How to Forecast the American Budget

"It's a $25 trillion economy, and moving it is hard. Most programs don't."

Dr. Doug Holtz-Eakin has worn many hats. He was Chief Economist on George W. Bush’s Council of Economic Advisors, and directed Domestic and Economic Policy for the John McCain presidential campaign in 2008. These days he runs the American Action Forum, a center-right think tank.

But today, we’re talking to him about his time as the 6th Director of the Congressional Budget Office (CBO), which provides budgetary analysis to the U.S. Congress. From 2003-2005, Holtz-Eakin supported Congress as it addressed major policies including the 2003 tax cuts and Medicare prescription drug bill, and the 2005 Social Security reform push.

What You’ll Learn

• Why do congressmen pretend to be mad at the CBO?

• What’s wrong with the American budget process?

• Why is the 10th year of a budget projection “worthless”?

This is the second in a four-interview series on Congress’s support agencies. You can read the first installment here, on the Congressional Research Service:

How to Teach Congress to Do Its Job

---

In 2011, then-presidential candidate Newt Gingrich called the CBO “a reactionary socialist institution which does not believe in economic growth, does not believe in innovation, and does not believe in data that is not internally generated.”

You responded at the time by calling that attack “ludicrous.” Why was it ludicrous? 

First of all, you have to consider the source. When I was CBO director, Mr. Gingrich came in, talked to me, and said, “You're doing a great job. It's nice to see a Republican being effective in this position.” And the next morning, the Washington Post ran an op-ed by him calling for the abolition of the CBO. It's a little inconsistent. 

More generally, the CBO is none of those things. The CBO works for the Congress, and so it doesn't have its own agenda, reactionary or otherwise. It has Congress's agenda. It operates and produces its research based on the large body of research that's produced outside of the CBO in the research community.

When the CBO has an estimate, it uses the consensus of the research literature. It doesn't have its own view that is somehow different from what's been found in the large body of research. Those are big misconceptions about how CBO does things. 

From an institutional perspective, why does Congress need CBO? 

So the Congressional Budget and Impoundment Control Act of 1974 created the CBO, the budget committees, and the modern budget process. It arose because then-President Nixon was not spending monies that Congress had appropriated for specific purposes. The executive branch was not executing laws that the Congress had passed and the executive had signed. 

The Office of Management and Budget was the only budgetary institution in town. Congress was utterly reliant on the executive branch for all its budgetary information.

And in the aftermath of this dispute, which went all the way to the Supreme Court, they decided they wanted to be able to do their own work. And in what was just a flat “let's balance the powers” move, it created its own budgetary institutions, including the CBO, to have data, make projections, and allow them to assess the budgetary outlook and alternatives to it and different pieces of legislation. That seems appropriate. If we're going to have an executive branch that has its budgetary analysis, Congress is entitled to an independent view, and that's what they created.

Speaking in broad strokes, how do the CBO and the OMB tend to differ? Are there common fact patterns that they see differently? 

They're different institutions. I've worked in the White House twice, and I've run the CBO. And if you are in the White House or at OMB or the Council of Economic Advisors (CEA), your only objective is implementing the president's agenda. OMB is there to formulate the president's budget and implement it to the best of its ability in the actions of the agencies, and it has oversight of the agencies through their management functions. That's a very particular kind of advocacy role. You're trying to get something done. 

CBO is not that at all. It’s nonpartisan and does not make policy recommendations. It is not trying to get something done; it's providing the information to its client, the Congress, as the Congress decides what it wants to get done. 

They have some overlap in that both are interested in annual discretionary appropriations and defense and non-defense decisions. But OMB doesn't analyze entitlement programs or mandatory spending. CBO has both discretionary and mandatory spending, so it's an entirely different institution. 

I'm editing an interview right now [spoiler] with Trump’s OMB Director, Russ Vought, who takes Nixon's side of that debate. He thinks presidential impoundment should be used much more often than it is, and that the executive has a role in these processes that Congress very clearly does not want the president to have.

Having been in both the White House and working for Congress, where do you fall on this debate? 

So, I first want everyone to understand that this is quantitatively irrelevant. Historically, the total impoundments have been in the double-digit billions. This is less than Medicare fraud in a single year, so it's not going to change the budgetary outlook. 

So it really is simply a power issue. I think that the current situation is the correct one. I don't agree with Nixon. Nixon and any president has every chance to shape the appropriations bills during their deliberation and passage. White House staffs exist to do that logrolling and campaigning for people and all the other ways that you can influence folks. If you didn't get it done, then it's over; the law passed, and you signed it into law. You do not then get to turn around and say, no, it's not right. So I'm not a big fan of those positions. 

---

Why does CBO matter procedurally for Congress?

First, CBO supports Congress in its budget process. I'm going to describe the cycle as it was envisioned in the ‘74 Act, with the caveat that it doesn't happen this way in practice. In January, CBO would release the budget and economic outlook, the starting point for the budget cycle.

That's a projection of what the economy will look like if we put current law on autopilot. What will happen to the budget? What revenue will we collect? What spending will we have? 

It would then receive the president's budget and analyze it. The President would have some claims, and CBO would reanalyze the same proposals and give Congress its view of them.

At that point, Congress would begin deliberations on a budget resolution in the House and Senate, ultimately agreeing on a budget plan, a joint budget resolution that said, “What are we going to do to change taxation? What are we going to do to change spending?” And off it would go to execute the laws, the appropriations bills and the entitlement and other reforms that it might want.

In that process, CBO would provide scores on each piece of legislation, answering the question, “If this legislation became law, how would revenues into the Treasury and spending out change over the course of the 10-year budget window?” It's been different lengths over time. So it's supporting that, on a regular basis, all those cost estimates. 

In August, when Congress is presumably on recess, CBO would publish the summer updates of the budget. “You guys did all this stuff. Where do we stand?” Congress is supposed to close the books by September 30th.

CBO then spends time looking back, redoing its budget and economic outlook, updating for new laws and new economic conditions, and starts over in January. So there's a cycle of budgetary products. There's a daily flow of requests for scores on different versions of legislation to change those budget outlooks.

In many people's view, that's the primary function of CBO. When I was there, there were 230 people there, 190 of whom had advanced degrees. There's an enormous amount of expertise on that floor.

CBO also supports Congress by doing studies to answer questions like: “What are options for single-payer health care? What are the implications of a carbon tax for the budget and the emission of greenhouse gasses into the economy?” There's no question you can't ask CBO to analyze.

A number of studies from CBO are really laying the groundwork for future legislation. They are educational products for when Congress wants to get smart on this or that. When I was there, at the request of then-House Budget Chairman Jim Nussle, we did a whole series of studies on what happens if we have to make Fannie and Freddie wards of the state.

That turned out to be true, but we were studying it well before it happened. “How much would it cost taxpayers if the Pension Benefit Guarantee Corporation goes belly up?” Which it probably will. That's laying the groundwork for fixes to both of those institutions.

In addition to all that work, the scores that CBO generates for bills have concrete procedural implications for getting into certain bills, right?

Often you can't just have any budgetary impacts. You have limits on it. Maybe there's been an agreement in a budget resolution that says, “We're not going to spend more than $800 billion this year.” So you're doing the scores, and if they add up to more than $800 billion, Congress is going to have to change something. 

Or they’re using special budgetary procedures like reconciliation, which allows you to bypass the filibuster in the Senate and pass things with a simple majority (and not 60 votes). Reconciliation instructions are very precise. They tell each committee in both the House and the Senate what they're allowed to do to the budget. The CBO scores essentially bless the legitimacy of the product; otherwise, they cannot be voted on.

I'm assuming that most of the antipathy from congressmen and women towards CBO comes in those moments. Is that fair? 

Eh, no. First of all, I would say there's a fair amount of faux antipathy, because CBO is a convenient institution to hide behind. You complain, “CBO is moving slowly, we'd like to have this done, but we're waiting for the CBO score,” and often that's not true: You don't have the votes yet. But that's what you’re telling people. Or you can't get what you want, so you blame the CBO score.

I came to understand I had this psychosocial role, of being beaten up and getting kicked like the dog. That's just allowing Congress to vent, and that was fine. 

There is real antipathy sometimes. The interesting thing is that it did not come on the big bills. I was there for the tax cuts in 2003, the Medicare Modernization Act that gave us Part D, and some really big moves. The real anger came over pet projects, often relatively small dollar numbers that we said were not small enough, and they didn't get into the bill. Congressmen or Senators had promised it back home, and it wasn't happening, and they were just angry.

You're saying the antipathy comes largely when it's somebody's baby, rather than a big coalitional product. 

Yeah. People care most when they own something. No one owns really huge reforms, and there are many actors, but, you know, “That dam project, boy, I want to build that dam, and CBO’s wrong.” 

Is there any obvious money on the table to make the budget process go faster or better, or is it fairly intractable as it stands? 

I think it's simpler than that. I think Congress should pass a budget process reform bill, and I don't care what's in it. It could modify the Budget Act so that we have a debt-to-GDP target or something as small as that. 

But in the process of deliberating on that bill and then turning it into law, they would now own the budget process. It would be theirs. This Congress. So it's not the dead hand of some guy from 1974 telling me to do something. They don't care. But if they agreed, then at least for a couple of years, they're going to feel obligated to do that, and we'd have a process that functions.

But this is not an interesting concept to the current Congress. 

It seems like so much of the decay between 1974 and today is due to the fact that you can't run this yearly cycle anymore. But you think a yearly cycle would work if people felt like they owned it? 

Yes, and especially if the leadership said, “October 1st is a real deadline and we’ve got to be done by then.” They wouldn't be thinking we can always do this on Christmas Eve, which they do a lot. They take the time because they feel they can. That's all. That's all that's happening. 

You've alluded to this already, but talk to me about the formal modeling constraints on CBO. CBO can't just pick up any tools in the economic modeling toolkit, right? 

Not exactly. A word about scoring and models probably more than any of your listeners ever wanted to know.

Please go as deep as you can. They're nerds. 

Scores are the budgetary implications of a piece of legislation. The answer to the question, “How much will revenues into and spending out of the Treasury change over the next 10 year by year as a result of implementing this Scores are computed relative to a baseline.

As I said, CBO puts out a baseline in January, it updates it after doing the president's budget. At that point in time, the baseline is frozen, so all proposals are compared against the same benchmark, even if the economy, as it did say in 2020, collapses shortly thereafter. Okay. The baseline is there. So you're comparing against the same baseline that gives you some consistency and the whole purpose of scores is to rank things appropriately. 

They're not forecasts. If you were a good forecaster, you would update the starting point as you went through the year to be more accurate. That's not the primary purpose. The primary purpose is to find out whether Proposal A costs more than B and B costs more than C, and then not conclude that C costs more than A.

You don't want to send conflicting information like that to Congress. They'll go crazy. So getting the right ordering is the primary purpose. Putting the outcomes somewhere close to reality is a goal, there's no question about it. But in many cases, your hands are tied in ways such that you're not going to get that goal.

Big pieces of legislation often span over a year. You hold on to the original baseline and keep scoring against that so that you don't upset everything. By the time some of these final scores get done, they are well divorced from reality. 

Further down the road, Congress changes a bunch of things, and then people look back and say, “CBO wasn't even close.” Yeah, there are reasons for that. Scoring is a judgment science. We hire, train, and pay the CBO experts to give us their best judgment about the path of revenues and spending over the budget window. 

Models can and do inform judgments. If you have formal models, they help you figure out what's going to happen. If you have experience with Medicaid expansions — and they've been done a lot before — you can look back and see how they happened. You can go to lots of sources to inform your judgment but in the end, we're asking people to make judgments.

I was asked to score terrorism risk insurance, the federal budget backstop to the private property and casualty insurance industry, in the event it went bankrupt from the implications of an unknown terrorist attack at an unknown time in an unknown location with an unknown weapon. There is no model. We made a judgment about the cost of that act and that was our job.

I think people often have the misconception that CBO just drops the legislation into a computer and out it comes. That's not true. If you've got to make judgments, you make them based on what we know from the research literature about, in that case, terrorist events and the probability distribution of them. It turns out there's a lot of terror in the world over time, and people went and studied it. CBO is quite serious about the scoring of bills. 

It's not that they can't pick up any models. It's that as a matter of their professional practice and pursuit of the mission, they will not use just any old model, they'll use the one that is best suited for that purpose.

I'm paraphrasing here, but for scores in particular, the CBO has to sacrifice some amount of live updating to fulfill a formal role. 

Yes, the formal role is to provide these cost estimates and to allow Congress to make good decisions and you have to give them consistent information. First and foremost, A is bigger than B and B is bigger than C. If you're updating all the time, you might flip those. 

Sometimes there's a tough call to be made on something like that. For example, after the pandemic in March 2020, CBO reissued a baseline, because we lost 20 million jobs in the month of April, and that was 10x worse than any month in U S. economic history. It didn't make sense to score a stimulus bill of two and a half trillion dollars against a baseline that said there was no recession, right? That would have led to bizarre things. 

When that happens, CBO doesn't reissue baselines unilaterally. It works for the budget committees, first and foremost, and consults with them. The budget committees might sometimes say, “What do you think is the right thing to do?” But in the end, the budget committee's blessed the direction. 

We interviewed Jason Matheny about his time running IARPA. He's now running RAND.

How to Predict the Future

He’s a big advocate of internal prediction markets as a way to improve the intelligence community's forecasting. Has that idea ever crossed your desk? 

No. CBO has a great track record of internal accountability. And I just want to say right now what any CBO director would tell you if they were sitting where I am. We owe all of this to Alice Rivlin and Bob Reischauer, who started CBO and set it on such a good course that, essentially, we're all still doing what they did. And the staff is the key to that.

The nonpartisan professionalism is in the DNA. No director can mess with it. And among the things they did was to say, “All right, we do a baseline. They do a bunch of legislation. We then need a new baseline. Let's go back and look and see how accurate we were. How good were the scores? How good was our baseline? What do we need to change? Who really missed it? Who on this staff, and why?” That's what sharpens the reliability of the scores, the internal review process that really is supposed to be occupying the time between October 1st and January 1st. When you start a new cycle, Congress is supposed to be done.

Now, they've blown that up in recent years, but that process still goes on. There are internal reviews of every single budgetary account, and every account is owned by an analyst, and the analyst is quizzed on “Why were you doing this and why did we end up there?” That's one of the things that no one can see, but about which CBO should be quite proud.

I've got a Brookings paper here in front of me from my former colleague Phil Wallach. He suggests that at its inception, CBO, in some ways, proved its nonpartisan status by incurring criticism from both sides of the aisle under Rivlin.

Is that your read of the early years of CBO?

Certainly, yes. There was not agreement on what CBO should be, and Alice made great choices about what it should be and how it should behave. Alice decided, even though it's not in the law, that it would be nonpartisan, not bipartisan. You don't give each side 50%. You call shots correctly to the best of your ability. She's the one who said CBO should refrain from making policy recommendations. You work for the Congress, you take their proposals and you analyze them. And she, first and foremost, decided that CBO should have its independent economic forecast.

Particularly during the Reagan years, CBO criticized Republicans. It then blew up the Clinton health plan. It has, over time, drawn the ire of everybody. And I was quite sensitive to this because I was the first and only CBO director to come directly from the White House. I had been working at the Council of Economic Advisors. 

Most Democrats on the Hill had no idea who I was, other than a Republican hack from the White House, and they thought CBO was deeply endangered by my presence. We had to just, again, prove our independence and nonpartisanship through the quality of the work. That's the only thing you can do.

Was it a risky strategy for a new congressional support agency to consciously anger partisans on both sides? 

A little bit. But Alice was a great believer in the protection you were afforded by the facts. And she believed she had the facts on her side every time she said something because they'd done their work. And that's the risk she took. 

I've got a 2010 quote from you here to that effect: “Every advocate comes to CBO, and their null hypothesis is that they are right, and they think CBO’s null hypothesis is that the CBO is right. That's not CBO’ null. CBO’s null is: ‘You're wrong.’”

This has changed a little bit over time, and I'll come back to that. By and large, in my experience, people only came to CBO to lobby. They came in with a point of view, and it was not the one CBO had chosen as its estimate or its reading of the situation based on the research that was out there. So the null was, they were wrong. We'd already made a decision based on the facts and they were coming in to tell us something else. And it's because they were lobbying.

They wanted to really move CBO, so I was quite skeptical. Now, the thing that was true then but has changed a little bit now is that we were always happy to have people provide us with information. If you know something, please bring that information to us. And do it well before the legislation is on the floor getting voted on.

Because if you show up when it's getting voted on, you're there to lobby and get the outcome. But if you came in the January before and said, “Look, you're about to do work on a prescription drug work program: We're a drug development company. This is our pipeline. This is what we'll have in the way of branded drugs. These are the generic substitutions we see.” That's huge information for us. We had all of that. Companies gave that to us. We signed nondisclosure agreements so that their research was secure. It informed the drug benefit estimates enormously, to say the least. So that was always welcome. Showing up to complain about the outcome was not.

And what I've seen Phil Swagel, the current director, do is announce requests for information about things. I didn't do that. I didn't know anyone before Phil who did that and just said, “We're in the business of understanding something better. If you know something about that, please come talk to us.” And that's a good idea. 

What other kinds of information might CBO proactively ask for in preparing for its work? 

Who knows? We did a whole lot of work on something that never came along, which was essentially an asbestos fund to provide awards to victims of lung disease from asbestos. We talked to every manufacturer who would have been liable and to the existing private funds that do that work. “How do you adjudicate the claims? How fast would you adjudicate the claims? What’s the standard of evidence, and how expensive is it to produce it?” 

The prescription drug law passed in 2003 and it started in 2006. We allowed time for it to start. There's a lot of setup involved: how fast you adjudicate claims, request reimbursements, etc. There's a million details, and people in business know that stuff. It's helpful to hear from them. 

You were CBO director from 2003 to 2005. How have interactions between Congress and the CBO changed since then?

I came in right after the Clinton era, in which there were four balanced budgets, the only balanced budgets we've seen. There've been none since. There was an enormous amount of pride in that accomplishment, especially from the Democrats. The budget committees were held in great esteem and they were prized spots.

And that's all gone south in the 21st century: the budget committees don't have the same standing they had then. I got enormous amounts of pressure from different folks on budget committees about how CBO was doing its business because they cared a lot about how it came out. 

I think there's a lot more benign neglect now by Congress, and CBO is left more on its own devices. I'm not sure that's a good idea. It'd be better if they cared more. I think that's the biggest thing. The environment has changed dramatically. 

We had some pretty testy moments because budgets did go from balanced to deficit. We were fighting wars in Afghanistan and Iraq. The Bush administration was financing them not through regular appropriations, but separately, and Democrats were accusing the administration of hiding the ball on the cost of the war and not budgeting for it. They felt it was our responsibility to set the record straight and show all this.

So I spent a lot of time testifying on things like Iraqi troop rotations. “Was it plausible that we could, with the force of the size we had, rotate people through active duty and then back to rest and rotate back in and prosecute the war at the cost they claimed? Or was it going to be a lot more?” Congress really wanted to know about this stuff, and we did all that work. 

When you refer to “benign neglect” from Congress toward CBO, presumably there are ways in which you don't want Congress breathing down CBO's neck. So how should Congress actively manage its relationship with CBO?

The budget committees are Congress's official scorers. People often don't realize that. Everything CBO does is advisory. The budget committees should care enough about the scores they have to deliver that they're keeping an eye on CBO, protecting it when it's being attacked unfairly and unduly, and prodding it when they need information faster or in a different format.

That's the most important relationship. And with pretty limited interest in the budgetary outcomes, budget committees don't have as much to do. So they're not responding to their colleagues in the same way and it's not getting passed down to CBO. 

I think the setup is fine. It's just that we need the leadership at the top about the fiscal outlook, and that would translate all the way down the food chain to CBO. 

Talk to me about the difference between static and dynamic scoring. Why does CBO do the former and not the latter? 

It does both. I did the first dynamic score for CBO. In scoring, you do an economic forecast and then freeze it in place. In a conventional score, you do not let the size of the economy change. You take into account a lot of behavior. A classic example is that if you had a policy, a tax exemption or subsidy, or something that caused employers to provide more health insurance over time, we expected that would come at the expense of cash wages.

So you'd see less in the way of cash wages, less income tax on cash wages, and less payroll tax on cash wages. The composition of the economy is changing, but the size of it is not. That's a static score. It doesn't mean that no one's responding to the programs — you're taking into account all the responses you think are important — but you're not letting it affect the size of the economy. 

There are some things whose whole point is to change the size of the economy. The CARES Act, in the midst of the pandemic, was designed to do that. Or a big Social Security reform, you’d wonder what it's going to do to national saving and growth in the economy and things like that. 

Immigration reform, you're trying to change the size of the labor force and you expect it to change the economy. So there are places where it makes sense to give Congress the information with a dynamic score on the impact on the growth of the economy, positive or negative, as a result of the program. But it's only when you've got large proposals in magnitude that this is really going to matter. Now, that's the only difference, is allowing the size of the economy to change.

I was under the very strong impression, although it was unspoken, that a litmus test from my appointment as CBO director was that I would be willing to do dynamic scoring. And I have no objection to dynamic scoring. If you're going to do some really big tax cut or some really big spending program, you would want to know this information, and it's our job to deliver it.

So I didn't have any particular objection to that. Nevertheless, when we finally did the first Bush budget in 2004 after my appointment, we analyzed the macroeconomic consequences, and it showed that the 2003 tax cuts, when combined with a big new prescription drug spending program, offset each other, and the impact was imperceptible. It was a little above or below zero.

I thought this would teach Congress how, if you really care about pro-growth policy, you have to be disciplined and not provide a whole bunch of spending. What they concluded was that CBO did it wrong. Among Republicans, it had become this article of faith that tax cuts paid for themselves and you get these big growth impacts. 

That's not supported by the research literature, and it's not true. My feeling was that for years, dynamic scoring had been the forbidden fruit. I thought let's just give it to them, let them have a bite, see what they think.

So it's now done. We did that in 2004. It's been done on pieces of legislation that merited it since both CBO and the Joint Committee on Taxation have the capability of doing it. It takes longer. It's a bigger undertaking from an academic perspective, from a computational perspective.

You're asking about the growth impacts on the economy, and you have to think through that. If the economy is growing faster or slower, you're essentially saying we're changing the baseline. So you have to update the baseline estimates of every program. It's just a lot of work, so you don't want to do it all the time. It's been done where appropriate, I think. And the world didn't end when it was done. 

In my view, it hasn’t changed the legislative process at all because, as it turns out, it's a $25 trillion economy, and moving it is hard. Most programs don't. 

---

You mentioned dynamic scoring being a lot more work. Is the size of CBO’s workforce a constraint on its ability? If you gave it 100 more employees, would that meaningfully change CBO's output or effectiveness? 

It wouldn't have done so in my time. I had the right number of people, by and large.

Congress launched into something new every now and then, and we had to get smart on prescription drugs fast, for instance. So the composition and expertise might feel tight every now and then, but overall, there were enough people to do the work that we were asked to do. We could never have done — and CBO still cannot do — every request that it gets.

A big part of your job as director is triaging these things. CBO, unlike CRS, doesn't work for individual members. It works for committees. The Budget Act lays out the hierarchy: work for the budget committee, the appropriations committee, the authorizing committees, and in that order.

No sensible director would ignore the leadership, so the Speaker, the majority leader, sure, we'll do it. And then, you have some freshman member on the budget committee who wants their legislation scored. You just have to say, “Yeah, I hear you. I would love to do that. If you can get the chairman to make the request, we'll do it. But I can't just do it for you.” 

How does the CBO make sure it has surge capacity so that when, say, prescription drug pricing comes up, there's room in the tank to tackle that immediately? 

Alice's original design of CBO broke it into two pieces. There's the budget analysis division, which has the baseline and scoring responsibilities. And then there are the studies divisions, which were like health and labor and macroeconomic analysis and national security and financial. And those have been reshaped over time. Different directors have different study divisions.

The studies divisions are surge capacity for the budget analysis. You can pull them in, and we did. When the Medicare Modernization Act was up, everyone in the health area was doing it. We didn't have the luxury of walling them off to do a longer-term study. 

Maybe this is a bald question, but what is the pay scale for CBO staff? 

I have no idea. I'm way out of date. It's congressional staff, so the Director could not make more money than the fourth-highest-paid staffer in the House. There was some position against which my salary as Director was pegged, and everything went down from there. It is not a lucrative place. It is a well-compensated government job, but no one gets rich at CBO. 

Would increasing that pay scale improve the ability or quality of CBO? 

There are some targeted places where I think it would make a difference. I'm not sure it's worth it. Honestly that's something Congress should have to decide.

In my era, for example, there were economists who were much more expensive than any other academic discipline. This was widely understood, to the chagrin of every other academic discipline. And then there were finance economists. I couldn't afford a real finance economist.

It wasn't going to happen. It would have been nice to have a few more finance guys on the staff, but we just couldn't do it. So there are places where it does show up. Do you want to change the pay scale for everything to accommodate those kinds of situations? I don't think so. Because you can get outside help. One of the nice things is you can send CBO analysts to research conferences where those folks are presenting their research, and they can talk to them and get smart on those things. And there are other avenues for getting the assistance. 

---

What do people think is hard about CBO's work that is, in fact, not that hard? Is there an overrated difficulty? 

I think they think the whole scoring budget projections baseline operation is harder than it really is.

Initially, the first ones must have been really hard, but there's a lot of history there and a lot of practice, and a lot of it's routinized in ways that don't make it that hard. And there's also a temperament. If you're Doug Holtz-Eakin sitting at Syracuse University [where Holtz-Eakin chaired the economics department] trying to forecast the outlays for a particular program, it's the one time I'm going to do this in this professional paper I'd like to get published, and I'm sweating bullets over every dime out there in the fifth year.

If I'm doing this for the 20th time at CBO, and I know we're going to be wrong, the mental anguish is not nearly the same. They're going to do a good job. They're going to put out a highly professional piece of work. But they know the limits of anyone's ability to do these projections. 

But I often tell people that the 10th year of a budget projection is worthless. The whole world is going to change by the 10th year, but Congress wants 10 years, and it wants 10 numbers. It doesn't want 10 confidence intervals. It wants 10 numbers, and we give them 10 numbers, but we know what we're doing when we do it.

Should Congress rely more on confidence intervals, or is that beyond the scope of possibility?

It is not beyond the scope of possibility. We published the outlook for the deficit. So remember, we’d had balanced budgets. Now I'm on duty. We have a $430 billion deficit and people are losing their minds. And so we published the 10-year budget projection for the deficit and the 95% confidence interval around it. And the confidence interval consumes the entire page.

And Jim Nussle takes one look at it and says, “This makes it look like you have no idea what you're doing. Don't you ever publish that again.” Sure. So fine. When they were doing Social Security reform, there was an enormous amount of debate over individual accounts, private equity, and individually owned private investments in risky assets.

That's all about risk. That's all about confidence intervals. That's it. And giving people single numbers for risky investments makes no sense, but they didn’t want it.

And that's just a function of the way political incentives and demands work? That if you're a congressman, you can't do anything with a confidence interval? 

It's funny. I talked a lot with senators on this and senators are unusual people. The senators I worked with all literally, not all but out of 100 senators, there were probably 30 millionaires.

Every one of them told me a story about how they'd been told they were going to lose. They were down six with three days to go in the campaign and they won. None of them believed in downside risk. They had done nothing but win. 

So I'm telling them, gee, if they make this investment and lose their money, they're probably going to be pressures to provide them with more benefits from somewhere.

And they're like, “They're not going to lose. I've never lost.” Okay. That's great. So some just didn't have the temperament to be interested in that. Others just didn't know how to use it. They knew, “This number is bigger than that one. I want that one.” 

What do people think is not that hard about CBO's work that in fact is? 

Scoring the Terrorism Risk Insurance Act. 

That seems pretty impossible. 

People send CBO all sorts of requests and they don't know what they've done to them. They sent me a request to give them the score on a bill that would have paid X dollars for additional benefits to the families of servicemen killed in Iraq.

You're asking me to tell you how many people are going to die in Iraq. No thanks. I don't want to be that person. This is prior to the invasion. 

Before we had boots on the ground at all?

Before we had boots on the ground. I said, look, we've got this request. I'm going to go talk to them and ask them to withdraw this request because I do not want it to be in this business. But separately, the CBO national security guys went off and they studied casualties in Grozny, which was door-to-door, block-by-block siege warfare.

Brutality.

Horrible. And then they looked at the first invasion of Kuwait and things like that, so we knew what the worst case scenario of an invasion would look like. We did our work, we could have produced a number if they made us, but I talked them out of making this.

So that's hard. You spend a lot of time on stuff that should never appear and doesn’t. 

I'm assuming that over time and tenure, Congressmen and women get better at deciding what to send to CBO? 

Yeah. New budget chairs are less seasoned. Over time, they improve.

Good budget chairs run off colleagues’ bad requests. They say, “You don't want to ask them that. You don't want that answer. I don't want them to give you that answer. We're not going to do that.” 

---

What's your biggest win as director? 

We scored the Medicare Modernization Act: This was an intellectual accomplishment of enormous scale and scope. There did not exist in nature an insurance product against the costs for seniors of outpatient prescription drugs. And we had to build a world in which people sold that insurance, the federal government subsidized it, and people paid the premiums and bought it. 

It covered certain drugs, which ones at what price over for how long, so you're inventing an insurance market, a new sales market for the manufacturers, a negotiation on that, and seniors changing their plans. It was enormously complicated.

And I think we did a fantastic job. I do. 

What kinds of research by academics tend to be most helpful for CBO as it's trying to build expertise in a given subject or group? 

Hard to predict. Honestly, people change tax law a lot. There's always interest in empirical work on magnitudes of responses, different tax changes on labor supply, saving, portfolio allocations, all these different things. There are some things that Congress does predictably.

But CBO is most important when Congress is doing something new. So go to the Medicare Modernization Act again. There wasn't any research on outpatient prescription drug insurance. When Congress is doing something big and new, that's when there's the least research, and that's when the CBO is most important, because people are pointing at it. Those are the hard jobs right there. And the academic research helps in ways that the academics never envisioned.

In finance, a seminal moment was the development of the Black Jensen Scholes options pricing formula. How do we value the option to buy or sell a stock? They published that paper and people then started simulating it and estimating it using computational empirical models. 

Wall Street firms could actually operationalize this stuff and they did. We had a situation where there would be an option to put Fannie and Freddie to the taxpayers. They would be in financial distress.

And it was the belief that the treasury would rescue them that's essentially a put option. What's the value of that put option? How much should the taxpayers see themselves as being a risk? So we just took an option pricing model and plugged in the Fannie and Freddie numbers. And so that was enormously healthy, and we got a fantastically accurate estimate. We thought that option had a value of about 20 billion a year for 10 years, 200 billion, that's about what it was when the crisis hit. It's one of the better estimates we did. So who would have known that particular piece of research would be so valuable?

Do you see any risks to the CBO, as an institution?

It's a good question. CBO and the Office of Technology Assessment (OTA) were born at the same time, and the Office of Technology Assessment does not exist anymore.

And CBO is keenly aware of this. I certainly was aware of it as director. But I think its biggest risk is that it's been too good. It performs its mission extraordinarily well, and when people see that, they think, “Maybe they could do my mission really well too.” So they want to give CBO another mission. 

The cost of regulations comes up a lot, and I regularly evinced my interest in the cost of regulations, but I thought someone else ought to do it, because CBO is really good at what it does. It should stick to that. That's its biggest danger: being given another mission and degrading the existing mission. 

Next week’s interview is [spoiler] with Peter Douglas Blair, who wrote a monograph about the rise and fall of OTA. One of the proposals he discusses is to make a new OTA, and stick it in the CBO. You think that would be bad for both parties? 

I think it's bad for both parties, I do. When I became director, I looked at CBO and what I saw was an extremely risk-averse institution. It knew that a sister agency had gone away and so, given its druthers, the staff would rather take two years and get the perfect answer than move quickly in time for the actual decision. 

I was the productivity hawk. It was one of these weird situations, I was the academic, but I was there saying, “We're doing this now. We're doing this now. And I would rather be 90% right and on time than perfect and two years late for the debate. It is our job to be on time.” 

I harassed people unmercifully about the timeliness of things. and they were afraid we’d go too fast, we’d make a mistake, and we’d get roasted by Congress.

I said, “That is my job. I defend every product. I own it. You get the work done. And if someone's going to get axed, it's going to be me but we're going to do this.” And that was some of the fallout of the OTA experience. They had watched this happen. This wasn't hypothetical to them.

They understood Congress has been around a long time, and CBO had only been around for some 20-odd years. 

It’s funny you say that, because some of the criticism OTA got in 1995 was that its products didn’t show up on time.

I didn’t know that, is that right? 

Yeah, that it’d have a great report on an issue, but it would come out two years after the relevant bill was voted on. So who cares? 

At CBO, and now running a think tank, I always believe it is better to be 90% right and on time than late. We're going to make mistakes if we move as fast as I think we have to, but I think the value is informing the decisions. And you can’t inform them after the fact. 

Anything I should have asked you about CBO?

“Is it really nonpartisan, or is it a bunch of crazy lefties in disguise?” The answer is, it’s genuinely nonpartisan. Every one of those staffers has their political views and they're voting, but that's their business. The work is nonpartisan, and it's really in the DNA of the place. And no director is going to change that. I was under a lot of pressure when I showed up, and I realized that the best allies were the staff on the floor because they knew exactly what to do.

And I only got in trouble when I wasn't smart enough to ask their advice. If you're surrounded by smart people, ask them and you'll be fine. And it's when you forget to do that, they get in trouble. So it genuinely is nonpartisan and it prides itself on the quality of its work. Congress should be extraordinarily proud of what they did. It turned out very well.