poor are transient poor. They should be separated from the chronic poor, not only because of presumed different nutritional resources but also because their food purchasing habits may differ with equivalent resources. The specific differentiation of these two categories of poor cannot be accomplished with the data from HANES I. A second HANES program is planned to begin data collection in 1976 in order to measure changes in nutritional status over time, and that new data will complement the HANES I data relevant to defining more precisely the groups at risk of malnutrition.

(2) Will the data for different demographic groups suggest clear policies for targeting food stamp outreach efforts?

If a national demographic subgroup has a sufficiently large number of persons who are undernourished (greater than 1 to 2 percent of the total United States population) the present HANES survey sample sizes can identify them sufficiently to provide data relevant to targeting policies. For some demographic subgroups which probably include relatively high proportions of persons who are malnourishedpoor, rural "Chicano" preschool children, for example-the numbers who are malnourished probably make up much less than 1 percent of the population, and the presently designed HANES surveys cannot identify those groups. The foregoing comment addresses only the matter of sample size. There is also the problem discussed in answer to your first question and the further difficulty presented by the fact that clear-cut early measurable malnutrition is not common in the United States. We are exploring ways to best use the several separate indexes we have obtained of high risk of malnutrition.

(3) Would it be possible to project the demographic findings regarding nutritional deficiency of your present sample to geographic areas by use of Census data? What would be the probability range of such projections?

I cannot give a definite "yes" to the first question, nor can we yet assign any probability range to projections which might be developed by such a method. This approach, however, is the one in which we are now actively engaging, and we believe it holds the most promise. Of course, such projections are limited by the extent to which the pertinent demographic variables related to the risk of malnutrition. are actually available from the Census. All of the problems mentioned earlier, such as limitations of sample size for subgroups at risk of malnutrition, of course, also are relevant.

(4) If such projection is not feasible, what would it cost to expand the sample to a county level on some sort of cycling basis, such as once every five years? Would the county level be sufficiently homogeneous to indicate geographic priorities for outreach efforts?

The question suggests that there may be a misunderstanding as to the frequency of the HANES programs at the present time. The HANES I program began data collection in 1971, and the nutritional component was completed in 1974. The second HANES program is scheduled to begin in 1976, so that we are currently on basically a five-year cycling basis with a sample size which permits us to make national and regional estimates but does not permit estimates at the State, let alone county, level.

The cost of expanding the sample to a size that would permit estimates at the State level is far greater than seems feasible. One other survey program conducted by the National Center for Health Statistics (NCHS), the Health Interview Survey, does plan to expand the sample size in fiscal years 1977 and 1978 so that by the latter date the sample size will have been quadrupled. The present sample size of that program is more than five times as great as the size of the HANES program. Of course, a calculation of appropriate sample size to detect malnutrition at the State level requires consideration of other matters, such as expected levels of nutrition, but these affect the calculation in the direction of requiring an even greater expansion than the twentyfold expansion suggested by the comparison with the other NCHS program. We believe that the kind of approach you suggest in your question 3 is the one which should be explored, and we are doing this. Some future expansion of the HANES surveys' sample size-perhaps a doubling or tripling-may eventually be demonstrated to be desirable in order to collect nutritional data on adequate numbers of persons in the demographic subgroups at risk of malnutrition. This could be accomplished, however, only if required additional resources-both money and personnel-were to be made available to the NCHS for that purpose. Identification of such subgroups will be done in part through use of present HANES data, and the process will be facilitated through using some information from other sources such as data collected in connection with surveillance programs of the Center for Disease Control. Once relationships of malnutrition to Census type variables are better understood, it is hoped that some local level projections may be made and then validated. Another advantage of this approach is that characteristics identifying high risk subgroups may be used for targeting nutritional programs and allocating

resources.

If we can be of further assistance to you or members of your staff, we will be happy to oblige.

Sincerely yours,

ARTHUR J. MCDOWELL,
Director, Division of Health
Examination Statistics.

APPENDIX D

MAY 2, 1975.

Hon. LAWTON CHILES,

Chairman, Subcommittee on Federal Spending Practices, Efficiency and Open Government, Government Operations Committee, U.S. Senate, Washington, D.C.

DEAR MR. CHAIRMAN: At the hearing before your Subcommittee, Monday, April 28, 1975, on the efficiency of the Food Stamp program, I promised to submit for the Record certain additional information. Pursuant to that commitment, I am submitting herewith:

(1) A chart, with explanatory material, indicating the success of the program's effort to reach the "poorest of those eligible";

(2) Quality Control in the Food Stamp Program, NonpublicAssistance Households, January-June 1974;

(3) Food-Aid Program Effects on Food Expenditures and Levels of Nutritional Achievement of Low-Income Households;

(4) Families in the Expanded Food and Nutrition Education Program; Comparison of Food Stamp and Food Distribution Program Participants and Nonparticipants;

(5) Program Evaluation: Food Stamps and Commodity Distribution in Rural Areas of Central Pennsylvania;

(6) The Food Stamp Program: Del Norte and Humboldt Counties, California;

(7) Options for Improving the Equity and Efficiency of Benefit Determination Procedures for the Food Stamp Program.

If additional information is desired, please contact me and we will be pleased to cooperate.

Sincerely,

EDWARD J. HEKMAN,

Administrator.

REACHING POOREST OF THOSE ELIGIBLE

The Food Stamp Program is designed to assist the poorest of those eligible with maximum benefits. Benefits are reduced proportionately as income rises to provide work incentives and to give vertical equity with higher income people. At the point of eligibility, the program is designed to be only marginally attractive. For this reason, participation is relatively high at the lowest income levels, and low at the highest levels.

Comparison of data from our recent National Survey of FSP recipients with population data from the Current Population Survey shows that:

-Nearly 72 percent of the U.S. population earning less than $2,000 a year received food stamps in 1973.

-Approximately 51 percent of the U.S. population earning less than $4,000 received food stamps.

The above comparisons are lower limits of the percentages of people being reached by the program that are eligible to participate, for the following major reasons:

-Some people at low current income levels are not needy and not eligible to participate because of large savings and other

assets.

-Many people participate on an infrequent basis. The program reaches an estimated 70 percent more people over a 12-month period than in any given month.

REACHING POOREST OF THE POOR

The Food Stamp Program is designed to assist the poorest of poor with maximum benefits. Benefits are reduced proportionately as income rises to provide work incentives and to give vertical equity with higher income people. At the point of eligibility, the program is designed to be only marginally attractive. For this reason, participation is relatively high at the lowest income levels, and low at the highest levels.

Comparison of data from our recent National Survey of FSP recipients with population data from the Current Population Survey shows that:

-Nearly 72 percent of the U.S. population earning less than $2,000 a year received food stamps in 1973.

-Approximately 51 percent of the U.S. population earning less than $4,000 received food stamps.

Only about 13 percent of total food participants were from high income households-$7,000 or more a year.

-In 4-member households approximately 20 percent of the participants had cash incomes that exceeded the FNS income threshold, but were eligible to participate in the Program due to the income disregards that lowered net income.

The above comparisons are lower limits of the percentages of people being reached by the program that are eligible to participate, for the following major reasons:

Many people at low current income levels are not needy and not eligible to participate because of large savings and other assets. We estimate this figure to be as high as 15-20 percent of the low income group.

--Many people participate on an infrequent basis. The program reaches an estimated 70 percent more people over a 12-month period than in any given month. On that basis the program now reaches 30 million different people at least once during the year.

APPENDIX E

PRESENT DEDUCTIONS FROM INCOME

In determining a household's eligibility for the food stamp program, some deductions from total income are allowed for:

(1) Federal, State, and local taxes, Social Security payroll taxes, and mandatory union dues;

(2) 10 percent of gross earned income up to $30 per month; (3) Medical expenses, if they are more than $10 a month; (4) Child care, or care for an incapacitated adult, when such care is necessary for a household member to work;

(5) Educational expenses, including tuition and mandatory fees, but not books, living expenses, or other education-related costs;

(6) Court-ordered alimony and child support;

(7) Unusual expenses due to disaster or casualty losses;

(8) Shelter expenses (mortgage payments, rent and utilities) above 30 percent of income after all other deductions have been taken.

The purpose of these deductions is to disregard the portion of a food stamp applicant's income which is spent in the specified categories when calculating total resources available to purchase food, and, therefore, in determining the applicant's eligibility.

COST SAVINGS FROM SUBCOMMITTEE'S RECOMMENDATION

Projecting forward the effects of proposed changes, the following is expected, based upon the simulations in the Mathematica, Inc. studies:

1. A reduction of 12% in the total number of households participating; that is, the 646,000 households with the highest gross income levels would cease to be eligible. These would not be elderly persons. The elimination of these households from the program would reduce the total monthly bonus costs by 5-7 percent, producing savings of $250 to $350 million for fiscal year 1976. 2. The simplified certification process and tighter quality control procedures should greatly reduce the costs of errors. The 8 percent of program participants reported by GAO to have been ineligible for reasons relating to income, assets and deductions calculations should be reduced toward zero. The underestimate of charges for stamp allotments should be reduced from 26 percent of cases to 3 percent, yielding a substantial reduction of bonus payments. These savings should be somewhat offset by enrolling the 7 percent of cases reported to have been improperly denied benefits.

4. Simplification of procedures should substantially reduce the costs of administration, which are estimated to exceed $150 million for fiscal year 1975. This will free funds to be used for increased outreach to high risk groups and better evaluation of nutritional status mandated by our legislation.