03-6652. Proposed Agency Information Collection Activities; Comment Request-Thrift Financial Report: Schedule CMR

AGENCY:

Office of Thrift Supervision, Treasury.

ACTION:

Notice and request for comment.

SUMMARY:

The Office of Thrift Supervision (OTS), as part of its continuing effort to reduce paperwork and respondent burden, invites the general public and other Federal agencies to comment on proposed and continuing information collections, as required by the Paperwork Reduction Act of 1995, 44 U.S.C. 3507. OTS will submit the proposed information collection requirement described below to the Office of Management and Budget (OMB) for review, as required by the Paperwork Reduction Act.

Further, OTS requests comments on the replacement of Schedule Consolidated Maturity/Rate (CMR) of the Thrift Financial Report (TFR) with a new schedule to be known as Risk Exposure Data (RED). Schedule RED will reduce the data collection burden on institutions while at the same time increase the flexibility and granularity of the data collected. The proposed Schedule RED will also increase the flexibility of the agency's Net Portfolio Value (NPV) model and assist the agency in better monitoring individual institution and system-wide credit risk.

DATES:

Submit written comments on or before May 19, 2003.

ADDRESSES:

Send comments, referring to the collection by title of the proposal or by OMB approval number, to: Information Collection Comments, Chief Counsel's Office, Office of Thrift Supervision, 1700 G Street, NW., Washington, DC 20552. Send a facsimile transmission to [phone]. Or send e-mail to: [e-mail]. OTS will post comments and the related index on the OTS Internet Site at www.ots.treas.gov. In addition, interested persons may inspect comments at the Public Reading Room, 1700 G Street, NW., by appointment. To make an appointment, call [phone], send an e-mail to [e-mail], or send a facsimile transmission to [phone].

FOR FURTHER INFORMATION CONTACT:

Mark Flood, Senior Financial Economist, [phone], Economic Analysis Division, or Teresa A. Scott, Counsel (Banking and Finance), [phone], Regulations and Legislation Division, Office of the Chief Counsel, Office of Thrift Supervision, 1700 G Street, NW., Washington, DC 20552.

SUPPLEMENTARY INFORMATION:

Title of Proposal: Thrift Financial Report: Schedule CMR.

OMB Number: 1550-0023.

Form Number: Schedule RED of the Thrift Financial Report.

Abstract: Currently, Schedule CMR provides all the institution-level inputs to OTS's NPV model, the agency's key resource for measuring interest-rate risk. The NPV model: (1) Complements and supplements on-site exams with quarterly off-site monitoring; (2) helps identify aggregate patterns unapparent at the level of the individual institution; and (3) by employing scenario analysis, draws attention to thrifts with unusually large risk exposures. Further, the NPV model effectively offers thrifts a quarterly risk-management consultancy function. It is therefore centrally important to the agency and the thrift industry that the NPV model operates reliably, is understandable, and adapts itself regularly to advances in the analytical state of the art, as well as to changes in market conditions, including new financial instruments and other innovations.

A. Why Replace Schedule CMR?

Future improvements on the NPV model should focus both on its benefiting savings associations and improving OTS's oversight responsibility. Both of these objectives can be met by replacing Schedule CMR with Schedule RED.

Proposed Schedule RED reduces the data collection burden by changing the manner in which data is collected while at the same time increasing the flexibility of the NPV model. Further, Schedule RED addresses concerns raised by some institutions about shortcomings in the NPV model that are caused by the nature of data collected in the current CMR format. Lastly, Schedule RED permits OTS to collect certain new data to aid in calibrating the NPV model and measuring credit quality.

1. Simplification and Burden Reduction

Schedule RED will simplify, and thereby reduce the burden of, the reporting process for both OTS and reporting thrifts. Switching to the proposed new schedule will result in a substantial net reduction in the number of field definitions. Indeed, Schedule RED has roughly half the number of fields as the current CMR (and therefore half the field definitions to implement, and field instructions to understand). Since the data burden falls both upon OTS and the submitting institutions, both parties benefit from this reduction. To see better how this will work, interested parties can find the full details of the proposed new form and its instructions on the OTS Web site at: http://www.ots.treas.gov/​docs/​r.cfm?​84259.html.

The current CMR collects data in 535 individual CMR cells (numbered between 001 and 903) plus 26 additional fields in the supplemental tables, for a total of 561 fields. Each of these fields potentially has a separate, idiosyncratic definition in the CMR instructions, although there are currently some shared definitions.^[1] Each separate definition must be implemented by programming logic and/or data-entry training. In contrast, the number of position attributes defined by the proposed RED is 262 “ less than half that of the CMR. Among these, there are considerable overlaps in definitions (e.g., position balance is defined identically for most positions), so that the total number of distinct instructions for RED fields is currently 84 (with 116 instructions overall “ including those not attached to specific input fields).

Another example of Schedule RED's simplification is reflected in the collection of fixed-rate mortgage (FRM) data. Schedule CMR collects FRM data in three sections on two pages of the schedule: balances, coupons, and maturities (CMR 001-125); miscellaneous aggregate memoranda (CMR 501-508); and warehouse loans (CMR 578). In all, there are 107 separate cells on CMR collecting data on FRM loans and mortgage-backed securities (MBS). Because many of these cells collect the same sort of information for different aggregation buckets, only 18 instruction paragraphs are needed to describe these cells. In the latter two sections “ aggregate memoranda and warehouse loans “ FRMs are commingled with adjustable-rate mortgages (ARMs), multifamily mortgages, and non-residential mortgages.

Schedule RED collects FRM data in two separate sections: FRMs (10 fields, 4 of which are new “credit-risk” attributes); and loan memoranda (8 fields). Ignoring the credit-risk attributes, there are in all 14 separate fields on RED collecting data on FRM loans and MBS, and only 14 instruction paragraphs are needed to describe these cells.

A simplistic comparison of the CMR cell count for FRMs (107) to the RED attribute count (14) would overstate the benefits of the RED because the CMR is restricted to one data point per cell, while the RED will typically take multiple observations of each attribute field. A more informative comparison is between the instruction counts (18 CMR vs. 14 RED), since these represent the business logic that must be written, tested, deployed, and maintained in reporting software. By this measure, the RED represents a 29% reduction in reporting burden relative to the current CMR.

2. Flexibility

The current CMR format is rigid in defining the data it accepts. For example, the definition of the five coupon buckets for fixed-rate mortgages currently covers the following ranges: <7%, 7-8%, 8-9%, 9-10%, and >10%. With few exceptions, mortgages above 7% have not been issued for some time, and refinancing to the current lower rates has been intense. As a result, CMR's data bucketing has less value in the current environment, as nearly all new mortgages and refinancings fall into the first bucket. Addressing this problem by redefining the bucket ranges requires reprinting the form, editing the instructions, and testing the edit and NPV model software. This process may take several filing cycles to complete. When rates rise again, the process would have to be repeated.

These transitions would be unnecessary with Schedule RED, because flexibility is built into its structure. The basic structure of Schedule RED is similar to that currently used to collect supplemental positions (for example, the supplemental OBS or supplemental assets and liabilities tables). All data in the RED would be entered into tables, the rows of which represent financial positions held by the thrift, and the columns of which are the attributes of those positions. For example, here is one of the proposed RED tables, for ARM servicing rights:

Adjustable-Rate Mortgage-Servicing Rights

Type	Balance	Original maturity	Remaining maturity	Rate code	Margin	Service fee	Sub-serviced	FHA VA	Conventional

OTS believes that proposed Schedule RED eliminates most of the rigidities present in the current CMR layout while still providing pertinent data for input into the NPV model.

3. Increased Data Detail

a. Increased Granularity

While slashing the field count in half, the RED would increase the number of data points by collecting finer-grained observations on each field. All reporting schemes entail bucketing with some degree of categorization, i.e., granularity. By definition, all loans within a given bucket are treated identically—typically by assuming that all loans lie at the center of the bucket. For example, suppose loans in the 2-to-5-year rate-reset bucket have an average rate-reset frequency of exactly 3.5 years. Thus, the NPVs and sensitivities of loans at one end or another of a bucket will tend to be less accurate. These estimation differences do not necessarily disappear when the whole bucket is averaged. To the extent that the NPV model's results guide supervision policy, this less accurate estimate of the loan characteristics may adversely affect some thrifts, while arbitrarily rewarding others.

Reducing bucket sizes to shrink the potential measurement error can alleviate this “bucketing burden.” This increases granularity and necessarily reduces the average magnitude of the estimation error in measuring loan characteristics. However, changing bucket ranges requires flexibility in the structure of the reporting schedule.

In the case of Schedule CMR, increasing granularity means adding cells. Because the cells are indexed sequentially, this requires renumbering and/or redefining some cells. For example, in the case of FRMs, there are 5 coupon buckets.^[2] Inserting a new column (i.e., a new bucket) on this page of the form would require either: (a) inserting cells with non-sequential numbers; or (b) renumbering all subsequent cells in the form. Either solution necessitates a redefinition of certain cells, and either is likely to create confusion and implementation difficulties. As a result, increases in data granularity are infrequent under the current Schedule CMR.

In the case of Schedule RED, the number of observations collected is open-ended, and all bucketing is handled through the aggregation rules defined in the instructions to the form.^[3] Thus, increasing or decreasing granularity in any particular dimension involves only a change to the instructions. Of course, the reporting institutions (or their vendors) must still implement this change. However, these changes in aggregation rules do not redefine the attributes collected, but only the number and composition of the observations reported. Increases in the number of observations (the granularity) for a particular instrument have no impact on reporting elsewhere on the form. At the same time, it is similarly possible to add (or remove) to the list of RED attributes collected without affecting reporting elsewhere on the form.

b. Aggregation Rules

Using Schedule RED simplifies the aggregation of data thrifts supply. At present, the conversion of accounting data to CMR fields often involves awkward aggregation rules. For example, balances of fixed-rate fixed-maturity deposits (FRFMD) are reported in buckets that depend on both original and remaining maturity. At the same time, balances on brokered FRFMD (a subset of the total) are bucketed only by original maturity. This reduces the number of cells on the form by requiring differential aggregation procedures be applied to total vs. brokered balances.

Moreover, many aggregations must be “unwound” within the NPV model by applying assumptions about how the aggregation should, or might, have occurred. Such derived disaggregations inevitably result in less accurate estimates. The crucial difference between proposed Schedule RED's system of aggregation and that applied on the CMR is that the RED Schedule's aggregation rules (described below) are set in the written instructions, rather than being built into the structure of the form itself. The upshot is that adjustments to the aggregation rules—to increase or decrease the level of detail collected—will be significantly simpler, as they would not affect the structure of the form or the definitions of the fields.

Schedule RED allows a wide range of possibilities for aggregating position information. For example, at one extreme, if institutions so chose, RED could allow contract-by-contract reporting of all loans in the portfolio and account-by-account reporting of deposits. At the other extreme, Schedule RED could accept highly aggregated positions, representing large segments of the portfolio as single position entries.^[4]

Schedule RED will however constrain the degree of aggregation. In other words, Schedule RED imposes a maximal degree of position aggregation (or, equivalently, a minimum degree of granularity). Under this proposal, reporting institutions could potentially break positions down into more detail than required by the aggregation constraint (possibly down to the contract-by-contract level), but never less. Here are the proposed maximum aggregation limits for fixed-rate mortgages:

Maximum Aggregation Constraints

Aggregation Rule: Mortgages and MBS that match simultaneously on *all* of the following criteria can be aggregated together and reported as a single position:

1. Mortgage and MBS:

• 30-year mortgage loans
• 30-year MBS backed by conventional mortgages
• 30-year MBS backed by FHA or VA mortgages
• 15-year mortgage loans
• 15-year MBS
• Balloon mortgage loans
• Balloon MBS

2. Coupon buckets (in quarter-point increments, as follows):

• 0.00 to 0.25%
• 0.26 to 0.50%
• 0.51 to 0.75%
• 0.76 to 1.00%
• 1.01to 1.25%
• Etc.

3. [IF SUBMITTED] Borrower credit rating type

4. [IF SUBMITTED] 5-digit zip code

Mortgages or MBS that differ in *any* of the above criteria cannot be aggregated together into the same position. NOTE: The aggregation (i.e., bucketing) rules are subject to change.

There are two reasons for such constraints. First, to support current legacy applications, and to track industry trends over time, it must be possible for OTS to convert data submitted in the new RED Schedule back into the legacy CMR format. As a result, Schedule RED must always be at least as detailed as Schedule CMR. Second, OTS would like to see the benefit of the increased reporting detail that Schedule RED allows.

OTS is contemplating allowing any filing firm to submit non-aggregated data (that is, account-by-account position data). OTS anticipates that many filers may find this latter option very attractive, as it alleviates the burden of maintaining programming logic and operator intervention necessary to calculate the aggregations. It may also result in cost reductions when providing the requested information.

4. New Attributes for Loans

Schedule RED includes several new fields measuring basic loan attributes, such as loan-to-value (LTV), borrower credit rating, and collateral. The three largest potential benefits from this innovation are the improvement in OTS's ability to calibrate the NPV model,^[5] assess interest rate risk in relation to portfolio risk attributes, and the possibility that the NPV model could someday provide OTS institutions with an analytical toolkit that approximates the Basel II internal-models approach. This could, after further comment and review, and consistent with systems developed by the other federal banking agencies, open the door for OTS-regulated institutions to qualify for more risk-sensitive capital treatment for their mortgage and retail assets in a manner analogous to the evolving standards of Basel II. There is clearly much to be done before such analytics could be deployed, but the underlying credit-quality data would be needed during the development and testing phases, well before any deployment.

One area where the availability of these new dimensions could improve the quality of the interest-rate risk measurement involves credit spreads on loans. Currently, we are forced to assume a fixed, one-size-fits-all credit spread for all institutions, implicitly assuming that none of the observed differences in interest rates across institutions is due to risk. With credit-quality information, we can realistically assign credit-risk-adjusted discount rates to cash flows in the model, improving the quality of the final NPV measurement.

OTS recognizes that some reporting institutions may be reluctant or unable to provide new loan attributes, as these have not been reported heretofore on the TFR. As a result, OTS proposes that reporting of these attributes under Schedule RED be optional.

B. Side-by-Side Comparison of Schedules RED and CMR

To assist the industry in assessing the impact of the proposed RED schedule, this section uses FRMs to exemplify the differences between current CMR procedures and the proposed Schedule RED. The relevant sections of Schedule CMR (pages 30 and 34 of the TFR) and of Schedule RED (FRM and loan memoranda tables) are attached here for reference (See http://www.ots.treas.gov/​docs/​78155.pdf for the full CMR form). In addition, included is a table that provides a full side-by-side comparison of the two schedules.

Schedule RED

Fixed-Rate Mortgages

Type	Balance	Coupon	Original maturity	Remaining maturity	Amortization period	LTV	Credit rating	Credit rating type	Zip code

Loan Memoranda

Type	Warehouse	Non performing	Accrued interest receivable	Advances for taxes and insurance	Unamortized yield adjustment	Valuation allowance	Unrealized gains (losses)

CMR/RED Comparison Table

	CMR	RED
Information structure
Data inputs	535 cells + 26 fields	262 fields.
Distinct instructions	220 paragraphs	116 paragraphs.
Granularity increases possible	No	Yes.
Credit-risk measurement
LTV	No	Yes (optional).
Borrower credit rating	No	Yes (optional).
Commercial loan ratings	No	Yes (optional).
Commercial borrower ratings	No	Yes (optional).
Commercial loan SNC status	No	Yes (optional).
Consumer loan collateral	No	Yes (optional).
Non-performing loans	Limited	Limited.
Valuation allowances	Limited	Limited.
Miscellaneous
Geographic exposure data	No	Yes (optional).
Bucketing
Fixed-rate mortgages:
Mortgage vs. MBS	No	Yes.
Coupon range	Five 100 bp ranges	25 bp ranges as needed.
Borrower credit rating	No	Yes.
Location	No	Yes.
Adjustable-rate mortgages:
Mortgage vs. MBS	No	Yes.
Reset frequency	2 or 3 buckets	monthly.
Rate index	No	Yes.
Teaser vs. non-teaser	Yes	Yes.
Current vs. lagging index	Partial	Yes.
Distance to lifetime cap	4 buckets	100 bp ranges as needed.
Periodic caps	Partial	Yes.
Periodic floors	Partial	Yes.
Borrower credit rating	No	Yes.
Location	No	Yes.
Fixed-rate other real estate loans:
Balloon multifamily/amortizing multifamily/2nd mortgage/land	Yes	Yes.
Coupon range	Five 100 bp ranges	25 bp ranges as needed.
Borrower credit rating	No	Yes.
Location	No	Yes.
Adjustable-rate other real estate loans:
Balloon multifamily/amortizing multifamily/2nd mortgage/land	Yes	Yes.
Rate index	Limited	Yes.
Distance to lifetime cap	No	100 bp ranges as needed.
Borrower credit rating	No	Yes.
Location	No	Yes.
Commercial loans:
Adjustable vs. fixed-rate	Yes	Yes.
Rate index (adjustable rate)	No	Yes.
Consumer loans:
Adjustable vs. fixed-rate	Yes	Yes.
Rate index (adjustable rate)	No	Yes.
Borrower credit rating	No	Yes.
Location	No	Yes.
Fixed-rate mortgage servicing rights:
Servicing by vs. for others	Limited	Yes.
Coupon range	Five 100 bp ranges	25 bp ranges as needed.
Adjustable-rate mortgage servicing rights:
Servicing by vs. for others	Limited	Yes.
Rate index	No	Yes.
Current vs. lagging index	Yes	Yes.
Money market assets:
Instrument type	Partial	Yes.
Fixed-rate fixed-maturity deposits:
Deposit type	No	Yes.
Coupon range	No	25 bp ranges as needed.
Original maturity	3 ranges	12-mo. ranges as needed.
Remaining maturity	4 ranges	3-mo. (short-term) or 12-mo. ranges as needed.
Variable-rate fixed-maturity deposits:
Deposit type	No	Yes.
Rate index	No	Yes.
Remaining maturity	3 ranges	3-mo. (short-term) or 12-mo. ranges as needed.
Non-maturity deposits:
Deposit type	Yes	Yes.
Other liabilities:
Liability type	Yes	Yes.
Commitments to buy, sell or originate:
Firm vs. optional	Yes	Yes.
Buy/sell/originate	Yes	Yes.
Underlying type	Yes	Yes.
Mortgage subtype	Yes	Yes.
MDP subtype	Yes	Yes.
Long vs. short	Yes	Yes.
Construction loans in process (LIP):
Coupon range	No	25 bp ranges as needed.
Interest-rate derivatives (swaps, swaptions, caps, collars, floors, futures and options):
Position-level	Yes	Yes.
Self-valued instruments:
Instrument type	Limited	Yes.

C. Request for Comments

OTS invites comment on all aspects of the proposed Schedule RED and, in particular, whether the proposal will in fact reduce reporting burden, aid in more flexible data collection, and provide an opportunity for more accurate analysis of institution-specific and industry-wide interest rate risk. Consideration should be given to the amount of data collected and the ease of obtaining the data. Moreover, comments are requested on the amount of transition costs to convert from Schedule CMR to Schedule RED and the extent to which cost savings would be realized over time as a result of change.

Further, OTS requests comments on:

a. Whether the proposed collection of information is necessary for the proper performance of the functions of OTS;

b. The accuracy of OTS's estimate of the burden of the proposed information collection;

c. Ways to enhance the quality, utility, and clarity of the information to be collected;

d. Ways to minimize the burden of the information collection on respondents, including through the use information technology.

Type of Review: Revision.

Affected Public: Business or for profit.

Estimated Number of Respondents: 915.

Estimated Frequency of Response: Four times per year.

Estimated Burden Hours per Response: 12 hours.

Estimated Total Burden: 43,920 hours.

Clearance Officer: Marilyn K. Burton, [phone], Office of Thrift Supervision, 1700 G Street, NW., Washington, DC 20552.

OMB Reviewer: Joseph F. Lackey, Jr., [phone], Office of Management and Budget, Room 10235, New Executive Office Building, Washington, DC 20503.

Dated: March 14, 2003.

Deborah Dakin,

Deputy Chief Counsel, Regulations and Legislation Division.

Footnotes

1.  For example, CMR 006 through 010 are all weighted average remaining maturities on different buckets of fixed-rate mortgages, with the same rules for calculation.

2.  See Schedule CMR page 28.

3.  See discussion infra.

4.  Again, this is in contrast to Schedule CMR, for which the degree of aggregation of position information is built into the structure of the form itself. This inflexibility is one of the motivations for moving to Schedule RED.

5.  For example, cash-out refinancing tends to create higher prepayment speeds for low-LTV mortgages.