For example, our belief is that in the basic social security program we collect back perhaps a half or more of overpayments so that the net error would be less than half.

Senator NELSON. Well, what is the most common one? The most common one that I see from my constituents is some person who, I would assume almost always, has a dependent who is now no longer a dependent and the person receiving benefits for the dependents does not realize that the dependent is no longer eligible and draws a check for 3 more months or 6 more months. Or they were dependent in college and then got out-is that the most common?

Mr. Ross. Yes. I would say that the majority of errors are caused by changes in beneficiary status or characteristics which are not reported to us.

Senator NELSON. So to get at that point, there may be a better-I do not know what the methodology for correcting it is-but it is not mistakes being made by employees of social security since there is no way they are going to know about a change of status and eligibility for a payment except being notified by the beneficiary, is there? Mr. Ross. That is correct, sir.

The one thing we can do, and I have described it in my testimony, we have launched a number of projects to do a better job of educating beneficiaries to the things that they must report to us in a timely fashion. I think that we do have the burden of doing the best job we can to educate individual members of the public on when they must come back to us with changes in status or characteristics so that we can get the payment [rolls] corrected.

We do not regard the fact that it is a beneficiary caused error as an excuse for not trying to improve performance. We think we can do a better job on helping beneficiaries to help us to make this system run better.

Senator NELSON. Is overpayment the major part of the so-called error, or whatever?

Mr. Ross. Yes, it is, sir.

I was just going to refer to a pilot study which we have recently completed which surprised us with some preliminary data on what the major causes of error are. We did a test involving 3,000 cases primarily to do a serious trial run on this overall quality assurance program that we are implementing and in which we are investing a great deal of

resources.

Because it was a trial run, the measurement techniques do not meet rigorous statistical standards. However, it was a serious pilot study and the data, with those qualifications, is helpful to tell us where the problems are and where we can get working on corrective action even pending the full operation of the system.

This pilot study identified an 0.5-percent rate of overpayment errors and an 0.08-percent rate of underpayment errors.

Senator NELSON. 0.08?

Mr. Ross. 0.08.

Senator NELSON. Less than 1 percent on underpayments.

Mr. Ross. On underpayments-less than one-tenth of 1 percent. Senator NELSON. And one-half of 1 percent of all payments are overpayments?

Mr. Ross. Yes.

Senator NELSON. Of the total dollars, is that it?

Mr. Ross. If that percentage were applied to all of our payments in 1979 it would represent $445 million in overpayments and $72 million in underpayments, and this is on a base of almost $90 billion of benefit payments in these programs in 1979.

Senator NELSON. And did you say a few moments ago that you collect back 50 percent of overpayments?

Mr. Ross. We believe that is right; at least that amount.

Senator NELSON. So instead of-what was it, $400 million in overpayments?

Mr. Ross. Yes, sir. It would come down to more like $200 to $250 million. That might be net error, if you follow out these numbers and, indeed, that is the amount in the Inspector General's report that is our target for corrective actions to produce cost savings, roughly $250 million.

Senator WALLOP. Excuse me, Mr, Chairman. His hindsight is much greater than you would suggest by your pilot study. The Inspector General's hindsight of his forecast is much greater than

Mr. Ross. His figures, as well as GAO's figures, are based on extensions and speculations of the data we have, so that the data I am giving you is really the basis of the Inspector General's speculations, just as it is the basis of mine.

There really is not an independent data base that is available, apart from what is in our records.

Senator WALLOP. Would the reporting suggestion that Senator Schmitt and I and others are talking about have any effect on providing that data base?

Mr. Ross. The matter of posting of earnings is one that has concerned me, and with your permission, I would like to have inserted into the record a statement I released on posting of earnings which I think contains the numbers and some further detail that was mentioned in your statement.

[The statement referred to follows:]

STATEMENT BY STANFORD H. Ross, COMMISSIONER OF SOCIAL SECURITY

One of my first priorities as Commissioner of the Social Security Administration was to review the agency's performance in posting individuals' earnings to their correct name and social security number. These are the records which determine a worker's entitlement to benefits under the program. They are vital to the efficient operation of the social security system.

My basic finding is that SSA's performance in crediting these earnings is excellent. A national rate of better than 99 percent accuracy has been reached, which I believe is a remarkable achievement. The American public can be confident that we are keeping accurate track of their earnings records.

I have found that the agency accurately credits all earnings that are reported with a correct social security number and name. Problems arise when employers or the self-employed report incomplete or inaccurate information. Over the years about 5 percent of the earnings items submitted to the Internal Revenue Service and subsequently to the Social Security Administration by employers have an incorrect social security number or name, and an additional 1 percent have no number or name. The name and social security number on the earnings report must agree with our records before the item will be credited. This 6 percent equates to about 6 million items per quarter.

The Social Security Administration is able to correct all but 2.6 percent of the items incorrectly reported by employers and the self-employee through the use of modern computer technology. The remaining 2.6 percent must be manually corrected by asking the employer or self-employed individual to furnish the correct social security number or, if necessary, the wage earner's address, thereby enabling us to contact the person directly.

After additional corrections are made from this procedure, the remaining 1.4 percent improper reports are recorded in a "Suspense File" where they are readily accessible. They are never destroyed. They are available, for example, if a person advises us that all of his or her earnings have not been posted. This file is then checked as part of the effort to complete someone's earnings record.

Social security deals in large volumes. For example, this year it will pay out more than $100 billion to some 35 million beneficiaries. This figure represents almost one quarter of the entire Federal budget. It has paid $768 billion in benefits from 1937 through 1977.

In the area of earnings reports, the numbers are truly mind-staggering: Since 1937, when social security began, employers have submitted over 10 billion individual earnings items involving $9.7 trillion in wages and self-employed income. The Social Security Administration has properly credited all but 142 million of these 10 billion items reported since 1937. While these items represent almost $69 billion in wages, they also represent only the seven-tenths of one percent of wages for which we have received inadequate documentation which we were unable to correct. Usually these incorrect wage reports are for short-term and transient labor, such as restaurant workers or carwash employees. Even with these incomplete reports, social security has properly credited 99.3 percent of all wages.

A very significant change took place in the reporting system for wages earned in 1978 and subsequent years. Employers now report annually, rather than quarterly, eliminating an enormous amount of paperwork for them. I am convinced that the introduction in 1978 of this new annual wage reporting system will enable us-employers and the Government-to further improve the accuracy of wage reporting. Nevertheless, I am concerned to find that .7 percent of all wages had been reported with incorrect information that could not be credited to any specific individual. Accordingly, I have initiated a five-point action plan to further improve performance in this area:

1. SSA contacted the news media and asked their help in publishing stories to emphasize to employers and employees the importance of making sure that W-2 forms contain accurate information.

2. Through improved computer processing routines we have identified 6.4 million unposted wage items from previous years that were 1 digit social security number errors. This involves $3 billion in unposted wages which are now being posted.

3. Similarly in March-April 1979, we will credit an additional 1.2 million unposted items from previous years that involved errors in the surname. This involves about $560 million in unposted earnings.

4. SSA is currently designing a computer system to aid in the correct identification of erroneously reported earnings items.

5. We are doing a statistical analysis, using the records of earnings from 1937 to 1955. In general, most benefit payments calculated for claims filed today do not use monies in the computation that were earned prior to 1956 to establish the correct benefit amount. With this study as a basis, recommendations will be made in April about reducing our inventories in the suspense file.

As a result of these five actions, I am confident the Social Security Administration will overcome reporting errors to the maximum extent possible to come as close as possible to 100 percent accuracy in posting individuals' earnings.

Mr. Ross. We have instituted, with the help of GAO, a number of actions to try to improve our performance in this area. We are presently at 99.3 percent accuracy in that area. We hope that

Senator WALLOP. At what point?

Mr. Ross. 99.3 percent. That $69 billion figure represents all of the accumulated unposted earnings since the beginning of the program in 1937; we maintain them in a suspense account, and we find ways to reconcile some of that when beneficiaries come in and complete their earnings records.

But, beyond that, we have instituted some new computer methodologies and other steps to try to come as close to 100 percent accuracy as

we can.

Senator WALLOP. Excuse me. I am a little lost and on new ground. The 99.3 percent figure is related solely to the posting of earnings? Mr. Ross. Yes, sir.

Senator WALLOP. Other errors are outside that?

Mr. Ross. Yes. The other errors, which I am going to talk about now, are strictly on the payment side as opposed to the posting of earnings.

This pilot study identified to us where these payment errors were. A good number involved late notification of a beneficiary's death.

Another major cause of error is where students receiving benefits were not in full-time school attendance. Now, we have already instituted a program to deal with the student problem and we are, this spring, requiring verification that students are enrolled full time. We will do matching of our student rolls against the Office of Education's rolls. We feel we already have put in an action program that will help us save perhaps $50 million or more in the student benefit area. Now, another major problem

Senator NELSON. May I ask a question?

Mr. Ross. Yes, sir.

Senator NELSON. Is my memory wrong on this? I had thought the Social Security Administration thought they lost very much more on that, on people who were-well, for example, who got paid through the summer and intended to start college in the fall and did not start. Therefore, they would not have been eligible for the summer payments. Is that correct?

Mr. Ross. That is one of the problems, yes, sir. If the student does not go to school, the summer payments then become an overpayment. Our overall figure from our latest study is that the amount of error in this program, which now pays $1.6 billion a year in student benefits, is about $150 million. Our first thrust at it, we hope, will save about $50 million. We will not stop there; we will continue to try to improve, and we have some action plans that we think will close the gap on the errors in dealing with students.

The disability insurance program is another area where we know we have errors which require that we improve. The major source of payment error occurs when beneficiaries fail to notify us of successful returns to work. But the problems in the DI program run much deeper and have led us to conclude that we need to restructure the entire administrative system in order to get higher quality initial determination decisions, and we have been working with the Social Security Subcommittee of the House Ways and Means Committee on these matters. The Ways and Means Committee is going into a markup of that bill today and tomorrow, and we hope that out of that legislative change and the complementary administrative change we will be making, we will be able able to run a much tighter and more efficient disability insurance determination process.

Senator NELSON. This subcommittee will be conducting hearings on that after the House's actions. Are you satisfied that the disincentives have been removed from the system in the proposal that the House will be acting on?

Mr. Ross. I am satisfied that if the subcommittee bill basically follows the lines that it is, it makes very good progress on providing incentives for rehabilitation and going back to work and also for helping to hold down the number of people that go on the disability rolls. I think that it has real progress in it.

But, of course, you will have to make your own judgments when we come over here following the House action on whether we have gone

47-524-79-3

far enough, not too far-there are no easy answers in that area, and I would not propose to prejudge the feelings of this committee on some of the very difficult problems that we have.

One of the problems we have is that, over the years, the program has become more and more complex. At one time, in the early days, the retirees' basic benefit involved a simple computation applied to the worker's average monthly wage. The beneficiary might well, in days gone by, be able to understand enough to compute his or her own benefits.

That simplicity has given way over the years to an extremely complex series of computation methods and interacting conditions that make benefit computations highly error prone. I believe that today it is doubtful that any beneficiary would have the knowledge and sophistication required to compute his or her own benefit amount. Thus, beneficiaries are even more dependent on the accuracy of our efforts and our responsibilities are enlarged. To meet that challenge, we are bending every effort to improve our accuracy so that the public can have a feeling of confidence and can avoid the frustration that comes from not getting the right amount on time.

I lay out in my testimony, and will not repeat it here, some critical examples of how the programs have developed with overlays of computation methods, one on the other; and one of the things I am trying to stimulate is a review in today's world of this complexity to see if we cannot come up with legislative and administrative actions which would allow a reduction of this complexity without unduly disadvantaging beneficiaries.

I really think that it behooves us all in the executive branch as well as the legislative branch to try to see if there are not ways to simplify our programs so that they will be more understandable to the public and less error prone in terms of administration.

Finally, Mr. Chairman, I would like to say that the HEW position is-and the Secretary has emphasized this throughout all of our programs is that it is terribly important to try to run them as efficiently as possible, because the program dollars saved can then be used by the program for the benefit of the vulnerable people of our society. We believe that by being efficient managers we are showing compassion for the people served by our programs. Only when the integrity of the programs earns support from the American people as a whole, will we have the kind of public support that continues to support these massive social programs, and we hope that we can continue to work together with you to improve these programs and insure the support of the American people.

I thank you and I will be glad to answer your questions. Before I do, I would like to address the one point of Senator Wallop's suggestion which I have been doing something on. I also share the belief that, if the taxpayer who pays into this system every year could receive some tangible example of our concern in that individual taxpayer's interest in this program, there would be much broader support for these programs.

Therefore, I have put together a team being directed by the Deputy Commissioner for Programs, Mr. Robert P. Bynum, to study the feasibility and cost of supplying an annual notice to each taxpayer.

It requires some study. There certainly must be a balance of how much information you can supply. You do not want it to be so much