Christopher Bliss's 1997-98 Annual Research Report

Christopher Bliss (Professorial Fellow) has continued to work on economic growth, convergence and non-convergence. He has learnt much more about the limiting (or ergodic) distribution that results from common wealth distribution behaviour combined with random shocks. Some results in this vein are given in ‘The Ergodic Distribution of Wealth with Random Shocks’, Nuffield College Economics Discussion Paper, 145, July 1998. These include some surprising conclusions. For example, Danny Quah noted that the distribution of national per capita wealth levels exhibits a ‘twin peaks’ feature, where the distribution shows a mass of poor observations and a mass of rich, with far fewer in between. The paper shows that this outcome is consistent with the standard convergence model, given suitable non-linearities in optimal accumulation behaviour. Thinking on these lines informs the assessment of the empirical evidence, and a paper on the so-called Galton’s Fallacy issue will appear early next year.

Work continues on corruption, some again with Rafael di Tella. Further consideration of the JPE paper model has concentrated Bliss’s interest in the type of decision problem examined in that paper. This theory is being developed further, particularly where two agents are involved. The basic problem, which has a particularly simple and appealing structure, is quite realistic for many real life situations. The decision-maker chooses a target for the decision variable (an example could be courtship, and the variable the ‘quality’ of the partner). The higher the target the larger the utility reward for success; but the lower the probability that the target will be attained. Such problems are standardly non-convex, because cumulative probability distributions are non-convex, and this leads to some fascinating results and possiblities.