COWLES FOUNDATION FOR RESEARCH IN
ECONOMICS Box 208281
COWLES FOUNDATION DISCUSSION PAPER NO. 1771 The Mysteries of Trend Peter C. B. Phillips September 2010 Trends are ubiquitous in economic discourse, play a role in much
economic theory, and have been intensively studied in econometrics over the last three
decades. Yet the empirical economist, forecaster, and policy maker have little guidance
from theory about the source and nature of trend behavior, even less guidance about
practical formulations, and are heavily reliant on a limited class of stochastic trend,
deterministic drift, and structural break models to use in applications. A vast
econometric literature has emerged but the nature of trend remains elusive. In spite of
being the dominant characteristic in much economic data, having a role in policy
assessment that is often vital, and attracting intense academic and popular interest that
extends well beyond the subject of economics, trends are little understood. This essay
discusses some implications of these limitations, mentions some research opportunities,
and briefly illustrates the extent of the difficulties in learning about trend phenomena
even when the time series are far longer than those that are available in economics. |
