Rating agencies have played a prominent role during the on-going Global Crisis. In principle, agencies constantly update their sovereign credit ratings on the basis of new economic information, and changes in ratings offer investors valuable information (Masciandaro 2011). This constant updating is one of the reasons why financial markets rely on agencies (Cantor and Packer 1994). Needless to say, understanding the performance of agencies is crucial given the important economic implications of sovereign ratings. For instance, ratings strongly influence capital flows and are a main driver of sovereign bond spreads (Cantor and Packer 1996), which in turn determine the financing costs of the public sector. But despite their importance, the agencies do not provide enough details on how they determine their ratings (Mora 2006), although some recent regulatory initiatives are intended to enhance the agencies’ transparency (EU Commission 2013).
In practice, the evolution of sovereign ratings seems to be strongly asymmetric (Koopman et al. 2008), in the sense that downgrade periods tend to be shorter and deeper than upgrade periods. In other words, a country can be abruptly downgraded but it takes a long time for it to recover its rating. Moreover, the initial status may not ever be fully regained.
One explanation of this evolution could be that the information the credit ratings agencies use to update their ratings also exhibits the same different pattern during recessionary and recovery periods. An alternative explanation could be – simply – the particular way in which a rating agency incorporates its understanding of a country’s fundamental aspects and financial market conditions into its rating revision. There is no consensus as to which is the most convincing explanation for the asymmetries.
A less widely-held view implicitly maintains that rating agencies use a ‘point-in-time’ strategy (Hu et al. 2002), adapting to borrower countries’ current conditions in a manner that is constantly updated. However, most papers state that rating agencies do not adjust accurately to domestic indicators. The interpretation of this is that, having failed to predict a economic and financial downturn, agencies have reputational incentives to downgrade countries during crisis periods by more than the fundamentals would justify (Ferri et al. 1999).
Having confirmed the existence of this asymmetric pattern, our research provides an explanation based on the different reaction of ratings to domestic fundamentals during upgrade and downgrade periods (Broto and Molina 2014).
What do rating cycles look like?
As with economic variables, we can call the evolution of a country’s credit rating during consecutive downgrade and upgrade periods a ‘rating cycle’. Thus, a complete ‘cycle’ encompasses the period in which the rating goes from peak to trough and from trough to peak. Although these cycles are not directly linked to the business cycle, their duration (the number of days from peak to trough and vice versa) and amplitude (the number of notches gained or lost in each period) can be computed. As an illustration, the evolution of the sovereign rating of South Korea (see Figure 1) exhibits precisely these assumptions for the ratings path: the duration of the downgrade period is shorter than that of the upgrade period and the rating had not regained its initial level by the end of the cycle.
Figure 1. Sovereign rating of South Korea
Source: Standard and Poor's.
It is important to note that we focus on just one agency, Standard & Poor’s, in order to not mix data sources. Standard & Poor’s tends to be less dependent on other agencies and, of the three major ones, it provides the lowest and most volatile ratings (Alsakka and ap Gwilym 2010).
First, we check whether Standard & Poor’s ratings exhibit an asymmetric pattern. Table 1 shows the mean duration, amplitude and number of rating cycles for a representative set of countries (the G20 and the countries that were most distressed during the Eurozone crisis). These descriptive statistics confirm the existence of an asymmetric pattern. In those countries with at least one complete cycle, its duration is clearly asymmetric as, with very few exceptions, upgrade phases have a longer duration than downgrade ones. Amplitudes are also asymmetric, because for most countries the number of notches from peak to trough tends to be higher than from trough to peak. To put it another way, once downgraded by the agency, very few countries recover their previous status.
Table 1. Descriptive statistics of upward and downward rating phases
Source: Standard and Poor's and own calculations.
Finally, the countries with at least one complete cycle (from trough to peak and from peak to trough) are mainly emerging market economies and non-core Eurozone economies. The scarcity of upgrade and downgrade periods illustrates the persistence of ratings and their strong inertia, as once the first downgrade (or upgrade) takes place it tends to be followed by more downgrades (and vice versa). Figure 2, which represents the weighted averages of ratings for the complete sample of countries rated by Standard & Poor’s, shows that there was at least one complete cycle in most emerging market economies, whereas developed countries are on average still in the downward period.
Figure 2. GDP-weighted average sovereign rating
Source: Standard and Poor's and own calculations.
Our next step is to use panel data to analyse the main determinants of ratings during downgrade and upgrade periods, so as to provide an interpretation of the ratings’ asymmetries based on the reaction of the agency to domestic fundamentals. Our dependent variable is a non-linear transformation of the sovereign credit rating by Standard & Poor’s that allows us to consider different rating dynamics in the intermediate categories around investment grade.1
Our study suggests that previous downgrades have a strong influence in determining future rating downgrades while past upgrades do not encourage future improvements in the agency’s evaluations.2 This is in line with previous literature (see Ferri et al. 1999) which finds that for reputational reasons rating agencies may overreact once the downgrading phase starts.
To fit sovereign ratings, we also use the explanatory variables typically used in previous empirical papers that represent economic fundamentals, financial markets, domestic conditions and global shocks. Our estimates for these variables are in line with previous work on the subject.
Can authorities do anything to influence the ratings’ path?
Our study also considers the possibility of a country having a previous upgrade or downgrade and how this will affect its rating. Our results are striking. A favourable economic performance seems to be useful to soften the downgrading path.3 But once the agency starts upgrading a country, positive fundamentals do not necessarily entail an acceleration of the upgrading process.4
This result emphasises that domestic authorities are unable to speed up future upgrades by means of a good economic performance. This is bad news for those developed economies that have already lost their initial rating status during the crisis.
All in all, our findings support the view that during downgrade periods rating agencies exhibit an excess sensitivity to fundamentals, which can even exacerbate the downturn in the business cycle. During upgrade phases, authorities have little room for manoeuvre in accelerating the rating improvements, as agencies’ ratings remain steady.
Sovereign ratings seem to follow a strongly asymmetric path. Based on previous empirical work and using Standard & Poor’s ratings, we find that these asymmetries are the result of an overreaction to deteriorated economic performance during downgrade periods and an under-reaction to better economic conditions during the recovery phases.
On a more positive note, during downgrade periods, good domestic fundamentals can influence the rating agency so that national authorities are capable of smoothing out downgrades. However, the nature of upgrades is rather different. Previously downgraded countries have little ability to accelerate future upgrades through enhanced economic fundamental performance.
These results are useful for drawing some lessons about what the rating cycle will be like in those advanced countries that have been downgraded during the Global Crisis.
Disclaimer: The views in this column are the authors’ and should not be reported as those of the Bank of Spain or any official organisation affiliated with the Eurozone.
Alsakka and ap Gwilym (2010), “Leads and lags in sovereign credit ratings”, Journal of Banking & Finance 34: 2614-2626.
Broto, C, and L Molina (2014), “Sovereign ratings and their asymmetric response to fundamentals”, Banco de España, Working Paper 1428.
Cantor, R, F Packer (1994), “The credit rating industry”, Federal Reserve Bank of New York, Quarterly Review Summer-Fall, 1-26.
Cantor, R, F Packer (1996), “Determinants and impact of sovereign credit ratings”, Federal Reserve Bank of New York Economic Policy Review, October, 1-15.
Hu, YT, R Kiesel, W Perraudin (2002), “The estimation of transition matrices for sovereign credit ratings”, Journal of Banking & Finance 26: 1383-1406.
Koopman, SJ, R Krussl, A Lucas, AB Monteiro (2009), “Credit cycles and macro fundamentals”, Journal of Empirical Finance 16, 42-54.
Mora, N (2006), “Sovereign credit ratings: Guilty beyond reasonable doubt?”, Journal of Banking & Finance 30, 2041-2062.
Masciandaro, D (2011), “What If Credit Rating Agencies Were Downgraded? Ratings, Sovereign Debt and Financial Market Volatility”, SSRN Electronic Journal 09/2011, DOI: 10.2139/ssrn.1924859.
1 We transform the rating scale by means of a log function, which has a higher slope in the intermediate categories. For instance, this feature allows us to identify the difference between the impact of losing one notch from AAA to AA+ and the impact of losing the investment grade category (from BBB- to BB+) which triggers huge capital outflows as many pension funds limit the percentage of their portfolio that could be invested in the so called junk bond categories.
2 Our main variables of interest in this analysis are two dummy variables. The first one takes the value 1 if the country has been downgraded in the previous quarter and 0 otherwise, whereas the second dummy is 1 if the country has been upgraded in the previous quarter and 0 otherwise. The coefficient of the dummy for previous downgrades is negative and significant whereas that of previous upgrades is non-significant.
3 Our estimates also consider the interactions of the domestic variables with the two dummies that represent the existence of a previous upgrade or downgrade as rating determinants. The estimated coefficients of these interactions can be interpreted as the capacity of domestic fundamentals to influence the downgrading and upgrading path by smoothing or amplifying both phases. Our results are striking. First, the interactions of the domestic variables with the dummy that represents the existence of previous downgrades are significant for several cases (namely, GDP growth, GDP per capita and the reserves-to-GDP ratio).
4 In contrast, the interactions with the dummy that corresponds to previous upgrades are barely significant.