THE ROLE OF RATING AGENCIES IN SECURITISATION 1 INTRODUCTION
Diagram 4.1: The securitisation rating process
4. DATA EVALUATION IN SECURITISATION TRANSACTIONS
It would be an omission not to stress too heavily the importance of reliable and standardised data in the rating process. For most assets, the most easily securitisable portfolio consists of a homogeneous pool62, ideally with a large obligor and collateral base that generates a stable and predictable cash flow. Dommisse et al., (2004:6 - 8) describe the rating agency data requirements with respect to portfolio characteristics and portfolio performance.
62 Homogeneous pools are groups of receivables that are similar both in contractual terms and interest rates as well as in
4.1 Portfolio Characteristics
The rating agency requires a set of data that provides an understanding of the underlying pool of assets. The dataset varies depending on the asset class to be securitised, but generally needs to provide information on the following areas:
− the characteristics of individual obligors, including geographical location and creditworthiness as defined by internal or external risk rating systems;
− the structure of receivables, including original amount and term, interest rate, outstanding balance and remaining term;
− the characteristics of the underlying collateral, e.g. in the case of auto loans this would include information on the make, model and age of vehicles; and
− the obligor concentrations, e.g. the cash flow from a pool of residential mortgages with high concentrations in a specific area could be negatively impacted by an economic downturn in that area.
4.2 Asset Portfolio Performance
The rating agency also requires data to understand the asset pool’s historical performance, including delinquencies, defaults, recoveries and prepayments. Rating agencies believe a securitisation is impacted by the performance of an asset portfolio more directly than an originator would be when the assets have not been securitised. For example, a bank has access to a range of cash inflows and is therefore able to meet its liquidity needs even if expected loan payments are not received. By contrast, when the same loan pool is securitised, the cash flows from the specific loan pool are the only funds available to meet the repayment obligations to investors. Delinquencies and defaults increase the cost of securitisation to originators, as liquidity facilities and credit enhancement have to be provided for the transaction to cover potential cash shortfalls. For this reason, the trends in delinquencies and defaults have to be clearly understood.
Data provided to the rating agency should follow common definitions. Securitisation relies on international standards for the calculation of delinquency whereby, when a borrower is late on a payment, the entire outstanding principal exposure is classified as delinquent. This definition is used
as delinquencies are analysed as a leading indicator of potential defaults. A common understanding should also be achieved between originators and the rating agency on the definition of default and realised loss before recoveries. The rating agency looks for a set date beyond which a delinquent asset is classified as defaulted. This differs from the practice of many originators whereby write-offs are considered on a case-by-case basis. For example, originators may have outstanding delinquent receivables to government agencies, which, while frequently slow payers, generally do ultimately pay. Many originators, therefore, do not classify such debt as in default regardless of the length of time it has been classified as delinquent. The rating agency, however, views a portfolio in terms of its ability to make timely payments to investors in a securitisation. To size credit enhancement appropriately, the rating agency should be able to project accurately what percentage of cash flows from a portfolio may not be available, due to extended non-payment, to meet repayment obligations to investors.
Definitions aside, the format in which performance data is required for a securitisation transaction often differs from an originator’s normal reporting. Originators often track the performance of their asset portfolios based on dynamic analysis, e.g. current month delinquencies are measured as a percentage of current month total outstanding portfolio balance. Trends in these performance measurements are then tracked across time. However, the securitisation process requires static pool analysis, whereby an originator’s asset pool that is to be securitised is broken into static pools. A static pool, or vintage, is a group of assets originated during a specific calendar period, typically a month, quarter or year. Cumulative gross losses and recoveries for each pool are then tracked independently over the remaining life of the assets. Static data analysis allows the rating agency to move beyond an understanding of fluctuations in levels of credit losses over time towards a further understanding of the timing of losses within an asset’s life cycle and how the performance of assets may be changing. By analysing different vintages, the rating agency gains insight into how losses build up over the life of the assets and how the loss profile may have altered over time.
The rating agency should also be able to quantify the magnitude of ultimate losses in a portfolio caused by defaults. This necessitates static data on amounts and timing of recoveries. The rating agency needs to estimate, based on historical data, how long it takes for the originator to recover funds from defaulted assets. It uses this information to model the level and timing of cash flows a securitisation transaction can realistically expect to receive from defaulted assets to meet its obligations to investors.
Some asset classes such as residential mortgages are affected by prepayments and it is important for the rating agency to be provided with this information.
A rating agency typically expects a minimum of three to five years of historical data before it can draw meaningful default, recovery and prepayment assumptions. The goal of the data is to create a base case model of performance that can be used to forecast performance. The less data available, the more conservative the rating agency’s default probability and realised loss assumptions become, which will result in increased credit enhancement levels for the transaction.
4.3 Dynamic versus Static Analysis
A typical business relies on the cash flows from a revolving base of receivables to pay its debts. As long as the business is able to continue to generate new receivables, the corresponding cash flows can be used to cover losses caused by customers who defaulted on their obligations. A revolving pool of assets is referred to as a dynamic portfolio, and the performance of such a portfolio is tracked through dynamic analysis, whereby one current portfolio measurement is compared with another current portfolio measurement, e.g. current month defaults against current month total portfolio balance. Trends in these dynamic performance measurements are then tracked across time. However, dynamic data frequently masks certain trends owing to shifts in portfolio seasoning63 and overall size. For originators with growing portfolios, dynamic data is particularly problematic, as it tends to make the performance of an asset portfolio look better than it actually is. This because newly originated assets tend to perform better until they become more seasoned and delinquency and default trends develop (Dommisse et al., 2005:3).
By contrast, during the amortisation period of a securitisation transaction, when investors are repaid, the transaction relies on the cash flows from a static pool of assets, since no new assets are added to the pool. To size credit enhancement for the transaction appropriately, the rating agency must be able to accurately project what percentage of cash flows from a static pool of assets will be unavailable to the transaction’s repayment obligations. The risk in a securitisation transaction is the performance of a static portfolio of assets rather than that of a dynamic portfolio. A review of the originator’s assets by static vintages provides important insights into the potential performance of a securitised pool.
When preparing for a securitisation the originator’s total loan pool is divided into vintages based on the calendar period, typically a quarter, in which each loan was originated. A common understanding must be achieved between the originator and the rating agency on the definition of default. It is necessary to set a date beyond which a delinquent asset is classified as defaulted, for example, all loans
more than 150 days delinquent are classified as in default. The data reviewed should reflect this definition. Defaults are then tracked by vintage and the number of quarters after origination in which the defaults were recorded. The rating agency uses the static data relating to the historical performance of the asset pool to estimate base case levels of gross defaults and the projected timing of the defaults and recoveries (Dommisse et al., 2005:2).
An originator’s consumer loan pool can be divided into vintages based upon the calendar period in which each loan was originated. For example, as illustrated by Chart 4.1 below, the originator’s loan book is divided into 10 vintages reflecting the assets originated in each quarter, starting in the third quarter of 2001 through to the fourth quarter of 2003. The quarters in which the loans were originated are labelled consecutively as vintage A to vintage J. Vintage A thus reflects all the loans that were originated in the third quarter of 2001, being R612 000, vintage B all the loans that were originated in the fourth quarter of 2001 (R673 400), and so on till the last vintage J, during which quarter R1 569 557 in loans were originated.