makes use of the worst possible catastrophe in the mortgage marketthe early 1930's-and indeed the implicit loss rates are probably worse than the historical record indicates. (See Fisher & Rapkin; the book to which I allude is entitled "The Mutual Mortgage Insurance Funds," p. 86.)

The formula also makes use of a 21/2-percent rate on investments in determining income. If the worst catastrophe in the future is less than the one used, if a firm realizes more than 22 percent on its investment of its reserves, a lower premium on insurance or a smaller reserve is possible, even for the MMIF. With the same premium rate as MMIF, it seems reasonable to conclude that the proposed reserves are more than adequate, even though the proposed firms would incur some costs not borne by MMIF.

Liquidity problems will not arise from inability to dispose of liquid assets of the insurance firms. These assets will be largely Government securities; Federal Reserve policy at a time of such stress as to raise a problem of liquidity will be easy and security prices will rise. Nonetheless, a period of prolonged strain in the mortgage market would result in acquisition of foreclosed real estate and loss of liquid assets unless the properties could quickly be sold. Whether these properties can quickly be disposed of without damaging loss depends on (1) whether the Government intervenes to lessen the strain in the mortgage market, (2) the fall of property values compared with the average loan-to-value ratio on the insured properties, and (3) the quality of these properties, which in turn depends on the judgment of the management of the insurance firms. If good management is assumed (how can the contrary be assumed?) and if antirecissionary measures by the Government can also be assumed (a reasonable assumption?), the liquidity question is reduced to the fall of property values versus loan-to-value ratios, and this again to the adequacy of the reserves. That is to say, when the company sells the properties that have been foreclosed, is there enough difference in the reserves to make up any difference in the judgments that were made when the insurance was issued?

So the liquidity problem, it seems to me, is reducible once more to the question of reserves.

Those who consider establishing a marketing firm under the auspices of the proposed legislation will find the task of estimating its prospects a tricky one. For example, comprehensive statistics on mortgage rates and terms have only recently been compiled. Historical rates can only be approximated. Care must be exercised in the selection of alternative approximations and the unwary may be misled.

For example, portfolio yields of various types of financial institutions was used by one critic of the proposal to represent current mortgage rates. The critic to which I refer was the author of the article offered by Mr. Neel. He failed to recognize that such yields are the average of various contract rates at which mortgages were made during the entire time when the portfolio was accumulated. When mortgage rates show a strong tendency to rise during the period of portfolio accumulation as they have in the past 5 to 10 years, this method seriously understates current mortgage rates.

Another and better method, relating portfolio changes to changes in mortgage income, must also be used carefully as a measure of current

mortgage rates. When mortgage rates rise, current mortgage income also rises as older mortgages are paid off and the proceeds are reinvested. Thus, it is possible for the computed yield on changes in portfolios to rise when no changes in current lending rates have occurred. You simply refinance one that is at 4 percent at 6, and you get a higher yield on the change.

Still, this possible upward bias is less than the obvious downward bias resulting from the other method, especially when changes in the total portfolio are large. Moreover, when rates computed this way are compared with other fragmentary data, they do not seem to be far out of line with contract rates. And the ones I refer to are the surveys of typical rates charged by the savings and loan associations in the savings and loan fact books and, starting in 1960, the partial survey made of contract rates by the Federal Home Loan Bank Board, and now, since February, I believe, we do have a current comprehensive series to gage things by.

But, again, a private entrepreneur cannot very well get started at February 1963; to make his judgments he ought to go back further in time. And if he does this, he is going to run into this problem of approximating what current contract rates and fees were at the time. Institutional lending practices must also be taken into account when searching for as accurate an approximation of current effective rates as possible. For example, savings and loan associations as a group derive a notable amount of income from fees. Such income in considerable part is equivalent to an advance payment of interest or discount. Thus, the reported interest income on mortgages understates the true income received. In 1961, such income was 0.42 percent of the average portfolio of member savings and loan associations.

On the expense side, there are also pitfalls to look for. It would be easy to be overoptimistic in assessing the prospects of the firm and thus to underestimate the cost of debentures issued. And I am not here referring to the use of Aaa security yield data for an approximation of the cost of funds. Following such a method for a new firm would be too obviously unrealistic. Instead, I am referring to the fact that new issues of securities bear higher yields than seasoned issues of the same rating and maturity. For example, in July 1963 the yield on new Aa public utility securities was 5 to 8 basis points higher than seasoned issues of the same kind. And by "same kind" I mean maturity and the like.

The average spread from 1957 through 1962 ranged from 4 basis points in 1961 to 19 basis points in 1957.

On the other hand, it would be easy to overstate these and other costs. For example, it would be easy to disregard the fact that a firm the assets of which are so largely insured by a private firm with adequate reserves, or are liquid assets, and is regulated by a Government agency, might be able to raise funds more cheaply than the average Baa firm.

Another example: Since data on the mortgage servicing costs of FNMA at about one-half percent are readily available, it would be convenient to use such data and to overlook the fact that data processing equipment has so sharply reduced the costs of mortgage servicers that a rate of one-fourth percent can be obtained.

24-155 0-63—15

Summarizing to this point, the picture confronted by a prospective organizer of a mortgage marketing firm might look like this:

And I have a table here.

(The table referred to follows:)

1 Member savings and loan association data. change in interest on mortgages.

Obtained by relating the change in average portfolio to

2 Member savings and loan association data. Obtained by relating change in merger portfolio to income from fees, premiums, etc.

3 Approximate administrative costs of Fannie Mae as a percentage of portfolio.

4 For Aa utility securities. Seasoned current coupon issues compared with like new callable issues. Computed by Salomon Bros. & Hutzler.

Dr. CARR. In the first column, the one for indicated yields, is obtained by relating the change in the average portfolio of member savings and loan associations to the change in interest income on mortgages.

In the second: fees, premiums, and the like on new loans that savings and loan associations have secured.

The third one: estimated servicing costs; and I have used for this one-fourth of 1 percent.

Administrative costs: these are the approximate administrative costs of FNMA as a percentage of their portfolio.

I have netted the columns up to that point.

Then I have listed the cost of funds and used for that Baa yields. And the next column: the spread between new issues and seasoned issues. It has always to be remembered that a new issue is going to be sold at a higher yield, even by the same company and of the same maturity.

Senator SPARKMAN. Doctor, I do not know how pertinent that is to this, but I am curious. Why is that true?

Dr. CARR. Well, it is due in part to the fact that the investor knows what the issue was like that he already has in his portfolio. And there is some question in the market about that situation-even for that particular company-for a new issue. And here you could take, to make this clear, if you take a 20-year security to maturity of a current coupon, and you take a new issue with 20 years, that same company or like companies or rated companies, rated at the same credit risk, the investor seems to want a little margin to play with. That is about all I can say.

Now, the reason for this particular point is that, again, it is a matter of the data you are using to estimate what the cost of funds of the mortgage marketing firm would be.

Now, in the market you have securities outstanding and you have compilation of yields. These compilations are based on seasoned securities. So if you are going to be realistic about the cost of issuing

of a new firm, even if you call it a Baa firm, you have got to add to it the spread that they would have to pay because it is a new issue, and not merely because it is classified in a certain fashion.

Senator SPARKMAN. Thank you.

Dr. CARR. Well, in any case, in the eighth column it shows a net return. And these net returns range from 1.41 percent down to .96 percent; 1957 was the high one, 1959 was the low one.

Before computing the estimated after-tax return on invested capital, one final important point would have to be borne in mind by a prospective organizer. The historical data on mortgage portfolio yields of savings and loan associations are influenced by the fact that the portfolios are largely, indeed almost entirely, conventional. It may be speculated, then, that such yields as reported are higher than they would have been had the mortgages in the portfolios been insured. In other words, the historical yields probably contain an allowance for the fact that the associations were in effect carrying their own insurance. Under the conditions prescribed by the proposed act, the insurance firms would bear the risk. In all likelihood, the incidence of the premiums would be borne by the lender. Thus, the payment by the borrower, had these conditions prevailed in the past, would have been the same, but the gross yield on the lender's portfolio would have been smaller.

Figuring the cost of the insurance at one-half percent-and I would like to interpolate at this point that by prepayment of premiums this could be reduced to three-eighths of 1 percent-the net return before and after taxes earned by a marketing firm and the yield on capital invested this way would be as follows:

And there is another table showing that. (The table referred to follows:)

* Assuming maximum borrowing and maximum portfolio-$100 million.

The final column in that table shows the resulting yield on $5 million of capital. And that table assumes maximum borrowing of the firm and maximum portfolio at a hundred million. They range from 6.82 percent on capital in 1957 and 4.42 percent in 1959.

In addition to income from its debenture issuing activities, the marketing firm could earn income from its mortgage trading activities. Employing a method shown in the volume, "The Secondary Mortgage Market, Its Purpose, Performance, and Potential," by Jones & Grebler on p. 229), the potential annual volume of mortgage trading on uncommitted basis would be upwards of $1 billion. And I should add at this point when I say "uncommitted basis" I am not

referring to the participation activity of savings and loan associations; this is entirely apart from those activities.

At a charge of 1 percent, the marketing firms could obtain an income of $10 million.

That such a spread would be available to marketing firms seems likely from an examination of other markets. A spread of this size is often shown between bid and offer prices in long-term Federal land bank bonds and International Bank bonds. Even U.S. Government bonds are subject to a spread of one-fourth of 1 percent between bid and offer.

From these considerations, I have concluded that there is a reasonable prospect for the success of these firms had they been organized during the years cited. A group of investors considering applying for a charter would, of course, need to make a more detailed and searching study, but I believe that they should be given an opportunity to apply for a charter should their studies lead them to conclude that profitmaking opportunities exist.

In other words, once again I urge the passage of the proposed act. Thank you.

Senator SPARKMAN. Well, thank you very much, Dr. Carr, for a very illuminating discussion.

Let me ask just one question as a matter of curiosity on that last table that you showed.

You show the yield that would be expected in the last column of 1957 would be-is that 6.8?

Senator SPARKMAN. More than one-third less?

Dr. CARR. Yes.

Senator SPARKMAN. What would be the explanation for that?

Dr. CARR. Well, to answer that, Senator, we would have to go back to the previous table on the previous page.

Senator SPARKMAN. Well, I know; I just wanted you to tell me offhand.

Dr. CARR. The reason for that was that the cost of funds rose; that is to say, in the market Baa yields rose. And, indeed, the spread between new issues and seasoned issues also rose.

So that the money market relative to mortgage rates was increasing. It increased faster than the indicated yields on mortgages. So mortgage rates also rose. Fees on mortgages fell, but there was a fairly sizable rise in the cost of funds-32 basis points on Baa yields and 6 basis points on the spread between new and seasoned issues.

Senator SPARKMAN. Now, on page 7, that last table which you have, in the first and the second and third columns you give the net return before tax and after tax.

Now, how, then, do you get the figure for the last column if your net return after tax in 1957 was .341 and you get a yield of 6.82? Dr. CARR. Well, let me try to explain it this way.

Senator SPARKMAN. What is the connection between the net return after tax and the resulting yield?

Dr. CARR. The .341 can be thought of as a return per dollar. Consequently, since you have $5 million that you are working with, the