Tychism, Exchangeability, and Indeterminism in Economics
David Duane Dearmont
Advisor: Chair: DAVID A. BESSLER
Source: DAI, 56, no. 02A, (1994): 0648
An explanation for the difficulty in making
accurate long-range forecasts of economic phenomena is that the economy is in
the process of evolving. Hence, the relationships between economic variables
are constantly changing. This study presents the argument that economic laws
are indeterministic, probabilistic, and do not have a fixed causal structure,
and that economic models must be flexible enough to allow for changing
parameters. Forecasting requires only regularity in the relationships between
Charles Sanders Peirce
argued that chance operates at a fundamental level in Nature and that the laws
of Nature are evolving. Peirce called this doctrine tychism. Tychism does not
rule out scientific investigation of phenomena, but it implies that
"laws" are merely stable relationships between variables. The first
essay of this dissertation demonstrates how tychism can lead to stable relationships
between variables, and how tychism was instrumental in Peirce's founding of the
institutional school of economics.
The demonstration of
tychism leads to an exploration of exchangeability in the second essay.
Exchangeability is a property that results in the ability to aggregate and make
reasonable forecasts when independence of events is violated. In this essay
exchangeability is applied to subjective probability appraisal, the formation
of group consensus, separability in econometric modeling, the treatment of
residuals in time series analysis, calibration, monetary history, and the
formation of scientific disciplines.
The last essay demonstrates
the results of applying tychism and exchangeability to agricultural economics.
In this essay, the history and methodology of agricultural economics is
reviewed through a study of the "farm problem." A counteraction
analysis suggests the existence of the farm problem.
This dissertation attempts
to show that economic forecasting and theory are compatible with indeterminism.
Forecasting performance is enhanced if economists do not insist upon
discovering "true" models with fixed and unchangeable parameters, but
instead recognize that the economy is evolving.
Descriptor: ECONOMICS, AGRICULTURAL