Sunday, March 29, 2009

Model builders versus model users redux

This article on the failure of the Gaussian copula was a reminder of a previous post on this topic. The bottom line from the post was that when model builders and model users/decision makers are divorced from each other and that model users do not fully understand the model that they are using (or ignore the assumptions of the model because they have little incentive to question these assumptions) then this is the real recipe for disaster rather than the model itself. Moreover, after seeing a number repeatedly, this number takes on an element of being THE TRUTH and become lulled into complacency that this number is all that matters. The real human element that resulted in the financial crisis was neither ignorance nor stupidity but greed.

Some excerpts (emphasis mine):
In 1998, before Li had even invented his copula function, Paul Wilmott wrote that "the correlations between financial quantities are notoriously unstable." Wilmott, a quantitative-finance consultant and lecturer, argued that no theory should be built on such unpredictable parameters. And he wasn't alone. During the boom years, everybody could reel off reasons why the Gaussian copula function wasn't perfect. Li's approach made no allowance for unpredictability: It assumed that correlation was a constant rather than something mercurial. Investment banks would regularly phone Stanford's Duffie and ask him to come in and talk to them about exactly what Li's copula was. Every time, he would warn them that it was not suitable for use in risk management or valuation.

In hindsight, ignoring those warnings looks foolhardy. But at the time, it was easy. Banks dismissed them, partly because the managers empowered to apply the brakes didn't understand the arguments between various arms of the quant universe. Besides, they were making too much money to stop.

... Bankers should have noted that very small changes in their underlying assumptions could result in very large changes in the correlation number. They also should have noticed that the results they were seeing were much less volatile than they should have been—which implied that the risk was being moved elsewhere. Where had the risk gone?

They didn't know, or didn't ask. One reason was that the outputs came from "black box" computer models and were hard to subject to a commonsense smell test. Another was that the quants, who should have been more aware of the copula's weaknesses, weren't the ones making the big asset-allocation decisions. Their managers, who made the actual calls, lacked the math skills to understand what the models were doing or how they worked. They could, however, understand something as simple as a single correlation number. That was the problem.

... In the world of finance, too many quants see only the numbers before them and forget about the concrete reality the figures are supposed to represent. They think they can model just a few years' worth of data and come up with probabilities for things that may happen only once every 10,000 years. Then people invest on the basis of those probabilities, without stopping to wonder whether the numbers make any sense at all.

As Li himself said of his own model: "The most dangerous part is when people believe everything coming out of it."

End of innocence?

K1's grade and the 5th graders put on Update: Earth by Roger Emerson this Spring. It was a great show and brought some tears - not so much for the actual performance itself - more because it symbolized the beginning of the end of their innocence. This is an age where kids believe that they can do anything - "we're gonna change the world each day". Soon but hopefully not too soon, their views will become more cynical and jaded and perhaps in no small ways, we as parents are responsible for this.

Wednesday, March 25, 2009

What I've been reading

1. American Born Chinese by Gene Luen Yang: Am still not quite sure what to take away from the book - interesting and entertaining nevertheless.

2. Richistan by Robert Frank: The first chapter on certified household managers was certainly the most eye-opening and enjoyable. The book was an easy and entertaining read.

3. Chess Machine by Robert Lohr, a ficitionalized account of the Turk using the historical characters of the time. A bit of a slow start for me but once it got going it was entertaining.

Netbook versus laptop

The current crop of netbooks (the genesis of this is in the link) has driven down the price of laptops to the $400-500 range while the price of netbooks still range from $300-400. This difference has to come down before netbooks become a deal for me. I realize that this is contrary to what I had posted earlier on Kindle versus Sony Reader here and here but is a reminder how quickly things can change in the tech world. At the current rate, the $299 for Kindle is looking expensive compared to what a netbook can do. Perhaps Kindle can try to become a netbook at $299?

Friday, March 13, 2009

Did an increase in aggregate risk cause the financial crisis?

And where does aggregate risk come from? The basis for many real business cycle models is an aggregate shock to the technology parameter of the aggregate production function. This shock has to have large persistence in order to match the moments of various economic parameters. Unresolved is the source of aggregate shock. Various multi-sector models have not been able to generate the persistence required and have been abandoned.

One possible source of technology shock is an increase in risk so that the shock parameter is interpreted as an increase in uncertainty as measured by its standard deviation. Existing literature shows that large persistence needs to be assumed for this to work. This persistence using multi-sector models is revisited not as different sectors of the real economy (as was in early multi-sector models) but as different sectors of an economy that chooses to adopt a new innovation e.g. early versus late adopters on a continuous time scale.

Two innovations are considered: just-in-time inventory and financial innovations. Advances in computer technology and communications have made the adoption of just-in-time inventory easier. Early adopters benefit when it works. These early adopters take on the most risks and hence bear the fruits of early adoption. They are also the guinea pigs in the sense that these early adopters face idiosyncratic shocks to their production functions as they iron out the early difficulties. In the early stages, these random shocks only affect the early adopters and not the entire economy.

The gains by early adopters provide an incentive for more firms to shift into the new technology. As more firms adopt just-in-time inventory, they become more interconnected. Consider an idiosyncratic shock that affects for instance, the software provider of a small segment of the economy that is interlinked by just-in-time inventory management. While this shock is idiosyncratic in a sense that it only affects the firms that use the software provider, the interconnectedness of the economy is now such that a ripple effect is created as inventories in one sector affects that of another even though they do not share the same software provider. In this sense, an idiosyncratic shock has resulted in a systemic problem for the economy. A small shock can thus have a large propagation mechanism.

Consider the role of financial innovations in the current economic crisis. Assume that securitization has made some securities safer. Early adopters that buy and sell these securities reap large gains providing the incentives for firms to enter the market. Now throw in the additional wrinkle that the fact that these securities are deemed safe (even though they may not be) provide incentives for all firms (early and late adopters) to take on additional risks by buying more of these securities. As more and more firms adopt the financial innovation an idiosyncratic risk to one firm now becomes a correlated or systemic risk as they become more interconnected. Thus aggregate risks increase as more firms adopt an innovation.

Why haven't risks from adopting JIT translated into an economic crisis. An Economist survey points to several examples where they almost did yet the scale of these problems tend to remain localized in time and space and did not spread, i.e. the shocks were not persistent. What is the difference? Both innovations allow firms to take on more risks. JIT adopters run lean inventories and this increases the risk of stock outs. In a systemic crisis there would be economy wide shortages.

One may be tempted to say that firms can hoard inventory as soon as a shortage is likely. The best analogy to hoarding in the financial sector is short-selling and thus the ability to hoard is likely not the major difference. As most commentators have pointed out, the most likely culprit is leverage or excessive leverage. Consider a firm that funds its day to day operations from sales and pays for its inventories out of sales. A prudent firm might continue operating in this fashion or it might borrow a little in order to take advantage of the next cycle of inventories that it thinks might make it first to market for a new product. As long as the firm does not borrow too much, an idiosyncratic shock is likely to remain just that - at worst it may translate into a systemic shock but with little persistence. As long as only small numbers of firms take this risk and on different goods the problem of failure will remain local. But when all firms borrow heavily and take greater risks then this can lead to a systemic problem.

The insight from this is not too great - but it provides the link between innovation and how it mass adoption can make the economy more vulnerable. Some experts in the Economist survey claim that it is only a matter of time before JIT leads to large problems but perhaps there is a chance that it will not. What then the role of leverage and adoption of new financial products in the current crisis? Should the Fed regulate adoption or leverage or both? One can argue that the innovation was NOT really the product that firms were buying into but the ability to increase leverage. Should financial innovations that increase leverage be more regulated than others? And what of JIT? Should the government increase its oversight into this as well?

Monday, March 9, 2009

What is the truth of the Asian Miracle?

Bill Easterly's post criticized the UN's recomendation of industrialization via state intervention. His subsequent post challenged the "myth" that the East Asian economies grew via state intervention. His claims are that these economies grew because they adoped pro-market policies. This was interesting because Ha Joon Chang in Bad Samaritans claims the exact opposite. He claims that it is a "myth" that free-market forces created the economic growth experienced by the Asian Tigers.

This debate will go on endlessly because of the inability of economic models to reject anything. Economists will be free to interpret the evidence according to their biases even though they may claim not to be a free market idealogue. In the absence of evidence or rather in the lack of evidence that can proved one sided or the other conclusively, one would think that economists would remain agnostic rather than to ascribe to either faiths of free markets or state interventions.

What is innovation?

While The Economist assesses financial innovation, it may be useful to ask what exactly is innovation?
1. Tang
2. Hot pockets
3. Aeron chair

These can all be considered innovations but how much do they represent progress? Likewise, can all financial innovation be equated with progress? The traditional economics argument would be to let the markets decide. Assuming rationality useful innovations would die or survive based on the public's desire for them. But what if markets are given to bouts of irrationality as evidenced by a bubble in Beanie Babies?

Another feature that some economists would consider being intrinsic to innovation is spillovers (both positive and negative). None of the above cited innovations appear to have spillover effects. Yet the economic concept of positive spillovers is nebulous. It is often asserted rather than proven. (Negative spillovers on the other hand can often be shown, e.g. pollution.) In the macroeconomics literature a spillover is an aggregate technology parameter that leads to increasing returns to scale. In other economic disciplines, a spillover effect is a change in one industry or sector that leads to progress in another sector. An often cited example is the positive spillovers from education - for example, that it leads to an educated populace that can vote in a more informed manner (as opposed to uninformed voting).

Are financial "innovations" really progress? Some would argue that by allowing increased leverage, financial innovations should be considered progress since it expands the technological frontier and thus is evidence of spillover effects. Is the market the best method of deciding whether innovations survive or die when there is irrationality in the market? As can be seen by the financial crisis, the bubble in CDS, MBS, and the economy had more of a negative spillover than a bubble in the Beanie Baby market.

Finally, an analogy albeit imperfect. Innovations in cars have made driving safer. This increases moral hazard - I can drive faster to get to where I want to go quicker. This is a positive externality since I can get more done in a day. But speeding increases the overall risks in the economy. Thus, speed is regulated. Likewise, financial innovations lead to moral hazard in that it allows me to take more risk without considering the rise in overall risks for everyone else.

Will nationalization work?

The original argument for nationalization as the financial crisis was unfolding was that it would be seen as a bold preemptive move that would restore confidence in the financial sector. As arguments against (see Alan Blinder) and for nationalization (see Stiglitz) continue, the question that hangs over the economy is the following: Will it restore confidence and get the economy going again?

The assertion (without evidence) here is No - that if nationalization had taken place back in October then perhaps there was a chance. While public debate is a feature of a democratic society, the debate over nationalization has undermined what could have been a swift, bold, forceful action into one that can be viewed cautious, uncertain and perhaps even incompetent.

In this sense, the public, politicians and economists are like dogs and wolves while the regulators and those who undertake nationalization are ordinary people. Once the dogs and wolves smell the fear and uncertainty in the regulators, nationalization won't stand a chance.

As argued by Stiglitz however, it does not mean that nationalization is not a necessary feature for fixing the financial sector. Many would argue that based on Japan's experiences, fixing the financial sector is a necessary first step. Moreover, as argued by Richard Clarida the financial sector provides the multiplier for any kind of fiscal stimulus.

The time for nationalization has passed. While nationalization may be required for fixing the financial sector, the process will be long and arduous. The experiences of Fannie and Freddie point to how difficult this could be even though the takeover of these two institutions would at the time be considered bold and forceful. However, these institutions have a legacy of being an implicit part of the U.S. Treasury.