Foreign Ownership and the Credibility of National Rating Agencies: Evidence from Korea University of Bari Aldo Moro by G. Ferri (U. Bari), T.S. Kang (Bank.

Foreign Ownership and the Credibility of National Rating Agencies:

Evidence from Korea

University of Bari Aldo Moro

byG. Ferri (U. Bari), T.S. Kang (Bank of Korea), Punziana

Lacitignola (U. Bari) and J.Y. Lee (Yonsei U.)

Workshop “Curbing Volatility? Financial Markets, Credit Rating Agencies and Sovereign Investments Funds”

Bocconi University, Milan 17 November 2011

Bank of Korea

2

Outline of the presentation The role of rating agencies as institutions

helping investors to evaluate issuer/issue’s creditworthiness

Global Rating Agencies (GRAs) versus National Rating Agencies (NRAs)

The Korean rating agencies experience Analysis of the effects of rating activity of GRAs

and NRAs on local markets Our event study (downgrading) Conclusions

3

Main criticisms of Global rating agencies (GRAs)

1. Low degree of competition among GRAs due to NRSRO recognition by SEC (Partnoy, 1999; White, 2002) and reputation matters difficulties for NRAs to gain access to international markets

2. Lagging (IMF, 1999; Bongini, Laeven and Majnoni 2002) and procyclical ratings (Ferri, Liu & Stiglitz, 1999; Bhatia, 2002)

3. GRAs’ ratings in developing countries:- less accurate than in developed countries (Ferri, 2004; Ferri & Liu, 2003 - 2005), but still no specific evidence on this about NRAs;- excessively tied to sovereign ratings (Ferri, Liu and Majnoni, 2001)

4

4. Little price competition5. Entry restrictions due to regulatory

requirements to obtain a rating from an NRSRO rating agency, to date the NRAs officially recognized as NRSROs are JCR and R&I (regulatory franchise; Partnoy, 1999)

6. Unsolicited ratings practice to gain market share by GRAs

7. Sub-prime: poor performance in the evaluation of structured finance product (complexity and conflicts of interest)

Main criticisms of Global rating agencies (GRAs) (cont’d):

5

Perilous reliance onstructured finance

Race torevise

A sip of recent criticisms: subprime & structured finance

6

A potential trade-off GRAs versus NRAs

GRAs (established since 1909):- reputation, indep’nce & int’l recognition … but- apply a “standardized” rating methodology

across countries (regardless the environment in which firms operate)

NRAs (established since mid 1980s):- comparative advantage accessing more

appropriate knowledge of local environment- originated with the support of nat’l regulation- lower fees than GRAs’ … but- less independent (mostly owned by domestic

financial institutions)

7

Trade-off between higher reputation/indep’nce of GRAs and in-depth knowledge of NRAs on local basis

Complementarity between GRAs (larger, internationalized companies) and NRAs (small-sized/domestic focused companies)(Ferri, Kang, Lee, 2006; Ferri & Lacitignola, 2007)

Markets seem to take into account both NRA and GRA ratings when pricing securities; both may actively contribute to financial market development (Ferri & Lacitignola, 2007)

In Japanese local markets, NRA rating changes influential more than GRA rating changes (Lacitignola, 2007) (less for Li et al., 2006)

Market impact of GRAs and NRAs

8

Specifically for Japanese financial markets: A combination of NRA and GRA ratings

predicts spreads on securities’ secondary market trading more accurately than any of the two classes taken separately (Packer, 2000)

NRA ratings are more related to rated companies’ financial ratios than GRA ratings (Packer, 2000)

In the opinion of Japanese corporations, NRAs do not differ from GRAs in terms of market influence and recognition (JCIF, 2001)

Market impact of GRAs and NRAs (cont’d)

9

Ratings Across the Emerging World

More ratings in Asia, where most NRAs belong

Figure 1 Source: Our calculations on Financial Times Interactive data for Winter 1998-99 and Winter 2004.

10

Only Asia features a significant presence of NRAs, elsewhere (e.g. EU) NRAs have been acquired by GRAs

Figure 2 Source: Our calculations on Financial Times Interactive data for Winter 2004

11

In some Asian countries NRAs issue numerous ratings compared to GRAs

Figure 3 Source: Our calculations on Bloomberg data for 2004.

12

•GRAs:

Moody’s, S&P and Fitch

•NRAs (example in 4 Asian countries):

Japan: JCR, R&I, MikuniSouth Korea: KIS, KR, NICE and SCIMalaysia: MARC and RAMIndia: ICRA, CRISIL and CARE

Who are the GRAs and the NRAs?

GRA’s usually operate world-wide through affiliatessubsidiaries and branches

13Fig. 5 Source: Our calculations on Bloomberg data between 1990 and may 2005

NRAs: rating less internationalized (listed) companies

14

Relative (Average) Size of Companies With National Ratings Only

0,128

0,097

0,259

0,055

0,000

0,050

0,100

0,150

0,200

0,250

0,300

India Japan Korea Malaysia

Fig. 6 Source: Our calculations on Bloomberg data between 1990 and may 2005

NRAs: rating smaller-sized (listed) companies

15

The Korean rating industry

GRAs due to regulatory burden do not directly operate in the Korean rating market

KIS is affiliated to Moody’s which controls “50%+1 Stock” of it since 2001KR is affiliated to Fitch: Fitch Ratings Ltd. had 9.01% of KR shares since 2001 and increased its shareholding to 54.4% in April 2007, and again to 64.2% in September 2008NICE is wholly domestically owned

16

Hypothesis: KIS and KR due to their affiliation with GRAs benefit of its reputation capital and are perceived as higher information providers than NICE (domestically owned)

We expect larger effects of rating changes by KIS and KR on daily bond prices with respect to rating changes from NICE (Korean issuers)

Estimation window: 90 trading days; Event window 21 trading days (t-10, t+10;

where t is the event date) Bond prices & rating changes over 2001-2007

The event study set-up

17

The event study set-up (cont’d)

To calculate the normal returns we use the Market model

Rit = i + iRmt + it

E(it = 0) var (it ) = i2

As market portfolio we use the KOSPI index

Bond returns are assumed jointly multivariate normal

and independently and identically distributed through time

Normal return is estimated by OLS

18

The event study set-up (cont’d)

Abnormal return (AR) is calculated as deviation of the predicted return from the observed return:

oTo construct tests, ARs are aggregated bothacross bonds and across time:

To test the persistence of the effects of rating changes AARt are aggregated through time:

19

Because of previous evidence we focus on Downgradings:

rating upgrades have no impact on the bond and stock markets and downgradings areAssociated with significant negative returns inEquity and bond prices (Griffin and Sanvicente, 1982;Holthausen and Leftwich, 1986; Impson et al.,1992; Barron et al., 1997; Ederington and Goh, 1998; Goh and Ederington, 1993, 1999; Liu et al., 1999; Matolcsy and Lianto, 1995; Wansley et al.,1992; Zaima and McCarthy, 1988).

20

Our sample

21

Results:

Effect of downgrading on AARt:

on the event date is significant only for KIS and NICE with similar impact (AARs are -0.0076 and -0.0080, respectively; Table 3). In the day after the event date the downgrading is significant for all NRAs, but the effect is larger for NICE (AAR= -0.020), with respect to KR (AAR= -0.012) and KIS (AAR= -0.009). Downgrading effects are persistent over time.

22

Results - from Table 3:

23

Results (cont’d):

Effects of 1-notch downgrading on AARt:

AARt are significant the day after the event occurs for all NRAs but with different magnitude (Table 5);effects slightly bigger for NICE (-0.01276), then follow KIS (-0.01141) & KR (-0.00676);effects on bond prices are persistent in time;they are anticipated more days before by markets for KIS and KR, the same is not true for NICE. The effects (CAARs) of NICE’s downgradings after the event date are persistently larger than for the other two NRAs (Figure 1).

24

Results (cont’d) - from Table 5:

25

Fig. 1: Effects of 1-notch downgrading on bond prices

26

University of Bari

Conclusions:

We confirm the evidence about downgrading effects on bond markets

Our results cast some doubts on the superiority of GRAs as perceived by a financial market as important as the Korean one. NRAs affiliated to GRAs don’t seem to benefit of their reputational capital

A policy implication is that developing NRAs may be beneficial to the development of domestic financial markets