6. Understanding the tests

The previous chapter describes how waterfall rules ensure that senior debt tranches get priority of payment over more junior tranches. The other way that CLOs protect debt holders in rising order of seniority is through tests.

If the test passes, the CLO’s rules remain unchanged for the next period. However, if a test fails, then the rules of the CLO change in some way – usually to the benefit of debt investors and at the expense of equity investors.

There are, broadly, two types of test: those where failure results in the early paydown of some liabilities, and those where, if the test is tripped, the CLO manager is prevented from carrying out actions that make the test failure any worse than it currently is (these are often called maintain-or-improve tests).

The OC test passes if the relevant OC ratio for each class of notes is above its original pre-determined level. The OC ratio is the par value of the CLO’s collateral (after making some important adjustments) divided by the outstanding amount of that particular class of notes together with that of all the classes of notes senior to it in the capital structure.

The par value of the collateral is simply the par value of each of the loans or other assets owned by the CLO, including cash (plus the par value of any assets which the CLO has recently bought but which have not yet legally settled).

But this collateral par value is then adjusted to take account of poorly performing assets (the “haircut” in the real example, see right). One important adjustment is that defaulted assets are valued at less than par (see par adjustments below for more details of how they are valued). This means that a default reduces the par value of the portfolio for the purpose of the OC test. If there are too many defaults, the CLO will start to fail one of its OC tests.

The pass levels (or required ratios) for each OC test are set such that the most junior OC test will fail first. If defaults rise further, the next most senior OC test will fail and so on.

The effect of failing an OC test is that the CLO must stop using interest proceeds to pay equity distributions and interest on all liabilities that are junior to that particular tranche.

In other words, if the most junior OC test (let’s say, the test for the triple B class D notes) fails, the CLO must stop making distributions to equity investors, must stop paying junior management fees, must stop paying interest on the double B class E notes, and must use the funds to pay back the principal of whatever is the most senior class of outstanding notes at that time.

All being well, this diversion of cash will boost the OC ratio up to a point where the test is cured and the CLO can start to make equity distributions again. (Paying down debt should improve the ratio because it reduces the denominator in the OC ratio.)

However, if there are further defaults, the class C OC test will also start to fail. When that happens, payments will be suspended to the class D notes as well as to class E and equity. If the class B test also fails, then the CLO will stop paying interest on the class C notes, and so on. In this way, the series of tests gradually cuts off payments to the liabilities in rising order of seniority, preserving value for investors higher up the capital structure.

As well as diverting the interest received by the CLO, the failure of an OC test also has a parallel effect on the principal waterfall. If the CLO receives any principal payments, these must be used to pay down debt rather than to buy new assets.

In most post-crisis CLOs, there are OC tests for each class of notes with two main exceptions. US CLOs that include single B rated tranches generally do not have an OC test for this class. At the top of the capital structure, there is a combined test for the two most senior notes (triple A and double A). In other words, there is no scenario in which the CLO is allowed to skip payments on either the double A or triple A notes.

In addition to the OC tests, many CLOs include a test which triggers earlier than any of these OC tests. This is generally known as the interest diversion test or the reinvestment overcollateralisation test. If this test fails, distributions to equity investors are suspended. However, instead of paying down debt, the CLO uses diverted interest to invest in new assets. This also boosts the OC ratio, but this time by increasing the numerator in the ratio rather than decreasing the denominator.

The definition of default varies from deal to deal but typically covers the impact of events such as:

• Failing to pay interest or principal on the actual loan owned by the CLO
• Failing to pay interest or principal on debt that is pari passu or senior to the loan owned by the CLO
• Bankruptcy of the borrower
• Downgrade to D or similar of the loan owned by the CLO

If a loan or other asset is deemed to be in default, it is treated as being worth something less than par for the purpose of calculating the OC ratio. Again, the rules on this point vary from deal to deal. Typically, defaulted loans are treated as being worth either their market value or an assumed rating agency recovery rate, whichever is lower. (Rating agencies have standard assumptions for what assets will be worth post-default, based on the type of debt and its seniority.)

Imagine that a CLO has $500 million in unadjusted par value. One of its loans, which happens to be exactly 1% of the portfolio (or $5 million) is in default. Its market value is 80% of par, but its assumed rating agency recovery rate is 40%.

The par value of the portfolio adjusted to calculate the OC ratio will then be $495 million plus 40% of $5 million. In other words the haircut, or adjustment, will be a reduction of $3 million, giving an adjusted par value of $497 million.

This figure of $497 million will be used as the numerator for each OC ratio and is divided by the balance of the class of notes for which the OC test applies plus all tranches senior to it.

As a result, the concept of a triple C bucket has now become standard. The idea is that assets rated triple C or lower are treated somewhat like defaulted assets for calculating the OC tests. In practice, things are a little more complicated.

In most CLOs, the portfolio administrator is required each month to add up the par amount of assets that fall into the triple C ratings bracket and work out what percentage they make up of the total portfolio in par terms. If the proportion is greater than a predefined threshold such as 7.5%, then the portfolio is adjusted by a number which is the difference between par and the market value of those assets which are in excess of the threshold. (To be clear, the administrator assumes that it is the assets with lowest market values that spill out of the bucket rather than the most valuable ones.)

In most CLOs, “triple C” for this purpose is defined as having a rating of Caa1 (from Moody’s), CCC+ (from Fitch or Standard & Poor’s) or lower. However, some deals also include ratings one notch just above that level (that is, B3 or B-) if the rating agency in question has put that credit rating on watch for a potential downgrade.

Similarly, if a manager sells at 100 (or receives a principal repayment at par) and uses the proceeds to buy assets below 100, the par value will go up. This practice of buying below par to boost a CLO’s OC ratio is known as par-building. In the aftermath of the financial crisis of 2007, many CLO managers bought assets at prices well below par. This process is credited with having boosted CLO performance since many of these underpriced assets recovered quickly in price.

However, earlier in the history of the CLO market, par-building received a lot of bad publicity. In the wake of the spike of defaults in the early years of the millennium, debt investors accused managers of taking on unnecessary risk to boost their CLO par values. As a result, rating agencies encouraged market participants to introduce new rules which would penalise CLOs in ratings terms if they engaged in excessive par building.

These rules, known as the discount purchase rules, remain common in CLOs today. They add a further complication to the way par values and OC ratios are calculated.

Broadly, the rules say that if a CLO manager buys an asset for less than 85 cents in the dollar, that asset should be valued not at par but at its purchase price for the purpose of calculating OC ratios.

These rules were reassessed in the wake of the sharp fall in loan prices in 2008. At one point, most loans were trading below 85 cents in the dollar, making it hard for managers to buy any assets. So, the rules were tweaked again. Some deals allow a manager to own what are referred to as swapped discount purchase obligations up to a set percentage of the CLO portfolio. This allows the manager to treat a discounted asset at par if it replaces another asset which is sold for a similar price.

Other CLOs define discounted purchases by reference to an index, such as the S&P/LSTA Leveraged Loan index.

Both of these approaches attempt to tweak the original discount purchase rules for the previously inconceivable situation where most loans are trading at a significant discount to par.

As with the OC test, there is an IC test at numerous levels in the capital structure (usually from the double A notes down to the triple Bs). The numerator is the interest proceeds available at that point in the waterfall. The denominator is the interest payment due on the next debt tranche.