Table 1 Potential Provisions and Adjustments for Loss Allowances For the Twelve Months Ending December 31, 2004

	Low Estimate	Best Estimate	High Estimate
Provision Related to Future Failures (1)	$129 million	($42 million)	($1 million)
Adjustment for Closed Institutions’ Net Recoveries (2)	$14 million	$0	($14 million)
Adjustment for Litigation Losses (3)	$3 million	$0	($3 million)
Potential Provision for Losses	$146 million	($42 million)	($18 million)
Notes: (1) Includes provisions required to bring the contingent loss reserve to estimated December 31, 2004 levels after accounting for a) actual reserve losses sustained in the first quarter of 2004 ($0.7 million), and b) estimated reserve losses sustained through the remaining three quarters of 2004 ($2 million). Changes in the contingent loss reserve occur because of reductions in reserves due to failures, reductions in reserves due to improvement in problem institutions' conditions, and additions to reserves due to problem institutions' deterioration. (2) Assumes a range of -5% to +5% of the estimated net recovery value of institution resolutions receivables totaling $273 million as of December 31, 2003. (3) Range is based on the standard deviation of changes in the contingent liability for litigation losses for the period 1998 to 2003.

Staff believes that the range provided by the statistical migration analysis adequately represents the most likely range of additional provisions needed to cover insurance losses from future failures. However, the bounds of this range do not represent "best case" and "worst case" scenarios, and larger or smaller provisions could occur.

Nevertheless, SAIF-insured institutions in general appear to be well positioned to withstand considerable financial stress from unlikely economic shocks. For example, staff has considered economic stress events as they relate to specific risk concerns enumerated in industry outlook contained in Tab 1. To determine the potential insurance fund implications of these concerns, staff has developed several stress event simulations, each of which demonstrate that SAIF-insured institutions are well positioned to withstand a significant degree of financial adversity.

Subprime Lending Risk: Staff believes that subprime lending continues to be the most likely source of near-term losses to the insurance funds. Subprime lending has been a significant factor in 28 percent of the number of failures and 67 percent of the assets of failed institutions since 1997. In addition, the percentage of "problem" subprime lenders continues to be high, making up 36 percent of the assets of institutions on the contingent loss reserve list.

Using periods of historically poor performance for various categories of consumer loans, staff subjected subprime lending institutions to a two-year period of instantaneously higher consumer loan loss rates. For the fourth quarter, these simulations produced SAIF-insured failed assets of only $1.4 billion (less than 1 percent of institution assets considered). Simulations using year-end 2002 data produced similar results.

Mortgage Lending Risk: Prospects for rising interest rates may cause some concern over the future performance of institutions engaged in mortgage lending activities. Rising rates could place pressure on the net interest margins of some mortgage lending institutions by raising funding costs against fixed-rate loan portfolios and securities holdings. Higher rates could also suppress mortgage origination volumes and the value of home prices in the face of weaker sales activity.on these findings, combined with signs of improving overall economic conditions, staff believes that current industry trends do not foreshadow widespread deterioration in the bank and thrift industries.

Using periods of historically significant declines in both net interest margins and mortgage loan performance, staff subjected institutions with mortgage lending concentrations to a two-year period of higher loan loss rates and declining net interest margins. For the fourth quarter 2003, these simulations produced SAIF-insured failed assets of only $1.2 billion (0.1 percent of institution assets considered). Simulations using year-end 2002 data resulted in SAIF-insured failed assets of $5.8 billion or 0.7 percent of institution assets. The improvement in simulation results from year-end 2002 to year-end 2003 reflects generally improved performance measures of thrifts.

Commercial Real Estate Mortgage Lending Risks: Rising interest rates could also have an adverse impact on commercial real estate loan performance as debt servicing burdens on variable rate loans increase. This concern is compounded by the already weak fundamentals that exist in many metropolitan areas for commercial property types such as offices and hotels. Institutions with heavy commercial real estate loan concentrations are most vulnerable to any rise in commercial real estate loan losses.

Using periods of historically significant declines in commercial real estate values, staff subjected institutions with commercial real estate mortgage lending concentrations to a two-year period of higher loan loss rates. For the fourth quarter 2003, the worst case simulation, which drew assumptions from the experience of New England institutions during the late 1980s, produced SAIF-insured failed assets of $0.4 billion. Results using year-end 2002 data were similar. For comparative purposes, this same simulation produced just under $8 billion in failed SAIF-insured assets using year-end 1991 data.

Based on these findings, combined with signs of improving overall economic conditions, staff believes that widespread deterioration in thrift industry performance is unlikely in the next one to two years.

B. Interest Income and Unrealized Gains and Losses on AFS Securities
Staff relied upon expert forecasts as detailed in the Blue Chip Financial Forecasts to develop interest rate projections and analyze the potential effect of changes in interest rates on interest income and unrealized gains and losses on AFS securities. The forecasts defined as our "best estimate" were the consensus forecasts through the fourth quarter of 2004 as detailed in the March issue of the Blue Chip Financial Forecasts. Adopting the experts' consensus forecasts also allows for forecasted yield curves that change in shape over time.

Along with forecasting yield curves based upon the experts' forecasts, staff also calculated bounds within which interest rates are likely to fall using the historical differences between the experts' forecasts and the actual interest rates. These bounds vary over the assessment period and change in shape over time, as opposed to being parallel shifts in rates. The bounds are consistent with the notion that the projections represent the most likely scenarios and that the actual rates may be above or below the projections. In general, the projections indicate stable or slightly rising rates for the period under consideration. The low estimate (high estimate) generally reflects rates that are as much as 200 bp higher (30 bp lower) than current rates, with the range increasing over time. Charts showing the projected rates, upper bound, and lower bound are included as Appendix A to this case.

Table 2 shows projections for low, best, and high estimates for interest income and unrealized gains and losses on AFS securities using the forecast rates and upper and lower bounds. Because of the significant percentage of AFS securities held in the insurance fund portfolio at this time, when interest rates change, the magnitude of the resulting change in market value of these securities dominates the effect of changes in interest income.

Table 2 Potential Changes in Interest Income and Unrealized Gains (Losses) on AFS Securities December 31, 2003 to December 31, 2004 ($ in millions)
	Low Estimate (1)	Best Estimate (1)	High Estimate (1)
Interest Income	545	537	528
Unrealized Gain (Loss) on AFS Securities (2)	-173	-93	-10
Net Fund contribution from Investment Activities	372	444	518
Notes: (1) The Low Estimate is calculated using upper bound interest rates, the Best Estimate is calculated using the projected rates, and the High Estimate is calculated using the lower bound rates. Net estimated failure resolution outlays equal $31 million for the Low Estimate and $10 million for both the Best Estimate and the High Estimate. (2) Includes actual unrealized gains on AFS securities for the period January 1, 2004 through February 29, 2004 and projected gains/losses for the remaining period through December 31, 2004.

Staff's best estimate reflects a modest upward trend in bond market interest rates. Since the end of March, treasury yields have increased to six-month highs on news of strong payroll growth, strong retail sales, and an increase in inflation indicators. Based on the treasury yield curve as of April 30th, this recent shift in yields generally falls in between assumptions underlying the "best" and "low" estimates in Table 2 above. Accordingly, some depreciation in the value of AFS securities should be expected if current interest rate trends persist. As the remaining maturity of the existing AFS portfolio shortens, previously identified unrealized gains will also dissipate. Over the longer term, higher yields on treasury securities will boost overall interest earnings as securities reprice upward and as maturing securities are reinvested at higher rates.

C. Projected Fund Balance
Table 3 summarizes the effects on the fund balance of the low, best, and high estimates assumed for insurance losses, interest income, and unrealized gains and losses on AFS securities. The projection also assumes that the current assessment rate schedule will remain in effect through December 31, 2004.

Table 3 Projected Fund Balance (1) ($ in millions)

	Lower Bound	Best Estimate	Upper Bound
Assessments (2)	12	12	12
Interest Income (3)	545	537	528
Total Revenue	557	549	540
Operating Expenses (4)	137	137	137
Provision for Losses	146	42	(18)
Total Expenses & Losses	283	179	119
Net Income	274	370	420
Unrealized Gain (Loss) on AFS Securities (3)	(173)	-93	(10)
Comprehensive Income (Loss) (5)	101	277	411
Fund Balance – 12/31/03	12,240	12,240	12,240
Projected Fund Balance – 12/31/04	12,341	12,517	12,651
Notes: (1) Projected income and expense figures are for the twelve months ending December 31, 2004. (2) Assumes that the current assessment rate schedule remains in effect through December 31, 2004. (3) See also Table 2 for an explanation regarding changes in interest revenue and unrealized gain (loss) on AFS securities under these projections. (4) Operating expenses for 2004 allocated to the SAIF are estimated based on budgeted 2004 operating expenses inclusive of amounts budgeted for litigation expenses. (5) Comprehensive Income is used instead of Net Income due to the magnitude of the change in market value of AFS securities that occurs with fluctuations in interest rates. See note (3) above.

2. Insured Deposits
Since 1989, annual growth rate for SAIF-insured deposits has been as high as 7.2 percent and as low as an annual shrinkage of 7.1 percent (Figure 1). After declining during the period from 1989 through 1993, insured deposits grew 2.2 percent and 2.7 percent in 1994 and 1995, respectively. After contracting 3.9 percent in 1996, and minimal growth between 1997 and 1999 (1.0 percent, 3.8 percent, and 0.2 percent respectively), insured deposits grew by 5.2 percent in 2000, 6.1 percent in 2001, and 7.2 percent in 2002. Growth slowed down slightly in 2003 to 4.6 percent, and is projected to grow at a rate of 1.1 percent in 2004. An upturn in equity markets have factored into the recent slow growth in insured deposits.

     Figure 1
D

It takes approximately $6.6 billion in insured deposit growth to create a 1 basis point decline in the SAIF reserve ratio, all other things held constant. Based upon the December 31, 2003 fund balance, it would take about $82.7 billion in insured deposit growth (9.2 percent) to reduce the fund to the DRR level, all else being equal. Staff's estimate indicates that deposit growth over the next year will be far lower than this figure.

Based on projections using a statistical model, the best judgment of the staff is that SAIF-insured deposits are likely to experience a growth rate in the range of -2.2 percent to +4.3 percent between December 2003 and December 2004. This range represents the statistical margin of error in the estimated model. ⁴ Staff believes the most likely scenario is that insured deposits will grow at the midpoint of this range (1.1 percent), which will bring the total for SAIF-insured deposits to $906 billion. A scenario that could force insured deposits into the high range of our forecast would include a depressed stock market with high volatility. In contrast, an upturn in the stock market and in the U.S. economy as a whole could force insured deposits into the low end of the forecast.

3. SAIF Reserve Ratio
Based on the projected SAIF balance and the growth of the insured deposit base, the best estimate of the SAIF reserve ratio at December 31, 2004 is 1.38 percent (Table 4, next page). The best estimate assumes a baseline of a small increase in contingent loss provisions, a modest increase in treasury yields, and insured deposit growth of 1.1 percent.

Staff projects the lower bound and upper bound of the likely range to be 1.32 percent and 1.44 percent, respectively (Table 4, next page). The lower bound, which reflects a 5 bp decrease from the December 31, 2003, ratio, assumes a strong increase in the insured deposit base (4.3 percent growth) and a higher interest rate scenario, which results in a downward adjustment to the fund balance due to a reduction in the aggregate amount of unrealized gains on AFS securities (Table 3). The lower bound also incorporates the high loss estimate for insurance losses from possible near-term failures as projected by staff. Although the estimate reflects staff's view of a reasonably possible adverse scenario, it is not intended to represent a "worst case" scenario.

The upper bound produces a 7 bp increase in the reserve ratio relative to December 31, 2003 levels. This estimate assumes a contraction of 2.2 percent in the SAIF-insured deposit base, reverse provisions for failure-related losses, and a modest decline in interest rates, which results in a nominal adjustment to the aggregated amount of unrealized gains on AFS securities.

Staff's best estimate of the reserve ratio for December 31, 2004 is 13 bp higher than the DRR and 1 bp higher than the ratio at December 31, 2003. The most significant factor influencing this estimated increase is the projected slowdown in insured deposit growth. This slowdown in growth outweighs other factors that tend to place downward pressure on the ratio including the following:

• Interest rates remain at very low levels but have recently begun to move higher in line with improving economic conditions. Unrealized gains on AFS securities will decline even in a stable interest rate environment because these gains disappear as securities move closer to their maturity dates. With rates moving higher, reductions in unrealized gains can be expected to accelerate.
• Approximately 17 percent ($82 million) of comprehensive income in 2003 represented reversals of provisions for insurance losses due to reductions in estimated losses on prior failures and due to reductions in the contingent loss reserve. Although staff remains optimistic about industry prospects, reserve levels are already at nominally low levels precluding substantial reversals to the loss provisions going forward.

As a result of these factors, staff believes that the "best estimate" is for a slightly higher SAIF reserve ratio. Because the entire expected range for the SAIF reserve ratio is greater than the DRR of 1.25 percent, staff believes that it is reasonable to maintain the existing SAIF rate schedule.

Risk-Based Assessment System
Staff recommends retaining the current spread of 27 bp between the assessments paid by the best- and worst-rated institutions as well as the rate spreads between adjacent cells in the assessment rate matrix. The proposed assessment rate schedule appears in Table 5. The Board previously determined that the current rate spreads provide appropriate incentives for weaker institutions to improve their condition and for all institutions to avoid excessive risk-taking, consistent with the goals of risk-based assessments and existing statutory provisions. The current rate spreads also generally are consistent with the historical variation in institution failure rates across cells of the assessment rate matrix.

Table 5 Proposed Assessment Rate Schedule First Semiannual Assessment Period of 2005 SAIF-Insured Institutions

Capital Group	A	B	C
1. Well	0 bp	3 bp	17 bp
2. Adequate	3 bp	10 bp	24 bp
3. Under	10 bp	24 bp	27 bp

In setting assessment rates to achieve and maintain the reserve ratio at the target DRR, the Board is required to consider the effects of assessments on members' earnings and capital. The estimated annual revenue from the existing rate schedule is $12 million, which is $1 million less than the annual amount that was projected six months ago. In recommending that the Board maintain this schedule, the staff has considered the impact on thrift earnings and capital of the current rate schedule and found no unwarranted adverse effects.

The Assessment Base Distribution and Matrix Migration
Table 6 summarizes the current distribution of institutions across the assessment matrix.

With 98.5 percent of the number of institutions and 99.9 percent of the assessment base in the three lowest assessment risk classifications of “1A,” “1B,” and “2A,” as of July 1, 2003, the current distribution in the rate matrix reflects little fundamental difference from the previous semiannual assessment period. The current distribution reflects a slight increase in the percentage of institutions in the best-rated premium category. Since the previous assessment period, 22 institutions migrated into the "1A" risk classification (Table 7, next page), and 15 institutions migrated out of the "1A" risk classification. Only 103 institutions are currently classified outside of the best risk classification.

Overall, for all SAIF-insured institutions, the supervisory subgroup component of the risk classification was upgraded since the previous period for 35 institutions with an assessment base of $5.3 billion and was downgraded for 17 institutions with an assessment base of $2.6 billion.

Other Issues
FICO Assessment. The Deposit Insurance Funds Act of 1996 (Funds Act) separates the Financing Corporation (FICO) assessment from the FDIC assessment, so that the amount assessed on individual institutions by the FICO is in addition to the amount paid according to the SAIF rate schedule. All institutions are assessed the same rate by FICO, as provided for in the Funds Act, and the FICO rate is updated quarterly. The FICO rate for the first quarterly payment in second semiannual assessment period of 2004 will be determined using March 31, 2004 Call Report and Thrift Financial Report data.

Staff Contacts
For information about deposit insurance assessments, please contact Steve Burton, Acting Chief, Fund Analysis Section, Division of Insurance and Research, at (202) 898-3539, or Joe DiNuzzo, Counsel, Legal Division (202) 898-7349. For FICO assessment information, please contact Richard Jones, Chief, Deposit Insurance Pricing Section, Division of Insurance and Research, at (202) 898-6592.

4 TThe model is a regression model where the current growth rate in insured deposits is estimated as a linear function of the previous growth rate in insured deposits, the current and previous growth rates of total (insured and uninsured) domestic deposits, as well as the current yields on 3 month and 10 year Treasury Bills. The range (-2.2%, +4.3%) corresponds to a 95% confidence level. In other words, if the process generating insured deposit growth in the future is the same as in the past, we can be sure with 95% confidence that the actual growth rate in insured deposits, over the year 2004, will lie within this range. The growth rate predicted by the model (thus, the most likely rate) is the midpoint of this range (1.1% annual growth).

D

D

D

D