Sunday, October 15, 2006

Accumulate, accumulate! That is Moses and the prophets!

The data in Tables 1 and 2 are from Kenickell (2003), as quoted in Cagetti and De Nardi (2005). Table 2 is in 2001 Dollars. Raw data are from the Survey of Consumer Finances (SCF), I guess. Unlike the Panel Study of Income Dynamics (PSID), the SCF does not allow researchers to follow households over time, but is better in tracking the richest households. Cagetti and De Nardi survey a variety of publications by researchers examining trends in wealth ownership.

Table 1: Distribution of Wealth By Year

Table 2: Absolute Distribution of Wealth By Year
< $07.3%7.2%7.1%8.0%6.9%
> $1,000,0004.

The data show ownership of wealth is extremely concentrated in the United States. The top one percent own approximately a third, while the top five percent own more than half of total United States wealth.

Cagetti and De Nardi examine three types of models commonly used in mainstream economics:
  • Models with agents with an infinite lifespan
  • Overlapping generations models
  • Models that combine features from both the above types of models
Agents in the model are heterogeneous because different agents receive different shocks to, say, wages. Cagetti and De Nardi show that none of these model types is able to generate inequality as extreme as is seen in the data. (Cagetti and De Nardi do not put their conclusion so stark.)

Cagetti and De Nardi suggest extensions to these mainstream models to solve this empirical puzzle. But I do not know why one would not consider other types of models. I can think of other places to look.

I don't fully understand the literature Cagetti and De Nardi survey, but I read them as confining themselves to models with analytical solutions. I think agent-based simulations are an interesting approach. One uses simulation, when analytical solutions are not available, to understand long term dynamics in such models. These models are incompatible with methodological individualism inasmuch as model parameters include macroeconomic distributions over, say, tastes. Offhand, I have only one reference (Wright 2004) for such models, but I understand diverse researchers, for example, Alan Kirman have been exploring such models.

I consider the Kahn-Kaldor-Pasinetti-Robinson theory of income distribution to be a classic Post Keynesian approach. This theory has received increasing criticism within the Post Keynesian community. Nevertheless, one might consider trying to explore how analytical updates to this model fit the empirical data. Moore (1975) is a somewhat old exploration of the Kaldor-variant of this theory. Surely more could be done here.

One might also consider models with complex dynamics. Goodwin (1967 and 1990) provides some interesting models in this vein. Perhaps some work by Dumenil and Levy might be interesting here, too.

I'm not sure how to delimit the range of the outcomes of these models. I guess one would like to limit the parameters to empirically reasonable ranges. This, like Cagetti and De Nardi's, is not necessarily a simple research program.

By the way, some of these references can be downloaded from somewhere or other.

  • Cagetti, Marco and Mariacristina De Nardi (2005). "Wealth Inequality: Data and Models", Working Paper 2005-10, Federal Reserce Bank of Chicago
  • Goodwin, Richard M. (1967). "A Growth Cycle", in Socialism, Capitalism & Economic Growth: Essays Presented to Maurice Dobb (edited by C. H. Feinstein), Cambridge: Cambridge University Press
  • Goodwin, Richard M. (1990). Chaotic Economic Dynamics, Oxford: Clarendon Press
  • Kennickell, Arthur B. (2003) "A Rolling Tide: Changes in the Distribution of Wealth in the U.S., 1989-2001", Mimeo (Sep.)
  • Moore, Basil J. (1975). "Equities, Capital Gains, and the Role of Finance in Accumulation", American Economic Review, V. 657, N. 5 (Dec.): 872-886
  • Wright, Ian (2004). "The Social Architecture of Capitalism", arXiv:cond-mat/0401053v1 (Jan.)

No comments: