Mr. CARDWELL. You are confusing error rate

Mr. EARLY. I am not trying to draw an analogy between the error rate and not being able to identify the eligibles. Those are different things. Back when you incorporated this program with the States, you assured the States that they would notice that the figures were probably very underestimated, but they would know within a year. We don't seem to be getting closer to it.

FEDERALLY ADMINISTERED STATE SUPPLEMENTATION

Mr. CARDWELL. The number of persons we are paying for on behalf of the States-that we can pinpoint right to the number. We have two kinds of relationships with the States. In the case of 28 States and the District of Columbia, we administer a basic benefit for a finite number of people. We know the number at any point in time. We can tell you the number of States and tell you the State.

Mr. EARLY. What States dropped out?

Mr. CARDWELL. Illinois was one; Minnesota, South Carolina, and Utah have also dropped out.

Mr. EARLY. How many States have gone over the hold-harmless provision originally incorporated in the program?

Mr. CARDWELL. Large States such as California and New York have. I could give you an analysis for the record. [The information follows:]

1/A State may choose to exceed its CY 1972 expenditures by the optional supplementation levels it sets.

Mr. EARLY. I thought Massachusetts went over.

Mr. CARDWELL. Massachusetts may have or may be very close to it. Massachusetts is one of the most generous States in the entire field. Mr. EARLY. Several of the commitments that we made by the Federal Government when they went into SSI aren't being fulfilled.

Mr. CARDWELL. I would like you to tell me which ones.

Mr. EARLY. I hope to identify some of them. You speak of a 25percent error rate. That is in dollars. How much of the $5.278 billion? Mr. CARDWELL. The 24.4 percent relates to the number of cases. In dollars it equates to about 10 percent of the $5 billion for combined Federal and federally administered State supplementary benefits paid during fiscal year 1975.

Mr. EARLY. That is $500 million.

APPROPRIATION ESTIMATE

Mr. CARDWELL. Yes.

Mr. EARLY. In 1976, you appropriated $5,518 million and you had an unobligated balance of $242 million.

Mr. CARDWELL. Yes.

Mr. EARLY. Would you explain that for us?

Mr. CARDWELL. When the 1976 estimates were made we estimated. that at the beginning of fiscal year 1976, 4 million people would be receiving Federal benefits by the end of fiscal year 1976 they would increase to 4.5 million. Actually we started fiscal year 1976 with 3.7 million individuals receiving Federal benefits. Based on experience we thought recipients would increase to 4.2 million by the end of 1976. The lower number of recipients accounts for the decline in estimated 1976 Federal payments. Well, it looks as though our estimate of Federal benefits is off by about 5 percent.

ERROR RATES AND OVERPAYMENTS

Mr. EARLY. In your comments, you say that you don't think you can reduce the 25 percent error rate; or you are not too optimistic it is going to be anywhere near the OMB's suggested 15 percent.

Mr. CARDWELL. No. I say 15 percent is our target not OMB's and we would reach that estimate by the end of 1977; that to go below that, in our experience, I suggest that you would have to modify the law. Mr. EARLY. I misunderstood you. You think you can attain 15 percent?

Mr. CARDWELL. Yes, sir. I would remind you that that would probably be about a third, maybe a little more than a third of the Massachusetts rate for public assistance, which is one of the highest.

Mr. EARLY. I take from your last comment that this might be onethird of the error rate of Massachusetts.

Mr. CARDWELL. For AFDC.

Mr. EARLY. For AFDC. That was a 52-percent error rate in overpavments and underpayments.

Mr. CARDWELL. I am saying 15 percent is significantly below that. Mr. EARLY. It certainly is. You have to look at that error rate that has been reduced. There is no justification of an error rate of that size. That is an error rate of overpayment, underpayment, and total mismanagement.

Mr. CARDWELL. Ours is the same, overpayment, underpayment, payments to ineligibles, all three.

Mr. EARLY. Could you tell us how much of the error rate is overpayment?

Mr. CARDWELL. Yes. On the 24.4 percent, 11 percent of it is overpayments, 7.7 percent is payments to ineligibles. Those two categories add up to 18.7 percent.

Mr. EARLY. So the underpayment is 7 percent?

Mr. CARDWELL. No; underpayment is 5.7 percent.

Mr. EARLY. Could you tell us the dollar factor of the 18-percent overpayment? I take the ineligibles as overpayment because they are not entitled to it.

Mr. CARDWELL. Well, it is in the neighborhood of 10 percent of the Federal benefits and federally administered State supplementary payments.

Mr. EARLY. That would make the underpayment the same amount. Mr. CARDWELL. No.

Mr. EARLY. I thought we identified $500 million.

Mr. CARDWELL. $500 million is an extrapolation of erroneous payments identified by the quality assurance case samples. These are total dollars in error-Federal and State.

Mr. EARLY. I thought the Department reported that its audited system overpayment tally was $547 million and that your quality appraisals system reported it was $579 million. Would you explain the difference or would you reconcile the difference? That discrepancy is great. Which is correct?

Mr. CARDWELL. That's correct; the $547 million figure is a correct figure.

Mr. EARLY. For 2 years?

Mr. CARDWELL. Yes.

Mr. EARLY. What about the $579 overpaid in 1 year?

Mr. CARDWELL. Let's see if we can get something straightened out as a basis for conversation. First, we are talking about a period of more than 1 year. Next, there are two sources of estimating the dollar value. One is the so-called computer tally, which reflects the actual differences that have literally occurred in the system that have been identified and recorded. They have actually taken place, and the system is tracking them. It is accumulating them all the time. That figure if we carried it into infinity would be so large it will just frighten anybody in the world. I am not sure it would be wise to carry it forward. Mr. EARLY. As we carry it on, I would think it would be diminishing Mr. CARDWELL. No. What is diminishing is the $579 million. The QA system does not measure dollars; it measures cases. It identifies a number of cases in error. It is a carbon copy of the system that was applied by HEW in 1972, and it prescribes a formula for converting the number of cases to the number of dollars and that will produce a much higher figure, and what it suggests is that either the formula itself needs to be reconsidered, and it is not our formula, it is a formula that we adopted from the original one, or that the QA system is identifying, and this is to be expected, I think, as a general proposition, and as a measurement, more errors than are actually being detected during the day-to-day performance.

Mr. EARLY. That figure that you couldn't identify-$579 million— which I said was overpaid in a 1-year period, from the report which you sent us, page 5.

Mr. CARDWELL. I don't think it says a 1-year period.

Mr. EARLY. It says from July to December 1974, from January to June of 1975, which I take it is a 12-month, 1-year period. You had $304 million plus $275 million.

Mr. CARDWELL. That is taking the case error rates interpreted into dollars. It does not represent the actual dollars that would have been paid out.

Mr. EARLY. Would you run that by me again?

Mr. CARDWELL. It is complicated.

Mr. EARLY. That isn't a dollar rate?

Mr. CARDWELL. It is an estimated dollar value of all the errors that the QA system itself identified; but remember, during the normal course of daily transactions, all those cases are not actually occurring. The QA system tends to look at the whole program after the fact and say, if every provision of law were carried out literally, everybody reported the way he is supposed to, if every claim was calculated the way it was supposed to be calculated, this is what the error rate would be. It tends to overstate real life, and if you took that and extrapolated the dollar values, you would get a much higher figure than the actual error rate that we are identifying and correcting as we go about our daily business.

Mr. EARLY. Along with those that you are actually identifying, aren't you also missing some?

Mr. CARDWELL. Yes.

Mr. EARLY. That is a fairly conservative figure also.
Mr. CARDWELL. $547 million.

Mr. EARLY. You identified $547 million.

Mr. CARDWELL. This is where expectations come into play. It is unreasonable to expect the law to be executed so literally as to be absolutely error free. The unanswered question is, what is the reasonable tolerance? What would be acceptable performance? The QA system that was designed and imposed on the States back in 1972 assumed that tolerance level should add up to about 13 percent, and that though one would never have to or expect to account for the dollar value of that difference.

Mr. EARLY. It is hard to think of one out of every four cases being in

error.

Mr. CARDWELL. It depends on how you define error. I think if we have made any mistakes in our public relations and congressional relations, it is by being too general in the use of that term. We count as an error, for example, the strange case that involves this particular law's definition of the payment period. The law says, the payment shall be calculated in advance per quarter, but the payment amount will be paid out to the individual in three installments monthly. We enroll you on the assumption that your income for that future 3-month period will entitle you to $100 a month. We pay you the first check of $100. You were acting in good faith and we were efficient in this hypothetical

case.

In the second month your nephew gave you an odd job that gave vou $50 worth of income and reduced your entitlement by $50, not for that