Tuesday, August 24, 2010

Complex Adaptive Systems

An ecosystem is in a complex equilibrium with diurnal and seasonal patterns, and also random trends in geology in species distributions. One thing most people understand is that trying to make an ecosystem better is really, really hard, because everything is connected in complex ways, and not obvious what 'better' means.

For example, before the middle of the 20th century most forest managers believed that fires should be suppressed at all times. By 1935, the U.S. Forest Service's fire management policy stipulated that all wildfires were to be suppressed by 10 A.M. the morning after they were first spotted. Early posters of Smokey Bear misled the public into believing that western wildfires were predominantly human-caused. In Yellowstone park human-caused fires average about 8 annually, while about 35 wildfires are ignited by lightning. Fire is a natural process, and helps clean out the understory and dead plant matter, allowing economically important tree species to grow with less competition for nutrients. Scientists knew a little about that, so starting in 1972 the National Park Service began allowing natural fires in Yellowstone to burn under controlled conditions, yet of a total of 235 prescribed natural fires burned under the directives of the new policy, only 15 spread to more than 100 acres. With all that kindling, in 1988, a total of 793,880 acres (3,213 km2), or 36 percent of Yellowstone park burned, an unprecedented disaster.

It should come as no surprise that trying to fix the economy--also a complex, endogenous, adaptive system--is also difficult. The cash for clunkers program that gave a $3,500 voucher to those who traded in old cars for more efficient ones (and then destroyed the old cars) seemed likea great deal, but it boosted used car prices by as much as 30% , basically giving a one-time boost to car sellers, as the subsidy mainly went into a rise in auto prices. By literally destroying over 500k cars, the price rationing system mercilessly punishes the poor slobs who did not have a car before the cash-for-clunkers started.

Alas, an esteemed economist was behind this boondoggle. Alan Blinder helped popularize the idea of a scrappage program, and the moniker "cash for clunkers", with his July 2008 op-ed piece in the New York Times. Blinder argued that a cash for clunkers program would have a tripartite purpose of helping the environment, stimulating the economy, and reducing economic inequality. Solving for these three objectives simultaneously is harder than the three-body problem of Newtonian mechanics, an impossibly difficult task. Simple carbon taxes, tax cuts, even straight out checks to the poor, would be more effective, but then the results would be too obvious. Apparently the impossibility of the objective isn't a bug in these programs, but rather, a feature, because if you can't say something worked, you can't say it didn't work.

Credit card interest rates have risen dramatically, as new federal rules intened to help credit consumers have backfired. The changes required under the Credit Card Accountability, Responsibility and Disclosure Act prohibits lenders from raising rates on outstanding card balances. Card issuers also won't be able to change the terms of a contract so long as the cardholder makes a minimum payment on time. In response, banks raised rates across the board for everyone. The average rate is now 11.45 percentage points higher than the prime rate, the benchmark against which card rates are set. That's the largest difference in at least 22 years. What did regulators think would happen?

Lastly, there's the US Treasury trying to figure out how it can still help new homeowners, but without implicitly paying as much as Fannie and Freddie appeared to have cost US taxpayers. Bill Gross writes that "nine in 10 new loans are currently backed by Fannie, Freddie or government agencies", and further that if the government got out of the business, mortgage rates would "rise 300 basis points to 400 basis points, crippling any hopes of a housing-led revival to the economy.” The Treasury will have a new plan announced in January, but they have a tiger by the tale. One thing the government does not do well is price for risk, and as Joe Stiglitz knows, this leads to pooling equilibria, where good and bad risks are priced the same. In such events, the good credits leave because they don't like subsidizing their free-riding brethren, so its a race to the bottom that eventually blows us (i.e., it's not an equilibrium). So, either we let the government-guarantees gestate to a much bigger problem at some later date, or we let housing prices fall dramatically soon.

As long as the Chinese keep giving us money and inflation remains low, I see nothing forcing US policy to change. This is not sustainable, however, and when the US turns into a Greece-like situation, it will be a very bad decade. When you try to micromanage a complex system, the most important virtue is humility.


Ashwin said...

Eric - Not that this changes your message on the need for humility, but it's far from clear that the Yellowstone fires were a disaster. The ecosystem started regenerating almost instantly after the fire which hints that it is essentially adapted to such "catastrophic" fires. Many ecologists believe that given how old the lodgepole pines in Yellowstone were compared to their usual burn-by date (250 yrs compared to usual lifespan of 100 years) such a fire was inevitable and in fact exacerbated by the combustible load built up by the half century of fire suppression.

Consistent with a well-known thesis in ecology that stabilisation of a complex adaptive system reduces its resilience in the long run which has relevance to economic systems as I note here http://www.macroresilience.com/2009/12/06/minskys-financial-instability-hypothesis-and-hollings-conception-of-resilience-and-stability/

nick gogerty said...

On complex systems. Eric correct the component variables in a complex are such that amplification or attenuation of the desired system change state over the long term is impossible to predict.

All one can hope for is a steady and stable state, an attactor if you will. extreme amplification would approximate hyperinflation and extreme attenuation of a critical variable would approximate economic death.

Tweaking the components in such a system is expensive and/ or dangerous. Blinders metaphor is so simplistic it makes the Laffer curve look rigorous and falsifiable (good economic science).

Anonymous said...

I visited Yellowstone during the 1988 fire, and then again last year. One impact, from a tourist's point of view, is that moose haven't returned to the park, because (a ranger told me) their preferred habitat hasn't grown back yet. In 1988 we saw moose all over Yellowstone, but last year saw none. However there were a fair number of moose scattered around Teton.