Q: Today we are interviewing Ken Sander who works in the Office of the Actuary and who will be retiring next month in December, after how many years at SSA?

Sander: Thirty-two years.

Q: Okay Ken, tell me how it is you came to work for SSA.  There are so many unusual stories of about how people get here. Tell me a little bit about your background and then how you ended up here?

Sander: Okay.  I started work for the government in July of 1963 in Washington.  I was living in Washington. I came to Washington because of a guy I had met in graduate school in Indiana University and... 

Q: That is where I went to school, by the way.

Sander: Oh really?

Okay. I stayed at his house for the first few days in Washington and actually the first night I was there we went out to College Park and met his girl friend, and she lived in Frederick, Maryland, which is about 45 miles from either Washington or Baltimore, sort of like a triangle type thing; but in any case one thing led to another and I  eventually married his girl friend.  But before that I was living between Scott and Thomas Circles in the fall of 1963, well Summer and Fall of '63, and that is as about as city as you can get.  That is between 14th and 16th street.  I originally came from a relatively small city in Indiana called Evansville.  It is on the Ohio river.  It never grew while I was there, in fact it shrank, that is one reason why I was in Washington, looking for a job.  But in any case I wasn't really all that happy living in Washington, although it did provide me with some unusual opportunities, 1963 being 1963.

I attended the March on Washington, I believe that was in August, and heard the "I Have A Dream" speech. I was still there in November, and was part of the crowd for President Kennedy's funeral procession; but I'll have to admit I didn't see much. But in any case, by that time I was  involved with my friend's ex girl friend and I moved up to Frederick in, oh I think in February of 1964, and proceeded to commute to Washington for a couple of years as a rider on a Greyhound bus or what have you.

Q: Now were you working for the Federal Government at the time?

Sander: I was working for the Federal Government, for the Department of Commerce at that time. What was then The Office of Business Economics, it is now the Bureau of  Economic Analysis.  It does the GDP and National Income Estimates.

Q: Were you trained as an economist or an actuary?

Sander: Yes as an economist.  The fact that I am an economist in the Office of the Actuary is a very long story that I think that we will get to at the end.


In any case, the commute to Washington was fairly long, it was fairly awkward, although the traffic was nothing like it is now around Washington, I think it was still--since we were still working in downtown Washington--it was still an hour plus, significantly more than an hour just to get in and out.  And this is actually relevant to how I came to Social Security.

I got tired of the commute and I was looking for alternative agencies that wouldn't involve me in a commute from Frederick--I was definitely tied to Frederick--to downtown Washington.  And there were a few choices. There was the Atomic Energy Commission, I guess it is the Department of Energy now, in Germantown, which was only about halfway down to Washington. Then there was the National Institute of Health, which was in Bethesda. 

I was certainly not qualified for doing anything in NIH.  And I actually did interview at the Atomic Energy Commission, but what the heck, you know that they were actually making atomic bombs then, I mean I wouldn't, it was not the Department of Energy, believe me.

But there was also the Social Security Administration, which happened to be on the outskirts of  Baltimore at the end of a nice duel highway called US 40.  And so I put in a "cold" job application to Social Security and a Branch Chief, who had recently come aboard there named Robert Brunner--I really don't know how it works, maybe Personnel directed applications at times--but in any case I got a call for an interview, came over, was hired, and started here in, I believe, in January of 1966.

And I want to tell you  the commute was great.  I got to recognize individual trucks.  This is unbelievable if you have ever commuted on what was then 70S, is now 270.  There was a Seagrams truck that started out in Lawrenceburg, I believe it is, Indiana, must have started out the evening before or perhaps the afternoon before and it drove to Baltimore, I suppose with a load of whiskey. And I saw that truck several times because we were basically on the same schedule.  He started out in Lawrenceburg about the same time everyday I guess and he wound up on that stretch of US40 between Frederick and Baltimore about the same time I was commuting to work.  And to be able to pick out an individual truck, to not see a car, either way, is amazing, but that is the way it was back then, virtually nobody was commuting long distance.

Q: It is not that way today?

Sander: Not that way today, no.  It's a nightmare compared to...

Q: Let me ask you. Did you have any background in social insurance?  Did you study it at a school?  Did you have any inclination about Social Security or about anything related to that or was it just a practical consideration?

Sander: It was a  purely practical consideration.  Despite the best efforts of my parents I wound up as something of a liberal democrat.  There were Republicans in retrospect, for no good reason, but I had sort of, I would say "Rooseveltian leanings," but I had no training per se in social insurance.

Q: And no particular attraction to the Social Security Program?

Sander: And no particular attraction to the Social Security program, other than it was a government agency and it had a wonderful location.

Q: Now did you come to work in the Office of the Actuary?

Sander: No, I started in the Office of Research and Statistics.  Actually we have only very recently been in the Office of the Actuary.  We, being the Revenue Estimating Function.

Q: Ah, okay.

Sander: I think it's within the past three years.  For most of my career I have worked in the Office of Research and Statistics and its many manifestations of name.  I think right now it is ORES and for a while it was ORSIP and you know how the organizations go.

Q: Now was this in the days when Ida Merriam was the head of ORS?

Sander: Yes.  Ida Merriam was the--I am not even sure exactly what her title was--she was the head of ORS at that time. Not that I met her all that many times.

Q: Alright, so you showed up for work at Social Security, and did you start right out in the revenue estimating business?  Or what did you do?

Sander: Well I think I had two early assignments.  One was, find out how that woman is making the revenue estimates and see if  you can improve on it; and the second was go through these printouts and see if you can come up with a good research paper out of them.

For the first two years, from 1966 to 1968, I did a paper on the retirement earnings test and spent time trying to figure out how the incumbent who was doing the revenue projections did it and how it could be improved. 

The paper was published very timely.  I think I have enough ego to say that in a sense my career peaked early, like about two years after I started here. The paper coincided  with a requirement on the agency to give a report to Congress on the Retirement Earnings Test.

Q: This is the impact if we were to modify the test, what would be the financial impact on the Trust Fund?

Sander: Well, no, rather more what was the impact on the beneficiaries of the provisions of the Retirement Test at the time.  I ascertained, and I honest to gosh don't think anybody has moved that research much further forward, that what was really important to the beneficiaries was the exempt amount, that is, how much they can earn without losing any benefits.  At the time there was a two-stage wedging out of  the exempt amount.  Where if you earned two dollars you lost one dollar of benefits and that was for a range and if you earned one dollar you lost one dollar of benefits.  I also have to say that this was one of the earlier supply-side pieces of research that I think was ever executed in the government.  The point I made was that you would not expect anybody to be in the one-for-one area of the retirement test because you are trading a tax free dollar for a taxable dollar and it is foolish to be there.  The more you earn the less you get in the terms of income.  That is, you go out and work you at least have to pay Social Security tax on it where as if you don't work in this stretch, if you hold your earnings down enough you can keep all of your earnings minus Social Security taxes of course, and all of your benefits; and that's what I found in the data.  There was a big cluster of people right around the exempt amount and the report to Congress emphasized pretty much what I had said in the paper that the really important thing is the exempt amount and if you want to do something for the beneficiaries get rid of the one-for-one, and the thing to focus on is the exempt amount.  I don't think anybody's really advanced this knowledge about the test much further than that.

Q: Well, that's probably because the fundamental truths remain about it, I suspect?

Sander: That's true, yes.  They did get rid of the one-for-one.

Q: Now was there any talk at that point about eliminating the test at all?  Was there any ...?

Sander: There has always been talk of eliminating the test.  There have always been people who were absolutely convinced that you would get a flood of labor if you just got rid of the test and turned the program into a straight annuity program.  Well there were some other old printouts around there that dated back into the 1950's, that we unfortunately no longer have.  Nobody wants to go back and look at 1950's data, but what they suggested to me was if you got rid of the earnings test you were at least as likely to reduce the work effort of most people, as you were to increase the work effort.  There is something about unearned income that reduces the "fire in gut" for getting out there and working.

Q: Let me see if I understand your point.  If they could collect their benefits without working . . .

Sander: The way it really works is, in a sense, the way I'd like to handle my job.  I don't necessarily really want to completely retire.  I'd love to work part time and keep my same income.  Part being annuity, part being earnings.

Q: Okay.

Sander: Lets take a worker who is earning, oh gee, I have not kept up with the test so I can't really say what the relevant amounts are.  Lets take a worker who is earning about twenty or twenty-five thousand.  I would suspect at the twenty-five thousand dollar level he has probably lost part if not all of his benefits. Let's say that we get rid of the test.  What I am saying is, the worker is no longer going to earn twenty-five thousand if he can have ten or fifteen thousand of unearned income dropped into his lap, the Social Security benefits, he's going to cut back to twenty. He's got the same income, and he is going to work less for it.

A very strongly held opinion of mine, that getting rid of the earnings test is not going to elicit a tidal wave of sixty-five year old's to go out there and work their little tails off.  No, it's going to go the other way. People who are earning well above the exempt amount now are very likely to pull back on their work effort.

Q: And you say that there was data going back to the 1950's that suggested this?

Sander: Yes.  In the 1950s the participation rate for workers 65 and over just fell tremendously. And the 1950s was also the time where not only were benefit amounts increasing significantly, but the amount you could earn without losing all your benefits for the month, I think it was a monthly test purely back then, were also increased dramatically.  Like in the 1940s, I think it was, that if you earned fifty bucks a month, bye-bye benefits.  The exempt amount items as I recall, boy I could be really wrong on these details, they introduced the notion of an annual amount in the 1950s and I think by the end of the 1950s they had increased it to twelve hundred dollars a year and although it is really difficult to sort out exactly what was going on then, since both relative benefits and the exempt amount were increasing simultaneously, I think an awful lot of it had to do with the fact that you could earn some amount of money and also get your benefits, that really pulled back the participation rate for workers age sixty-five and over.

Bluntly, if you want workers sixty-five and over to do a lot more work you institute another retirement test like they had in the 1940's where if you earned twenty-five or fifty bucks in a month, you don't get a benefit.

Q: Then you will get a surge in work?

Sander: That's right, then you'll get a surge in work. 

Q: All right, let me take you back again to the start there as you as you came into ORS and got into the revenue estimating business.  Now you alluded to the fact that there was some woman there already doing that business?

Sander: Right.

Q: And I want to ask you about that, but I also want to ask you the earlier question. Naively it would seem to me that you could make an argument that the Treasury Department ought to be the one doing the revenue estimating for this business.  Do you know any of that early history?  Why it is we do the revenue estimates instead of the Treasury and has that issue come up during your time here?

Sander: I think that if you asked Treasury, they would probably agree with your naive assessment that they should be doing the assessments, yes.  It hasn't really come up in the sense the Treasury has not brooked a proposal to take over the numbers.  I think that has to do with the history of  the Social Security program. And let's face it, with a real low tax rate and a real low tax-max, there wasn't all that much money that Social Security was pulling in the 1940's and 1950's.  It is only beginning in the 1960's and later that the money really became big.  Before that it was not all that important, I think, to Treasury that they do the numbers.

Q: I'll just tell you for the record also, I'll give you Harry Ballantyne's  answer to this same question, Ballantyne said to me in conversation that the reason that Treasury doesn't want to take it over is that they have so much confidence in Ken Sander that they don't have to think about it.  That's what he told me.

Sander: Harry has other hallucinations!

Q: So that is Harry's explanation.

Sander: I think a more recent explanation is, since the time of the disinvestment of the trust funds, when was that?  Back in the late 1980's when the government ground to a halt a couple of  times and one thing or another, at one time Treasury did cash-in some trust fund assets in order to keep the government functioning.  They got so much static on that from Congress that they would agree with Congress that Social Security is a "tar baby" to them and they want to sort of keep a nice distance away from it in terms of not rocking any boat unnecessarily.

Ah, let's see that is a very good question of how we actually really did wind up.  It's unusual, I think Treasury is probably virtually responsible for every other large chunk of revenue.  It is possible that OPM does the CSRS estimates.  In that sense if OPM did that as in effect a separate trust fund starting in the 1920's, but that might have been a precedent, if you viewed Social Security in effect as a separate trust fund and I guess there was some of that mystique even going back to the 1930's, that this was something of a separate operation, not a general revenue type of thing and Treasury does the General Revenues and maybe the retirement funds do their own, that is one possible explanation. It was probably historic accidents.

Q: All right so let's go back to that period in 1966 when you came here.  Tell me just a little about in general here what it was like to come to work here in 1966 at SSA, how you found the culture of the place and the operation of the place, just kind of in a general way?

Sander: I guess my career has gone long enough, I have one foot in the old Social Security and I'm retiring with the other foot out of the new Social Security.  When I first started to work here I worked in the fourth floor link.  At that time, I believe, they still did breaks in the Division of Data Processing, with bells at ten-fifteen, that was still going on.  There were fixed shifts that you came in.  Now there were several shifts, I guess starting maybe at seven-thirty and ending at eight-thirty or eight forty-five?    Such that  not everybody was going to start trying to get into the parking lot at eight-thirty in the morning. Traffic was much denser in terms of getting in and out of the parking lots, because everybody was leaving pretty much simultaneously and I think that there were possibly even fewer entrances and exits.

As far as the culture is concerned I suspect it was in transition then.  It was obvious that the mini-skirted women and the long haired men that were being hired then were not going to tolerate the regimentation that seemed to be part of the history of Social Security. For example, there was a fairly ridgid dress code that was eased up in the very early years of  my working here. I think it was probably eased up by around 1970. I am not sure of the dates on that. But you know, we had bells for breaks, virtually nobody went off the complex for eating lunch, because the lunch half hour was a little more rigidly adhered to than now.  That meant very crowed long lines in the cafeteria as compared to now.  Everybody coming and leaving at roughly the same time.

Based on my experience with a older secretary; heck she was probably younger then than I am now, but that's a different story.  I would say there is a Social Security way of doing things, that I never really got too terribly involved in, because definitely things were transitioning then.

And 1966 was protest time for the Vietnam War, Civil Rights, the whole thing.  It was a transitional period, but I had a bit of a flavor of the "older way" of doing things at Social Security one might say.

Q: Now Bob Ball was still here at that time as Commissioner. Did you see anything of Ball's influence at that time?

Sander: I never met him.  The only contact I had with him was very peripheral in the sense of seeing some of his famous "dictated but not read" notes or memos to respond to, answering some questions.  At my level, God could have been the Commissioner I would have been unaware of him.

Q: Some people thought that he was!

Sander: That may be the case.  As long as he is living he may be.

Q: All right.  You mentioned that Ida Merriam was the head of ORS when you got here.  I don't remember who took over ORS after that?

Sander: John Carroll.

Q: Ah that's right.

Sander: And then it turned into a revolving door.

Q: Okay.  Talk just a little bit about organizational stuff.  Is there anything in particular that stands out in your mind about any of those organizational transitions.  I especially want you to talk about the move of the estimating function into the Actuary's Office, but not yet, let's come to that. Is there anything before that?  Any of those leaders of  ORS that stand out in your mind for some reason or anything about organizational moves during this time that stands out?   If not that's okay?

Sander: I don't even remember all of the names of the Commissioners.  You are interviewing fairly far down.  I really didn't have that much contact with the Senior Staff.  I suppose at that time Larry Alpern was Senior Staff, he was Deputy Chief Actuary and that was about as high as I ever got.  As a matter of fact, you know I worked here for thirty-two years.

How many Commissioners have I personally met?  One, the present one.

Q: Okay.  Well let me ask you this then.  How about the relationship between ORS and the Office of the Actuary during those early years before you became part of the Actuary's Office.  Talk about that working relationship.

Sander: All Right.  Going back a little bit in history, ORS and the Office of the Actuary, I believe used to be in the same office.  I believe there was a reorganization around 1960 or 1963 that set up ORS and set-up what was then was OPEP, and set-up the office of the Actuary out of one very large--I think, it is called the Division of Program Analysis or something like that.  And the Actuary's office in Washington, some of that old organization still exists in the sense that the long-range actuaries on the seventh floor of Altmeyer were the Washington actuaries and the short-range actuaries on the fourth floor of the Operations, in the Link, were DPA people. 

Q: And DPA was what?

Sander: The Division of Program Analysis, I believe.  This is before my time, so I may be getting the names wrong.  What I really find interesting is,  I'd only been in OACT for three years.  I had contact with OACT for many, many more years--I can still see after thirty-plus years a certain level of "differentness" between the seventh floor and the fourth floor.  I mean it is just amazing how long this has lasted.  I don't know if it is culture or what have you that lasts even after organizations are joined together.  And I bet that is not at all that uncommon either.  Like forced marriages remain forced marriages.

Q: Now isn't there a similar distinction between part of ORS, which is in Washington to this day and the part of ORS that's here in Baltimore?  Isn't that the same issue?

Sander: That is true. Tying into what went on with DPA, the revenue projects were actually in the old DPA and I am not exactly sure of the logic of the time other than it appeared to be dealing with economic issues, rather than projections, they were put in the Office of Research and Statistics at the time. Go back far enough, you know, the revenue projection and OACT were actually together.  Now, as far as what can only be described as "friction" between the ORS Washington and ORS Baltimore, yes it goes on to this day.

In fact ORS Washington is known around Baltimore as "The University of Social Security." ORS Washington, has, at least up here, the reputation of being more interested in putting out more scholarly pieces of research than doing anything that might be really relevant for the program.  They have a life of their own.  This is not to deny that scholarly pieces of research are very useful to the agency, because they can be cited.  Perhaps it is a certain level of misunderstanding, but without a doubt, ORS Washington by choice, to some extent, wound up being disconnected from an awful lot of  the life of  Social Security.  Whereas ORS Baltimore produces the "Social Security Annual Statistical Supplement" with all of those huge tables, you know,  the history of the agency in numbers to some extent, and much more direct program-relevant research.  I've got a feeling, that probably Jack Schmulowitz had to have said roughly the same thing.

Q: He did, he did.  We talked about exactly the same issue and I just wanted to get your perspective on it.

Sander: It's more than Jack, I mean it's wide-spread.  You go over and start to talking to just about anybody you choose to in ORS Baltimore and I think that you will get roughly the same story.  There is a feeling that it is the step-child versus the glamour guys down in Washington.

Now of course Washington, once again, ORS Washington, ORS Baltimore was a forced marriage as you quite correctly note.  I am not exactly sure where they were, but once again it was that realignment in the early 1960's between two rather disparate groups that continues to have a distinction, or the two pieces continue to have a distinct life.

With regard to the revenue projections, I thought, given that the very high economic content of the revenue projections, there were no other economists in Social Security other than in ORS.  It was very logical to put them there and in a sense it was logical to keep them there.  However, we were ultimately tied to the Washington people as opposed to the Baltimore people.  I would say that the function was something of  a "step-child."

From the viewpoint of the Washington people, it was rather more of a published thing as opposed to a operational thing. Their natural inclination was to produce publications; my natural inclination was to provide the budget shop and the Office of the Actuary with the numbers that they needed to do the budget and the Trustees Report.  Writing came way, way down on my list of priorities when I finally took over the Revenue Estimates, took over doing them. We were essentially left alone--one could almost say, neglected.

Q: When did you take over doing the new Revenue Estimates?  Did that happen right away at the beginning of your career or was it a few years in?

Sander: Well I started doing the Revenue Estimates in 1968.  At the time this was under the direction of Bob Brunner.  He was arguably in charge of the estimates then.  I think in 1973 I more or less had absolute control over them,  my word was the final word on it.

Q: Okay.

Sander: I may be wrong about that, but that's so long ago that I can't remember the exact sequence.

Q: Now did you work on any other issues besides the Revenue Estimates in those first few years that we want to talk about.

Sander: Other than the Retirement Test?

Q: Other than the retirement test which was an excellent one and I am really glad that you talked about that one.  Anything else that came up during that time?

Sander: No.  Those were actually pretty busy years.  In the first place I was young and I had the fire in my gut, which I don't have anymore.  Setting up a method for doing the estimates, for improving the estimates, that took most of  the time.  And I think I only got away from doing most of anything in the revenue numbers in the, I guess,  in the 1970's, and that was rather more working with the Office of the Actuary on the Trustee's Report assumptions, the economic assumptions per se, as opposed to the revenue estimates flowing from them.

Q: Maybe now would be a good time for you to explain to us what the heck a revenue estimate is and what's involved in doing revenue estimates.  At some point we have to know about that.

Sander: All Right.  A revenue estimate to me is a payroll tax estimate, that is, the FICA,  Federal Insurance Contributions Act and SECA, Self-Employment Contributions Act,  tax that is levied on the earnings of the covered people.  It does not include General Revenue contributions; it does not include interest earnings on the Trust Funds.  Since the 1983 Amendments that is another piece of revenue for the program, the taxation of benefits. Interestingly, that one is in the charge of Treasury, not in charge of Social Security.  Now they get data on benefits from Social Security, for making their estimates, but those are Treasury estimates,  not Social Security estimates.  This could be because it in effect is a piece of the income tax as opposed to anything else.

Q: Which just happens to get credited to the Trust Fund?

Sander: Yes, that's right.

Q: But it is an Income Tax?

Sander: And also obviously OACT has to make the seventy-five year projections, and for the ten-year Trustee's  Report Projections, projections of the taxation of benefits, but the official numbers are Treasury's and that's an interesting thing, it never crossed my mind until now.

Q: So you have to estimate how much we are going to take in from the payroll taxes?

Sander: That's correct.

Q: For purposes of preparing the Trustee's Fund Report and projects, and for what else?

Sander: For appearing in the budget document. Our estimates are part of the determinations in the budget; projections if we are going to have a unified budget surplus or deficit.

Q: We need to talk some more about the mechanics of this, but you just mentioned one of the buzz-words that I want to ask you about and that's unified budget versus off-budget treatment of the Trust Funds.  Do you get into those kinds of issues in any way?

Sander: Let's see, yeah, they have an impact on what we produce.  The Revenue Projection print-out shows FICA Taxes, SECA Taxes, a line we call Receipts From The Public and then there's Federal Employer Taxes and then Trust Fund Collections. And that Receipts From The Public really amounts to the unified budget measure of revenues.  So yeah, to the extent that we have to refine our output to make a distinction between intra-governmental, that is Federal Employer Taxes and Receipts from the Public, there is a distinction in the revenue estimates made between the Unified Budget and the Trust Funds.  Did I get involved, did the Office get involved in the questions as to whether the Trust Fund should or should not be included in the unified budget, the answer is no.  That was decided at the OMB level.  That was decided, I believe by a Budget Commission back in the late 1960's, gosh what was the name of it, I can't remember.  But anyway it was a special Commission deciding what the Budget presentation was going to be.  HI (the Medicare Hospital Insurance Trust Fund) is off-budget by the way. We make the revenue estimates for the Hospital Insurance Trust Fund, which is not even a part of SSA, because the HCFA (Health Care Financing Administration) people have no capability, that I am aware of, for making their own HI Revenue Estimates.  They depend entirely on SSA, specifically us, for their hospital insurance payroll tax revenue estimates.

Q: So even though they have their own Actuaries, and even though HCFA and not SSA runs the Medicare program, they don't do their own revenue estimates.  Both Treasury and HCFA depends on your revenue estimates.

Sander: That's correct.

Q: That is good.

Sander: Around 1951 there was a change in the law that specified how the Trust Fund receipts were to be credited on the books of  the Treasury. Prior to 1950 or 1951, the actual collections were distinguished and these actual amounts were credited to the trust funds. Beginning in, I believe 1951, in a way to make life simplier for employers, the employers had only to report the lump-sum of withheld income and Social Security taxes and the law specified that from time to time--and I loved these things, they know how to write laws, just so totally ambiguous you could declare war with the Social Security Act I think sometimes, but anyway--from time to time the Treasury or the Secretary of the Treasury would transfer the receipts representing the estimated amount of Social Security taxes and at a later date, time unspecified of course in the law, the amount of  the transfers on an estimated-basis would be compared with the amount of the taxes that would be computed from the wages that Social Security processed and put either on a workers earnings record or in the suspense file.  That still is the system that the Trust Funds work under in getting their receipts credited.

What I didn't know, when I took over the job in 1968, and what I didn't know until sometime in the early 1980's, because it wasn't really wasn't relevant to me, is what happened to these numbers.  My job was to produce the numbers as best as I could, not to follow their happy way through wherever they go.

Treasury used the Social Security estimates of  FICA and SECA taxes to make the initial transfers. Harry is very fond of saying, without my estimates the Trust Funds wouldn't be getting credited for the taxes.  That is, of course, nonsense.  Treasury would find some way of doing preliminary estimates and transfers without my estimates. 

Q: But they use yours anyway?

Sander: They use our estimates.  And it is a very staggering thought to realize that right now, I think it is on the order of six hundred billion dollars, OSDI and HI, is transferred on the basis of one's work.    

Q: That is pretty remarkable.

Sander: It is fortunately transparent. I mean one sort of lives with it I guess, as viewing it all as "funny money." I have nothing whatsoever to do with the actual tax collections.  With the actual cash flowing from employers to the Treasury, this is all an accounting entry.  However I will say this......

Q: But it's hundreds of billions of dollars?

Sander: It is hundreds of billions of dollars.  Even without knowing that this was going on in the early times that I was doing these estimates, there were butterflies in the stomach when those numbers left the office.  What if there's mistakes in them?  That sort of thing.  This is without knowing if they were being used that way, however it hasn't happened like that for years.

I just do it now.

Q: So let me recap what you have just said to make sure that I have got it.  The way it works now is that basically the Treasury collects the taxes from the employers basically in a big pot of money?

Sander: That's correct.

Q: You do an estimate of what portion of that income should be credited to the Social Security Trust Funds in effect?

Sander: Yes I do an estimate of the amount of Social Security taxes--FICA taxes and SECA taxes.

Q: And the Treasury uses that estimate to allocate money to the trust fund?

Sander: To make the initial transfer.  That is correct.

Q: And at some point down the road they do what I would call a reconciliation, to get the books straight and to make sure that the actual receipts equal as close as possible ....?

Sander: The taxes that one would get by applying the tax rate to the wages and employment income.

Q: Okay, all right.  Now you said that before 1950 we didn't do it that way.  That they actually somehow waited until they found the actual receipt and used that to do the crediting rather that doing the estimating?

Sander: That is correct, they were separately identified by employers at that time.  This was Social Security taxes.

Q: And because they simplified the burden on employers, a new task came our way, to do Revenue Estimates?

Sander: That's right.

Q: Now can you explain to me just a little bit about the mechanics involved, how you do this?  It is sort of mysterious to me.

Sander: Okay.  Well, we start out with a set of economic assumptions.  The economic assumptions would be the real GDP, that is, the Gross Domestic Product, Total Wages and Salaries and Proprietaries Income, the Labor Force and the Unemployment Rate, what else? The Consumer Price Index and these are all assumptions that are sent over by the Office of Management and Budget in conjunction with the fiscal year budget. Like we're coming up on the fiscal year 2000 budget.  Interestingly, I am leaving just in time not to be doing the 2000 year budget. What we do with those assumptions is we apply relationships that we have established with historical time-series and judgements regarding the behavior of the relationships in the future, to the assumptions.  For example, let us take total wages and salaries; that is in the assumptions.  What we have to come down to eventually is taxable wages and salaries, and maybe I'll just follow that all the way down.

The first step is to split that total wages and salaries bill into that part which is military, that part which is Federal Civilian, state and local, private, farm, currently we've also got to keep track of private households separately, and as a historical artifact, we also keep track of the amount that is tips, and we project those.  Now this is all at the national income-products level--this is not even at the covered level.  We then apply, based on, what I was talking about, these historical relationships established, based on historical relationships and our judgement as to the behavior of the relationships in the future, what we call "coverage ratios," that is, ninety-eight and a half percent of the private wages are covered, seventy-five percent of State and local wages are covered, about seventy-percent of military wages are covered, a growing amount, of course, Federal Civilian wages are covered, because we have new hire coverage for OASDI going on in the Federal government.  And I am actually, by retiring, contributing to that.  There's one less Civil Service person and in the office there is one more FERS person who is covered under Social Security, already. So we then have the wages that are covered under Social Security, by a final coverage ratio.

The next step then is to get the taxable wages.  Now in order to get the taxable wages, taxable wages are determined by the taxable maximum, that is, wages up to the so-called "Contribution and Benefit Base," that which I will call the Taxable Maximum or even the Tax-Max for short. For ascertaining the amount of Taxable Wages, you have to go to earning distribution information, that is, how many workers have earnings and how much of their earnings that are equal to or less than the Tax-Max, and for those workers who have more total earnings than the Tax-Max, how many are they and there is yet another little piece, how many people have two jobs and wind up paying more taxes than the Tax-Max and will get their money refunded on their income tax return--so-called multi-employer refund wages.

So, we're at the covered wage stage.  We need to get the proportion of covered wages that are taxable.  The way we do that is we take a historical wages distribution and, I am not a mathematician, I don't know if this is the right word, but, I think it is that we normalize it with respect to the mean of the earnings distribution; that is earnings levels in the earnings distribution are expressed as some multiple of the mean and let me see, I think right now the taxable maximum is on the order of 2.3 times the average covered wage, which is the mean of the earnings distribution.

We then express the earnings distribution in formulation, that is, at any point in the earnings distribution how many workers are earning that amount or more.  One of the really outstanding things in mathematics is that you can take a distribution expressed in that form and compute the percentage taxable like finding the integral distribution when it is expressed as an equation. Are you sure you want to get into some of this?

Q: No.  Now you have lost me.  Up to now we have been doing fine, go ahead, keep going.

Sander: Well, in any case it is one of the really neat things that what happens is you can express a historical earnings distribution in equation form.  We do it typically piece-wise; that is, three different pieces, one for the very low end, one for the middle end, and one for the very high end.

You can find the integral, you have got to make sure that you can integrate the curve that you fit to the data, and you can find the integral of that and you can compute the percentage taxable.  And, since we have the earnings distribution normalized with respect to the mean of the earnings distribution, all we need to do to move that earnings distribution forward ten years, nine years, how many years that you want, is to project the mean of the earnings distribution in the future.

With the mean of the earnings distribution of the future you can then in effect blow up the historical one into an earnings distribution for the future year.  It is a beautiful system, beautiful system, elegant.  It's, I am sure, primitive as compared with what they have going on right now with modeling the atmosphere and modeling the atmosphere/ocean interaction, but it is in the same vein as that.  What you do is that you take what amounts to a physical phenomenon and express in it mathematical terms and it suddenly becomes free of its historical bases and you can do all sorts of wonderful things with it.  So what it really comes down to then in order to get the taxable wages we need to project the average covered wage and we also need to project the Tax-Max, after that it is rolling off a log.  I mean it is "child's play."  The key thing is setting up the system to work that way.

Q: And it sounds like a couple of the key variables of course are coverage variables. When there is a change in the law that changes coverage that will make a big change in your job. And the other one is the change in the tax base every year, which is now predictable, right?

Sander: To the extent that we have ever been able to project the average indexed wage very well it is predictable.  Like this past year it was just outrageously high and in fact I thought it was so high. Somebody out there, has got to be noticing this, I mean, there is not another economic series that the government has put out that says that average wages increased by 5.8  percent in 1997.  Well I thought that was so different from the 4 percent figure that people were batting around, as that is what you find when you look at the Bureau of Labor Statistics data, that I persuaded Harry to prepare a Q and A for the Commissioner in case somebody asked, "where did this number come from?" Well, one person, a Federal employee from somewhere out in the Midwest, sent an E-mail to OACT, wanting to know "where did this 5.8 come from when my pay isn't going up anything even remotely resembling that?"

Q: What is the answer?  Is there an simple answer?

Sander: There's not so simple answer to it.  Basically what happened, I could spend until tomorrow afternoon on this.

Q: Don't.  Don't tell me the formula.  Don't spend the whole afternoon on the formula.  Just give me the simple answer.

Sander: Basically what happened in 1997 is the unemployment rate was very low.  Apparently one way that employers have been getting the labor that they need to keep the economy growing at 3 percent plus is there is a very eager group of people in the labor force right now who are willing to extend their hours worked.  So you had not only the average weekly hours increasing in a job, I think, you had people taking more than one job, you had, if they are part-time, you had people working more weeks over the course of the year.  So I would say most of the difference has to do with the hours-effect due to the very low unemployment rate at 4 and a half percent, the remaining part is yet another manifestation of the "rich get richer" that's going on in the wage and salary distribution.

There were apparently, as usual,  some very large bonuses and some very large pay increases going to workers earning well above the taxable maximum; to the extent that, I believe, the percentage taxable in 1997 fell about 7 tenths of a percentage point.  And that brings us ever closer to the day that the ad hoc increases in the taxable maximum that occurred in 1979, 1980 and 1981 are going to be wiped-out by the increasing tendency toward "inequality," if  I may use a good Democratic word, in the Wage Earning Distribution.  Basically what is happening is the people earning above the taxable maximum are getting larger pay increases then people earning below the taxable maximum, and  that's causing the percentage taxable to fall overtime.

Q: So to put it in summary, the percentage of the American payroll that is covered by Social Security is falling?

Sander: That is correct, and I think it's fallen by something close to 4 percentage points now, maybe closer to 5 since 1983.

Q: You mentioned the ad hoc increases. Now it is only in recent years that we have had the wage base determined by the growth in average wages, or however it is determined.  Previously the wage base only changed when Congress enacted an increase in the wage base.  They weren't automatic?  They weren't on automatics the way they are now.  Am I right?

Sander: The automatics go back, I believe,  to the 1972 Amendments.  The wage basis increased automatically I think from 1974 through 1978 and then with the 1977 Amendments there were some additional ad hoc increases to the percentage taxable. Because the 1977 Amendments are one of those times when the program needed money.  So they got it up close to 90 percent.  There has not been an ad hoc increase, perhaps, there might have been sort of one around 1990, when deferred compensation was being introduced into the automatic increase computation, but roughly speaking, since 1974, I think,  we have been on automatic.

Q: All right, now tell me, how well do you do with your revenue estimates? Tell me how that works--the reconciliation part at the other end?

Sander: Oh I don't know, it depends on the year.  It also depends on the administrative quirks of the data too.  You really know that you are dealing with an accounting manifestation when it comes to those numbers, because since the quarterly number is adjusted about ten months after the end of the quarter, the rate at which the Internal Revenue Service processes Forms 941, gets them over to Social Security, and we tabulate the numbers from them, can vary.  I know historically, in particular I think it was 1983, 1984, 1985, right around, that was a terrible time for Social Security for them to do this, they got behind I think on income tax processing and they sort of delayed some of the processing of the 941's, like on ten or fifteen billion dollars of wages.  Well the taxes on ten or fifteen billion dollars of wages would have been damn useful in 1983, because that was at the time we were borrowing from the HI Trust Fund and the advanced tax transfers.  But getting back to your question, I have had some truly egregious errors on the order of 4 percent and I have had an adjustment that on the basis of I think on a hundred billion dollars it was off 15 million, and that's unreal. That is a quirk.

Q: Yeah that is pretty good.

Sander: Typically it's less than one-and-a-half percent.

Things were getting crowed in the Meadows East Building, this was around 1981 I think it was, and we were chosen, that is the Revenue Projection Staff, all four of us,  to move over to the Operations Building in what was a computer room.  It was frigid!  You know, it had these computers in it and even joking we had an early small computer put out by IBM, it was obsolete even by then, but it had the advantage of generating a little heat, which we appreciated.  When ORS came over from the Meadows, I can't remember if Meadows East was declared sick by then, well in any case, they reclaimed the computer room we were in and we moved over to the Annex Building and once again we were geographically very isolated from the rest of ORS, but that was okay, we had virtually no interchange with them anyway in working relationships.  The hall we were on, one of the guys in the office named Dave, called it the Hall of Death, because there was this office a little bit down from us that went through a sequence of people at the SES level who had gotten bumped out, washed over, what have you, and that was their office until they either retired or got a better offer.  I suppose that either outcome was what the agency wanted.  The shortest anyone ever stayed in that office was to be moved in on a Friday, we saw her, I don't even remember her name anymore, and she never showed up after that.

Q: I like that.  Now let's go back to what we were talking about before which was the process of doing your estimates.  Now, did I understand you to say that there is about a ten-month lag between the time you produce a  Revenue Estimate and the time the Treasury gets around to doing this reconciliation?

Sander: Right.  It is unavoidable, the actual mechanics of it is the employers don't have to file the 941's,  which is what the initial estimate is reconciled to, initially, until 30 days day after the end of the quarter.  There are, of course, always late 941's coming in so if you tried to adjust any earlier than the ten months you would be dealing with a very incomplete number.  So that's a built-in lag, I don't see is ever going to change. 

Q: What is the working relationship and interaction between your staff and Treasury?  Do they have a red phone over there?  Do they have a hotline to you, is there a close working relationship?                   

Sander: There is a close working relationship.  We generally do things by phone. With the advent of the Internet, better faxing capabilities, we send the material over now either by fax or by Internet.  In the old days it was a real problem getting things, old days!  God, I had better stop this.  It was a real problem getting the printouts down to Treasury.  We used the shuttle and sometimes they were specially delivered, it depended on what Treasury's time line was.  I have been working at this job long enough, that as compared to the Revenue Projection Staff up here both Treasury and OMB and CBO (Congressional Budget Office) look like revolving door operations.  I have gone through quite a number of people that I have worked with, both in Treasury and at CBO.  We do have a very close working relationship with CBO as well.

Q: Now do you do these estimates every quarter?

Sander: That used to be the case in the old days (I did it again), earlier we...

Q: Are you talking about quarterly reporting versus annual reporting?

Sander: No budget estimates.  Back before it was so terribly important that everybody use the same numbers over the whole sequence of the budget exercise, there were four different revenue estimating exercises: there was the primary one in December, for the fiscal year; there was the Spring review, which was in April; and then there was the Mid-Session review, which we still have, which tended to be in July then; and then there was the Fall Directors Review, which was in October.  Now we have the primary budget exercise in December and the Mid-Session Review so that we are really down to two budget exercises plus the Trustees Report and anything special that might come along during the course of the year. But it used to be quarterly, now it is semi-annual.

Q: Now, what controls that schedule?  The requirements of the budget process?     

Sander: The requirements of the budget process.

Q: Let me put it to you this way.  Could you do it more frequently, technically, if for some reason you were asked to do so?

Sander: We could do it every week.  That is one of the things that has really changed over the years in doing this. When I started out we had a clerical pool, it was roughly eight people, and the computations were done manually with Freeden's, and Monroe's and Olivetti's, you know, the mechanical calculators.  Very shortly after I got involved we started putting things on the computer and we developed computer programs and now ...

Q: Are you talking about the late 1960's, and early 1970's?

Sander: The late 1960's and the early 1970's, right.  This was very early. The data for running the programs was put on, do you recall that yellow tape that you could read into a terminal looked sort of  like a teletype?  That's how we first started working with computers.

Q: Punch tape?

Sander: Yeah, punched tape, right.  Even had some as recently as three years ago, but in the move some stuff just had to go, and it went out.  I was halfway wanting to keep it as a keepsake, you know.  But in any case now we do the estimates on PCs on the desk and if we've got the program in shape we can do five estimates a week if need be, just new assumptions, you know.  It's gotten just that routinized.  For the longest time there were only three people working on it, now there are going to be four.  And as compared to the way it was done at the time I first started it, the productivity of the branch just had to have increased thousands-fold. I mean you would never have figured on doing an estimate in under two weeks, I think, back in the mid 1960's, because you had to do everything by hand, everything had to be checked, the results then had to be typed up and put on tables and distributed that way. The whole thing, and the fact that it was all hand calculation, restricted you very much to what you produced, basically you produced as very little as you could get away with.

Not only have we speeded-up that which was produced, we are producing a fantastic amount more.  Like we routinely produce labor force projections; we produce the Average Indexing Wage, the Tax-Max, we produce old law Tax-Max's, we produce Legislative Growth; that is, what would the tax earnings have been at the old Tax-Max held constant, this goes into a budget table, all of that would have been absolutely impossible the way it was done earlier and we're doing it with far fewer people. And yet in the National Income and Products accounts the Government shows no productivity increase.  The Government's output is defined as wages and salaries.                       

Q: That's crazy.

Sander: That's simply not true.

Q: Can you tell me any significant milestones in that evolution of technology and methodology.  You talked about Olivetti's mechanical calculators, and then you had punch tape at some point? Were there other steps between then and now? Can you kind of give me the highlight of that?

Sander: I can't give you any exact dates, but I think as early as 1968 we got our  first time-sharing service where we could dial up a computer over telephone lines and this service specifically provided economic and econometrics oriented data series and statistical routines. Where previously if you wanted to fit a regression, it was agony.  You had to square a huge number of numbers and get the cross products, take a square root, oh God!  You did the minimum you possibly could with that.

Well with the time sharing, it was around 1968, you read the data in, you said LS which stands for Least Squares and specified the time period over which  you wanted the regression to be computed and within a matter of a few seconds or a minute you got the results back.  This opened things up for new ways to do the Revenue Estimates.

The way the Revenue Estimates are done now, the moving away from the old way, was enabled by computers.  I can't imagine doing anything remotely resembling what we do now with a clerical staff.  Okay, that was the first milestone.

The second milestone is when we brought a good deal of  that work in-house and had assess to the Univac computer.  At that point we went to, instead of paper tapes, we went to card decks and you submitted batch runs, and that was agony.  You submit a batch run and you might get the results back the following morning. However, if they had something more important to do you might get back until later, gosh! So it was still very slow, but that eliminated a lot of the paper spreadsheets.

Q: So in some sense that was a step backward for you?  Was it less efficient service?

Sander: Ah, that was a step forward, really.  I recognize that I didn't make it clear.  The time-sharing service enabled us to use regression in a lot more historical relationships in doing the revenue projection methods, and also allowed us to solve some of the equations for doing revenue projections themselves.  However, it did not produce nice neat formatted output that you could give to people as a print out and say here are the revenue projections.  The results came on a yellow sheet from a teletype machine, so that you still had to do the typing for that.  Well, taking it in-house, being able to use Fortran on the in-house computer, enabled us to do a lot more formatting and really integrate the equations and the output, the final output a lot more.  So that was definitely a step forward.

It was agony waiting overnight sometimes, because if you made an error, or if you wanted to change something, you got into this kind of recursive thing, that you submitted a batch run, you waited hours, you got the results back, it errored off because  there was a problem in the input, one thing or another, and you had to submit another batch run.  It was still a week, a good week to do a revenue run.

The next step would be when the in-house computer service got better and you could, instead of submitting card decks, you could work from a terminal and submit the runs remote; that automatically reduces the chance for error.  You are not sitting punching a card deck.  And eventually, oh gosh, I guess it was the late seventies, although I may be wrong on the time.  There was something approaching a time sharing mode within the UNIVAC computer in-house where you could submit a run and you could get the results back very, very quickly.  This was probably in the eighties when that finally came about.

The next step was the mid-eighties, when we started getting PC's. Up until that point there was still some spreadsheet or paper work involved in the computation of  the Revenue Estimates, but with the computer spreadsheet packages, that vanished, and the entire process was really automated.

The next and final step, was when we got computers big enough, PC's big enough, to move off of the UNIVAC and do all of the work in the office itself.  That was around 1991.  We no longer rely on any outside service to do the Revenue Estimates, it's all totally contained in the office.  We have, as computer speed increased, we have for all practical purposes instantaneous results coming back from a run.  If there is a mistake you know it in thirty seconds and you can just hopefully correct it and put it back and that is why we can do a revenue estimate a day if it that was ever required, just really fast turn-around time.

Q: Now, one big change for the agency in the business process around Earnings Reports was in the late 1970's when we went from quarterly reports to annual reports.  Did that impact on the Revenue Estimates business in any way?

Sander: Oh yes, and it still has.  We continue to get the Forms 941. Previously the Forms 941 included individual workers earnings, now it's just the summary, in fact, just the first page of the 941.  The intent at the time was that the W2's that were actually being used to post the workers earnings to the workers records would be the arbiter of the amount of the liability and hence the amount revenue of the trust funds.

I am sure that you have had somebody sitting in this chair talking about the long time it took to get us off of an interim basis, I mean we were on an interim basis for trust fund adjustments. I think it was 1992, or thereabouts, when there was finally a GAO decision that said that we could use both of the 941's and the W2's, whichever was higher.  Now there's a classic case of having your cake and eating it too.

Q: All right, let me make sure that I understand this.  Let's go back to the days of Quarterly Earnings Reports.  We used to get a quarterly report from the employer that would have both the individual wages and the...

Sander: The employers report is to how much taxable wages they would get.

Q: So you would in effect have the information that you needed for the reconciliation adjustment process, every quarter?

Sander: Every quarter, that's right.  Every quarter was a final. That's right.

Q: Now, when we shifted to annual reports what we get quarterly from the employer is just a summary total that includes both the income taxes and...

Sander: It says just the amount of  the taxable wage, but it doesn't show the breakdown of those taxable wages by employee.

Q: And that affects you how?  I can see how that affects the earnings process, now we can't post the earnings for that worker yet.

Sander: Okay, the way that the law reads, the final arbiter of Trust Funds Revenue is wages processed by Social Security.  The understanding at that time was the wages being processed by Social Security that would determine the final amount was from the W2 wages.

Q: So the earnings posting  has to be finished before you can do your adjustment, that's what you are saying?

Sander: Before you can do your final adjustment.  In 1978, when we went to annual reporting, the adjustments were called interims, because the expectation was that we would go to the W2's someday.  Well, as it turned out, the W2 totals typically never equaled the 941 totals and every year that went by and that was the case, I think that generally speaking, on an average, they were missing about 10 billion dollars of wages. Put a 15 percent tax rate on that and that is a billion and a half for every year that you don't adjust to the final.  It got to the point, I guess,  where nobody, nobody wanted to take responsibility for moving that much money out of the Trust Funds and putting them into the General Fund, which is what you would do if you went to the W2's. And it looked like there for a while annual reporting was going to have the effect of putting us permanently on interim adjustments.

Q: Okay.  Now, there are a lot of issues here we need to untangle and try get clear on this.

Sander: There are.

Q: Well one of the things that you just said was, if we did this adjustment based on the W2's we would have to give money from the Trust Fund to the General Fund.  I take it that implies what is going on there is that the 941's generally report more wages than the W2's end up showing?

Sander: And it is actually logical that it should be that way if you think about it.  In fact, I sort of look ascance at any year where the W2's are greater than the 941's.  It works this way: the 941 is  quarterly.  There is a fantastic turnover of businesses going under every year in this economy and they don't happen to go under on December 31st of each year, such that they will be preparing the W2's very nicely for their employees. But what it amounts to, for firms that go out of business in the course of the year, they may or they may not produce W2's for their employees.  They typically will either produce or have produced for them, since the 941 is a tax compliance document, a 941.  So that it should have come as no surprise in retrospect, that W2's adamantly refused  to come up to the 941 total.

Q: How did we resolve that by the way just in general in terms of wage reporting?  I don't remember now where we were with that ?

Sander: Well okay it was resolved, I think it was a Controller General's decision.           

(Editor's Note: This interview ends abruptly due to the unfortunate loss of a second cassette tape on which the end of the interview was recorded. Approximately 15 minutes of additional discussion was lost due to this mishap.