Re-inventing economics: Part 1

One of my favorite bloggers, Fred Wilson, recently questioned the relevance of economists, especially after their inability to predict, as well as (arguably) correct this current global economic crisis. Fred cites Umair Haque, who griped that “We can't fix today's problems unless we change yesterday's rules. But economists -- and the models they rely on -- are bounded by yesterday's rules.”

Now I’ve got no patience for people who try to ignore the past because they of sensationalism or some naive idea that this time it’s all different. However, we can always improve on the past, so in this post is the first of two very potent ideas I’ve read about recently pertaining to a shifting opinion in modern economics.

Nobel Laureate Joseph Stiglitz (well, Economics isn’t one of the REAL Nobel Prizes, but everyone treats it that way) has put out a paper with a handful of other clever clogs describing a fresh look at economics based on modern networks theory. They developed their paper using data pulled from banks and certain firms in the 2004 Japanese credit market.

The paper basically argues that the old way of looking at markets through “the average, or most probable, behavior of the constituent” does not best describe the true “dynamics of the system,” when that system is made up of autocatalytic processes.  Autocatalytic processes are simply processes that grow on their own (self-perpetuating), and in this case, they become very important when they grow faster than the average, or most probable process. This type of growth can be “scale-free” or “scale-invariant,” which means the bits and pieces that make up the whole all grow at different rates, so after time some processes become more important than others.

So to put it simply, Stiglitz & co basically said that we should use a theory that doesn’t ignore the rare (i.e. not average or most probable) processes, when those processes can grow at astonishing rates, and become very important (e.g. processes that caused this crisis). “The real world is controlled as much by the tails of distributions as by means or averages” (page 2).

And not that this is new: apparently “the relevance of scale free distributions in economics (e.g. of firm size, wealth, income, etc.) is now extensively recognized, and has been the subject of thorough investigation in the econophysics literature” (page 2). I don’t even want to imagine what econophysics is. Regardless, people apparently haven’t given much consideration to how this type of thinking relates to credit markets, UNTIL NOW! The authors purport “…Japanese credit market shows that the credit relations between banks and firms are scale-free, and the standard representative agent plus normal distribution framework is badly equipped for dealing with it” (page 2).

I could be wrong, but I’m pretty sure this is basically chaos theory (which has been around for decades) meets the credit market.

An example the authors give is “the failure of a firm heavily indebted with a bank may produce important consequences on the balance sheet, or the financial status, of the bank itself. If a bank’s supply of credit is depleted, total supply of loans is negatively affected and/or the rate of interest increases, thus transferring the adverse shock to the other firms. Therefore the study of structure of the links and their weights allows to gain some insights in the financial stability of the economic system and to develop new economic policy tools” (page 2). This all looks to be pretty obvious – surely anyone with a passing interest in economics or finance recognizes the interrelatedness of these processes. Joe and Co just look like they’re one of the first guys to decide to map out and measure those links and their weights, this time in a controlled experiment (2004 Japan).

Network theory is an analysis of the interrelationship between nodes and links. In the case of this paper, nodes are banks and firms, while links are debt/credit contracts those banks and firms hold.

I don’t have the patience (or probably the intelligence) to wade through the actual research, but the conclusion offers some interesting opinions on how lessons learned from this data can maybe produce useful tools to stem future crises.

Basically Joe & Co think that “instead of a helicopter drop of liquidity, one can make “targeted” interventions to a given agent or sector of activity.” Presumably, if you use network theory to understand how problems of, say, liquidity arise, then you can surgically fix the problem at its source. Of course, in order to put this into practice, economists would both need to satisfy a tall list of demands:

1. 1.      Have at hand all of the relevant data (this probably means every balance sheet for every relevant institution as well as every debt contract [both of which of course must include homeowners’ personal finances and their mortgages]
   
2. 2.      Intimately understand all of these autocatalytic processes (of which in a modern economy there must be potentially millions)
   
3. 3.      Be able to act quickly enough to solve the problem before it gets out of hand
   

Sounds like the job for a totalitarian, dystopian, and super-intelligent government. But let’s be honest, wouldn’t access to that degree of information be every economist’s dream? Or perhaps this will all be possible, now that we are building petaflop computers that can probably manage the work, and we own all of the banks anyway…