The effects of a sovereign credit
rating change on domestic banks’
stock prices
Abstract
Credit rating agencies are a crucial entity in the financial markets of today. This thesis analyzes the effects of sovereign credit rating changes on domestic bank’s stock prices. An event study was conducted and the results were that both upgrades and downgrades do not have a significant effect on domestic banks’ stock prices. Furthermore a larger magnitude of the credit rating change had no significant stronger effect on domestic banks’ stock prices.
Name: Barend Boelaars Student number: 10437827
Bachelor program: Economie & Bedrijfskunde Specialization: Financiering & Organisatie Supervisor: Pascal Golec
2 Statement of originality
This document is written by Barend Boelaars who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no other sources than those mentioned in the text and its references have been used.
3 Table of content
1. Introduction ……… 4
2. Literature review ……… 6
3. Data & Methodology ……….. 9
4. Results ……… 14
5. Conclusion ………. 21
6. Reference list ……….. 22
4 1. Introduction
Credit rating agencies have a big role in the financial markets of today. This is needed in a market with a lot of information asymmetry due to the increased complexity of financial markets and products (Rijksoverheid, 2008). Credit rating agencies (form now on referred to as CRA’s) try to lower the information asymmetry between lenders and borrowers (An & Chan, 2008). They are independent third parties that review the creditworthiness of
companies and countries and thus are an important source of information for credit risk (Cole & Cooley, 2014). CRA’s are an essential information source for unsophisticated investors because investors can judge the creditworthiness of a company or country by using their ratings. This makes it easier for investors to make investment decisions. Investors use these ratings as a screening tool to decide which investments should be added to their portfolio (Elkhoury, 2008). Credit ratings are also important for regulatory purposes for banks and pension funds. These institutions are sometimes restricted to only invest in high rated debt. When companies or countries receive a high credit rating investors believe that this is a safe investment. Investors all over the world accept these ratings as a reliable source of the creditworthiness of companies.
The importance of credit ratings is increasing even though there has been some discussion about the role of credit rating agencies in the financial crisis (SEC, 2013). CRA’s have a lot of influence on financial markets because they can affect the financial position of many institutions. This is mainly because there is not enough competition in this market due to the high entry barriers in this business. An example of how powerful these CRA’s are is when Moody’s downgraded Japan from A1 to Aa3. This downgrade caused the Yen to hit a seven year low against the dollar within a few months (BBC, 2014). Another example is when Moody’s downgraded the United Kingdom from Aaa to Aa1 for the first time in the history of the country. Following this downgrade the Sterling dropped to a two year low against the dollar. The pound dropped to a 16-month low against the Euro (The Telegraph, 2013).
Sovereign credit ratings provide information about a countries economic outlook. These ratings affect the whole nation and could have an effect on the entire market of a country. Previous research has found that sovereign credit rating changes influence
government bonds. This can be explained by the fact that a change in a sovereign credit rating reflects the default risk of a country. This default risk in turn affects government bond prices (Codogno, Favero & Missale, 2003). One can also look at the effects of a sovereign credit rating change on stock prices. There has been some research on this topic but there is no
5 definitive conclusion about the effect of a sovereign credit rating change on financial markets’ stock prices (Pukthuanthong-Le, Elayan & Rose, 2007).
CRA’s provide the market with valuable and reliable information about the
creditworthtiness of institutions. Even though there are some criticisers about this system, credit ratings are still a key instrument for every market participant and therefore the accuracy and reliability are crucial for financial institutions. Therefore it is interesting to investigate the effects of changes in sovereign credit ratings on stock markets. This research will investigate weather sovereign credit rating changes affect the stock prices of domestic banks. This will be done by conducting an event study.
The main findings of this thesis are as follows. First, sovereign upgrades and downgrades do not have a significant effect on domestic banks’ stock prices. Second,
sovereign credit rating changes do not react significantly stronger if the change of the rating is larger.
Hereafter in section two the literature review and the relevance of this thesis will be discussed. In section three the methodology will be explained and the hypothesis will be stated. Section four will describe the data used for this research. Also the empirical results will be shown and discussed in this section. Thereafter in section five a conclusion will be made based on the results.
6 2. Literature review
First the theoretical framework will be explained in this chapter. The efficient market hypothesis will be explained and also its relevance for this research. Thereafter previous research about the effects of credit rating changes will be analyzed. Some background information about credit rating agencies and how they base their ratings can be found in Appendix I.
2.1 Efficient market hypothesis
The efficient market hypothesis (EMH) states that all relevant available information should be reflected in stock prices (Fama, 1970). Due to the fact that all relevant information is
available there should be no arbitrage opportunity and so the EMH predicts that it is
impossible to make a profit by trading (Jensen, 1978). There are three forms of EMH namely weak, semi strong and strong. The weak form states that prices reflect all historic publically available information. The semi strong form states that prices reflect all historic publically available information and that prices react to new available information. The strong form states the same as the semi strong form but in addition to that the strong form states that prices react immediately to private information.
If CRA’s base their credit ratings on publicly available information, the EMH predicts that stock prices will not react to a sovereign credit rating change because this information should be reflected in the stock price already. Though there are many researches that found that sovereign credit rating changes do have significant effects on stock prices. This implies that there is evidence against the semi strong form of EMH or that CRA’s do have some form of private information that indirectly becomes available through a credit rating (Brooks et al, 2004).
7 2.2 Previous research
There have been many studies that investigate the effects of corporate credit rating changes on stock prices. Holthausen and Leftwich (1986) found that downgrades have a significant effect on stock prices but upgrades do not. Li, Visaltanachoti en Kesayan (2004) conducted a research about the effects of credit rating changes on Swedish listed companies. Their results show that both downgrades and upgrades had a significant effect, though the downgrades only became significant in the long run. Further they found that downgrades have a larger
significant impact on stock prices than upgrades. Freitas & Minardi (2013) analyzed the effects of a credit rating change on domestic markets in Latin America between 2000 and 2009. They found that a downgrade had a significant negative effect. Upgrades had an upward effect but this effect was not significant. In general, corporate downgrades have a significant downward effect but upgrades mostly do not have significant effects.
Research about the effects of a sovereign credit rating change can be done in two ways. One way is to look at the effects of a sovereign credit rating change on government bond prices. Cantor et Al (1996) found that sovereign credit rating changes affect bond markets due to the fact that new information becomes available to the market and therefore significantly affect the bond yields. According to Stancu & Minescu (2011) government bond prices react more severe to sovereign downgrades than sovereign upgrades. They also found that a sovereign credit rating change not only had an effect on domestic markets but also on neighbouring financial markets. Steiner & Heinke (2001) researched the effects of a sovereign credit rating change on international bond prices. Their results show that downgrades and negative outlooks have significant effects but upgrades cause no effect.
Another way is to look at the effects of a sovereign credit rating change on stock prices. Brooks et al. (2004) concluded that a sovereign downgrade has a significant negative effect on both domestic stocks and the exchange rate to the dollar. In the markets investigated by Norden & Weber (2004) they found that the financial stock markets significantly dropped due to a sovereign downgrade but an upgrade had an insignificant upward effect. Kaminsky and Schmukler (2002) researched the effect of a sovereign credit rating change on country risk and stock returns. Their results show that sovereign credit rating changes affect stock markets and bond yields and that these effects are stronger during crisis. They also found that stocks significantly drop due to a downgrade but an upgrade does not result in abnormal high returns. According to Martell (2005) stock markets only react to a sovereign credit rating downgrade. He also concluded that firms in more developed countries are affected less by a sovereign downgrade than firms in emerging countries. Hooper et al (2008) conducted a
8 research to look at the effects of a sovereign credit rating change on international financial markets. They found that CRA’s provide the market with new information and that markets react to this information even before the credit rating change. This could be due to the fact that the market already anticipates a credit rating change. They also found that markets tend to react more severe to downgrades than to upgrades. Markets also tended to react more severe when a rating change occurs during a crisis and when countries have relative high debts. Gu, Jones and Liu (2014) researched the effects of CRA’s outlooks’ and sovereign credit rating changes on equity markets. They found that a larger magnitude of the change whether it is an outlook change or a credit rating change had a significant different impact on equity markets. Gan et al. (2014) found that a sovereign credit rating change had no significant effect on the stock price or volume traded of banks in emerging markets. Though they do mention that this insignificant result could have come due to the fact that they only included the largest bank of each country. They recommend adding more banks per country because it is
insufficient to only include the top bank in the sample. This comment will be taken into account in this research.
There has been a lot of research done about the effects of a credit rating change. In general a corporate credit rating downgrade has a significant effect on the firms’ stock prices. Upgrades on the other hand mostly do not have a significant effect. For sovereign credit ratings the same results are found. Empirical evidence shows that sovereign downgrades have a significant effect on domestic stocks but sovereign upgrades do not. This thesis will
investigate if there is an effect on stocks of domestic banks after a sovereign credit rating change. A cumulative abnormal return analysis will be done to examine the effects of a sovereign credit rating change. This research will also look if there is a different effect for upgrades and downgrades and if a larger magnitude of the change has stronger effects on the stock prices.
This thesis will differ from existing research by looking at the effects of a sovereign credit rating change on domestic banks rather than the total domestic market. This research will also include many observations from different countries across the world rather than a specific market, such as emerging markets or the European market. Also this research will look if there is a significant difference in effects if the magnitude of the credit rating change is larger. For example this research will investigate whether a credit rating change of two or more notches has a stronger effect than a one-notch credit rating change. Also many of the empirical studies focus on the effects of S&P credit rating changes. This research will only study the effects of Moody’s credit rating changes.
9 3. Data & Methodology
In this section the hypotheses of this research will be stated. Second, the data used for this research will be described. At last the methodology of the analysis will be discussed.
3.1 Hypotheses
As stated in the previous paragraph the general consensus is that sovereign credit rating downgrades have a significant downward effect on the domestic stock market and upgrades sometimes have an effect but this effect is rarely significant. Based on previous literature and the EMH theory, the hypotheses for this research will be made. The following hypotheses will be tested in this research:
1. H0: DOWNGRADE = INTERCEPT = 0
H1: DOWNGRADE ≠ 0, INTERCEPT ≠ 0
2. H0: MAGNITUDE = 0
H1: MAGNITUDE > 0
3.2 Data
This research will investigate the effects of a sovereign credit rating change on stock prices of domestic banks. The long-term credit rating changes are analyzed because these tend to have the most significant effect on financial markets. Moreover, only credit rating changes of Moody’s will be taken into account in this research. This is because many empirical studies already have investigated the effects of S&P credit rating changes. In this research credit rating changes of many different countries around the world were added in the sample to look if there is a common effect of a sovereign credit rating change. Furthermore the largest banks of each country by market capitalization were included in the dataset.
The data for this research has been collected through the Moody’s website and Datastream. Data has been collected over a 20-year period from 1995 to 2015. For each country sovereign credit rating changes that followed a previous sovereign credit rating change within 130 days were left out of the sample. This was done because otherwise the estimation window and event window would overlap of different credit rating changes. Furthermore only countries were included for which Datastream had sufficient data of the domestic banks. For each bank the daily stock prices and stock volume traded were collected.
10 Also the exchange rate to the dollar was added in the dataset. This is added as a
control variable because the exchange rate could have an effect on the abnormal returns. Next the daily index values of the S&P500 were collected to use as a benchmark for the calculation of the abnormal returns. The S&P500 is used as a benchmark because the benchmark should not be affected by the sovereign credit rating change. Furthermore by using the S&P500 as a benchmark the world events that are unrelated to the sovereign credit rating change are controlled for.
The dataset consists of 27 downgrades and 33 upgrades. In the sample two upgrades and seven downgrades had a credit rating change of two or more notches. Appendix II shows an illustration of the descriptive statistics of the dataset.
3.3 Methodology
To research if a sovereign credit rating change has an effect on stock prices of domestic banks an event study will be conducted. The basic idea of an event study is to analyze if an
announcement has had an effect on firm value. In this research the event is a sovereign credit rating change. To determine if the announcement has had an effect on firm value, a
cumulative abnormal return analysis will be done. An event study can be defined in three steps. First the event must be identified and the dates of the event should be determined. Second a suitable benchmark should be chosen to calculate normal stock return activity. Lastly the abnormal returns should be calculated and analyzed in the event window. In figure 1 a graphical image of the estimation window and the event window is shown.
11 In this research the choice is made of an event window of 30 days prior and after the event to analyze the abnormal returns. According to Freitas & Minardi (2013) the choice of the event window is arbitrary and should not be too long. An event window that is too long can have a bias due to other events in the event window that have an effect on stock prices. An
estimation window that is too short could maybe not capture the effects of the event. For the estimation window this research will use 100 days. There is no definition of how many days should be included in the estimation window but according to Peterson (1989) 100 days is a decent amount.
This research will make use of the cumulative abnormal return analysis. The abnormal return is found by the actual return minus the expected normal return.
AR
i,t= R
i,t– E(R)
i,tWhere AR is the abnormal return of the stock (i) in period (t), R is the actual return of stock (i) in period (t) and E(R) is the expected normal return of the stock (i) in period (t). There are two types of models to calculate the expected returns namely economic models and statistical models. In this research the statistical type of model will be used because it is relative easy to apply and previous research suggests that statistical models provide good estimates
(MacKinlay, 1997, pp. 16-19). The market model is the most commonly used statistical model to estimate expected returns (Joo & Pruitt, 2006) and therefore this research will also use the market model to estimate the normal expected returns. The market model defines the expected return of an asset by comparing it to a market index return. As stated before the market index returns in this research are based on the S&P500. This is because the estimated normal returns should not have an impact of the event. Therefore a national stock market cannot be used because this index could be affected by a sovereign credit rating change. The expected normal return is calculated as follows:
E(R)
i,t=
α
i+
β
i*Rm
i,t+
ε
iWhere E(R) indicates the estimated return of an asset (i) at time (t). Rm indicates the return of the market at time (t). The alpha and the beta are estimated by conducting an OLS regression. The alpha is the expected return the stock would have if the market has a return of zero. The beta indicates the sensitivity of the assets return to the market. The error term shows how
12 much the chosen asset reacts to new available information, meaning that this indicates the abnormal returns between the expected normal return and the actual return of the stock. The alpha and the beta will be estimated by conducting an OLS regression from 100 days in the estimation window. After the alpha and beta have been estimated, the abnormal returns can be calculated.
For this research the return of the stock (Ri) is a self-created index. This index was
calculated by taking the weighted average return of the all the banks in the sample per country. A minimum of three banks per country were included in the self-created index. This was done because then the average effect of a sovereign credit rating change on the banking industry could be analyzed instead of analyzing each bank individually. One could also look at the effects of each bank individually but this thesis wants to research the effects of a sovereign credit rating change on the overall domestic bank industry and not individual domestic banks.
After calculating the daily abnormal returns in the event window they will be added together to get the cumulated abnormal return (CAR) of the event. For each individual event the CAR is calculated between the event date and the end date of the event window. In the
illustration below t1 indicates the beginning date of the event window and t2 indicates the end
date of the event window. Also an analysis will be performed on the CAR for the total event period.
After calculating all the cumulative abnormal returns for each event a model is built to test if a sovereign credit rating change had an effect on the abnormal returns. In this model the variables stock volume traded, exchange rate and magnitude of the credit rating change are added. Also interaction variables were added of each of these variables to see if these affected the CAR. A dummy variable is used to identify if the event was an upgrade or a downgrade.
13
CAR = α + β1DOWN+ β2VOL + β3ER + β4MAG + β5(VOL*DOWN) + β6(ER*DOWN) + β7(MAG*DOWN)
In this model DOWN is a dummy variable, which equals 1 for a downgrade and 0 for an upgrade. VOL is the percentage difference in total volume-traded stock before and after the event. ER is the exchange rate of the domestic currency to the dollar. MAG is a dummy variable, which is one for credit rating changes of one notch and equals two for credit rating changes of two or more notches. The interaction variables were added to see if stock volume traded, the exchange rate and the magnitude of the change have a different effect on upgrades and downgrades. The variable volume was defined in 3 different ways namely the absolute volume traded, log volume traded and the percentage change of volume traded before and after the event. This was done to see if different definitions of volume traded would have a different effect on CAR. All different types of volume variables were regressed in separate models and the volume percentage change was the only one that was significant. Therefore this definition of volume will be used in the rest of the models. A regression analysis will be conducted to see if the effects of a sovereign credit rating change are significant on domestic banks’ stock prices. After the regression a t-test will be conducted to test the significance of each variable. For the significance the model an F-test will be conducted.
14 4. Results
In this section the results of the regression analyses will be discussed. Prior to the regression analyses the assumptions of normality, linearity, homoscedasticity and no multicollinearity have been checked. The assumption of normality has been checked by observing the
histogram of the dependent variable and checking the normality of the standardized residuals. The histogram of the dependent variable and the probability plot showed a normal
distribution. Appendix III shows the histograms of the CAR after the event. Appendix IV shows the histograms of the CAR after event without the outliers. Appendix V shows the histograms for the CAR of total event period and appendix VI shows the histograms for the cumulative returns. After the regression analyses the assumption of linearity and
homoscedasticity were checked with the Z-predictor versus the Z-residual plot.
Multicollinearity was checked with the help of the VIF-score (VIF > 5.00 = multicollinearity). All four assumptions were met for all the dependant variables.
4.1 Descriptive statistics
Table one presents the descriptive statistics of the CAR for upgrades and downgrades.
Table 1. Descriptive statistics of the CAR for upgrades and downgrades.
Mean (SD) Minimum Maximum N Upgrade -‐0.0403 (0.17) -‐0.55 0.27 33
Downgrade 0.0120 (0.13) -‐0.30 0.36 27
Note. N=60.
The average CAR for upgrades was -0,0403 (SD = 0.17) and for downgrades it was 0.0120 (SD = 0.13). The CAR is in a different direction than expected, downgrades are expected to have a downward effect and upgrades an upward effect. This will be discussed later in section 4.3.
15 4.2 Regression results
Car after event
The results of the regression analyses for the CAR after event are shown in table two. The first regression model in table two presents the full model, which is the initial model that was built to see if a sovereign credit rating change would affect domestic banks’ stock prices.
Table 2. Regression results of the influence of a credit rating change, volume change, exchange rate and magnitude change on CAR after the event.
All observations Without the outliers
Model 1 Model 2 Model 1 Model 2
B p-‐value B p-‐value B p-‐value B p-‐value Intercept -‐0.231 0.166 -‐0.055 0.434 -‐0.162 0.238 -‐0.018 0.762 Downgrade A 0.250 0.187 0.037 0.373 0.181 0.246 0.010 0.766 VPC C -‐0.195 0.016** -‐0.112 0.025** -‐0.196 0.004*** -‐0.111 0.008*** Exchange rate -‐0.551 0.550 -‐0.017 0.965 -‐0.472 0.532 -‐0.017 0.946 Magnitude change B 0.189 0.230 0.018 0.772 0.150 0.247 0.010 0.849 Downgrade* VPC C 0.138 0.176 0.139 0.101
Downgrade * Exchange rate 0.566 0.562 0.487 0.543
Downgrade * Magnitude change -‐0.198 0.249 -‐0.159 0.260
Observations 60 58
Adjusted R-‐square 0.059 0.052 0.092 0.070
F-‐stat 1.531 1.813 1.829 2.076*
*P<0.10, **P<0.05, ***P<0.01.
A: Downgrade is a dummy variable, 1= downgrade, 0 = upgrade. B: Magnitude is a dummy variable. 1= one notch, 2= two or more notches. C: VPC= volume percentages change.
The intercept in the model indicates the intercept of CAR when all variables are equal to zero. This is the normal return of the banking industry prior to the sovereign credit rating change. The coefficient downgrade has a p-value of 0.187 and is therefore not significant. This implies that downgrades do not have a significant different effect on the CAR after the event than upgrades. Furthermore the intercept is not significantly different than zero (p-value = 0.166) and therefore in combination with the insignificant coefficient downgrade, the CAR after the event is not significantly different from zero. This implies that upgrades and downgrades do not significantly affect the CAR after event. Moreover the exchange rate
(p-16 value = 0.550) and the magnitude change (p-value = 0.230) are also both not significant. This implies that the exchange rate and magnitude change do not affect the CAR after the event. However there was a significant effect of volume percentage change (p-value = 0.016). This indicates that an increase in volume percentage change leads to a significant decrease in the CAR after the event. Further, no significant interaction effects were found between
downgrade and volume percentage change (p-value = 0.176), downgrade and exchange rate (p-value = 0.562) and downgrade and magnitude change (p-value = 0.249). This indicates that the effects of volume percentage change, exchange rate and magnitude change are the same for both upgrades and downgrades.
The second model in table two presents the model without insignificant interaction variables. The significant effect of volume percentage change (p-value = 0.025) remained. The coefficients exchange rate (p-value = 0.956) and magnitude change (p-value = 0.772) and downgrade (p-value = 0.373) were still insignificant. Although the F-statistic increased, the full model was still not significant (F (4,55) = 1.813).
Next, two outliers were detected in the dependent variable of the CAR after event. These were removed and the regression analyses were performed again. In table two, model three and model four present the results of these analyses. These results are almost identical to the results with the inclusion of the outliers. The only significant effect found, though slightly stronger compared to the model with outliers, was the effect of volume percentages rate (p-value= 0.004). All other variables remained insignificant at a five percent level.
CAR prior to event
Next, the same regression was performed for the dependant variable CAR before event. The initial model was built again to see if the market expected the sovereign credit rating change before it actually occurred. All variables, including the interaction variables, were found to be insignificant at a ten percent level. This indicates that there was no significant effect of the variables included in the model on the CAR before the event. Thereafter the interaction variables were left out to see if the model improved. Again all the variables remained insignificant at a ten percent level.
17 CAR total event & Cumulative returns
Thereafter the regression was done with the CAR of the total event period as the dependent variable. Also a regression was done with the cumulative returns as the dependent variable to eliminate the market model from the return calculation. The results of these regressions are shown in table three.
Table 3. Regression results of the influence of a credit rating change, volume change, exchange rate and magnitude change on CAR and the cumulative returns.
CAR total event period Cumulative returns
Model 1 Model 2 Model 1 Model 2
B p-‐value B p-‐value B p-‐value B p-‐value Intercept -‐0.182 0.435 -‐0.027 0.467 -‐0.089 0.581 0.023 0.376 Downgrade A 0.182 0.494 0.060 0.276 0.132 0.472 -‐0.028 0.475 VPC C -‐0.275 0.016** -‐0.287 0.006*** -‐0.163 0.038** -‐0.175 0.018** Exchange rate -‐0.852 0.513 -‐0.681 0.449 Magnitude change B 0.150 0.497 0.108 0.479 Downgrade* VPC C 0.291 0.046** 0.301 0.027** 0.252 0.013** 0.259 0.008*** Downgrade * Exchange rate 0.831 0.547 0.222 0.816
Downgrade * Magnitude change -‐0.123 0.610 -‐0.136 0.418
Adjusted R-‐square 0.045 0.103 0.073 0.082
F-‐stat 1.395 3.247** 1.666 2.767**
N=60. *P<0.10, **P<0.05, ***P<0.01.
A: Downgrade is a dummy variable, 1= downgrade, 0 = upgrade. B: Magnitude is a dummy variable. 1= one notch, 2= two or more notches. C: VPC= volume percentages change.
The first regression model in table 3 shows the initial model that was built to find an effect of a sovereign credit rating change on CAR. The results of this regression are similar to the results from the table two. The coefficient downgrade is insignificant (p-value = 0.494) which indicates that downgrades and upgrades do not have a significant different effect on the CAR. Again the intercept in the model is insignificant (p-value = 0.435) which indicates that the CAR does not significantly differ from zero. This implies that upgrades and downgrades do not significantly affect the CAR. The exchange rate (p-value = 0.513) and the magnitude of the change (p-value = 0.497) were insignificant. Also the interaction variables of downgrade with exchange rate (p-value = 0.547) and downgrade with magnitude change (p-value = 0.610) were insignificant. However there was a significant effect of volume percentage
18 change (p-value = 0.016). This indicates that an increase in volume percentage change leads to a significant decrease in the CAR. Furthermore the interaction variable downgrade with volume percentage change is significant (p-value = 0.046). This indicates that volume
percentage change has different effects of downgrades and upgrades on the CAR. An increase in volume percentage change leads to large decreases in the cumulative returns for upgrades. For downgrades an increase in volume percentage change leads to a small increase.
The second model presents the results without the insignificant variables of model one. The results are similar to the results of the first model, however the F-statistic is now significant at a five percent level (F (4,55) = 3.247). Within this model the coefficient
downgrade (p-value = 0.276) and the intercept (p-value = 0.467) are still insignificant. Again this indicates that there is no different effect of an upgrade or downgrade on the CAR and the CAR does not significantly differ from zero. This implies that a downgrade or upgrade does not have a significant effect on the CAR. Furthermore the volume percentage change is significant at a one percent level (p-value = 0.006). Also the interaction variable between downgrade and volume percentage change is significant (p-value = 0.027). This indicates that the effect of volume percentage change on the CAR is different for upgrades and downgrades. An increase in volume percentage change leads to relatively large decreases in the CAR for upgrades. An increase in volume percentage change lead to a relatively small increase for downgrades.
Cumulative returns
The third model presents the results of the regression analyses for the cumulative returns. These results are similar to the results from the previous regressions. The coefficients
downgrade (p-value = 0.472), exchange rate (p-value = 0.449), magnitude change (p-value = 0.479) were all insignificant. Also the interaction variables between downgrade and exchange rate (p-value = 0.816) and between downgrade and magnitude change (p-value = 0.418) were insignificant. The coefficient volume percentage change (p-value = 0.038) and the interaction variable between downgrade and volume percentage change (p-value = 0.013) were
significant at a five percent level.
The fourth model presents the results without the insignificant variables from the third model. The results are comparable to the results of the second model in table three. The intercept (p-value = 0.467) and the coefficient downgrade (p-value = 0.276) are insignificant. This implies that downgrades and upgrades do not have significant different effects on the cumulative returns and that the cumulative returns are not affected by upgrades or
19 downgrades. The coefficient volume percentage change (p-value = 0.006) is significant. This implies that an increase in volume percentage change results in a decrease in the cumulative returns. Furthermore the interaction variable between downgrade and volume percentage change was significant (p-value = 0.027) and this indicates that volume percentage change has different effects on the cumulative returns for upgrades and downgrades. An increase in volume percentage change leads to relatively large decreases in the cumulative returns for upgrades. For downgrades an increase in volume percentage change lead to a relatively small increase.
4.3 Discussion
The results of this research show that sovereign credit rating changes do not have a direct significant impact on domestic banks’ stock price. This is not in line with previous researches such as Brooks et al (2004), Norden & Weber (2004) and Kaminsky & Schmukler (2002). Though their researches studied the effects of a sovereign credit rating change on domestic stock markets, not specifically domestic banks’ stock prices. They all found that sovereign downgrades had significant effects on stock markets but sovereign upgrades did not. Though the findings of this thesis are in line with what Gan et al (2014) found. They researched the effects of a sovereign credit rating change on domestic banks’ stock prices and they also found no significant effect.
Furthermore the exchange rate is insignificant which is in contrast to what Brooks et al (2004) and Gan et al (2014) found. Brooks et al (2004) found that a downgrade had a
significant effect on the exchange rate of the domestic currency to the dollar. Gan et al (2014) found that the exchange rate had a significant effect on the CAR. In this research the
exchange rate variable is insignificant and therefore no conclusions can be made of this. The expectation was that an appreciation of the domestic currency should positively affect CAR. This is because domestic banks could purchase relatively cheap foreign assets and create firm value (Lin, Officer & Shen, 2014). Also foreign denoted debt decreases due to an appreciation and therefore making it easier for banks to pay-off their foreign debt (Gan et al, 2014).
The magnitude change was also insignificant and it has a positive coefficient. The expectation was that the magnitude of the change would have a significant effect on the stocks return (Gu, Jones and Liu, 2014). No conclusions can be made about this variable because it was insignificant.
The volume percentage change had a negative coefficient and was significant. The expectation was that volume percentage change would move in the same direction as CAR,
20 therefore having a positive effect on CAR. This is because an increase in stock price could arise due to an increase in the volume traded, which indicates a higher demand for a stock (Gan et al, 2014). In this research the coefficient was negative which implies that an increase in the volume traded has a negative effect on CAR.
A possible explanation for these insignificant results could be that the relationship between banks and a sovereign credit rating change is not strong enough to affect banks’ stock prices directly. For instance firm specific credit rating changes have a stronger direct relationship with firms’ stock prices than sovereign credit rating changes do. Besides that it could be that banks themselves had gotten a credit rating change by a CRA before the sovereign credit rating change. This would mean that the banks’ stock prices already were affected by their own rating and therefore no significant results were found after a sovereign credit rating change. Another reason could be that other CRA’s changed a sovereign credit rating before Moody’s and therefore the effects could have taken place earlier. Furthermore if a country had already received a negative or positive outlook prior to the sovereign credit rating change the market could have anticipated the credit rating change. Therefore the effects of the actual sovereign credit rating change could be weakened.
The set up of the research could also be a reason for the insignificant results. Event studies heavily rely on the dataset used. For instance a different benchmark than the S&P500 could affect the expected normal returns and therefore also the abnormal returns. Also the chosen estimation window could affect the results. A different estimation window could give different values for alpha and beta, which in turn also affect the estimated normal returns.
21 5. Conclusion
This research investigated the effects of a sovereign credit rating change on domestic banks’ stock prices. Based on previous literature the expectation was that sovereign downgrades would have a significant downward effect and upgrades would not have a significant effect. Furthermore a larger magnitude of the change was expected to have a stronger effect on banks’ stock prices.
This research used the event study methodology to determine if a sovereign credit rating change would have an effect on domestic banks’ stock prices. To determine if there was an effect the abnormal returns were calculated. The abnormal returns were calculated in an event window of 30 days prior and after the sovereign credit rating change. These
abnormal returns were calculated by the actual return minus the expected normal return. The market model was used to estimate the expected normal returns. As a benchmark for the expected normal returns the S&P500 was used.
This research found that sovereign credit rating changes do not have a significant effect on domestic banks’ stock prices. Sovereign downgrades and upgrades do not have significant effects on the CAR or the cumulative returns. Furthermore a credit rating change of two or more notches had no significant stronger effect than a one-notch credit rating change. All interaction variables were also not significant, which indicates that there were no different effects for sovereign upgrades or downgrades by the exchange rate, volume traded and magnitude of the change. The variable exchange rate was also insignificant. The volume percentage change was significant. This coefficient was negative, which implies that a larger increase in volume traded after the event has a negative effect on CAR. Furthermore the interaction variable between downgrade and volume percentage change was significant. This implies that volume percentage change has different effects for upgrades and downgrades. For upgrades an increase in volume percentage change lead to relatively large decreases and for downgrades an increase in volume percentage change lead to relatively small increases.
This research could be improved in the future by adding outlook announcements to the dataset. Furthermore researches could look at different effects of sovereign credit rating changes for financial and non-financial firms.
22 6. Reference list
An, H.H., Chan, K.C. (2008). Credit ratings and IPO pricing. Journal of Corporate Finance, 14 (5), 584-595. Armitstead, L. (2013, February 25). Sterling hits two-year low after Moody's UK downgrade. Retrieved December 17, 2015, from http://www.telegraph.co.uk/finance/currency/9893746/Sterling-hits-two-year-low-after-Moodys-UK-downgrade.html
Japanese yen hits 27-month low against US dollar - BBC News. (2012, December 27). Retrieved December 14, 2015, from http://www.bbc.com/news/business-20849204
Boot, A. W. (2005). De toegevoegde waarde van credit ratings. Amsterdam: Universiteit van Amsterdam. Brooks, R., Faff, W. F., Hillier, D., & Hillier, J. (2004). The National Market Impact of Sovereign Rating Changes. Journal of Banking and Finance, 233-250.
Cantor, R., & Packer, F. (1996). Determinants and Impact of Sovereign Credit Ratings. Economic Policy
Review, Vol. 2, No. 2
Codogno, L., Favero, C., & Missale, A. (2003). Yield spreads on EMU government bonds. Economic policy,
Oxford University Press.
Cornell, B., W. Landsman and A. Shapiro, 1989. Cross-Sectional Regularities in the Response of Stock Prices to Bond Rating Changes. Journal of Accounting, Auditing, and Finance, 4: 460-479.
Cole, H., & Cooley, T. F. (2014). Rating Agencies. Cambridge: National Bureau Of Economic Research. Elkhoury, M. (2008). Credit Rating Agencies and Their Potential Impact on Developing Countries. United
Nations Conference on Trade and Development. Rep. Vol. 86. Geneva: United Nations
Fama, E.F. (1970), Efficient capital markets: A review of theory and empirical work”, Journal of Finance, Vol. 15.
Freitas, A. D., & Minardi, A. M. (2013). The Impact of Credit Rating Changes in Latin American. BAR. Gan, S., Ngo, L., Say, S., & Tan, R (2014). The Effects of Change in Credit Rating to the Returns of Banking Industry of Different Emerging Countries. De La Salle University.
Gu, J. , Jones, J. and Liu, P. (2014) Do Credit Rating Agencies Sacrifice Timeliness by Pursuing Rating Stability? Evidence from Equity Market Reactions to CreditWatch Events. Theoretical Economics Letters Guzman, T. (2015, January 31). Politics, Financial Fraud and the. Retrieved January 5, 2016, from http://www.globalresearch.ca/politics-financial-fraud-and-the-big-three-credit-ratings-agencies/5428603 Holthausen, R. W., & R. W. Leftwich (1986). The Effect of Bond Rating Changes on Common Stock Prices.
Journal of Financial Economics. Vol 17, 57-89.
Hooper, V., Hume, T., & Kim, S.-J. (2008). Sovereign rating changes: Do they provide new information for stock markets? Economics Systems, Vol. 32, 142-166.
Jensen, M.C. (1978). Some anomalous eveidence regarding market efficiency. Journal of Financial Economics, Vol. 6, 95-101.
Joo, S., and S. Pruitt. (2006). Corporate Bond Ratings Changes and Economic Instability: Evidence from the Korean Financial Crisis." Economics Letters. Vol. 90, 12-20.
Kaminsky, G., & Schmukler, S. L. (2002). Emerging Markets Instability: Do Sovereign Ratings Affect Country Risk and Stock Returns. World Bank Economic Review.
23
Li, H., Visaltanachoti, N., & Kesayan, P. (2004). The Effects of Credit Rating Announcements on Shares in the Swedish Stock Market. International Journal of Finance.
Lin, C., Officer, M.S., & Shen, B. (2014). Currency appreciation shocks and shareholder wealth creation in crossborder mergers and acquisitions. 27th Australasian Finance and Banking Conference 2014 Paper. MacKinlay, A. C. (1997). Event Studies in economics and finance. Journal of Economic Literature, Vol. 35. Martell, F. (2005) The Effects of Sovereign Credit Rating Changes on Emerging Stock Markets. Purdue
University.
Moody's. (2012). Rating symbols & definitions. Retrieved December 7, 2015, from https://www.moodys.com/researchdocumentcontentpage.aspx?docid=PBC_79004
Norden, L., & Weber, M. (2004). Informational efficiency of credit default swap and stock markets: The impact of credit rating announcements. Journal of Banking and Finance. Vol. 28 (11).
Peterson, P.P., 1989. Event Studies: A Review of Issues and Methodology. Quarterly Journal of Business &
Economics.
Pukthuanthong-Le, K., Elayan, F. A., & Rose, L. C. (2007). Equity and debt market responses to sovereign credit ratings announcement. Global Finance Journal.
Rijksoverheid Bijlage: Credit ratings als indicator. (2008, October 6). Retrieved December 15, 2015, from https://www.rijksoverheid.nl/documenten/kamerstukken/2008/12/15/bijlage-credit-ratings-als-indicator
SEC. (2013). Annual Report on Nationally Recognized Statistical Rating Organizations. Retrieved December 15, 2015, from http://www.sec.gov/divisions/marketreg/ratingagency/nrsroannrep1213.pdf
Stancu, I., & Minescu, A.-M. (2011). The Impact of Sovereign Credit Ratings on the Issuance of Government Bonds in Central and Eastern Europe. Theoretical and Applied Economics.
Standard and Poor's. (2011). Guide to Credit Rating Essentials. Retrieved from http://img.en25. com/Web/StandardandPoors/SP_CreditRatingsGuide.pdf
Steiner, M. and Heinke, V.G. (2001), Event Study Concerning International Bond Price Effects of Credit Rating Actions. International Journal of Finance and Economics.
24
Appendix
Appendix I
Credit rating agencies
A credit rating agency reviews the creditworthiness of countries and companies. CRA’s provide an independent evaluation about the default chances of an institution. There are three internationally recognized CRA’s namely Moody’s, Standard and Poor and Fitch. The first two have a combined market share of 80% and Fitch has around 15% (Timothy Alexander Guzman, 2015). There is a lot of discussion about the fact that there are only three dominant players in this market and that they have too much power. Though CRA’s might be very powerful they still provide the market with a system to assess the creditworthiness of companies. This information is used in financial markets because it is an easy way for investors to compare the risks of different types of investments.
Credit rating
A credit rating is an opinion of a CRA about the ability of an entity to fulfil its total debts on time. The creditworthiness of an entity depends on two variables namely the probability of default and the loss given default. The probability of default is the chance that the entity could go bankrupt and the loss given default is the expected loss if the entity would go bankrupt. Also the historic debt payments are taken into account for the credit rating. For a sovereign credit rating there are many different factors taken into account such as GDP growth, income, inflation rate, foreign debt, economic development and currency reserves. For instance, if a country lowers their tax rate it could mean that the government has less income. This could result in a smaller chance of the debt repayment and thus could affect the sovereign credit rating (Standard & Poor, 2011).
A credit rating is of importance for lenders and investors. Lenders can assess whether the chances are high that the borrowers will pay their loan back. A high rating is given when the probability is high that the borrower will pay back the loan entirely. A high credit rating results in lower interest rates for the borrower. Thus lower credit ratings result in higher interest rates, which balance out the higher chance of default.
Moody’s expresses their ratings by letters and numbers with AAA being the highest rating and C the lowest. The numbers in the rating indicate weather a rating is in the top or bottom end of a class with one being the highest and three being the lowest. In appendix VII a table is shown with the ratings and descriptions of what they mean by Moody’s.
25 Appendix II
26 Appendix IV
27 Appendix VI
