Economic Policy Uncertainty: A Closer Look

Zachary Bartsch

Advisor: Carlos D Ramirez, PhD, Department of Economics

Committee Members: Tyler Cowen, Garett Jones, Scott Sumner

Carow Hall, Conference Room
June 19, 2018, 01:30 PM to 03:30 PM


The dissertation consists of three chapters: 

Responsive Agents: Economic Policy Uncertainty and Dollar-Pound Exchange Rate Return Volatility

The extent to which economic policy uncertainty (EPU) amplifies exchange rate volatility has been an important research question for at least a decade. Previous research has investigated this relationship using monthly data, concluding that EPU imparts an effect on exchange rate volatility either contemporaneously, or with a one month lag. The use of monthly frequency, however, may not provide an accurate causal interpretation, and may even compromise the accuracy of the estimates if the natural cycles of EPU are shorter than a month. To address this econometric concern I construct EPU measures at a daily frequency, and estimate a GARCH model using daily USD/British pound returns. The evidence indicates that EPU contributes to exchange rate volatility much more quickly than monthly data can detect. I also find that general market uncertainty increases volatility more than EPU does.


Separating Uncertainties 

I separate uncertainty into its underlying causes of market uncertainty and economic policy uncertainty. I identify and summarize the cause of major uncertainty shocks from the past 50 years. Using all of the major monthly broad uncertainty measures introduced in the past decade, I find wide variation in the series behavior. I find that uncertainty events that stem from market uncertainty alone are much more frequent than those that have their cause in economic policy uncertainty alone. I argue that non-policy uncertainty has larger negative effects on industrial production than does economic policy uncertainty - making “correct” policy more important than decisive policy. Surprisingly, I find that EPU briefly increases employment regardless of the broad uncertainty measure. Finally, I reveal that economic policy uncertainty shocks tend to persist only briefly and that there is mild evidence that they are associated with policies that decrease market uncertainty.


Economic Policy Uncertainty: The Problem of Endogeneity

I contrast the composition of modern EPU data, from 1985-2014, to historical data, from 1884-1913, in order to argue that the primary difference between the two series is the presence of expectations for economic intervention. The historical period had practically no precedence nor explicit expectation of recessionary intervention, while the modern period had both. The result is a modern EPU measure that is inflated prior to economic declines. The inflated measure underestimates the effect of exogenous economic political uncertainty on industrial production. Using a vector autoregression and Cholesky decomposition, I create orthogonalized impulse response functions that show the causative effect of EPU on industrial production to be about 3.75 times larger than previous estimates. This finding calls into question the reliability of many recent results that depend on EPU or broad uncertainty as explanatory variables.