What drives differences of opinion in sovereign ratings? The roles of information disclosure and political risk

• DOI: http://doi.org/10.1002/ijfe.1579
• Author: Rasha Alsakka,Owain Gwilym,Huong Vu
• Date: 01 July 2017
• Published date: 01 July 2017

RESEARCH ARTICLE

What drives differences of opinion in sovereign ratings? The roles

of information disclosure and political risk

Huong Vu

1

| Rasha Alsakka

2

| Owain ap Gwilym

2

1

Coventry School of Economics, Finance

and Accounting, Coventry University,

Coventry CV1 5FB, UK

2

Bangor Business School, Bangor

University, Bangor LL57 2DG, UK

Correspondence

Rasha Alsakka, Bangor Business School,

Bangor University, Bangor, LL57 2DG, UK.

Email: r.alsakka@bangor.ac.uk

JEL Classification: G15; G24; G28

Abstract

This paper investigates the causes of split sovereign ratings across S&P, Moody's,

and Fitch for 64 countries from 1997 to 2011. We identify that split sovereign rat-

ings are not symmetric, with S&P tending to be the most conservative agency. We

find that opaque sovereigns are more likely to receive split ratings. Political risk

plays a highly significant role in explaining split ratings and dominates economic

and financial indicators. Out‐of‐sample model performance is enhanced by captur-

ing political risk. Government information disclosure affects split ratings between

Moody's and Fitch in emerging countries. The study implies an incentive for gov-

ernments to reduce political uncertainty and to enhance transparency.

KEYWORDS

information disclosure, opacity, out‐of‐sample performance, political risk,sovereign split ratings

1|INTRODUCTION

Amid persistent concerns surrounding the international eco-

nomic outlook in the aftermath of the global financial crisis,

credit‐rating agencies (CRAs) have attracted a higher profile.

In Europe, there was a strong wave of negative sovereign‐

rating actions, which resulted in split ratings for many high‐

rated countries in this region. Split ratings arise when

different CRAs assign unequal ratings to the same issuer at

the same time (e.g., Livingston & Zhou, 2010). They may be

temporary due to the CRAs' asynchronous actions or may

become persistent if the disparity in ratings continues over time.

During recent years, split sovereign ratings no longer pre-

vail solely within low‐rated sovereigns in emerging countries

as documented by Cantor and Packer (1996a), but are evident

across different levels of economic development. Yet there is

very little evidence on the causes of split sovereign ratings.

Cantor and Packer (1996a) attribute split ratings to the CRAs'

lack of experience in rating sovereign default risk, but they

provide little empirical evidence on this. Hill, Brooks, and

Faff (2010) report differences in the credit‐rating models

across the largest CRAs, but they do not relate this to split

sovereign ratings. Alsakka and ap Gwilym (2012) analyse

macroeconomic factors, low creditworthiness, and home bias

as potential causes of split sovereign ratings, yet their

research is focused only on emerging countries.

We hypothesize that split sovereign ratings arise due to

information opacity rather than random errors by the CRAs.

1

The information opacity hypothesis for split ratings has been

tested for corporate issuers and across industries. Morgan

(2002) finds that split ratings occur more often in banks

and insurance companies because their assets are more

opaque. Livingston, Naranjo, and Zhou (2007) and Hyytinen

and Pajarinen (2008) evaluate corporate opacity by consider-

ing firm age, intangible assets, and other observables. We

contribute to previous literature by providing new evidence

on the information opacity hypothesis for sovereign issuers.

We contend that opacity arises when CRAs rely, to a great

extent, on subjective evaluations of sovereign risk. We focus

on two key sources of opacity in assessing sovereign ratings:

the quality of information disclosure by governments and

political risk.

1

Ederington (1986) concludes that corporate bond split ratings result from

random rating errors, which suggests that split ratings are symmetric between

any pair of CRAs. See Section 3 for more details on the random error versus

the information opacity hypotheses of split ratings.

Received: 13 May 2015 Revised: 7 March 2016 Accepted: 14 March 2017

DOI: 10.1002/ijfe.1579

216 Copyright © 2017 John Wiley & Sons, Ltd. Int J Fin Econ. 2017;22:216–233.wileyonlinelibrary.com/journal/ijfe

We first consider split sovereign ratings in the light of

governments' information quality and transparency. This

approach is original because we focus on the openness and

willingness of the government to release information in the

public domain. Information from transparent sovereigns is

richer, more accessible, and updated more frequently than

from opaque sovereigns. Split sovereign ratings could arise

when CRAs have limited access to high‐quality data with

which to base their credit assessments. Quality of information

is central to the quality of credit ratings. Recent developments

in the regulatory sphere suggest that CRAs should be held

accountable for taking appropriate measures to make use of

all the available information from credible sources (European

Commission, 2009). Policy makers also call on the CRAs

and governments to be more transparent in terms of informa-

tion disclosure (House of Lords, 2011). Therefore, material

changes in the quality of the data utilised by the rating ana-

lysts must be signalled to the users. Although regulators are

interested in the transparency of the CRAs, we emphasize

the potential benefits from governments' transparency.

Among the determinants of sovereign ratings, we focus

on the importance of political risk (see Afonso, Gomes, &

Rother, 2011; Butler & Fauver, 2006; Mellios & Paget‐

Blanc, 2006). In contrast with corporates, sovereign credit

ratings consider the governments' capacity and willingness

to repay. The latter factor is affected by political concerns.

Political risk represents the soundness and stability of the

legal and institutional systems of a country. From the per-

spective of financial market participants, it commands a sig-

nificant risk premium and raises stocks' correlations and

volatilities, particularly in weak economic conditions (Pástor

& Veronesi, 2013). Recently, political risk has attracted con-

siderable attention, not only in emerging markets, which are

commonly characterised by a lack of political stability, but

also in European countries. Europe is facing a confluence

of serious political challenges. The impact of challenging

political dynamics has been particula rly evident in Greece,

as well as the rise of new antireform political parties in some

countries (e.g., Portugal and Spain), and disagreements

within governments (e.g., France) reduce the governments'

willingness to embark on structural reforms and fiscal

consolidation.

2

However, the assessment of political issues

usually involves subjectivity and ambiguity. Therefore, we

expect that political risk tr iggers greater differences of opin-

ions and interpretation by CRAs than do the quantitative indi-

cators (typically used for economic and financial risks). By

addressing political risk, this study is clearly differentiated

from literature on split corporate ratings and from the very

limited prior work on split sovereign ratings.

We investigate the determinants of split sovereign ratings

from the three largest CRAs (S&P, Fitch, and Moody's) using

ratings, outlook, and watch information from 1997 to 2011.

Our data allow for consideration of split sovereign ratings

in many developed countries, including a focus on the Euro-

pean dimension. Further, prior studies on split ratings only

consider the rating notations, whereas outlook and watch

are ignored.

3

In fact, outlook and watch signals can be at least

as important as rating changes in their impact on financial

markets (e.g., Afonso, Furceri, & Gomes, 2012; Kim &

Wu, 2011; Sy, 2004). A CRA's complete opinion on an issuer

consists of a credit rating and a rat ing outlook/watch status,

so split ratings in our paper are expressed based on all these

elements.

Our empirical analysis supports the opacity hypothesis in

explaining split sovereign ratings. We show that split sover-

eign ratings are lopsided rather than symmetric, with S&P

tending to be the most conservative. We identify the impor-

tance of opacity inherent in political risk, par ticularly in

countries outside Europe. We use six worldwide governance

indicators (WGI) published by the World Bank as the politi-

cal risk proxy variables and find that they are significant

explanatory factors. Out‐of‐sample model performance is

enhanced by capturing political risk. We highlight that rating

splits that involve Fitch (vs. Moody's or S&P) are more prone

to the political factor. This may imply a more political risk‐

focused approach to assess sovereign risk by Fitch relative

to the other two CRAs. Further, we assess the information

disclosure quality by whether (and for how long) a Freedom

of Information Act (FOIA) has been in place. Our findings

reveal that an incentive exists for non‐European emerging

countries to provide the CRAs with updated, credible infor-

mation, but the evidence is only linked to Moody's versus

Fitch ratings.

We make several important contributions to the literature.

First, we provide evidence that opacity is highly relevant to

split ratings of sovereign issuers. Second, we distinguish

our study from the literature on split corporate ratings by

introducing a political risk factor to represent the “willing-

ness to pay”element of sovereign ratings. Third, we intro-

duce an approach to analysing the link between split ratings

and quality of information disclosure by governments.

Fourth, we identify that these effects differ between European

countries and emerging markets in the rest of the world.

Finally, the prior credit‐rating literature has defined split rat-

ings using only the rating notations; this paper defines split

2

A recent example of such political influence on ratings is S&P's downgrade

of Poland in January 2016, which was stated to be driven by concerns regard-

ing radical policies implemented by its new conservative government, for

example, steps taken by the newgovernment to seize control of Poland's pub-

lic media and to challenge the independence of its constitutional court.

3

Outlook changes indicate the changes between four statuses: positive, neg-

ative, stable, and developing (or evolving by Fitch). Watchchanges indicate

the changes between watch for possible upgrade, watch for possible down-

grade, watch with uncertain direction, and no watch assignment.

VU ET AL.217