Originally published on Aug. 24, 2017
Reviewed by Mark S. Rzepczynski
An investor who can predict when business cycle turning points will occur can build an effective model for dynamic asset allocation. Successfully switching to safety in advance of "bad times" and switching back to higher risk just before "good times" return is the holy grail of such tactical asset allocation.
Unfortunately, the critical element of identifying business cycle turning points has been a comparative backwater in macroeconomics. Since the pioneering research on the business cycle by Arthur Burns and Wesley Mitchell in Measuring Business Cycles, published by the National Bureau of Economic Research (NBER) in 1946, limited work has been done to unify research techniques and focus on the practical challenges of dating and systematically measuring cyclicality.
Identification of business cycles through quantifiable measures has been spotty. An NBER committee dates the beginnings and ends of recessions, but the announcements are made well after the fact and often seem ad hoc. Matching investment allocations with NBER announcements is a loser's game that adds little value.
Don Harding and Adrian Pagan, authors of The Econometric Analysis of Recurrent Events in Macroeconomics and Finance, are leading econometricians who have been working in this area for decades. This new book, which is part of the Econometric and Tinbergen Institutes Lecture series, walks through all their own research, as well as that of other scholars, to provide a synthesis on business cycle dating. The resulting framework for determining business cycles can be applied to any recurrent events in macroeconomics and finance.
The authors' chosen task is threefold:
- Describe cyclical events, such as peaks and troughs, through a set of clear statistics that can either be prescribed or made subject to model-based rules.
- Show how these statistics can be used to identify and measure such recurrent events as business cycles.
- Demonstrate how these cycle-related statistics can be applied to make useful predictions of economic events.
Peaks and troughs in business cycles can be described through either prescribed rules or model-based regime rules. Prescribed rules can be difficult to apply in identifying turning points in business cycles because economic variables do not all turn together, which is the basic problem facing the NBER dating committee. A more extensive, subtler approach is to use models to identify regimes, but Harding and Pagan document that such models have generated mixed results.
The authors present the pros and cons of each approach to help readers understand the complexities of finding peaks and troughs. They illustrate the difficulty of predicting peaks and troughs by discussing the simple problem of defining differences between cycles and oscillations. The noise around any macro series makes ex ante identification of a cycle extremely difficult.
A macro cycle is made up of many economic components that do not always move synchronously. Accordingly, Harding and Pagan discuss how to construct reference cycles through multivariate information. They show how measurement of recurrent events can address cycle amplitude and duration as well as phase shape. Because of the wide diversity in steepness and deepness combinations, what seems like a simple identification problem may have layers of complexity.
The authors describe how a range of models can quantify these aspects with probabilistic likelihoods. Such approaches are based on the chance of a digital (0, 1) event. Regime-based models, often using logistic regression, can give the user a useful probabilistic measure, such as "there is a 45% chance that we are at the peak of the business cycle." For example, a simple analysis based on a regime model that is often used by practitioners estimates the chance of a recession on the basis of the shape of the yield curve; a negatively sloped curve increases the likelihood of a recession.
This short book covers all the methodological issues encountered in dating business cycles in real time. It requires careful reading. The work necessary to understand and use the methodologies described by the authors is not, however, beyond most quant-focused readers. Moreover, the requisite quantitative techniques are available in any econometric toolbox or through some simple programming. Some of the approaches, however, will be unfamiliar to those with only standard statistical training.
Investors looking for a single model or approach that can effectively predict turning points or periods of expansion and contraction in the business cycle will be disappointed. Harding and Pagan provide invaluable insights into the critical issues in this area, but from a practical perspective, the uncertainty in predicting these recurrent events is still high, with many false positive results and frequent tardiness in calling the cyclical peaks and troughs. It remains for other researchers to determine whether the authors' insightful work will ultimately solve asset allocation problems.
Even so, those willing to imbibe the book's concepts will find useful information sufficient to develop a practical research agenda. The proposed framework can also be helpful in finding and measuring cyclical behavior in financial series other than the business cycle.
This book is not specifically intended for the investment professional, so it does not discuss the critical link between asset allocation and the business cycle. Nonetheless, investors' growing focus on time-varying risk premiums associated with macroeconomic factors makes it worthy of careful study.
Disclaimer: Please note that the content of this site should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute.