Identifying Uncertainty Shocks Using the Price of Gold

TLDR: This article proposed an instrument to identify uncertainty shocks in a proxy structural vector autoregressive model (SVAR), which equals the variations in the price of gold around events associated with unexpected changes in uncertainty.

Abstract: We propose an instrument to identify uncertainty shocks in a proxy structural vector autoregressive model (SVAR). The instrument equals the variations in the price of gold around events associated with unexpected changes in uncertainty. These variations correlate with uncertainty shocks because gold is perceived as a safe haven asset. To control for news‐related effects associated with the events we identify uncertainty and news shocks jointly, developing a set‐identified proxy SVAR. We find that the popular recursive approach underestimates the effects of uncertainty shocks and delivers responses for economic activity and monetary policy that have more in common with news shocks than with uncertainty shocks.

Figures

Figure D.10: Impulse responses to an uncertainty shock with uncertainty measure taken from Bachmann et al. (2013)

Figure D.6: Data used in the baseline specification of the VAR model

Table 2: Relationship between our proxy of uncertainty shocks and the instruments used in Stock and Watson (2012) for several structural shocks

Figure 3: Impulse responses to an uncertainty shock

Figure D.14: Impulse responses to an uncertainty shock with uncertainty measure ordered first in the recursive SVAR model (and baseline proxy SVAR)

Figure D.7: Impulse responses to an uncertainty shock with all variables expressed as deviations from HP trend

Full text

Discussion
Papers
Identifying Uncertainty Shocks
Using the Price of Gold
Michele Piffer and Maximilian Podstawski
1549
Deutsches Institut für Wirtschaftsforschung 2016

Opinions expressed in this paper are those of the author(s) and do not necessarily reflect views of the institute.
IMPRESSUM
© DIW Berlin, 2016
DIW Berlin
German Institute for Economic Research
Mohrenstr. 58
10117 Berlin
Tel. +49 (30) 897 89-0
Fax +49 (30) 897 89-200
http://www.diw.de
ISSN electronic edition 1619-4535
Papers can be downloaded free of charge from the DIW Berlin website:
http://www.diw.de/discussionpapers
Discussion Papers of DIW Berlin are indexed in RePEc and SSRN:
http://ideas.repec.org/s/diw/diwwpp.html
http://www.ssrn.com/link/DIW-Berlin-German-Inst-Econ-Res.html

Identifying Uncertainty Shocks Using the Price of Gold
∗
Michele Piffer and Maximilian Podstawski
†
February 11, 2016
Abstract
We propose a new instrument to identify the impact of uncertainty shocks
in a SVAR model with external instruments. We construct the instrument for
uncertainty shocks by exploiting variations in the price of gold around selected
events. The events capture periods of changes in uncertainty unrelated to other
macroeconomic shocks. The variations in the price of gold around such events
provide a measure correlated with the underlying uncertainty shocks, due to
the perception of gold as a safe haven asset. The proposed approach improves
upon the recursive identification of uncertainty shocks by not restricting only
one structural shock to potentially affect all variables in the system. Replicating
Bloom (2009), we find that the recursive approach underestimates the effects of
uncertainty shocks and their role in driving monetary policy.
JEL classification: E32, C32, D81.
Keywords: Economic uncertainty, external proxy SVAR, safe haven assets.
∗
We are thankful to participants to a seminar at the Bank of Italy and a seminar at Freie Univer-
sit¨at Berlin, together with Christoph Große Steffen, Michael Hachula, Helmut L¨utkepohl, Jonathan
Pinder, David Pothier, Morten Ravn and Malte Rieth, for insightful discussions and helpful comments
to this paper. Many thanks to Moritz Leitner for excellent research assistance. We also thank Alex
Pop for sharing his knowledge of the London Bullion Market.
†
Michele Piffer: German Institute for Economic Research (DIW Berlin); e-mail:
m.b.piffer@gmail.com. Maximilian Podstawski: German Institute for Economic Research (DIW
Berlin) and Freie Universit¨at Berlin; e-mail: mpodstawski@diw.de.

1 Introduction
Economic uncertainty, broadly defined as the difficulty of economic agents to make
accurate forecasts (Bloom, 2014; Jurado et al., 2015), is widely believed to have poten-
tially far reaching implications for the economy. Nevertheless, identifying the impact
of uncertainty on the economy is challenging, because uncertainty and the economy
are determined simultaneously. Since Bloom (2009), this challenge has been addressed
in the economic literature by developing strategies to identify uncertainty shocks, and
to estimate the impact of such shocks on the economy.
The economic impact of uncertainty shocks has been largely studied using Vector
Autoregressive (VAR) models. Their identification largely relies on the use of the
recursive approach (see, among others, Bloom, 2009; Baker et al., 2013; Scotti, 2013;
Bachmann et al., 2013; Caggiano et al., 2014; Jurado et al., 2015). Nevertheless, the
exclusion restrictions implied by the recursive identification have received substantial
criticism, because they impose that no other structural shock contemporaneously af-
fects the variables affected contemporaneously by the uncertainty shock (see Stock and
Watson, 2012; Baker and Bloom, 2013).
In this paper we propose a new strategy for the identification of uncertainty shocks.
We make use of the proxy SVAR methodology developed by Stock and Watson (2012)
and Mertens and Ravn (2013) to identify structural VARs using external instruments
and propose a new instrument for the identification of uncertainty shocks. In their
investigation of the macroeconomic dynamics of the great recession, Stock and Watson
(2012) highlight the challenge in isolating exogenous variations in uncertainty. This
paper attempts to fill this gap.
The identification strategy proposed in this paper relies on the dynamics of the
price of safe haven assets around selected events. It is reasonable to presume that
events generating unexpected variations in uncertainty are reflected in the price of
assets perceived by market participants as safe havens. For example, an increase in
uncertainty due to, say, an event that caused significant geopolitical instability might
1

materialize in a jump in the price of a safe haven asset. This could happen because
agents respond to the higher uncertainty by rebalancing their investments toward the
safe asset, or because those who hold such an asset are less willing to sell it, or both.
Accordingly, the price of safe haven assets can be a useful point of departure to build
an identification strategy for uncertainty shocks.
Since the price of a safe asset does not only reflect uncertainty shocks but also
many other structural shocks, not all variations in the price of safe assets can be used
to isolate uncertainty shocks. For this reason, we exploit the variation in the price of
safe assets around specific events. We consider events associated with variations in
uncertainty that occurred in an exogenous way relative to the state of the economy.
For example, we use the 9/11 terrorist attack to the World Trade Center, the invasion
of Kuwait by Iraq, the Chernobyl nuclear disaster and the fall of the Berlin Wall.
The use of events to isolate exogenous variations in variables of interest has a long-
standing tradition in the literature (see, for instance, Kuttner, 2001; Gurkaynak et al.,
2005). Having selected events that exogenously varied uncertainty, we construct an
instrument (or proxy) for uncertainty shocks by measuring the variation of the price
of safe haven assets around the events. While not measuring the shocks themselves,
these variations reflect the response of agents to the underlying uncertainty shocks,
and hence are correlated with such shocks, a feature that we exploit to construct of
an instrument. A battery of tests suggests using the price of gold to construct the
proxy, out of a wide range of candidate assets considered. For the price of gold, we
use intradaily data from the London Bullion Market, and Bloomberg News to address
when the news of each event hit the market.
The identification used in the paper has three main advantages. Firstly, it al-
lows for contemporaneous effects of the uncertainty shock on all variables, while not
restricting the uncertainty shock to be the only shock that potentially affects contem-
poraneously all variables. Secondly, it permits to build the identification approach on
high frequency data, instead of relying on monthly data as with identifications pursued
2

Frequently asked questions

Q1. What are the contributions in this paper?

The authors propose a new instrument to identify the impact of uncertainty shocks in a SVAR model with external instruments. The variations in the price of gold around such events provide a measure correlated with the underlying uncertainty shocks, due to the perception of gold as a safe haven asset. The proposed approach improves upon the recursive identification of uncertainty shocks by not restricting only one structural shock to potentially affect all variables in the system. 

Q2. What is the null hypothesis of the candidate proxy?

Rejecting the null hypothesis of no Granger-causality suggests that the candidate proxy indeed reflects variations in uncertainty. 

Q3. What is the recursive approach to identifying ut?

Under the recursive approach, identifying ut consists of first obtaining the Cholesky decomposition of the variance-covariance matrix of the reduced form shocks, and then selecting the column vector corresponding to the measure of uncertainty in yt. 

Q4. Why do the authors use a small selection of reliable events?

Since the instrument does not need to take values at every period, it is safer to use a small selection of reliable events, rather than a larger array of events that potentially pollutes the information captured in the proxy. 

Q5. Why is the fraction explained by the recursively identified shock small?

In addition, the uncertainty shock identified in the proxy SVAR explains a larger share in the variance of the real variables, while the fraction explained by the recursively identified shock is rather small. 

Q6. Why does the recursive approach of identifying uncertainty shocks improve?

Their identification strategy based on the constructed proxy variable improves upon the widely used recursive approach of identifying uncertainty shocks, because it uses outside intradaily information for the identification and allows the entire economy to react instantaneously to an uncertainty shock. 

Q7. What is the SVAR method used to identify ut?

under the proxy SVAR identification proposed by Stock and Watson (2012) and Mertens and Ravn (2013) and used in this paper, identifying ut consists of estimating the column vector bu that captures the correlation between the reduced form shocks and the proxy of uncertainty shocks (the position of this vector in the B matrix is irrelevant). 

Q8. What is the way to study the effect of measurement errors on the estimation of impulse responses?

As a proxy they use a dummy variable taking value 1 when the VXO peaks, and then employ a Monte Carlo to study the effect of measurement errors on the estimation of impulse responses. 

Q9. What is the correlation between the two series of structural uncertainty shocks?

Both series of structural uncertainty shocks exhibit higher volatility in the aftermath of the early 1980s recession, after the burst of the dotcom bubble in the early 2000s an during the recent financial crisis.