies and saved logins from Chrome and see if it helps.

Jun Okumura said...

I guess my response to that is, speed matters, but so does size. Think of lean production as a system that adapts to change by numerous and rapid but small adjustments, while large inventories dampen minor fluctuations at the cost of increasing vulnerability to major outlier events. The more access there is to relevant real-time information, the more desirable lean production becomes, and vice versa.

I’ll just post from Firefox; it’s easier for someone of my IT proficiency.