The title of the book - Origin of Wealth - is misleading. And that is a good thing. The origin of wealth could easily have been a history of money and wealth (not different from say ‘Ascent of Money’ – a great book but more descriptive than imaginative.) Instead it is precisely what the subtitle says – Evolution, Complexity and the Radical Remaking of Economics.
The book’s introduction talks in great detail about the failure of traditional economics and the challenges that has posed. The first chapter covers general question of how is wealth created and quickly moves on to describe economy as a complex system. The second chapter dwells on the traditional economics and its emphasis on equilibrium systems. So far nothing revolutionary and new happens. The fun starts with the third chapter entitled A Critique – Chaos and Cuban Cars. The similie is quite accurate. The basis of traditional economics seems as outdated as the Cuban cars. The experience of the Santa Fe institute dialogue is also very enlightening. It describes how natural scientists were nearly aghast at the assumption-making and theorizing of economists. The chapter goes through several “laws” of economics and describes how they don’t quite hold. It also goes on to describe why economics might have taken the ‘century long wrong turn’ by tracing the origins of this ill-fitting approach to Walras’ emphasis on using equilibrium models from half-baked theories of then available physics. The chapter ends with the coverage of what the author calls ‘misclassification of the economy’. That is apt. The Walrasian classification of an economy as a stable and equilibrium system is a gross oversimplification and fundamentally incorrect. An economy is a dynamic and non-linear and thus complex. This sets the stage of next section.
Chapter 4 is highly fascinating. It starts to get into the real complexity modeling. Although it covers a relatively simple model, it is quite illuminating. The sugarscape described in the chapter is quite an eye-opener. It talks of how markets, inequality, banking and so on emerge as properties of the system when modeled like a agent-based-system without specifying any of these things. Some steps seem like flights of fancy. However, the general tone is quite serious, believable and most importantly reproducible to anyone who bothers enough to model the sugarscape. I of course feel special affinity towards this approach since it gels well with my thinking about using agent based models and simulations to observe emergent properties rather than abstract those from intuition and black-box-like observation of the system as a whole.
Chapter 5 gets more general and still stays quite interesting. It talks about dynamics. The primary coverage is of static systems vs nonlinear systems. More importantly the idea of oscillating equillibria is discussed. Subsequently the chapter goes into the discussion of using nonlinear systems to explain economic phenomena such as business cycles. He exemplifies with the widget production case – which is itself quite interesting and hits home with the very real scenarios. The chapter also describes John Sterman’s attempt at nonlinear modeling of business cycles across industries.
Chapter 6 focuses on agents. This chapter also describes deductive learnings vs inductive learning and classifies the computer’s methodology as deductive and human ones as inductive. That is an interesting though known distinction. It still is brought home beautifully when the author notes that while Deep Blue can play chess as well as Gary Kasparov, the latter can also tie his shoelaces unlike the former. There are some things or skills which are very easily accessible to inductive learning but are very difficult for deductive thinking. Pattern recognition is a prime example. Human beings can reasonably read decently written hand-writing without much error and difficulty. Computers find it extremely hard to do so and have to go through a laborious process to get there – and still with errors.
Traditional economics assumes that human beings possess infinite deductive capacity and do not need inductive learning since they are already perfect in their decision making. The author then proposes that complexity economics would take the reverse view and try to model individuals as agents with inductive machinery and limited deductive powers – but a decent learning program. The frog example is quite illustrative in this regard. Subsequently the chapter goes through a more detailed agent based modeling of stock markets and describes how the simulation at Santa Fe institute led to indicating a close to real life stock market with the attendant volatility, booms and busts and so on. It boils down not to random noise but competing beliefs in the actors’ minds – different hypotheses about what makes money. The economy by extension can also be modeled using boundedly rational agent with inductive skills and competing hypotheses about how to achieve their goals.
The subsequent chapters cover emergence and evolution and are equally fascinating. I will cover them in another blog. The second part on evolution of physical and social technologies starts to look lot less exhilirating conceptually - so i might cover it briefly later.