We believe that both our original regression analysis performed January 10, 2007 and the later one you requested that is described in our April 23, 2007, letter represent a regression of changes of the combined underlyings against changes in the underlying which the SEC staff has interpreted to not qualify for the SFAS 133, paragraph 10(e)(2) exception. We believe that the FASB's focus on "settlement value" in paragraph 254 is consistent with, and not contradictory to, both paragraph 10(e)(2)'s reference to "behavior of the variables" and paragraph 58(c)'s reference to "changes in the combined underlying." The reason we believe all of these passages are consistent is that the underlying in every derivative is a forward price.

The contract residual value is originally set at our best estimate of a forward price based on Black Book. This estimate is based, in part, on data published by the Automotive Lease Guide, a provider of automobile residual value data. Note that there are no forward Black Book quotations or observable forward Black Book values in the marketplace, which calls into question whether a Black Book underlying actually does not qualify for the paragraph 10(e)(2) exception. But putting that argument aside, we do estimate a forward price for a vehicle at the expiration date of the contract, and CRV is set at that amount. In that sense, the option-style contract is struck "at the money" on a forward basis. Extending the analogy, forward Black Book represents the best estimate, assuming it is possible to do so for a non-financial asset, of the actual residual sales price of the vehicle. So the two underlyings we are comparing to each other all start at the same amount, but end at different amounts. T his fact may have been a major point of confusion up to this point.

Using the Row 1 example from our March 15 letter, let's assume the RVI contract lasted for four years and was entered into on May 1, 2003 to guarantee the residual value at lease termination on May 1, 2007. The CRV is set at the amount expected on May 1, 2003 to be the Black Book Value for that automobile on May 1, 2007.

Thus, these examples illustrate why we believe that a "settlement value approach" and a "change in underlying approach" is one and the same, and why we do not see a conflict in the literature between paragraphs 10(e)(2), 58(c), and 254.

Earlier telephone requests from the Division of Corporation Finance requested that we perform a regression of levels of the underlying, an approach that we believed would bias the result to imply that the two variables were highly correlated. A regression of levels for the two examples above would have plotted the data pair (74900, 79000) for the Row 1 lease and (6750, 6750) for the Row 4 lease. Note however that even under our regression methodology, the Row 4 data pair plots out as perfectly correlated. Under the methodology you requested at the April 18 teleconference, the Row 4 data pair does not plot as perfectly correlated. This helps explain why correlation declined in the "pro forma" example you requested we run in our April 23, 2007 response wherein we reduced CRV for a hypothetical contract that was required to be settled on Black Book (i.e. the 9% of contracts that were settled on Black Book in our initial analysis, and therefore perfectly correlated with Actual Sales Value, were no longer perfectly correlated).

In prior correspondence with the Staff, we stated that we believe we should not be performing a regression analysis of value levels and were encouraged when the Staff indicated in our April 18^th telephone conversation that it would not object to using a correlation analysis focusing on settlement values as outlined in paragraph 254 of SFAS 133.

None of the 116,000 leases in our analysis had terms that adjust settlement values for the effects of excess wear-and-tear or excess mileage. Our residual value contracts insure unique non-financial assets that are not readily convertible to cash; it generally takes 30 to 60 days to sell a vehicle. Each leased vehicle's residual value is based on a unique set of characteristics including make, model, style, year, mileage, color and condition. In accordance with DIG Issue C5, the non-financial asset (the leased auto) is owned by the party (the insured) who would not benefit under the contract from an increase in the price or value of the non-financial asset. If the leased auto increases in value, the insurance contract becomes less beneficial to the insured.

Please note that AFG's residual value business is in run-off. Using a discounted cash flow approach, we estimate that marking to fair value all of the remaining residual value contracts outstanding as of December 31, 2006, would result in a charge of less than $3.2 million (one-tenth of one percent of equity), net of tax. We have not attempted to analyze prior periods due to the onerous amount of time involved in the calculations.