Rating Transitions and Default Rates 2001-2011

Updated January 2012

Introduction

Transition matrices or credit migration matrices characterise the evolution of credit quality for issuers with the same approximate likelihood of default. They are constructed by mapping the rating history of entities rated by CI into transition probabilities. Hence, transition matrices indicate the probability of a credit rating being upgraded or downgraded, or remaining unchanged, within a specific time period.

Transition Rates for Banks

Table 1 shows the weighted average one-year transition matrix for long-term foreign currency issuer ratings by broad rating category (i.e. without the modifiers ‘+’ and ‘-’) assigned to banks by CI during the 11-year period 2001-2011.

The vertical axis shows the credit rating at the beginning of a period, the horizontal axis the rating one-year later. Credit ratings that do not change between the start and end of the year are captured along the diagonal of the matrix. Movements to the left of the diagonal indicate upgrades of foreign currency ratings and movements to the right indicate downgrades.

The matrix is diagonally dominant, meaning that most of the probability mass resides along the diagonal. In other words, banks tend to maintain their ratings on a one-year horizon rather than migrating to other ratings. For example, on average, almost 93% of banks rated ‘A’ at the start of the year were still rated ‘A’ 12 months later.

The transition matrix also shows that investment grade ratings (ie ‘AAA’ to ‘BBB’) are more stable than speculative grade ratings (‘BB’ and below) and also that upgrades exceed downgrades in most rating categories. For example, on average, 4.9% of ‘BBB’ banks were raised to an ‘A’ range rating while less than 3% were lowered to speculative grade.

Table 1 Average One-Year Transition Rates for Banks, 2001-2011

Rating One Year Later (% of issuers)

Tables 2 and 3 show the weighted average migration of banks’ long-term foreign currency ratings for three years and five years, respectively. The two matrices also indicate that higher ratings are more stable than lower ratings. However, the diagonal of the matrix becomes less strong as the horizon increases from one year to three years and from three years to five years; in other words the likelihood of a bank’s rating migrating to another rating increases over time. For example, while 92.06% of banks rated in the ‘BBB’ category retained their rating after one year (Table 1), a lower 63.00% remained in the ‘BBB’ range after five years (Table 3).

Similar to the one-year transition, a bank is more likely to be upgraded than downgraded over a three- or five-year time span, and the likelihood of a large movement in ratings is reassuringly low. For example, in the five-year transition shown in Table 3, only 1.9% of credit ratings in the ‘A’ category fell by more than one rating grade, 0.5% in the ‘BBB’ category, and 0.2% in the ‘BB’ category.

Sample size (i.e. the number of rating transitions, shown in the right-hand column) decreases as the transition horizon increases. This is because with each year that is added to the transition matrix, the number of entities with sufficient or commensurate rating history declines (see Methodology for Transition Matrices). It also reflects the declining growth rate in the number of rated banks due to broad coverage in the markets covered by CI.

Owing to the larger sample size, transition matrices for shorter time periods (one year, three years) are more accurate and reliable relative to matrices that measure transitions over long horizons.

Table 2 Average Three-Year Transition Rates for Banks, 2001-2011

Rating Three Years Later (% of issuers)

Table 3 Average Five-Year Transition Rates for Banks, 2001-2011

Rating Five Years Later (% of issuers)

Upgrade / Downgrade Rates for Banks

Table 4 shows the weighted average one-year upgrade/downgrade rates for long-term foreign currency issuer ratings by rating category (including the ‘+’ and ‘-’ modifiers) assigned to banks by CI during the 11-year period 2001-2011.

The vertical axis indicates the rating category at the beginning of the one year period, while the horizontal axis indicates the upgrade, unchanged and downgrade rates, along with the sample size. For example, 4.55% of banks rated ‘AA-‘ were downgraded on average within a one year time interval.

Upgrades dominate downgrades for all rating categories, with the exception of ‘AA’ and ‘AA-‘. In general, 20.12% out of all 6,288 rating transitions where upgrades while only 6.15% were downgrades.

The majority of ratings do not change on a one-year horizon, and investment grade ratings (i.e. ‘BBB-‘ and above) are particularly stable. As might be expected, the lowest unchanged rates are found at sub-investment grades.

Table 5 shows the upgrade/downgrade rates for the latest year, 2011. In contrast with the through-the-cycle average rates presented in Table 4, there were more downgrades than upgrades in almost all rating categories last year, particularly at investment grades. Of the 282 rated banks in 2011, 11.35% were downgraded and 6.74% upgraded, while 81.91% did not experience a change in their ratings. The same information on upgrade/downgrade rates for 2011 is presented in bar chart format in Figure 1.

Table 4 Average One-Year Upgrade/Downgrade Rates for Banks, 2001-2011

One Year Later (% of issuers)

Table 5 Upgrade/Downgrade Rates for Banks in 2011

One Year Later (% of issuers)

Figure 1 Upgrade/Downgrade Rates for Banks in 2011

The evolution of upgrades/downgrades in each of the calendar years from 2001 to 2011 is shown in Figure 2. With regard to upgrade rates, we observe an increasing trend from 2001 until 2004. Then the upgrade rate remains relatively stable until 2007 where it starts decreasing again. Downgrade rates generally follow the opposite trend.

Figure 2 Annual Time Series of Upgrade/Downgrade Rates

Default Rates for Banks

Table 6 presents the average cumulative long-term foreign-currency default rates for 2001-2011 by broad rating category, for banks rated by CI.

Default rates generally increase with the time horizon but nevertheless remain quite low. In some cases, the default rate becomes constant after a certain number of years due to the absence of any additional reported defaults. For example, the average cumulative default rate for banks rated ‘BB’ is 0.05% for the first year, it increases to 0.18% for the second year and remains at 0.18% thereafter. This is because there are no banks in CI’s dataset that have gone on to default on their financial obligations between three and 10 years after being assigned a ‘BB’ range rating. We emphasise that the reported default rates are cumulative meaning that they represent the probability of default at any time during the time period that they refer to (and not only at the end).

Default rates should increase as credit quality weakens (i.e. lower credit ratings should have higher default probabilities) but this trend is not observed in the Table. This is partly due to data limitations as the number of actual defaults is very low (just four during the 11 year period). In addition, two of the four defaults were associated with financial irregularities rather than an observable deterioration in debt-servicing capacity. Both of those banks were rated in the ‘BBB’ range prior to defaulting, hence the higher default rate for that rating category relative to other categories.

Table 6 Average Cumulative Default Rates for Banks, 2001-2011 (%)

Time Horizon (Years)

N/A= Not Applicable. There are no banks with the respective rating (vertical axis) at the beginning of the corresponding time period (horizontal axis) in the dataset.

Transition and Default Rates for Corporate Issuers and Sovereigns

Transition rates for corporate issuers (excluding banks) and sovereign governments are recorded in Tables 7, 8, 9, and 10. The tables show that higher credit ratings generally exhibit greater stability than lower ratings.

It should be noted that CI rates substantially fewer sovereigns and corporate issuers compared to banks, and there are some rating grades that have yet to be assigned to any sovereign or corporate. The small sample sizes and relatively low number of migrations from each rating category undermine the accuracy and reliability of the estimated transition rates.

Default rates are similarly affected. Among the population of corporate issuers rated by CI, only three defaults were recorded between 2001 and 2011. For rated sovereigns, no defaults were observed over the period. Consequently, cumulative default rates for corporate issuers and sovereigns are not tabulated owing to their limited usefulness.

Table 7 Average One-Year Transition Rates for Corporate Issuers, 2001-2011

Rating One Year Later (% of issuers)

Table 8 Average Three-Year Transition Rates for Corporate Issuers, 2001-2011

Rating Three Years Later (% of issuers)

Table 9 Average One-Year Transition Rates for Sovereigns, 2001-2011

Rating One Year Later (% of issuers)

Table 10 Average Three-Year Transition Rates for Sovereigns, 2001-2011

Rating Three Years Later (% of issuers)

Upgrade/Downgrade Rates for Corporate Issuers and Sovereigns

The weighted average one-year upgrade/downgrade rates for long-term foreign currency issuer ratings by rating category, including the ‘+’ and ‘-’ modifiers, assigned to corporate issuers and sovereign governments by CI during the 11-year period 2001-2011 are recorded in Tables 11 and 12 respectively.

With regard to corporates (Table 11), downgrade rates are generally higher than upgrade rates for investment grade entities while the opposite trend is observed at sub-investment grades. With regard to the overall sample, 16.60% of the 253 transitions are upgrades and 17.39% downgrades.

The average one-year upgrade rates are significantly higher than downgrade rates for each rating category of the sovereign ratings (Table12) and for the overall sample as well.

However, for both corporate and sovereigns the latest year’s upgrade/downgrade rates show that downgrades are significantly higher than upgrades. These statistics are not presented due to the limited sample size of both corporate and sovereigns in 2011.

Table 11 Average One-Year Upgrade/Downgrade Rates for Corporate Issuers, 2001-2011

One Year Later (% of issuers)

Table 12 Average One-Year Upgrade/Downgrade Rates for Sovereigns, 2001-2011

One Year Later (% of issuers)

Methodology for Transition Matrices

CI calculates transition rates by using the cohort approach. Cohorts (or pools) of rated entities with the same long-term foreign currency issuer rating are formed every six months, and the change in the ratings of the entities within each cohort tracked until the end of the chosen time horizon. Ratings that are withdrawn or suspended are excluded from the calculations.

For example, to calculate a one-year transition rate based on 11 years of rating history, we examine the change in an entity’s rating between the beginning of January in the first year and the beginning of January in the second year, as well as the change in the rating between the start of July in the first year and the start of the following July. We repeat this exercise for each 12-month period beginning January and July within the overall 11 year timeframe.

The 11 years of rating history can be divided into 21 time horizons of 12 months, starting in January and July each year. A cohort formed at the start of the 11 years can have up to 21 unique one-year transition rates; a cohort formed in the January of the last year would have just one.

The annual average transition rate is the weighted average of these unique rates, where the weight is the relative size of each cohort, i.e. the number of entities in an individual cohort divided by the number of entities in all comparable cohorts (e.g. the number of ‘BBB’ ratings in January 2001 divided by the total number of ‘BBB’ ratings measured in January and July of each year from 2001 to 2011). The denominator is referred to in the far-right column of the transition matrices above as ‘Sample Size’.

Methodology for Upgrades/Downgrades

Upgrades and downgrades are also calculated based on the cohort approach outlined above. As in the case of transition matrices, ratings that are withdrawn or suspended are excluded from the calculations. Upgrade/downgrade statistics may be considered as a generalized form of rating transitions that focus on whether a rating has been raised or lowered. In contrast the transition matrices reported above capture movements between specific rating categories.

Unlike the transition matrices methodology, the upgrade/downgrade calculations take into account the plus ‘+’ and minus ‘-‘ rating modifiers. As a result, a transition from ‘BB+’ to ‘BB’ is considered as a downgrade in the downgrade/upgrade calculation but is considered to be stable in the transition matrix calculations.

Upgrade and downgrade rates are presented in tables illustrating upgrade and downgrade rates per rating grade and overall, for specific time periods (i.e. specific years) and for specific time horizons (i.e. average annual, average two year etc) which consider various specific periods of the same length. For example, the total annual average upgrade rate is the weighted average of the 21 total annual upgrade rates calculated for the 21 annual periods in the overall 11-year sample. The weight given to each rate is the ratio of the size of the specific period (size in terms of number of transitions) over the size of all periods.

Methodology for Default Rates

As with transition matrices, cohorts of rated entities with the same credit rating (by broad category) are formed on January 1 and July 1 of each year.

We observe how many members of each cohort default in each six-month period from the time the cohort is formed. For each interval (i.e. up to six months, from seven to 12 months etc), we calculate the marginal default rate, which measures the frequency of default during each interval from cohort formation. The marginal default rate is the number of defaults divided by the size of the cohort at the beginning of each period, adjusted for any defaults or rating withdrawals in the previous period.

We then find the weighted-average marginal default rate of all cohorts belonging to the same rating category, using the number of entities in each cohort as a weight. Hence, the average marginal default rate takes into account the experience of cohorts of similarly rated entities over the same time interval from cohort formation, but in different years. For example, the one-period marginal default rate for the ‘BBB’ rating category, based on ratings data from 2001-2011, will take into account the six-month default record of the ‘BBB’ cohort formed in January 2001 and the six-month default record of the ‘BBB’ cohort formed in July 2011, and of all other ‘BBB’ cohorts formed in between those dates.

We use this information to calculate the average cumulative default rate, which measures the probability of default at any time over a given time horizon (one year, three years, five years etc).

The average cumulative probability of defaulting up to one year, for example, is equal to the weighted-average marginal default rate for 0-6 months, plus the product of the weighted-average marginal default rate for 7-12 months and the average proportion of entities in the cohort who did not default in the previous six-month period (i.e. the weighted-average marginal survival rate).

The use of average marginal default rates enables us to utilize all available rating history – for example, the calculation of the five-year average cumulative default rate is not based solely on entities that CI has rated for at least five years, rather it draws on all entities that have been rated for at least six months. An entity may also belong to more than one cohort if its credit rating migrates to other grades over time.

Limitations of Default Rates

It is important to note that the default rates referred to above and in Table 6 reflect the actual default record of entities rated by CI in the period 2001-2011. The default rates are not estimates of the future probability of default.

Data source

The dataset for this report comprises all public and private ratings assigned by CI to banks, corporate issuers and sovereigns between 1st January 2001 and 1st January 2012. The credit rating tracked is the long-term foreign currency issuer rating. The rating modifiers ‘+’ and ‘-’ are considered in the calculations of upgrade and downgrade rates but they are excluded in the calculations of transition matrices and default rates.

Definition of Default

CI considers a default to have occurred when:

1. An issuer fails to pay a material sum of principal and/or interest on a financial obligation in accordance with its terms;
2. An issuer files for bankruptcy or similar protection from creditors – unless there is reason to believe that debt service payments will continue to be made in a timely manner;
3. An issuer restructures (reorganises), reschedules, exchanges or in some other way renegotiates a debt instrument and the following apply: (a) there is an adverse change to the terms of the original debt agreement; AND (b) the renegotiation or exchange is considered by CI to be distressed or coercive.

Adverse changes to the terms of the original debt agreement may include the following:

• A reduction in the principal amount or coupon/ interest rate.
• An extension of the maturity date or loan tenor.
• A reduction in seniority or a substantial weakening of covenants.
• A cash tender for less than par.
• A decrease in the frequency of payments (e.g. to bullet from amortising).
• Swapping debt for equity or hybrid instruments.

A debt renegotiation or exchange is deemed to be distressed or coercive when one or more of the following apply:

• The issuer would, in CI’s opinion, be unable to honour its obligations under the original debt agreement due to its weak financial position.
• The issuer is unwilling to honour its obligations to those investors who choose not to participate in the renegotiations or exchange offer.
• The issuer threatens, explicitly or implicitly, to miss payments, weaken the governing indenture or to seek bankruptcy should the terms of its proposal or exchange offer not be accepted.

When an entity defaults on an instrument rated by CI, the issue rating assigned to that instrument is lowered to ‘D’.

An entity’s issuer rating is lowered to ‘SD’ (selective default) when it fails to service one or more of its financial obligations (rated or unrated), but CI believes that the default will be restricted in scope and that the entity will continue honouring other financial commitments in a timely manner. An issuer rating of ‘D’ is assigned when an entity defaults on an obligation and is expected to default on all, or nearly all, of its other financial obligations.

Reflecting the special nature of financial institutions, CI assigns an issuer rating of ‘RS’ (Regulatory Supervision) to banks that are placed under the supervision or administration of the authorities due to their weak financial condition. An ‘RS’ rating signals to investors the precarious repayment capacity of the institution and the extremely high likelihood of default absent regulatory forbearance and/or financial assistance from the state.

Attachments

• Rating Transitions and Default Rates