Mr. FAZIO. I appreciate that. This committee has been struggling with these points for a number of years. I give Ed Lombard a lot of credit for having educated me and others on the committee as to the areas where the rule of thumb you were using for the baseline didn't apply to personnel-intensive agencies, to the Legislative Branch agencies in general, or to many of the administration agencies in the executive branch. I really think that presents a far more accurate picture, which is really what we are all afterCBO's area of concern. It also has the effect of giving us a case we can make. Not that we are always going to succeed. I can tell you as concerns overall budgets, we are going to be tighter. We are going to be in trouble competing with all these other needs. At least we don't look to be overspending when, in fact, all we are trying to do is maintain a steady state operation in light of all the things that normally occur. We have been fighting that. I hope we no longer will have to. I want to thank you not only for your attention to this debate, but your lucid explanation of it. For those of us who understand generally what we were looking for, it is very meaty. DIFFERENCE BETWEEN NEW AND OLD BASELINES Off the top, what would have been the difference between CBO's budget this year had you not made these changes in the budget you submitted? Could you give us a feel in terms of what its impact on CBO would have been? Mr. REISCHAUER. I can give you a feel for what the difference is in the baseline under the two methodologies. That difference is about $414,000. It raises the baseline for CBO by $414,000. Mr. FAZIO. In the past, of course, without being vindictive at all, this committee was always more than happy Mr. REISCHAUER. Sorry. I misspoke. It is $260,000. Two hundred and sixty thousand dollars is the increase in the baseline between the old and the new. The difference between our budget request and the new baseline is $414,000. Mr. FAZIO. Not an insignificant amount. Mr. REISCHAUER. No, it is not. I can explain why we are above the baseline; which is really because of two factors. The first is that we have been operating at something below the 226 positions that we have available-namely, about 218. During the course of this year we will go up to the 226, which is our ceiling on positions. Over half of that $414,000 represents the full year's effect of going up to that level. It is the pay cost associated with it. In addition, we were asking for $221,000 for systems, data, and model development. That is clearly over and above the amount that would be provided from an inflationary adjustment. Mr. FAZIO. I have some questions I will submit to be answered for the record. [The questions and responses follow:] Question. One of your 1992 goals is savings in administrative costs (p. 9). What type of administrative savings do you project? How much will be saved? Answer. As a percentage of its total budget, CBO's administrative costs have been declining and the agency continues its efforts to hold down administrative costs by concentrating on those areas that are increasing faster than the general inflation rate. Specifically, CBO is looking at ways to offset the recent postal increase of 16 percent. And the agency is looking for alternatives to some of the administrative services currently provided by the Library of Congress, which raised its rates by 32 percent in fiscal year 1991. One of the savings CBO clearly hopes to make in 1991/1992 is a reduction in its inventory of previously published reports. These are stored in a warehouse at our mailing service; periodically excess inventory is disposed of by a GSA approved recycling company, saving us considerable storage expense. It will not be possible to project the amount saved until all of the alternatives have been identified and evaluated. Question. The workload data in your budget is highly detailed (p. 10). We ask for it that way. How accurate is it? For example, you have 1,716 projects (actual) started in FY 1990. How do your track that data? Answer. CBO accounts for its cost estimates with a computerized system that tracks written responses-the official ones provided to the Congress and the informal responses sent to Committee staff. Federal cost estimates are done by CBO for nearly all reported legislation that has a budgetary impact as well as for legislation as it is being developed and at later stages as it moves toward enactment. Since the CBO budget justification was submitted, final numbers for fiscal year 1990 have become available. Federal cost estimates rise slightly from an estimate of 849 to 855. Of the 855, 825 were counted by the correspondence system and 30 informal estimates were reported. State and local estimates fall from an estimated 743 to 720. Both federal cost estimate numbers probably underestimate the workload because omnibus legislation, such as budget reconciliation, is counted as one bill for which estimates may be done at several stages. In addition, undercounting of the informal estimates occurs because the system collects only those that are in writing. When working with committees, analysts often cost out options and transmit results by telephone because of time pressures and none of that work is recorded by our system. The remainder of CBO work products is made up of mandated and requested reports, and testimony, all of which flow from written requests that are controlled by a computerized correspondence system. DATA PURCHASES AND MODEL DEVELOPMENT Mr. REISCHAUER. In large measure, as you know, we have been starving data purchases and model development because they are the kinds of things, like buildings, that you do not have to do every year if you are up against a financial constraint, which we have been up against. We now have a situation where the 1990 Census data is going to become available in the course of this summer and next year. It really will be a time when we have to update these models and buy a lot of new data. I don't think we can Mr. FAZIO. You don't want to input all that new data under old models? Mr. REISCHAUER. Precisely. Mr. FAZIO. Perhaps you could explain for the committee's benefit what these models are and how they are used by CBO? Mr. REISCHAUER. There are so many models that we could be here all afternoon. For example, we have models that we use to project the baseline budget to determine what budget authority and outlays will be for every account in the government for the next five years. We also have models that we use to estimate what happens to food stamps, AFDC, and unemployment insurance when the economy strengthens and weakens. In addition, we have models that we use to estimate what would be the costs of various legislative proposals dealing with expanding health care to the uninsured. That is one area where we are doing a lot of work right now. In other words, what would it cost to cover the approximately 32 million Americans who do not have health insurance, either through some form of Medicaid, some form of Medicare, employer mandates, or whatever. These are all terribly complex models. Mr. FAZIO. These models that you cite are created for you? Do you do it on-site with your staff? Mr. REISCHAUER. The answer is all of the above. We develop a lot of them on our own, but in some cases we must obviously buy software and selected data from other people. We will buy special Mr. FAZIO. When you say a model, do you mean a software package? Mr. REISCHAUER. Right. We will ask a firm that has already done some work in the area to provide us with cleaned-up data in a form that we need. Obviously we have a limited staff, and they cannot spend their time editing and cleaning large data bases. We did a study of the impact of the movement toward capitalism in Eastern Europe, and the unification of Europe in 1992, and the reunification of Germany. What kind of impacts are all of those economic and political changes going to have on interest rates in the world, and in particular, in the United States and on U.S. growth? Would these changes help or hinder U.S. growth? Our analysis depended on three world macro-models. Mr. FAZIO. You are talking about widely varying levels of assumptions that go into those things? Mr. REISCHAUER. Correct. Mr. FAZIO. I have some questions I will submit to be answered for the record. [The questions and responses follow:] Question. There is a $211,000 increase for ADP systems and data development. What is the base budget for these items, and what percentage increase does this $211,000 represent? Answer. The $211,000 increase for ADP systems, data, and model development is 184 percent higher than the fiscal year 1991 base of $114,500. This spending was reduced to a minimum level in fiscal year 1991 in order to fund a full staff and the January 1, 1991 COLA of 4.1 percent. The fiscal year 1992 request returns this spending to its historical levels. Question. What additional data is required? Answer. Maintaining and updating computer systems and models is an ongoing process that requires both software and data enhancements. The funding level requested for fiscal year 1992 anticipates the availability of new data, including the 1990 census data, which is essential to CBO's ability to provide cost estimates in many areas. The record of CBO's and the Administration's economic forecasts over the past 13 years shows that, while the records are close, CBO has made smaller errors most of the time. CBO's edge is especially pronounced when attention is confined to the particular economic variables that have the strongest effect on the accuracy of the two agencies' projections of the budget deficit. CBO's record also compares well to those of the major commercial economic forecasting firms. CBO's 13-year track record in forecasting growth in GNP, inflation, unemployment, and interest rates is detailed in the attached tables (see pages 343 through 351), along with those of the Administration and of several large private forecasters. At the extreme right, the table also shows figures summarizing the accuracy of the forecasts shown. The most useful of these summary measures from a statistical point of view are the mean absolute error and root mean squared error (RMSE), which are shown in the third-to-last and second-to-last columns, respectively. The overall forecasting record is summarized in more concise form on the first table, which is entitled "CBO/OMB Forecast Rankings, 1978-1990". As that summary table shows, CBO made smaller errors than the Administration when forecasting for the current year for five of the six variables appearing in the table. When forecasting one year ahead, CBO's errors were smaller for three of the six variables shown. As for the two variables that matter most to accurate forecasting of the budget deficit-growth in real (inflation-adjusted) GNP and the real interest rate-CBO's record is better than the Administration's on forecasts for both the current year and one year ahead. The tables also show that CBO's forecasting record is quite similar to those of the private economic forecasters that are represented here. It should be noted, however, that all of the differences in forecasting records are small, and the length of the historical record is short. As a result, some analysts do not view them as significant. Question. For the record, list your expenditures for this item and indicate where the increases are budgeted. Answer. Major expenditures in this area in fiscal year 1990 include: Development of a model to measure the economic cost of deposit insurance guarantees-$30,250; Actuarial assistance in estimating long-term care by type and payment source$25,000; Analysis of the economic impact of defense spending reductions-$39,500; and Increases for fiscal year 1992 would be used to enhance our ability to analyze the reorganization of financial institution regulation, estimate Medicare changes by income group, update models for new census data as it becomes available, study health insurance premiums by type of plan, and update existing models for changes to legislative programs. Question. You describe a $270 thousand savings in interagency timesharing achieved in FY 1990 (p. 9). What are those savings associated with? Is the $270 thousand an annual, recurring savings, or a one-time savings? Answer. Most of the $270,000 savings in interagency timesharing achieved in fiscal year 1990 will be a permanent reduction. This savings can be attributed to several factors: 1) increased reliance on microcomputer-based applications, 2) closer monitoring of computer and storage use, 3) cancellation of the IBM 9370, and 4) agreement with HIS to provide 9370 replacement processing capabilities on the HIS mainframe at a fixed monthly cost. Question. Update the ADP contracts and interagency agreements tabulation on p. 388 of the FY 1991 hearings. |