DBRS Announces New SplitShare Rating Methodology

DBRS has announced that it:

has today published updated versions of two Canadian structured finance methodologies:
— Stability Ratings for Canadian Structured Income Funds
— Rating Canadian Split Share Companies and Trusts

Neither of the methodology updates resulted in any meaningful changes and as such, neither publication has resulted in any rating changes or rating actions.

DBRS’s criteria and methodologies are publicly available on its website, www.dbrs.com, under Methodologies. DBRS’s rating definitions and the terms of use of such ratings are available at www.dbrs.com.

Of interest in the methodology is the explicit nature of their rating categories (I have added the Asset Coverage Ratio, calculated from the Downside Protection):

Minimum Downside Protection Criteria by Rating Category
DBRS Preferred Share Rating Minimum Downside Protection*
(Net of Agents’ Fees and Offering Expenses)
Pfd-2 (high) 57% 2.3+:1
Pfd-2 50% 2.0:1
Pfd-2 (low) 44% 1.8-:1
Pfd-3 (high) 38% 1.6+:1
Pfd-3 33% 1.5-:1
Pfd-3 (low) 29% 1.4+:1
* Downside protection = percentage reduction in portfolio NAV before preferred shares are in a loss position.


Downside Protection Adjustments for Portfolio Diversifi cation
Level of Diversification Adjustment to Minimum Downside
Protection Level (Multiple)
Strong by industry and by number of securities 1.0x (i.e., no change)
Adequate by industry and by number of securities 1.0x to 1.2x
Adequate by number of securities, one industry 1.2x to 1.3x
Single entity 1.3x to 1.5x

Also noteworthy is:

The importance of credit quality in a portfolio increases as the diversification of the portfolio decreases. To be included as a single name in a split share portfolio, a company should be diversified in its business operations by product and by geography. The rating on preferred shares with exposure to single-name portfolios will generally not exceed the rating on the preferred shares of the underlying company since the downside protection is dependent entirely on the value of the common shares of that company.

They are, quite reasonably, unimpressed by call writing strategies:

DBRS views the strategy of writing covered calls as an additional element of risk for preferred shareholders because of the potential to give up unrealized capital gains that would increase the downside protection available to cover future portfolio losses. Furthermore, an option-writing strategy relies on the ability of the investment manager. The investment manager has a large amount of discretion to implement its desired strategy, and the resulting trading activity is not monitored as easily as the performance of a static portfolio. Relying partially on the ability of the investment manager rather than the strength of a split share structure is a negative rating factor.

They even have a table for the effect of cash grind (which is a special case of Sequence of Return Risk):

Impact of Capital Share Distributions on Initial Ratings
Size of Regular Capital Distributions (see note) NAV Test Likely Impact on Initial Rating
Excess income None None
5% or less per annum 1.75x coverage 0-1 notches lower
5% or less per annum 1.5x coverage 1 notch lower
8% per annum 1.75x coverage 1-2 notches lower
8% per annum 1.5x coverage 2 notches lower
The likely impact on ratings for these distribution sizes assumes a typical split share structure (preferred shares $10 each, capital shares $15 each). If a structure were to differ from this assumption significantly, the likely impact on the preferred share rating will not match what is shown in the table.

I consider their VaR methodology highly suspect:

The steps in the VaR analysis completed by DBRS are as follows:
(1) Gather daily historical performance data for a defined period.
(2) Annualize each daily return by multiplying it by the square root of the number of trading days in a year.
(3) Sort the annualized returns from lowest to highest.
(4) Using the initial amount of downside protection available to the preferred shares, determine the appropriate dollar loss required for the preferred shares to be in a loss position (i.e., asset coverage ratio is less than 1.0)
(5) Solve for the probability that will yield a one-year VaR at the appropriate dollar-loss amount for the transaction.
(6) Determine the implied long-term bond rating by comparing the probability of default with the DBRS corporate cumulative default probability table.
(7) Link the implied bond rating to the appropriate preferred share rating using an assumption that the preferred shares of a company should be rated two notches below the company’s issuer rating.

As stated, it’s nonsensical. Whatever one’s views on long-term mean reversion of equity returns, there is definitely short-term mean reversion, so annualizing a single day’s return is far too pessimistic. Using the square root of the days in the year to annualize the results implies that each day’s returns are independent.

There’s a big table titled “Maximum Preferred Share Ratings Based on Portfolio Credit Quality and Correlation”, which I won’t reproduce here simply because it’s too big.

I am not a big fan of this “base case plus adjustments” methodology and (not surprisingly) continue to prefer my own stochastic model, which is used in every edition of PrefLetter. Implications of my methodology have been discussed in my articles It’s all about Sequence and Split Share Credit Quality.

Leave a Reply

You must be logged in to post a comment.