I’m almost halfway through a wonderful econ class this quarter, the best I’ve had at UW. In contrast to all my past econ classes, it is small and discussion/writing oriented. Here are some highlights.
Modern growth theory starts from the work of Robert Solow in the mid 1950s, who in turn reacted to the Harrod and Domar model of the previous decade. Solow’s innovation was to use a different form for the production function, an important part of the model, yielding very different results. Harrod and Domar are nowadays only of historical interest, whereas Solow’s work underlies contemporary theory.
The most tricked out version of the Solow model we have covered accepts as exogenous parameters population growth (actually, labor force growth, though the growth rate is the same if the labor force participation rate is constant), the rate of depreciation of capital (how quickly machines wear out), the savings rate (proportion of income saved, which equals investment in capital by firms according to the basic macro identity), human capital (usually, average years of schooling), and the growth rate of technology (which has the effect of allowing the same number of workers produce more output), and alpha (determines the prominence of capital in the production function; sort of acts as another measure of technology). Based on these parameters, a steady-state equilibrium growth rate of average income is predicted. The bottom line in this model is that a country with a high savings rate, copious schooling, low population growth, and a high level of technology will in the long-run have high average income (GDP per worker).
What surprised me is how well the model fits the data. There is a scatter plot in our text of the predicted steady-state GDP per worker relative to that of the U.S. for each of 109 countries in 1997 versus the actual value for relative GDP per worker. Ideally, the data would fall along a 45 degree line, indicating that the model captures the trend in the actual values. In fact, it comes pretty close.
Another striking plot in the book (this one; note its brevity — it’s pretty concise and seems like it would be an excellent introduction for anyone who is interested) is one that compares growth in GDP per worker 1960-97 to deviation from predicted steady-state GDP per worker in 1960. There is a clear relationship, where countries far above their steady-state in 1960 grew slowly or shrank (e.g. Zambia, Kenya, Ghana) and countries far below grew rapidly (e.g. South Korea, Japan, Taiwan). This plot illustrates the concept of convergence. Not only does the Solow model tell us what the steady-state GDP growth per worker will be, it also tells us the path GDP growth per worker follows over time as it moves from its current value toward the steady-state equilibrium.
Technology turns out to be the key variable. It is exogenous in the Solow model. That is, technology is not determined within the model; we have to stick it in as a parameter. Most recently, we have been discussing technology in preparation for the next big step, which is attempting to model it. What makes technology hard to model is that it is made up of ideas, which have different properties than other economic goods. Ideas are the only goods that are truly nonrival — an infinite number of people could all use an idea at the same time, and the enjoyment of it by any one person in no way detracts from the enjoyment of it by another. Ideas also tend to be fairly nonexcludable, meaning that it is difficult to control access to an idea (which means it’s difficult to charge money for using it). Next week, we will be looking at how to model technology growth, so perhaps I will report on that at some point.