Commercial Mortgage Metrics Powered by CBRE/Torto Wheaton Research

Moody’s Investors Service

Frequently Asked Questions I. Overview What is CMM? CMM is a quantitative tool for assessing credit risk in commercial mortgage loans and portfolios. CMM utilizes loan level inputs (e.g., lease structure, NOI, DSCR, LTV) and property market forecasts to generate a sophisticated array of credit risk metrics.

How does CMM help in managing commercial mortgage portfolios? Outputs from CMM can be used by clients to more efficiently size capital allocations, price new loans, and maintain an early warning system for potential defaulters.

What are the deliverables and time frames?

Topics Of Discussion: ■■

Overview

■■

Analytics

■■

Relationship of CMM to other products of Moody’s

The web based version of CMM is now available to clients. Its functionality includes analyzing single loans from all of the asset classes (multifamily, retail, industrial, office and hotel) and markets (list available) currently covered by Torto Wheaton Research (TWR).

and TWR ■■

Inputs

and stratify large portfolios of loans and have expanded market coverage.

■■

Outputs

How Is CMM priced?

■■

Demonstration/Performance

■■

Security, Architecture,

Enhancements expected to be delivered during Summer 2004 will enable CMM to analyze

A number of considerations are taken into account in CMM pricing including usage by department, the number of asset classes and markets selected, and customized set-up

and Data Storage

or integration within other systems. Please contact your CMM sales representative for more details.

■■

Contact Info

credit ratings

■

company analysis

■

opinion

■

data

■

market perspective

■

credit ratings

■

company analysis

II. Analytics How does CMM differ from other real estate risk analytics? CMM has an enormous advantage in being able to provide analytics that can be customized to the specific property backing a loan. The advantage comes from a bottom-up - as opposed to a top-down - approach. CMM uses forecast rents to build up NOI and value. Some other approaches, by contrast, start by forecasting NOI and value directly. In CMM, a significant portion of each loan’s risk profile is derived from its market’s underlying economy. Models that reduce this to a small number of scenarios, a base case and a recession scenario for example, distort the distribution of default risk. CMM generates an entire probability distribution, incorporating hundreds of possible recession and growth scenarios. The difference is between having just a few points on a probability distribution curve, without even knowing which points they are, and CMM providing you with the entire curve. CMM also recognizes that future risk can diverge from historic risk depending on many factors. Most importantly, the amount of risk can depend on a market’s current position in the credit cycle. Some other approaches address risk by referring to a historic standard deviation which remains constant throughout the cycle.

In CMM, what triggers a default? Is it debt service coverage ratio (DSCR), loan to value ratio (LTV) or a combination of the two? CMM has two default triggers that can be based on DSCR and/or LTV. The client can choose to use their own settings or the CMM recommended settings in which the parameters for risk of default during the loan term are set entirely on DSCR. This means that the “trigger” for probability of default (PD) is entirely based on potential reductions in collateral income - if the income on the property declines, the probability of default increases as the income approaches or falls below the debt service payments.

What is “conditional default probability”? CMM derives its estimates of probability of default from the joint effect of two different probabilities. The first is the probability that in some future year the condition of the real estate market and collateral results in a certain level of income and value, or a certain future DSCR and LTV. CMM uses the market and collateral forecasts and the distribution around those forecasts to determine this probability. The second probability is the default probability if a given DSCR and LTV occurs (i.e., the likelihood that a loan, its DSCR having fallen to 0.90, defaults at that point). The total probability of default is a joint probability or the combination of the two probabilities. Since the default probability depends on the distribution of possible collateral income and value it is called a “conditional” probability, i.e., conditional on the outcome of the other distribution.

What parameter settings for PD should I use? Certain parameters (or coefficients) in CMM drive the estimates of default probability at all points across the range of possible real estate market and collateral forecasts. These parameters dictate the impact of DSCR and LTV on the conditional default probability. However, in some cases the characteristics of a client’s underwriting criteria or portfolio might not match CMM’s initial setting on these parameters. In order to accommodate this, the relationship between loan conditions (DSCR and LTV) and probabilities of default can be adjusted by CMM clients. For example, in general a loan with a DSCR of 0.90 is more likely to default than one with a DSCR of 1.00. However, not all loans with a

2 to “carry” a loan, and some borrowers might be more willing to do DSCR of 0.90 will default. A better-capitalized borrower might be able so if the loan is in a market that is only temporarily under stress and where recovery is in sight. So CMM clients can refine the relationship between loan parameters and default probability to reflect their experience with borrowers and properties. Please contact Torto Wheaton Research (see back page) for a technical paper that provides a detailed explanation of the estimation process for the initial parameters.

How is loss given default (LGD) determined? LGD is primarily determined by changes in collateral value and LTV. In the event of default the loss on a loan is determined by the difference between the current balance of the loan and the current value of the asset, less workout costs and lost interest. CMM clients can stipulate workout costs, as well as a minimum LGD.

2 credit ratings

■

company analysis

■

opinion

■

data

■

market perspective

■

credit ratings

■

com

s

■

opinion

■

data

■

market perspective

■

credit ratings

■

company analysis

■

opinion

■

data

■

market perspe

How is balloon refinance risk determined? Balloon refinance risk is analyzed on the basis of how a “take out” lender would view a loan’s risk at maturity. However, a different conditional default function is used in order to recognize that the nature of a balloon default differs from that of a term default. For example, while a DSCR of 1.05 might not trigger a default during the term, it might trigger a balloon default as it falls below a threshold DSCR of 1.30 that many lenders use. However, the LGD in such a case would likely be minimal and would reflect the property’s value at balloon maturity.

How does CMM take interest rate forecasts into account? CMM uses an interest rate scenario approach to drive its cap rate and floating rate risk components. Simulations can be run on interest rate paths provided by TWR or specified by the client. The TWR interest rate scenario is used in the base case version of CMM and is consistent with the Torto Wheaton Research real estate fundamentals forecasts. Alternatively, clients have the flexibility to express their own interest rate views, as well as the ability to run multiple interest rate scenarios. This approach allows clients to run CMM with the same interest rate scenarios they use to assess credit risk for other fixed income sectors, thus comparing sensitivity to interest rate movements on an apples-to-apples basis.

If we use our own interest rate forecast does it affect CMM’s cap rate forecast? Yes. Alternative interest rate forecasts will affect CMM’s pricing of real estate cash flows.

What standard deviation is used for forecast variables? In CMM the standard deviation of each forecast variable is derived from the econometrics of the forecasting models. The forecasting equations incorporate the uncertainty of the entire system of econometric equations into a confidence band for each variable the model forecasts. These confidence bands measure the impact of all aspects of uncertainty, including: uncertainty of the local economic climate, volatility of the local real estate market, the current state of the local real estate market (relative to the cycle), measurement errors in the data, and the ability to forecast the market. Thus, the standard deviations are different for each market and property type, for each year, and for each critical variable driving the collateral valuation (market rent and market vacancy). In other words CMM’s measurement of future risk varies by location, by time period, and by property type.

What are “cones” and how are they generated? “Cones” are graphic representations of the distribution of possible outcomes around the forecast NOI and value scenarios that are used by CMM. Since a time-series graph of the distributions looks like cones, we call them by that name. They generally widen out as they go further into the future (i.e., next year can be predicted with greater accuracy than 10 years ahead). The cones are also wider in markets that are more volatile or that are subject to greater uncertainty. The shape of the distributions that the cones represent is well based in statistics and econometrics.

Has CMM been back-tested? Back-testing is an ongoing process. Moody’s and TWR are presently conducting research on defaulted CMBS loans in order to continue calibrating CMM’s conditional default functions. Additional CMM back-testing will be conducted on a series of seasoned CMBS loan pools, utilizing TWR’s forecasting models for the related time periods. TWR’s track record serves as a back-test for the forecast component of CMM. Please contact your CMM sales representative for copies of the latest test results.

What is TWR’s track record in forecasting real estate market performance? TWR has been providing real estate market forecasts for 20 years with a high degree of accuracy. Research documenting their track record is available at www.tortowheatonresearch.com.

mpany analysis

3 ■

opinion

■

data

■

market perspective

■

credit ratings

■

company analysis

■

opin

credit ratings

■

company analysis

■

opinion

■

data

■

market perspective

■

credit ratings

■

company analysis

III. Relationship Of CMM To Other Products Of Moody’s And TWR How will TWR’s ongoing quarterly forecasts affect CMM’s risk metrics? Updated data is constantly coming in, and TWR continually incorporates this new data into its forecasts and will keep CMM clients fully informed of material changes (see next question). In addition, sometimes fundamental changes in data occur that have the potential to alter the outlook of the future. Some alterations are unusual (for example, the translation from the SIC system to the NAICS system for classifying industries), and some occur on a scheduled basis, but have an uncertain effect (such as the annual re-benchmarking of the employment data). Incorporating new data into market forecasts and examining the effect of underlying changes in the nature of that data is an on-going effort - hence TWR’s commitment to quarterly forecast updates.

How will CMM clients be advised of material changes in economic data and forecasts? Two reports will be made available to CMM clients on a quarterly basis to highlight significant changes. The “Default Risk Index Flash” will highlight markets and asset classes in which TWR’s forecasts have changed enough relative to previous forecasts to significantly alter the view of loan default risk. The “Overview and Outlook Property Market Report,” compiles TWR’s quarterly reports by property market and by asset class.

How does Moody’s CMM compare with Moody’s CMBS rating approach? Moody’s CMM and Moody’s CMBS rating approach share analytical underpinnings, although there are several important distinctions. Moody’s CMM uses a dynamic approach that models a property’s income over time and tests it against a contemporaneous default function. Moody’s CMBS approach, while also incorporating a forward-looking view of risk, utilizes a static pro-forma of property income at the time of issuance. The CMBS default function is based on a loan’s initial DSCR, taking into account both actual and stressed interest rates, and is intended to address risks during the term and balloon from all potential causes. Real estate market outlooks and risks are addressed in part by adjusting a property’s pro-forma cash flow to sustainable levels and in part by credit support at the bond level. Both approaches are designed to be flexible, allowing analyst input to reflect subjective judgments as to the quality of a property, its borrower and tenants. Moody’s CMM and CMBS approaches recognize that real estate credit analysis is a combination of art and science. Moody’s expects that over time the two approaches will converge, and will use CMM side-by-side with its CMBS approach for monitoring and rating commercial mortgage backed securities.

How does Moody’s CMM compare with Moody’s KMV? Are the products integrated? Moody’s CMM and Moody’s KMV produce similar credit risk metrics, but do so utilizing different analytical frameworks. Credit risk in Moody’s CMM is determined through econometric forecasting, while in Moody’s KMV it is derived from market observables such as stock prices. At present the systems operate independently, but Moody’s is exploring ways to integrate them.

IV. Inputs How much input is needed to get a conclusion? The inputs should be as thorough as possible to get the most robust credit analysis. However, CMM can supply clients with market information from the TWR data base for many of the input items if property specific data is not otherwise available. TWR can provide guidance on which data items have the most impact.

Our portfolio has thousands of loans. How can we efficiently analyze them in CMM using our current information systems? CMM incorporates a function in which information on loans and collateral can be imported. You may be able to use your loan servicing or other systems to obtain the data for the import file. In the current version of CMM individual loans or small groups can be analyzed at a time. During Summer 2004 CMM is expected to include functionality that will allow for large portfolios to be imported and analyzed in batch processing mode and provide a sophisticated reporting package.

4 credit ratings

■

company analysis

■

opinion

■

data

■

market perspective

■

credit ratings

■

com

s

■

opinion

■

data

■

market perspective

■

credit ratings

■

company analysis

■

opinion

■

data

■

market perspe

Does CMM handle all types of commercial loans? CMM is appropriate for most loans that are primarily secured by commercial real estate. Loans to corporations for operations (to a hotel operating company, for example) would not be appropriate for CMM.

Can CMM handle construction loans? Yes. CMM directly models construction loan risk related to lease up and market conditions upon completion. Certain areas of risk, such as construction delays or cost overruns are not directly measured in the model, but clients can run scenarios to test a loan’s sensitivity to these risks.

Can CMM model blanket loans? An enhancement to CMM expected to be delivered during Summer 2004 will enable it to handle blanket loans in which one loan is secured by multiple properties, even when the properties are in different cities. CMM will model the performance of the properties in aggregate, with the overall risk influenced by each property’s risk and by the correlation among the properties by market and asset class.

Are multiple loans on a single property supported? CMM supports multiple loans on a given asset. CMM models the default risk of all loans on a single property as a unit, as the event risk for all loans is tied to the performance of the most junior loan. CMM apportions losses according to the specified loan’s position in the hierarchy, impacting each loan’s recovery rate.

Some loans in my portfolio fall outside of current TWR market coverage. Can CMM handle loans in smaller markets? TWR is constantly expanding its coverage of small markets. There are also several alternative strategies. TWR has developed a model that allows clients to develop forecasts for smaller markets by providing a few inputs. Clients may also choose to run non-covered markets using the national average profile for a given asset class. Small market forecasts are currently under development that will extend coverage to all MSA’s, but they will be less robust than the forecasts for TWR covered markets given small market data limitations.

Does CMM factor in borrower quality? Certain tangible aspects of borrower quality may be directly modeled in CMM, such as letters of credit and reserves. Less tangible measures of borrower quality can be addressed in several ways. For example, clients can use a matrix approach to assess the credit quality of a loan, in which CMM results are one element and the client’s other risk assessments form the rest of the matrix. Many lenders use credit scores for various aspects of a loan, such as the current performance of the collateral, the borrower’s credit history, and the borrower’s other assets. The scores on these and other characteristics are combined in a matrix to yield an overall loan rating. Clients can continue to use the matrix approach, but substitute CMM results for those loan characteristics that reflect real estate market based risk.

Does CMM factor in tenant quality? Tenant quality is not directly modeled in the current version of CMM, but is a high priority for future versions. CMM focuses on real estate market risk, with the initial setting having tenant leases remain in effect until maturity. However, clients can modify the analysis to quantify the impact of tenant default risk. For example, an analysis can be performed in which tenants leases terminate earlier than the contractual dates. If the resulting early exposure to the leasing market significantly affects a loans default risk, this information can be useful in the rating of the loan.

Will we have the ability to use our own market forecast scenarios? Yes. Many institutions have “house” views on the largest markets in which they operate, and in some cases these views may differ from the TWR forecasts. Torto Wheaton Research can work with CMM clients to create scenarios that incorporate their house views. Additionally, the resulting credit metrics can be contrasted with those obtained using the TWR view.

mpany analysis

5 ■

opinion

■

data

■

market perspective

■

credit ratings

■

company analysis

■

opin

credit ratings

■

company analysis

■

opinion

■

data

■

market perspective

■

credit ratings

■

company analysis

If a property’s current income is temporarily high or low, is there flexibility to change the cash flow for a given time? If a property has an unusually high or low current income relative to its value, CMM can handle it in one of two ways. First, CMM can be run with sufficient data to incorporate the reason behind the income/value disparity, say an unusually high current vacancy. Second, if such data is not available CMM will start the property income cones from a value within a band around the market average cap rate over time, effectively “stabilizing” current income

How do I model other property types, such as nursing homes, or mixed-use properties? Customized modeling is available. Please contact TWR to discuss.

V. Outputs Does this system give the user the ability to generate reports on loan and portfolio risk? Yes. This attribute is a significant benefit of CMM. With all data centralized, the creation of reports is straightforward.

Can I document performance over time? Yes, a centralized database is maintained that includes the history of the analysis. There are several important implications of this. For example, CMM allows you to: ■■

Look at how a loan was analyzed in the past, so you can compare CMM’s output to actual outcomes.

■■

Access previous analyses to provide easier input of updated information for a current analysis.

■■

Rerun previous analyses with current inputs and current forecasts.

■■

Audit the analysis process over time in a manner that is readily available for viewing.

Can I have an audit trail of changes made to the default inputs and parameter files? Yes. This audit trail is available in a centralized database. The database is developed in such a way as to allow easy reporting on these changes.

VI. Demonstration/Performance How can I see firsthand how the system works? On-line demonstrations of CMM utilizing Microsoft Live Meeting are available by appointment. The demonstration is hosted by a TWR economist, live, on the phone. The demonstration, including questions and answers, normally lasts about 2 hours.

Can you run a sample portfolio through CMM so we can see the output? Yes, a demonstration can be run on your loans. We recommend a consultation beforehand to ensure that the information you provide matches the requirements of the system. We provide specifications for data delivery in our loan and collateral import templates.

Is CMM “Basel compliant”? The standards for banks to be Basel compliant are still evolving. Moody’s and TWR intend to update CMM over time to take into account evolving regulatory considerations. However, the core principals of Basel call for banks to have a robust credit approach with which to assess capital adequacy and a sophisticated reporting and monitoring system, both of which are features of CMM. Only banks themselves can be Basel compliant, but CMM can be an important part of the process.

Can we demo the product? A working demo can be arranged when it is a helpful part of the evaluation process. However, we encourage you to first make use of the on-line demonstration as the product is complex.

6 credit ratings

■

company analysis

■

opinion

■

data

■

market perspective

■

credit ratings

■

com

s

■

opinion

■

data

■

market perspective

■

credit ratings

■

company analysis

■

opinion

■

data

■

market perspe

What model enhancements and upgrades do you envision making? We expect to enhance CMM on an ongoing basis. Moody’s and TWR will continually calibrate and refine CMM by having its analytics reflect commercial mortgage default research and other relevant data and quantitative tools that become available. In addition, we plan to add market and asset class coverage to CMM to the extent possible. We intend to be responsive to the needs of clients and encourage their input as to enhancements.

Can we compare our results with those of our peers? Moody’s and TWR intend to develop life company and bank data bases for peer group analysis. Data remains confidential and the property of the client. Clients that agree to contribute data would receive the benefit of a peer group analysis in which all loans are analyzed with CMM as a common credit benchmark.

For institutions with limited staffing and resources, what alternatives are there to licensing CMM? Moody’s offers analysis of portfolios using either or both of the CMM and CMBS approaches. Results and interpretation can be provided for the portfolio as a whole and for individual large exposures. Moody’s can also provide indications of potential CMBS capital structures for loans or groups of loans as if they were securitized.

How much support can I expect? What is considered additional? TWR will assist clients in setting up their operating environment with user ID, user permissions and parameter settings. TWR will also provide training for your current employees when you subscribe to CMM. Additional training can be arranged if desired. CMM is a sophisticated quantitative tool for which users will need occasional assistance. A support line will be available from 9 am to 6 pm EST. During this time support is provided as part of the subscription price to registered users. TWR is available for consulting support to develop specialized forecasting models or to extend current models to unique property types. Moody’s is available to assist clients with interpreting CMM results.

VII. Security, Architecture And Data Storage What is the basic architecture of the system? CMM is an enterprise program housed on TWR servers (or behind your in-house firewall, if desired). Users access the system with a browser from any computer (except in the case of the in-house version). Client data stored on the TWR system is secure and segregated from that of other clients. Data stored on in-house systems are as secure as the in-house firewall.

What is the administrative security model for this system? Access to data and loan analysis are controlled by permissioning within your organization. CMM works on a “role based” model, where, for example, your institution could have administrators, supervisors and analysts. Administrators could adjust parameters, supervisors could approve loans for inclusion in the rating system and analysts could enter data and analyze loans. Loans would viewable only by the owner analyst, his/her supervisor and any administrator. Groups of analysts can work as a team, if desired by the institution.

How is the software updated? One of the biggest advantages of web-based applications is that new or improved features can be updated without disruption to your use. Minor improvements will be introduced as available. Improvements that produce material changes in the results or the interface experience will be introduced cautiously. In either case clients will be notified in advance.

mpany analysis

7 ■

opinion

■

data

■

market perspective

■

credit ratings

■

company analysis

■

opin

Contact Us for More Information To find out more about Moody's Commercial Mortgage Metrics, please contact: Jason Schwarz

Mark McKenna

Product Specialist

Product Specialist

+1-201-915-8702

+1-212-553-4829

[email protected]

[email protected]

To find out more about Torto Wheaton Research, please contact: Ken O’Brien SVP, Director of Client Services Torto Wheaton Research 973.635.2604 [email protected]

SP0790 / April 2006