Macroeconomic Uncertainty in South Africa

• Author: Chris Redl
• Published date: 01 September 2018
• Date: 01 September 2018
• DOI: http://doi.org/10.1111/saje.12198

© 2018 Economic Society of Sout h Africa .

South African Journal of Economics Vol. 86:3 September 2018

doi :10.1111 /saj e.1 2198

MACROECONOMIC UNCERTAINTY IN SOUTH AFR ICA

CHRIS RE DL*

Abstract

This paper develops a new index of economic uncertainty for South Africa for the period 1990–

2014. The index is constructed from three sources: (1) Disagreement among professional

forecasters, (2) a count of international and local newspaper articles discussing economic

uncertainty in South Africa and (3) mentions of uncertainty in the quarterly economic review of

the South African Reserve Bank. The index shows high levels of uncertainty around the period of

democratic transition in 1992–1994, the large depreciation of the currency in 2001 and the

financial crisis of 2008. The uncertainty index is a leading indicator of a recession. An unanticipated

increase in the index is associated with a fall in GDP, investment, industrial production and private

sector employment. Contrary to evidence for the U.S.A and the U.K., uncertainty shocks are

inflationary. These results are robust to controlling for global uncertainty shocks, consumer

confidence and financial shocks.

JEL classification: D8 0, E32, E31, E66, N17

Keywords: Economic un certainty, business cycles, infl ation, South Africa

1. INTRODUCT ION

The Great Recession has renewed interest in the question of the sources of business c ycle

fluctuations. I study a new d river proposed by Baker et al. (2016): fluctuations in un-

certainty. These authors develop a proxy for economic policy uncertainty based on news

articles discussing policy uncertainty, the number of federal taxes set to expire and dis-

agreement among professional forecasters over the future values of government purchase s

and inflation. They show, using a vector autoregression, that an increase in their proxy

equivalent to the rise seen during the financial crisis is associated with a loss of around 2

million jobs and a decline in industria l production of 2.5% in the U.S. Studies following

Baker and Bloom (2013) have provided similar evidence that uncer tainty shocks are im-

portant drivers of the business cycle, e.g. Dendy et al. (2013) for the U.K.

Despite some cross-country work relating uncertainty to growth by Baker and Bloom

(2013), there is relatively little evidence of the effects of such proxies for economic uncer-

tainty for developing countries. Given that developing countries experience much hig her

levels of realised volatility t han developed nations (Fernandez-Villaverde et al., 2011a;

Bloom, 2014) it is plausible that fluctuations in uncertainty are important drivers of

business cycles in these regions. It has been argued by Leduc and Liu (2012) that shocks

* Corresponding author: Ba nk of England, Threadneedle Stree t, London, EC2R 8AH, United

Kingdom of Great Britain a nd Northern Ireland. E -mail: redl.chris99@gmail.c om

This work was conducted at Queen Ma ry, University of London before the author joined the

Bank of Engla nd. The views expressed in this paper a re therefore those of the author and do not

necessarily re flect those of the Bank of Engla nd.

361

South African Journal

of Economics

© 2018 Economic Society of Sout h Africa .

South African Journal of Economics Vol. 86:3 September 2018

362

to uncertainty have a centra l role to play in understanding business cycles as t hey are

prototypical aggregate dema nd shocks, with lower output and inflation. However, recent

papers by Popescu and Smets (2010) and Gilchrist et al. (2013a) have challenged the rel-

evance of uncertainty shocks once their correlation with credit spreads is accounted for,

suggesting th at uncertainty shocks only matter when acting th rough a financial channel.

Extending studies of uncertainty beyond developed nations can help disentangle the

effects of fina ncial shocks from uncerta inty shocks. During the Great Rec ession many

developing countries experienced increases in uncertainty, as the impact of a large reces-

sion in trading partner countries took hold, yet they did not experience the same levels

of financial stress and instability as in t he developed world.

This paper makes t wo contributions to this literature. Firstly, it extends the evidence

that uncertaint y shocks generate drops in real activity to a developing country. Secondly,

it provides evidence that uncertainty shocks have real effects e ven when controlling for

financia l stress (credit spreads).

I construct an index of economic uncerta inty following Baker et al. ( 2016 ) a nd

Dendy et al. (2013) for the period 1990–2014. The index is constructed from three

sources: (1) Disagreement among professional forecasters about macroeconomic condi-

tions using novel data from a forecasting competition run by a national newspaper, (2)

a count of international and local newspaper articles discussing economic uncerta inty

in South Africa and (3) mentions of uncerta inty in the quarterly economic review of

the South African Reserve Bank (SAR B). The index is positively correlated with other

available domestic proxies for uncertainty, i.e. realised and option implied volatility of

the stock market, however, it exhibits significant independent variation from widely

used international measures of uncertainty (e.g. VIX and U.S. uncertainty indices).

The index shows high levels of uncertainty around the period of democratic transition

in 1992–1994, the large depreciation of the currency in 2001 as well as the financial

crisis of 2008.

To measure the impact of uncertainty shocks I use a st ructural VAR. The results show

that economic uncertainty is a leading indicator of a recession in South Africa. An un-

anticipated increase in the index is associated with a fall in GDP, investment, industrial

production, capital inflows and private sector employment. Contrary to evidence for the

U.S.A. and the U.K., uncertai nty shocks are inf lationary. I show that this result is robust

to the inclusion of a proxy for credit spreads and global uncertainty shock s as well as

alternative methods of construction for the index.

The remainder of the paper is organised as follows. Section 1.1 reviews the literature

on uncertainty shocks. Section 2 describes the construction of the index and compares

it to alternative proxies in South Africa. Sect ions 3 and 4 presents the VAR results and

Section 5 concludes.

1.1 Literature

Why should uncerta inty matter for business cycles? There a re (at least) three broad reasons iden-

tified in the t heoretical literature: real options, risk aversion and g rowth options effec ts (Bloom,

2014 ).