[House Hearing, 105 Congress]
[From the U.S. Government Publishing Office]



 
                   OVERSIGHT OF STATISTICAL PROPOSALS
=======================================================================

                                HEARING

                               before the

                 SUBCOMMITTEE ON GOVERNMENT MANAGEMENT,
                      INFORMATION, AND TECHNOLOGY

                                 of the

                              COMMITTEE ON
                           GOVERNMENT REFORM
                             AND OVERSIGHT
                        HOUSE OF REPRESENTATIVES

                       ONE HUNDRED FIFTH CONGRESS

                             FIRST SESSION

                               __________

                             JULY 29, 1997

                               __________

                           Serial No. 105-87

                               __________

Printed for the use of the Committee on Government Reform and Oversight













                       U.S. GOVERNMENT PRINTING OFFICE
46-777                     WASHINGTON : 2003
_____________________________________________________________________________
For Sale by the Superintendent of Documents, U.S. Government Printing Office
Internet: bookstore.gpo.gov  Phone: toll free (866) 512-1800; (202) 512-1800  
Fax: (202) 512-2250 Mail: Stop SSOP, Washington, DC 20402-0001










              COMMITTEE ON GOVERNMENT REFORM AND OVERSIGHT

                     DAN BURTON, Indiana, Chairman
BENJAMIN A. GILMAN, New York         HENRY A. WAXMAN, California
J. DENNIS HASTERT, Illinois          TOM LANTOS, California
CONSTANCE A. MORELLA, Maryland       ROBERT E. WISE, Jr., West Virginia
CHRISTOPHER SHAYS, Connecticut       MAJOR R. OWENS, New York
STEVEN SCHIFF, New Mexico            EDOLPHUS TOWNS, New York
CHRISTOPHER COX, California          PAUL E. KANJORSKI, Pennsylvania
ILEANA ROS-LEHTINEN, Florida         GARY A. CONDIT, California
JOHN M. McHUGH, New York             CAROLYN B. MALONEY, New York
STEPHEN HORN, California             THOMAS M. BARRETT, Wisconsin
JOHN L. MICA, Florida                ELEANOR HOLMES NORTON, Washington, 
THOMAS M. DAVIS, Virginia                DC
DAVID M. McINTOSH, Indiana           CHAKA FATTAH, Pennsylvania
MARK E. SOUDER, Indiana              ELIJAH E. CUMMINGS, Maryland
JOE SCARBOROUGH, Florida             DENNIS J. KUCINICH, Ohio
JOHN B. SHADEGG, Arizona             ROD R. BLAGOJEVICH, Illinois
STEVEN C. LaTOURETTE, Ohio           DANNY K. DAVIS, Illinois
MARSHALL ``MARK'' SANFORD, South     JOHN F. TIERNEY, Massachusetts
    Carolina                         JIM TURNER, Texas
JOHN E. SUNUNU, New Hampshire        THOMAS H. ALLEN, Maine
PETE SESSIONS, Texas                 HAROLD E. FORD, Jr., Tennessee
MICHAEL PAPPAS, New Jersey                       ------
VINCE SNOWBARGER, Kansas             BERNARD SANDERS, Vermont 
BOB BARR, Georgia                        (Independent)
ROB PORTMAN, Ohio
                      Kevin Binger, Staff Director
                 Daniel R. Moll, Deputy Staff Director
         William Moschella, Deputy Counsel and Parliamentarian
                       Judith McCoy, Chief Clerk
                 Phil Schiliro, Minority Staff Director
                                 ------                                

   Subcommittee on Government Management, Information, and Technology

                   STEPHEN HORN, California, Chairman
PETE SESSIONS, Texas                 CAROLYN B. MALONEY, New York
THOMAS M. DAVIS, Virginia            PAUL E. KANJORSKI, Pennsylvania
JOE SCARBOROUGH, Florida             MAJOR R. OWENS, New York
MARSHALL ``MARK'' SANFORD, South     ROD R. BLAGOJEVICH, Illinois
    Carolina                         DANNY K. DAVIS, Illinois
JOHN E. SUNUNU, New Hampshire
ROB PORTMAN, Ohio

                               Ex Officio

DAN BURTON, Indiana                  HENRY A. WAXMAN, California
          J. Russell George, Staff Director and Chief Counsel
                         Mark Uncapher, Counsel
                 John Hynes, Professional Staff Member
                          Andrea Miller, Clerk
           David McMillen, Minority Professional Staff Member















                            C O N T E N T S

                              ----------                              
                                                                   Page
Hearing held on July 29, 1997....................................     1
Statement of:
    Katzen, Sally, Administrator, Office of Information and 
      Regulatory Affairs, Office of Management and Budget, 
      accompanied by Katherine Wallman, Chief Statistician of the 
      United States..............................................     2
    Sondik, Edward J., Director, National Center for Health 
      Statistics; Jay Hakes, Administrator, Energy Information 
      Administration, Department of Energy; Everett Ehrlich, 
      former Under Secretary for Economic Affairs, Department of 
      Commerce, Clinton Administration, ESC Corp.; Mark Wilson, 
      Rebecca Lukens fellow in labor policy, Heritage Foundation; 
      and Mary Susan Vickers, research director, Interstate 
      Conference of Employment Security Agencies, Inc............    30
Letters, statements, etc., submitted for the record by:
    Ehrlich, Everett, former Under Secretary for Economic 
      Affairs, Department of Commerce, Clinton Administration, 
      ESC Corp., prepared statement of...........................    46
    Hakes, Jay, Administrator, Energy Information Administration, 
      Department of Energy, prepared statement of................    42
    Katzen, Sally, Administrator, Office of Information and 
      Regulatory Affairs, Office of Management and Budget:
        Director's review........................................    17
        Information concerning requests per year from departments    23
        Information concerning survey requests turned down in the 
          average year...........................................    25
        Prepared statement of....................................     5
    Sondik, Edward J., Director, National Center for Health 
      Statistics, prepared statement of..........................    32
    Vickers, Mary Susan, research director, Interstate Conference 
      of Employment Security Agencies, Inc., prepared statement 
      of.........................................................    62
    Wilson, Mark, Rebecca Lukens fellow in labor policy, Heritage 
      Foundation, prepared statement of..........................    51
















                   OVERSIGHT OF STATISTICAL PROPOSALS

                              ----------                              


                         TUESDAY, JULY 29, 1997

                  House of Representatives,
Subcommittee on Government Management, Information, 
                                    and Technology,
              Committee on Government Reform and Oversight,
                                                    Washington, DC.
    The subcommittee met, pursuant to notice, at 2:03 p.m., in 
room 2154, Rayburn House Office Building, Hon. Stephen Horn 
(chairman of the subcommittee) presiding.
    Present: Representatives Horn and Sununu.
    Staff present: J. Russell George, staff director and chief 
counsel; Mark Uncapher, counsel; John Hynes, professional staff 
member; Andrea Miller, clerk; David McMillen, minority 
professional staff member; and Ellen Rayner, minority chief 
clerk.
    Mr. Horn. The meeting of the Subcommittee on Government 
Management, Information, and Technology will begin.
    The economic statistics gathered and analyzed by the 
Federal Government are integral to public and private 
decisionmaking. The financial markets rise and fall based on 
the data provided by the government. Federal aid is determined 
and distributed using this information. Businesses make a wide 
variety of decisions with reference to these statistics. 
Although sound statistics and analysis do not by themselves 
produce sound public policy, they do provide the necessary 
foundation from which to identify problems, to evaluate 
options, and to monitor results.
    We are here to consider three initiatives intended to 
encourage greater cooperation and coordination between the 
Federal Government's statistical agencies.
    The first is the consolidation of the three main 
statistical agencies into a single entity. Introduced last 
Congress as the Statistical Consolidation Act, this measure 
would create the Federal Statistical Service as an independent 
agency. The Service would incorporate the Bureau of the Census, 
the Bureau of Labor Statistics, and the Bureau of Economic 
Analysis. This proposal directly addresses the need for better 
coordination and planning among economic statistical agencies.
    The second initiative under discussion today is the 
creation of a commission. Senator Daniel Patrick Moynihan of 
New York has proposed that a commission be established to 
provide ``a comprehensive examination of our current 
statistical system and focus particularly on the agencies that 
produce data as their primary product.'' The commission would 
be charged with recommending a strategy to maintain a modern 
and efficient statistical infrastructure.
    The third initiative is a data sharing proposal put forward 
by the administration. This measure would designate eight 
statistical agencies as statistical data centers and establish 
new laws to assure confidential treatment of statistical data 
in this environment. The bill would allow agencies to propose 
data sharing projects and would provide protections for the 
confidentiality of information.
    All of these proposals share the goal of improving Federal 
statistical systems by reducing the organizational and legal 
barriers to greater coordination. Each seeks to address the 
fragmented nature of the Federal Government's statistical 
agencies. Our challenge now is to build a consensus for 
concrete steps toward reform. A very productive discussion has 
been under way for several months now between the House and the 
Senate. We are working for a rough outline for reform and are 
confident that an agreement will soon be reached between the 
two legislative bodies.
    We are fortunate enough to be joined by experts who can 
help us move toward a consensus proposal. We begin with the 
view from the administration: Sally Katzen, Administrator, 
Office of Information and Regulatory Affairs, Office of 
Management and Budget. After Ms. Katzen, Dr. Edward J. Sondik, 
Director, National Center for Health Statistics, and Mr. Jay 
Hakes, Administrator, Energy Information Administration, 
Department of Energy, will round out the testimony from the 
administration.
    We will continue the discussion by turning to some of the 
top minds working on statistical issues at the Federal level. 
They will help us to sort through the many complexities 
involved in improving Federal statistics, and we look forward 
to their testimony.
    So welcome to all witnesses. We will begin with this panel. 
Since this is a separate hearing, we should swear you in.
    [Witnesses sworn.]
    Mr. Horn. We will note both Ms. Katzen and Ms. Wallman, the 
Chief Statistician of the United States, did affirm the oath.
    We will begin with Ms. Katzen.

STATEMENT OF SALLY KATZEN, ADMINISTRATOR, OFFICE OF INFORMATION 
   AND REGULATORY AFFAIRS, OFFICE OF MANAGEMENT AND BUDGET, 
  ACCOMPANIED BY KATHERINE WALLMAN, CHIEF STATISTICIAN OF THE 
                         UNITED STATES

    Ms. Katzen. Thank you very much, Mr. Chairman. It is a 
pleasure to return to your subcommittee and to return to a 
discussion of this issue.
    Your counsel provided me the opening line that I used the 
last time I was here to testify on this subject, when I said I 
hope this is the first of many opportunities to engage with 
your subcommittee on how best to enhance the efficiency and 
effectiveness of our Federal statistical activities. I was 
prescient then, and I am delighted to return to this subject.
    In your opening statement you reflected on the goal--it is 
a goal we share--to improve the quality and usefulness of our 
Nation's statistics, even as we seek to maximize the 
effectiveness of scarce resources for statistical activities 
and minimize the burden we impose on the American public.
    When I testified before you last March, I outlined a number 
of initiatives that we were pursuing to achieve this goal; and 
I would like to report briefly on the progress that has been 
made since that time.
    First, with respect to the strengthening of confidentiality 
protections, we have now issued the Federal Statistical 
Confidentiality Order which officially took effect yesterday, 
July 28, 1997. In brief, this order establishes a consistent 
confidentiality policy for a dozen agencies initially and which 
other agencies may aspire to meet in the future. The analyses 
required by the order will provide a comprehensive assessment 
of current disclosure policies and practices for consistency 
with this governmentwide policy and will result in corrective 
action, if necessary.
    Second, we have completed our work with the statistical 
agencies to prepare and submit to Congress the administration's 
proposed Statistical Confidentiality Act, which you and Mrs. 
Maloney introduced during the last session as H.R. 3924. In 
brief, enactment of this legislation would: create a credible 
governmentwide confidentiality umbrella that would guarantee 
that the entire government stands behind pledges of statistical 
confidentiality; create the legal presumption that data 
collected for any purpose may be used in a safe environment for 
statistical purposes; provide a consistent policy in all 
statistical data centers for treatment of confidential 
statistical data under the Freedom of Information Act; permit 
the data sharing authorities of the Paperwork Reduction Act to 
work without compromising confidentiality; and provide a 
privacy sensitive alternative to the creation of universal data 
bases that different departments have proposed at one time or 
another to support their own policy interests.
    In short, the Statistical Confidentiality Act that we have 
put together permits the designated data centers and 
statistical agencies working with them to share both expertise 
and data resources in order to improve the quality and reduce 
the burden of statistical programs while preserving 
respondents' privacy. Moreover, however the organizational 
boxes for the ideal Federal statistical system may be drawn, 
this bill will permit the components of the statistical system 
to manage their data as if they were a single, functionally 
integrated organization. The administration's bill was 
transmitted to the Congress again on June 5 of this year, and 
we are eager to continue our work with you toward its 
introduction and passage.
    Third, there has been progress to enhance coordination and 
collaboration among the statistical agencies in other areas as 
well. Perhaps most notably, when Director Raines took the helm 
at OMB last summer, he made clear the importance of greater 
integration of statistical agencies as a priority for budget 
review; and, to that end, he held for the first time in 20 
years, a formal Director's crosscutting review of Federal 
statistical programs. This review served to identify a set of 
high-priority, cross-agency initiatives that would ameliorate 
the increasing inability of our statistical system to mirror 
the current economy and to foster accurate allocation of 
increasingly scarce Federal resources.
    In addition, Director Raines challenged statistical 
agencies to demonstrate further evidence of collaboration and 
to propose interagency initiatives that would address important 
national priorities.
    I am pleased to report that the agencies have accepted that 
challenge and identified several activities to further improve 
the overall performance and efficiency of the Federal 
statistical system, including addressing the significant 
statistical issues associated with improving the measurement of 
income and poverty; organizing our efforts to meet emerging 
welfare and health data needs; strengthening data on national 
and personal income; and managing the transition to the new 
North American Industry Classification System. These and other 
collaborative initiatives will more closely link the 
statistical agencies and address important national priorities.
    Mr. Chairman, the rest of my comments are set forth in my 
written statement; and I will not repeat them here.
    I would like to use this opportunity to stress the 
importance of enacting legislation that would permit limited 
sharing of data among the principal statistical agencies for 
statistical purposes. I cannot overemphasize how critical this 
effort is, regardless of what decisions may be made about 
reconfiguring the statistical agencies.
    I thank you, Mr. Chairman, for your patience; and I would 
be pleased to answer any questions you have.
    Mr. Horn. We thank you for that statement.
    [The prepared statement of Ms. Katzen follows:]



    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
    
    
    Mr. Horn. What I would like to know is, with the Federal 
Statistical Confidentiality Order which took effect yesterday, 
July 28, 1997, how much of the matter of limited sharing of 
data between agencies has been covered by that particular 
Executive order?
    Ms. Katzen. It does not cover sharing, but sharing is one 
part of a two-part process. The inhibitions to sharing, whether 
they be legislative or administrative, to date have been 
concerned about how different agencies will treat confidential 
information. So before we got to sharing, we wanted to ensure 
that agencies would keep confidential information confidential 
in a consistent, governmentwide approach; and the order that I 
signed was designed, to that end, to establish a governmentwide 
set of standards, or set of practices to ensure 
confidentiality.
    With that in place, it seems to us that it is now easier to 
begin talking about sharing, since you know that, if you give 
data to one of the other agencies, these data will be treated 
with the confidentiality they deserve.
    Mr. Horn. As I read your testimony, this really related to 
carrying out the Paperwork Reduction Act of 1995. Am I wrong on 
what you are saying on page 1?
    Ms. Katzen. We can use some of the authority under the PRA 
to do administratively the kinds of steps that we took in the 
order. So we used that authority to do administratively.
    In terms of sharing data, which would be the second step, 
that requires legislation; and that is why we have sought to 
work so closely with you.
    Mr. Horn. Give me an example of some of the data sharing 
that is needed, cannot be done now because it hasn't been 
authorized by the Congress. Give me an example of what you are 
worried about here. What are you trying to achieve?
    Ms. Wallman. Do you want me to get into the act here?
    Mr. Horn. Fine.
    Ms. Wallman. There are a number of areas where we believe 
that progress could be made. The one that is best known to 
people, of course, has been the issue of lists or list frames 
for drawing samples, both for economic and other kinds of 
activities in the survey area.
    We also believe that there are areas where having agencies 
capable of sharing one another's information could allow them 
to check out some quality issues that have become of increasing 
concern; but the agencies are currently, because of their 
current confidentiality laws, unable to undertake that sharing.
    There are other areas where we see new prospects that 
merely haven't been examined previously given the history of 
the earlier legislation--areas in health, for example, where 
there are new opportunities for looking at issues related to 
industries that have health activities--health-related 
activities and hospitals that have health-related activities--a 
collage of activities--for example, benefits under these 
different programs.
    My colleagues, actually, from the Energy Information 
Administration and the National Center for Health Statistics 
will provide better examples when they talk, and we have a 
number of other examples.
    Ms. Katzen mentioned the work that the Interagency Council 
on Statistical Policy has been pursuing over the last several 
months. That initiative has come up with a fairly detailed list 
of opportunities that the agencies see; and, actually, that is 
what underscores our concern that we move forward on this 
legislation. Many of the opportunities for increasing the 
quality of information and restraining our costs at the same 
time lie in the passage of this legislation.
    Mr. Horn. Well, is the sharing provisions of that 
legislation adequate to what you are talking about here? Does 
that give you the authority if Congress were to pass it?
    Ms. Wallman. The legislation that we have proposed gives us 
the authority that is needed to carry out the kind of 
activities we are talking about, yes.
    Mr. Horn. Are there any agencies that are exempted from 
that?
    Ms. Wallman. The legislation actually works in the opposite 
direction, if you will. It empowers eight agencies to become 
statistical data centers. They are the core of the opportunity 
for sharing. In addition to that, other agencies would be able 
to work with the designated--excuse me, that comes from a 
historical part of my life--the statistical data centers.
    So the opportunity is broader than the eight, but the eight 
are the ones that would be the centers of the activity.
    Mr. Horn. So, in other words, as you just suggested, the 
sharing could go beyond the eight with other Federal agencies.
    Ms. Wallman. With specific agreements with those agencies, 
yes, sir.
    Mr. Horn. Do you have the eight in front of you there?
    Ms. Wallman. I do have the eight in front of me.
    Mr. Horn. Would you mind reading it into the record?
    Ms. Wallman. I would be happy to.
    Mr. Horn. What groupings are we talking about?
    Ms. Wallman. We are speaking specifically about the Bureau 
of Economic Analysis, the Bureau of the Census----
    Mr. Horn. Is that all one? Give it to me one, two, three, 
four.
    Ms. Wallman. I am going to enumerate eight.
    Mr. Horn. OK. Tell me when you move to two.
    Ms. Wallman. No. 1 was the Bureau of Economic Analysis. I 
am doing these in alphabetical order, I believe. No. 2 is the 
Bureau of the Census; No. 3, the Bureau of Labor Statistics; 
No. 4, the National Agricultural Statistics Service; No. 5, the 
National Center for Education Statistics; No. 6, the National 
Center for Health Statistics; No. 7--it is slightly more 
complicated--the Energy End Use and Integrated Statistics 
Division of the Energy Information Administration; and No. 8 is 
the Division of Science Resources Studies at the National 
Science Foundation.
    Mr. Horn. There are two obvious ones that are not there. 
Let me raise them. One is Social Security and the other is 
Immigration and Naturalization Service.
    We have legislation in which would provide access, if they 
wish to, to chief State election officers, to county registrars 
of voters, if they are checking whether a person is a citizen 
of the United States. Since 1982, my understanding is that the 
Social Security Administration asks for documentation of 
citizenship.
    Go ahead, if you want to take Ms. Katzen's advice. I will 
stop there and get to the next one.
    So why wasn't Social Security included? This is a basic 
resource, and it isn't a matter of revealing one's amount that 
they are paid or anything else, but we also have the problem of 
hunting for deadbeat dads.
    Ms. Katzen. The comment I was offering to her is that there 
are two different--they are not totally unrelated and not 
mutually exclusive, but there are two quite different areas 
that you can be talking about when you talk about sharing 
information.
    On the one hand, what we have been talking about to date 
and the reason the eight agencies were selected and what is 
reflected in our proposed legislation goes to the sharing of 
statistical information for statistical purposes. And that is, 
as we were mentioning the other day, we are talking about 
aggregate information and the ability to share information 
among those agencies that enable you to do basically two 
things:
    One is to check the quality of the data that you have. 
Again, on an aggregate basis, individual personal identifiers 
are irrelevant. You are verifying the quality of the data.
    The second purpose is to reduce respondent burden. In some 
instances persons will be responding to surveys twice, three 
times, or four times: if those to whom they are responding 
could share the data, we could minimize the burden on 
respondents. That is not just an end in itself, but that would 
then presumably heighten the respondents willingness to respond 
and that would provide us more timely and more accurate 
information. All of that is on a statistical information 
platform, if I could call it that.
    On the other hand, the references that you have made to 
Social Security, to INS, and indeed even to IRS go to 
individual data for individual personally identified persons or 
entities. That kind of information can be used, if it were 
available and for the most part it is not because of statutory 
restraints to check for eligibility for benefits, to check for 
appropriateness of citizenship status or other types of 
programmatic administrative or enforcement efforts that are not 
necessarily aggregate information, but individually based 
information.
    That type of sharing, which you can have under the Computer 
Matching Security Act, for example, and issues that are not the 
thrust of the confidentiality--I am sorry, the statistical data 
centers that we are seeking to establish here. The sharing of 
data that we have been talking about is the sharing of 
aggregate statistical information.
    As I said, these two areas are not necessarily mutually 
exclusive. In fact, there are people in my office who are 
looking to expand areas where individualized data can, in fact, 
be shared. But that is a separate issue from the issue that has 
driven the discussion so far about enhancing the effectiveness 
of the statistical agencies which are developing statistical 
information for statistical purposes.
    Mr. Horn. Well, you are talking about having accurate and 
adequate statistical information. It seems to me one way you do 
that, if you are worried about socioeconomic class, is you find 
out what the income was for those that answered the BLS bread 
basket questions. You see if they are eating certain lines of 
food or certain purchases out in the market beyond food, is 
there a relationship, obviously, between income. And you would 
have accurate income data then if you could run the tape of 
those interviewed against what they file for income. So that is 
a socioeconomic class usage.
    You can do the same with the health center. Maybe it is 
just too obvious that socioeconomic class, some people there 
with certain diseases and certain types of psychological 
whatevers, have it more than another socioeconomic class. So 
you can use that individual data.
    Ms. Katzen. Absolutely.
    Mr. Horn. It isn't just a statistical problem on the board. 
It is the reliability of the government's statistics, which 
some of us have real doubts about, by the way, because we 
wonder why they do not do that.
    To finish with the Immigration and Naturalization Service, 
obvi- ously one way to check citizenship on voting--and I would 
hope the administration would agree that the sanctity of the 
ballot box--is something that all parties ought to agree about, 
that you want only citizens voting. I would hope the 
administration would take that view.
    If they do, then we need to know the citizenship 
information held by Social Security and we need to know the 
natu- ralization information. So a person can be made eligible 
to vote through the naturalization process, and we ought to be 
encouraging that.
    Ms. Katzen. I am not saying that it is not desirable to be 
able to link those data bases and provide that information. I 
am saying that is a different subject than determining how many 
in the ag- gregate of one socioeconomic class or another--how 
many have how much income, or what type of health problems, 
what type of other types of issues, is very different from 
saying you are going to track a particular person through 
various data bases.
    That was the only distinction I was making. The two are not 
mu- tually exclusive, and to my mind there is enormous benefit 
to be derived from the linking of those types of data bases for 
the check- ing and cross-checking purposes. There are costs to 
that which are very significant.
    If you go anywhere near the IRS data base, there is a very 
strong wall that is built; and that is done because tax forms 
are filed, in effect, voluntarily. If people feel their tax 
information will be used to their detriment without their 
knowledge, there may be an ad- verse effect on the response 
rate, which conceivably could be very significant.
    So those kind of issues I think are very real. I am not 
saying they are dispositive or they are not worthy of 
consideration. I am just saying they are a different type of 
issue than what we have been focusing on in the preparation of 
our proposed legislation. We are working in the other arena as 
well, but they are two different arenas. That was the only 
point I wanted to make.
    Mr. Horn. As to the walls within the Immigration and Natu- 
ralization Service, let's say, we find for some people those 
walls are easily broken down and we will be checking that. One 
subcommit- tee has already. Some people get lists of 
naturalized citizens, and the other people don't get lists of 
naturalized citizens. So you could even write them a 
congratulatory letter.
    But it seems to me they make various decisions that don't 
nec- essarily apply to all of us, just some of us. So I am 
curious about that, and another Subcommittee on National 
Security is probably examining that one.
    Well, what do you think about the commission proposal that 
was outlined? I am sure you are familiar with it. Senator 
Moynihan in- troduced it. He and I agreed we will try to work 
something out, that this commission would perhaps come up 
almost like a base closure commission, after taking a lot of 
testimony, and come in with a plan on the integration; and we 
either vote it up or down, one way or another. That is sort of 
a popular device around here on controversial things that we 
otherwise don't put our fingers on.
    So whether that would be the outcome, it is hard to say. We 
don't know what the commission might do. But have you thought 
through some options of what a commission might do and what its 
agenda should be?
    We are searching for ideas, in brief. You have always had 
good ideas. So what are they?
    Ms. Katzen. When I appeared last time before your 
committee, we spent a lot of time talking about the costs and 
the benefits of rearranging the boxes, and I think a lot of 
work has gone into thinking about those issues.
    The benefits are fairly obvious; the disadvantages are less 
so, but nonetheless real. A commission might be able to 
approach it with bipartisan, professional individuals of 
stature committed to the same common goals and shed more light 
on the issue.
    We know that a lot of work has been done in the past, and 
that undoubtedly more work will be done in the future. Our 
attempt has been to try to make the most effective, efficient 
use of the resources that we have and, within the constraints 
that are established by the shape of the executive branch, once 
by the organization of the Congress in terms of different 
committees with different jurisdic- tions, to try to weld this 
group together in a way that is going to maximize the benefits.
    Any help that can be given to that effort, any support for 
our ini- tiatives, would be greatly appreciated. And therefore 
I would hope, if a commission were formed, it would think not 
just about rear- ranging the boxes, but also whether, short of 
reorganization, there are important steps that can be taken to 
enhance the accuracy, timeliness, and reliability of 
statistical information.
    The items that I identified in my written statement and my 
oral statement that we have undertaken are, I think, very good 
pros- pects. We can use all the help we can get in terms of 
support. And I am sure we do not have all of the answers. I am 
not even sure we'd know all of the questions.

    So bright minds focused on this issue could be highly 
desirable.

    Mr. Horn. On that point, you mentioned timeliness of data. 
Are there particular concerns in terms of the timeliness 
related to deci- sionmaking that any administration has to 
make? Or what is your feeling on that?

    Ms. Katzen. Statistical information is used not just by 
decisionmakers within the government, but also by those in the 


private sector. They run from very large and much anticipated 
an- nouncements of unemployment increases and decreases, or GDP 
numbers, to more focused, specific--sometimes sector-specific, 
sometimes geographic-specific--types of data.
    In some instances, there have been criticisms in the trade 
press and in the general press of whether they are able to--
notwith- standing all of our computer power, whether we are 
able to gather the information necessary, to make it available, 
particularly to the private sector as well as the public 
sector, on a timely basis.
    Mr. Horn. Director Raines, when he asked for the 
crosscutting study and the analysis, has that come in yet?
    Ms. Katzen. Yes, we did that--time flies when you are 
having fun; it was last October, in anticipation of the fiscal 
year 1998 budget. There is an annual budget review, a 
Director's review, pre- ceding the formulation of the 
President's budget each year.
    Traditionally, they have gone area by area as set forth in 
the President's budget. This year, Director Raines called for 
the inclu- sion in that process. It was actually October or 
November, he in- cluded a crosscutting review for the 
statistical agencies; and Ms. Wallman and I appeared at the 
Director's review with a series of proposals that we thought 
would enhance the effectiveness of the statistical agencies, 
looking for proposals that were not solely based in one agency 
but that would be able to assist a number of agen- cies. I have 
identified some of those in my written testimony.
    The last one that I mentioned, for example, has to do with 
en- hancing the use of information for the Government 
Performance and Results Act. We are looking for outcome 
measures from the various agencies as part of the new 
management approach that I think is extraordinarily important 
and productive for the executive branch; and the ability of 
agencies to provide the information that is necessary to make 
that work is one that no single agency would want to devote its 
particular resources from its statistical agency to that 
effort. But as a crosscutting matter, this is something where 
relatively few dollars could go a very, very long way to 
everyone's benefit.
    There are a number of other areas that we identified that 
in- volved different issues that we thought were very 
important. Some of those I remember were located in BLS, one of 
them having to do with the CPI and the additional work that was 
appropriate in that area that Congress has identified and the 
executive branch has identified and we wanted to support; 
moving to the North American Industry Classification System, 
implementing that, which would assist a number of different 
agencies. We identified five or six different initiatives, and 
the Director was very positive and responsive and approving of 
our proposals.
    Since then, Ms. Wallman has gone back to the Interagency 
Coun- cil and said, for this year's crosscutting review, why 
don't you all help us formulate the initiatives and proposals 
to present to the Di- rector? They have been very responsive 
and enthusiastic about that, and I expect this October or 
November we will have a bigger and better menu to choose from.
    Mr. Horn. Any comment, Ms. Wallman?
    Ms. Wallman. No. Ms. Katzen has it just right.
    I think the distinction she was trying to draw at the close 
of her remarks was that much of what was done in the fiscal 
1998 budget was done internally within the Office of Management 
and Budget, based on knowledge of the staff over the years. The 
activity that is about to come to completion this year toward 
fiscal 1999 and the future--it is not just a 1-year effort--I 
think I would underscore that--that that has been done in a 
much more outgoing way, if you will, with the agencies that 
form the Interagency Council on Statistical Policy, the 14, and 
that has helped us identify a broader, richer set of 
activities, some of which could start immediately, some of 
which we need to put some more thinking into, frankly, before 
we would come forward with specific proposals in a budgetary 
sense.
    The third group of activities, I underscore again, includes 
things that would be available under sharing capabilities we 
don't now have.
    Mr. Horn. When the crosscutting study was done, I take it 
was strictly OMB people on that, or were agency people also on 
that review?
    Ms. Katzen. Last year it was an internal OMB meeting.
    Mr. Horn. I gathered it was just OMB.
    Ms. Katzen. This year we expect to have more input, 
although all Director's reviews are internal only, for each of 
the departments and agencies. OMB staff presents the 
outstanding issues to the Director.
    Mr. Horn. Can we get a copy of that review to see what the 
recommendations were?
    Ms. Katzen. I will be happy to look into that.
    Mr. Horn. Look into it. We will put it at this point in the 
record. If they don't want it public, we will work out 
something. I would just like to see what some of the thinking 
is.
    [The information referred to follows:]



    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
    
    Mr. Horn. The Interagency Council, is that established by 
Executive order essentially?
    Ms. Katzen. It was established originally--or a variation 
of it was established originally--by our office, but it was, in 
effect, codified in the Paperwork Reduction Act of 1995, which 
gave it a congressional imprimatur and a formal name. Ms. 
Wallman is the chair of that organization.
    Mr. Horn. The 14 include you, Ms. Wallman, or are they in 
addition to you?
    Ms. Wallman. They are agencies, and I am the chair ex 
officio.
    Mr. Horn. Sure. So the 14 essentially recognize the major 
statistical agencies in the executive branch?
    Ms. Wallman. Yes.
    Mr. Horn. Should it be larger than that? Are there others 
that have their feelings hurt? What are a few more seats at the 
table?
    Ms. Wallman. If I might, the eight I enumerated for you in 
terms of the Statistical Confidentiality Act are indeed the 
first eight in that group. There were four more that were added 
actually by my predecessor in this position. And we have, in 
fact, entertained petitions, if you will, from agencies that 
particularly wish to be members and sit at this table; and we 
have embraced a couple of additional agencies on that basis.
    The legislation that the Congress passed in 1995 also 
suggested that we include on a rotating basis the heads of some 
of the smaller agencies. In effect, we have been doing that in 
adding the extra agencies we have embraced already and would 
certainly be willing to entertain others. Most recently, we 
brought in the Social Security Administration's office 
responsible for statistics.
    Mr. Horn. Are they smitten by their title or function 
within the agencies, or does the Administrator simply appoint 
them?
    Ms. Wallman. The members are the heads of the agencies 
themselves. Dr. Hakes and Dr. Sondik, who will join you 
shortly, are the heads of their respective agencies and sit at 
the Council table.
    Mr. Horn. Let me ask you one question. I will then call on 
Mr. Sununu.
    I understand from your testimony that under the 
administration's statistics 2000 initiatives, private 
respondents, individuals and businesses could save tens of 
millions of dollars in compliance costs and reduced burden 
hours from filling out Federal Government surveys. Are there 
any estimates of the burden hours that could be saved as a 
result of the administration's data sharing proposal? Do you 
have an estimate on that? And how did you find out what that 
burden-sharing was?
    Ms. Wallman. No, we do not have a specific estimate at this 
point of respondent burden hours that would be saved.
    I think that, in all fairness, the only place we could 
probably come even close at this point on giving you some 
estimates would be in the area of developing sampling frames.
    Some of the other opportunities that I mentioned briefly 
and that my colleagues may mention in more detail in a few 
minutes are new proposals where we really don't have a good 
base of prior experience to give you a very firm estimate, but 
it is the kind of thing we would hope to look into when this 
becomes a reality.
    Mr. Horn. Well, we would like to hear about that. Because, 
obviously, everybody that ever walks into a small business, a 
large business or a farm, they all hear about Federal paperwork 
and surveys, no matter how useful some of that might be.
    I recall in the Roosevelt administration Secretary of 
Agriculture Wallace went out visiting some farmers. When he 
came back, he was livid at all of the forms his own agency was 
sending to farmers; and that is when he invested in a fairly 
sophisticated statistical operation in the Second World War. So 
it is not a new problem. It is an old problem.
    Of course, that is why old BOB, and now OMB, your job is to 
clear statistical surveys; and, hopefully, that burden thing is 
somewhere in the mind and we can say do we really need to know 
this when somebody has a great idea. It might make a terrific 
doctoral dissertation, but do we need to burden the American 
people with that?
    Ms. Katzen. The burden is very much in the forefront of our 
mind.
    Mr. Horn. How many requests do you get a year from 
departments, a ball-park total, or file it for the record if 
you want.
    Ms. Katzen. I will be happy to provide a general statement.
    [The information referred to follows:]

    Over the past ten years, we have received between 2500 and 
3500 requests for information collection approvals annually. 
These totals include all types of collections--those required 
to carry out regulations and to administer benefit programs, as 
well as statistical surveys.

    Ms. Katzen. A wide variety of information collection 
requests that we receive, whether it be for statistical 
agencies per se, we have now again--as part of the GPRA 
approach, we have a series of departments and agencies who are 
interested in sending out something which might be considered 
customer service surveys to find out whether the department or 
agency is, in fact, serving the needs of the constituent 
interests.
    That is a very different type. Some of those are on a 
voluntary basis. They still would be counted under the 
Paperwork Reduction Act. We would still look at those to ensure 
that not only the burden is kept to a minimum, but they are 
structured in a way to produce accurate, responsible 
information.
    Then you have a lot of other types of information 
collections, ranging from tax forms to applications for 
passports, applications for loans from the SBA, or from an 
educational student loan. Each of these is called an 
information collection request, and they all come through our 
office, and they are very different kinds of animals.
    Mr. Horn. I would hope, since we are trying to get results 
under the Government Results Performance Act, that those 
surveys, if they make some sense, are just good public policy. 
I think they are useful in one agency head or more in 
fulfilling that law which we feel very keenly about. So I would 
hope those are not dumped. And if they are not very well-
designed, then I would hope the administration or OMB would 
just redesign them.
    Ms. Katzen. We are actually working toward that. We have a 
generic clearance process to enable us to quickly approve 
certain types of forms; and where there are issues spotted, Ms. 
Wallman's able staff has been of assistance to the departments 
and agencies to redesign them so that the resulting information 
is more valuable.
    Mr. Horn. How many survey requests do you turn down in the 
average year?
    Ms. Katzen. I would have to get you that information.
    Mr. Horn. Could you, please?
    [The information referred to follows:]

    In 1996, we received a total of 2433 requests for 
information collection approvals. Of those, 19 ultimately were 
disapproved. Details on the submissions and disapproval's by 
department were as follows:



    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]

    Ms. Katzen. Our objective is not to turn them down but 
rather to enhance their utility or point the department or 
agency to another place where they could get that kind of data 
without having to conduct their own survey. It is therefore a 
false measure to say, if we turn a certain amount down, it says 
something about either how careful or how creative we are 
being.
    Mr. Horn. Well, we will save a little place in the record 
for that exhibit. I would be interested, as stupid as I am sure 
it is, to see how many you turn down and say this shall not 
pass the gate.
    If it makes sense to share data, which you are into sharing 
data, to get more cooperation between agencies through data 
sharing, why not go one step further and just combine agencies? 
How do you feel about that?
    Ms. Katzen. As I mentioned to you the last time I 
testified, there may be certain benefits of efficiency of 
operation, although those would be more long-term, and in the 
short-term there is enormous disruption from moving agencies 
around. We have seen this in other areas of reorganization 
within the executive branch.
    In addition, as many in business have discovered with their 
R&D projects, for example, it is very different to have a 
centralized R&D office and R&D groups that are located in the 
heart of a product manufacturing area or service provision 
area, so that the R&D is actually tied to the products or 
services rendered. Using that analogy in the statistical 
agencies, it is clear that in addition to several major 
statistical agencies, there are a number of--I don't want to 
call them minor--but less major agencies in each of the 
departments that are responsive to the programmatic needs of 
that department.
    If one consolidated all of the major statistical agencies 
and each of the small pieces of statistical agencies found in 
the executive branch, there is, to my mind, a very strong 
likelihood that what would be created would not be responsive 
to the needs of the Department of Education or to the 
Department of Agriculture or the Department of the Interior; 
and those Departments would then, in effect, recreate to meet 
their own needs certain information-gathering processes.
    There is, of course, a distinction between the collection 
of information and the analysis and the application of that 
information. Those are three different steps. But once one 
speaks about consolidating, one has to think about the uses of 
information. That is why I said earlier I think it is a more 
complicated question than just taking an organization chart and 
moving a bunch of boxes around. And the agencies in many 
instances, have well served their departments and in turn have 
been enhanced in their approach to collection by being located 
in a department or agency that has a particular programmatic 
jurisdiction or objective. It can go either way, functional or 
programmatic.
    So my own view, as I expressed last year, was that I 
thought it would be more productive to form a, if I can use the 
term, ``virtual'' or functionally integrated statistical system 
from the existing decentralized system and not try to 
centralize it all.
    My experience in the past year, particularly with the 
efforts that Director Raines undertook, that Ms. Wallman has 
done with the Interagency Council, has reinforced that view, 
that we can make the decentralized system work more 
effectively, more efficiently, and we do not need to centralize 
it.
    Mr. Horn. On that note, I will yield to the gentleman from 
New Hampshire for as long as he likes, Mr. Sununu.
    Mr. Sununu. Thank you, Mr. Chairman. I have only one focus 
question, and that is on the issue of the States' interaction 
and the States' relationship to any data sharing that might 
occur.
    I would imagine a number of government surveys rely on 
State and local governments for data collection and that these 
collections would have certain confidentiality restrictions 
associated with them. With the data sharing legislation that 
you have been discussing, what would be the effect on the 
States doing the collection and what kinds of confidentiality 
agreements that would be in place might hinder the data sharing 
once it gets to a particular agency?
    Ms. Katzen. Well, I think there are three types of 
possibilities, and that is State information, which is then 
shared with the Federal Government; Federal information, which 
is shared with the States; and sharing among the States. Each 
raises different kinds of issues.
    With respect to State information that is given to the 
Federal Government, I think that with the confidentiality order 
that is in place now and the action that will be taken 
following on the heels of the effective date of that order, I 
expect that information that the States give to the Federal 
Government would be treated with confidentiality and that we 
could so persuade the States of the confidential treatment that 
will be afforded to that information so as to minimize, if not 
eliminate, the concerns that they may have from their 
respondents.
    Mr. Sununu. Would that simply require a rewriting of the 
confidentiality agreements to include all of the agencies 
involved in the data sharing agreement?
    Ms. Katzen. To some extent, many of those are already 
covered by the order that is in place. If there were others 
that would be affected, new agreements probably would have to 
be crafted. Although, as I indicated, the standards set out and 
the processes set forth in our order are ones which we hope 
other statistical agencies will aspire to meet, and that would 
solve that problem.
    With respect to sharing of Federal information with the 
States, that is a bit more problematic in that some of the 
States do not have the experience with what we call the 
functional separation principles that have often been at the 
heart of the discussions of these kinds of issues in which you 
are separating out statistical information for statistical 
purposes from the use of statistical information for 
programmatic, administrative, or enforcement purposes.
    If that functional separation principle is in place and 
followed, then there would be less concern with sharing Federal 
information with the States or among the States. All of this 
would depend in part upon the legislation that we have 
proposed. In fact, it depends in large part on the type of 
legislation we have proposed, because virtually all of the 
agencies and departments have existing provisions that preclude 
such sharing, and we would have to legislate that matter before 
we even begin to think in terms of contractual agreements to 
that end.
    Mr. Sununu. Thank you very much.
    Thank you, Mr. Chairman.
    Mr. Horn. I thank the gentleman.
    Let's now take a look at a few other areas. Let's discuss 
the dissemination of Federal Government statistical products.
    Where do we stand on creating a single point of access to 
the Federal Government information?
    Ms. Katzen. We stand beautifully. Several months ago we 
initiated Fedstats, which is a website that would enable 
someone who does not know ahead of time where information is 
deposited to access information in a variety of different ways. 
You can do it by topic, alphabetically, there is a finder's 
guide that shows different topics. You can do it thematically, 
and there are other types of user-friendly tools.
    Once you accessed the system and we showed how you did not 
need to know in advance whether automobile accidents by persons 
during working hours, was it BLS or Department of 
Transportation, you could just get the information. And once 
you got on to that, it would take you directly to the home page 
of the agency that had the information and cross-linked to a 
number of the other agencies.
    We got extraordinarily positive response, in addition to a 
lot of what they called hits on Fedstats when it first went up, 
and it has continued since then. We were chosen for a 
particular commendation by a number of the trade press, site of 
the week, site of the month, latest, greatest breaking events 
in town type commendations and awards.
    We have continued to watch that and are continuing to build 
that; and our objective is to have all 70 agencies up there 
with their information. This is something which, as you know, 
the Office of Information and Regulatory Affairs takes very 
seriously the responsibilities to disseminate government 
information. It is, in fact, a national resource and asset; and 
the American people should have access to it in the most user-
friendly way possible.
    Mr. Horn. Is it correct that the Bureau of the Census plans 
to put the year 2000 census on the Internet?
    Ms. Wallman. I guess----
    Mr. Horn. As Chief Statistician of the United States----
    Ms. Wallman. Who didn't know the answers this morning. I 
guess I was trying to better understand your question, Mr. 
Horn. If you mean the results of the census, the kinds of 
information that have previously----
    Mr. Horn. I mean the results, once you get them.
    Ms. Wallman. You are speaking of aggregate information 
again? Or information that can be released pending assurance of 
the confidentiality of the information?
    Mr. Horn. I am talking about simply the normal census which 
we now have in libraries.
    Ms. Wallman. Yes, sir. The answer is yes.
    Mr. Horn. Now, if you do that, are you also going to 
publish it in normal book form?
    Ms. Katzen. This has been an ongoing discussion we have 
had. Under Circular A-130, which is the bible for information 
dissemination, our position is we are clearly entering an 
electronic age, and information should be made electronically. 
But that does not mean that everybody has a computer or a 
website or modem, and that paper-based products remain a very 
important part of our law and literature and life in this 
country.
    One of the issues that we are grappling with is ensuring 
that the depository libraries, who receive many of these 
products, continue to have their shelves restocked, both 
electronically and in paper, as we sort through these kinds of 
issues. This has been an ongoing discussion that has had to do 
with the Government Printing Office and the printing policy and 
title 44 of the U.S. Code that we are involved in, even as we 
speak and at present I think are very challenging but great 
opportunities for us all.
    Mr. Horn. Well, I think you are absolutely correct on that, 
and I am not undecided on that question that we do need both. 
And it might change 20, 40 years from now; but, certainly for 
the depositories, I think it is needed to continue those series 
so the people that do not work the Internet, fine, although I 
am sure that some people will say all 5 years old and up over 
the next 50 years will have them appended to their body and 
carry around the key. But we haven't gotten there yet. So I am 
glad to hear rationality still prevails in this area.
    Let's see, do we have anything else we want to talk about? 
I think that is about it. If we find something that we wish we 
had asked and didn't, I know you are good enough to answer the 
question or have 60 people answer it, and you will at least 
clear it. So that is what we are going to work out.
    Sorry to keep you so long on the stand. Sorry I was a 
little late. But I think we are getting there.
    Thank you both for your help, as usual.
    OK, we now have our next panel, panel II: Edward J. Sondik, 
Director, National Center for Health Statistics; Jay Hakes, 
Administrator, Energy Information Administration, Department of 
Energy; Everett Ehrlich, former Under Secretary for Economic 
Affairs, Department of Commerce, Clinton administration, ESC 
Corp.; Mark Wilson, Rebecca Lukens fellow in labor policy, 
Heritage Foundation; Mary Susan Vickers, research director, 
Interstate Conference of Employment Security Agencies, Inc.
    Welcome to all of you.
    As you have gathered, since this is a Government Reform and 
Oversight Committee, we do ask you to take the oath when you 
testify.
    [Witnesses sworn.]
    Mr. Horn. The clerk will note that all five witnesses have 
affirmed, and I think we will just go down the line with Mr. 
Sondik at this side and work our way down. So the Director of 
the National Center for Health Statistics.

 STATEMENTS OF EDWARD J. SONDIK, DIRECTOR, NATIONAL CENTER FOR 
HEALTH STATISTICS; JAY HAKES, ADMINISTRATOR, ENERGY INFORMATION 
 ADMINISTRATION, DEPARTMENT OF ENERGY; EVERETT EHRLICH, FORMER 
 UNDER SECRETARY FOR ECONOMIC AFFAIRS, DEPARTMENT OF COMMERCE, 
CLINTON ADMINISTRATION, ESC CORP.; MARK WILSON, REBECCA LUKENS 
  FELLOW IN LABOR POLICY, HERITAGE FOUNDATION; AND MARY SUSAN 
VICKERS, RESEARCH DIRECTOR, INTERSTATE CONFERENCE OF EMPLOYMENT 
                    SECURITY AGENCIES, INC.

    Dr. Sondik. Thank you very much, Mr. Chairman.
    Protecting confidentiality is a fundamental value at NCHS 
and in all statistical agencies. It is impossible to overstate 
its importance.
    We also recognize that we have a responsibility to respond 
to the taxpayers to maximize the use of their data. We continue 
to develop new approaches to making data available and the 
detail needed without compromising our responsibility to avoid 
disclosures about individuals.
    I am pleased that the subcommittee is considering perhaps 
the most important of these approaches, data sharing. I would 
like, first, to address briefly two related topics: 
coordination within our departments and across the statistical 
system.
    Within our departments, we Federal statistical agencies 
work closely with programs, subject matter specialists, and 
policy-related offices. As NCHS Director and Senior Adviser to 
the Secretary, I work with the HHS Data Council to integrate 
the department's statistical efforts and bring a strategic 
focus to our information needs.
    Each of the statistical agencies also has a distinct role 
in coordinating statistical efforts across departments. I have 
been impressed by the agencies ideas and actions for 
coordinating our efforts, promoting system-wide efficiency and 
minimizing duplication. The strength of this decentralized 
system is that it allows relevance by keeping data close to its 
use. This makes for a solid foundation that allows us to tap 
into each other's unique expertise, resources, and 
technologies. It also enables us to serve multiple purposes--at 
the macro level, with national economic and social indicators, 
and at the micro or program level, where data is critical to 
the operation and accountability of innumerable Federal 
programs.
    Data sharing is important to many of the most important 
interagency initiatives. It can only be accomplished with 
legislative changes proposed by the administration and 
introduced by you and cosponsored by Representative Maloney of 
the 104th Congress.
    Many of the confidentiality statutes currently in place 
were written narrowly to address statistical agencies on a one-
by-one basis rather than apply to the system as a whole. As the 
agencies have evolved, efficiencies and analytic benefits could 
result from greater flexibility.
    Indeed, our efforts are increasingly interrelated with the 
other agencies. For example, with health at 13.6 percent of the 
gross domestic product, there is a considerable confluence of 
interest, you might say, between NCHS and its counterparts in 
economic statistics.
    In my written statement I have included several examples of 
how data sharing authority could be used to improve the design 
and sampling of our population, business and health care 
organization surveys, and to establish joint research data 
centers. Data sharing, for example, could foster the 
development of new longitudinal studies of children that would 
begin at birth and address the interaction of child health, 
education, development and so forth. With data sharing, NCHS 
and the National Center for Education Statistics could more 
readily collaborate on such efficient multipurpose studies.
    Regarding scope, I believe that data sharing should not be 
limited to just three agencies, the three involved with the 
greatest involvement in economics; and I have three reasons. 
First, as I noted, there is a growing confluence of interest in 
the subject matter addressed by all of the agencies. Second, 
the scope of coverage of BLS and Census make the agencies' 
resources--their comprehensive sampling frames, for example--of 
particular use to us at NCHS and other smaller agencies. Third, 
a limitation would simply not support collaboration between the 
smaller agencies such as the child health education example I 
described.
    Before closing, I want to comment briefly on another aspect 
of confidentiality, the privacy of medical records. The lack of 
uniform privacy protections in a new era of electronic medical 
records poses major potential risks to individuals.
    As mandated by the Health Insurance Portability and 
Accountability Act of 1996, HHS is nearing the end of a review 
of privacy protection. The Secretary will make recommendations 
for privacy legislation at the end of August. The implications 
of legislation in this area are important, both in providing 
for the protection of individuals and in providing for 
important statistical research and public health uses of 
medical information. Indeed, advances in biomedical research, 
the detection and control of disease, and in our health care 
system often derive from the aggregation of individual medical 
records.
    In closing, I want to reemphasize the importance of 
achieving dual objectives, assuring privacy protection, and 
assuring we maintain our ability to provide answers to 
important health questions through carefully controlled access 
to medical records.
    I am looking forward to working with the subcommittee as it 
considers the Secretary's recommendations; and, again, I thank 
you for the opportunity to discuss all of these issues; and I 
would be pleased to answer any questions. Thank you.
    Mr. Horn. We are most grateful for your testimony.
    There has been a bill that the subcommittee has worked on 
in both Democratic and Republican Congresses on the 
confidentiality of records. I would like to have the staff make 
sure that you and your staff have looked at it. Some in HHS 
have been advising us. I don't know if it is your direct staff. 
I would like to have your input.
    [The prepared statement of Dr. Sondik follows:]


    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
    
    Mr. Horn. Our next witness is Jay Hakes--am I pronouncing 
it right--Administrator, Energy Information Administration, 
Department of Energy. Welcome.
    Mr. Hakes. Thank you, Mr. Chairman.
    In addition to serving in the Administrator's position at 
the Energy Information Administration, I also serve on the 
Interagency Council on Statistical Policy; and I think during 
the relatively short period it has been a statutorily 
authorized body, it has accomplished quite a bit.
    The website that was mentioned in earlier testimony is a 
very substantial contribution to improving access to 
information. I would say a high school student in California 
today probably has better access to Federal statistics than 
many top analysts here in Washington had 2 or 3 years ago. I 
think that is a substantial achievement.
    Today we are looking at opportunities for data sharing 
which do have potential to reduce costs, improve quality, and 
reduce respondent burden. We feel that part of this is a 
management challenge. All of us have to sort of work on this 
and make it happen. But we feel that it will require removing 
the barrier of different standards of confidentiality.
    Different agencies are authorized to protect data under 
different statutes. The Energy Information Administration is, 
in some ways, kind of the poster child for confidentiality 
problems, because we have very weak statutes governing 
confidentiality; and, therefore, we have even more problems 
than other statistical agencies in getting access to 
information that we need.
    We view this bill's provisions on confidentiality very 
helpful, and we feel that there will be real benefits. In my 
written testimony I just referred to three types of benefits. 
One is, if we can share the frames from which samples are 
drawn, there is a lot of efficiencies to be gained from that. 
Second, when there are discrepancies in data, we can get into 
the individual respondent level and identify the causes of 
those discrepancies. I think that is very important. Third, 
when a new question comes along, we don't have to develop a new 
set of data. We can often use the data we already have.
    I think the Bureau of Economic Analysis makes a number of 
good points about how it can use existing data. We have one 
example at the Energy Information Administration. We actually 
have many, but I know there is an interest in seeing how this 
works in the concrete world. We go out every 4 years and survey 
residential energy customers. We are interested in things like: 
are they using double or triple pane windows, what percentage 
of the square footage is heated, how many showers do they have 
installed. We have a sample of about 6,000 people, which 
doesn't give us State-level data, but gives us regional data at 
a national level.
    It would make a lot of sense for us to use the census 
information as the frame for that study, but we cannot do so 
under the current situation. It forces us to make a lot of 
compromises. It costs us more money, it has an adverse effect 
on quality, and, in this case, there is a little bit more 
respondent burden, although I think in this case that is not as 
big of an issue. There is some real, practical, quality 
enhancements that would come if we could share data with 
census.
    Then just one other example occurs that may stimulate your 
interest. When I was talking with Steve Landefeld last week, 
the Director of the Bureau of Economic Analysis, he mentioned 
that in the 1980's there was a lot of interest in the Congress 
in foreign direct investment, and the Congress couldn't get the 
type of statistics that it wanted. There was one proposal that 
BEA or somebody go out and collect a whole new data series. 
What the Congress did at that time was, in the Foreign Direct 
Investment Act, took care of the confidentiality problem, so 
new data series were created without having to collect new 
data.
    Although that effort has lapsed a little bit in recent 
years as interest in that area has dropped, I think it shows 
that legislation in this area can make a difference. There are 
real savings that can be realized, and we look forward to 
working with you to expand these opportunities.
    Mr. Horn. That is very helpful comment. That might be an 
interesting model to see what really happened there.
    [The prepared statement of Mr. Hakes follows:]

    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
    
    Mr. Horn. Mr. Ehrlich, we are glad to see you again. You 
testified last year when you were still at the Commerce 
Department. It is nice to have you back.
    So, Mr. Ehrlich is former Under Secretary for Economic 
Affairs, and I take it you are now with the ESC Corp.
    Mr. Ehrlich. That is right.
    Mr. Horn. Based in Washington?
    Mr. Ehrlich. Here in Washington.
    Mr. Horn. We are glad to have your views.
    Mr. Ehrlich. I appreciate that, Mr. Chairman. I appreciate 
the opportunity to be here, and it gives me an opportunity to 
thank you for your role and your constructive assistance during 
my tenure in the previous 4 years, which were very much 
appreciated. I learned as Under Secretary to treasure the 
friends you had, and I did.
    Let me respond to, or amplify on one or two other points 
that had been made, particularly those you asked of the 
previous panel.
    In 1993, when I first arrived at Commerce, I asked the 
leading economic survey managers at Census, BEA, and BLS, which 
was then under the direction of Acting Director Bill Barron, to 
put together a list of efforts that they could pursue if they 
weren't obstructed by the obstacles that this bill would 
remove.
    That group was called the Mesenbourg Commission after Tom 
Mesenbourg, who was the Associate Director at Census for 
Economic Programs. It gave estimates in their report. They 
listed 40 initiatives with the burden hours that would be saved 
if those were to be implemented.
    I don't have a copy of it with me today, but it is an 
official Census document, and it would be available to your 
staff. Particularly in the area of the business register, which 
is the largest area of duplication between Census and BLS, the 
savings could be in the tens of millions of dollars in private 
sector burden reduction.
    My colleague, Mr. Hakes, also made the point about BEA and 
BLS sharing data in the link project, which allowed us to look 
at the characteristics of plants that were owned by foreign 
affiliated capital. That was extraordinarily productive and 
another good example of what might be done.
    Data sharing legislation is good and long overdue, and the 
movement to a virtual agency is also long overdue. But, as I am 
fond of saying, virtual is nineties speak for ``not.''
    There are still remaining opportunities to pursue 
aggressive coordination, and I think for that reason the 
Congress should still consider the various dimensions of the 
consolidation issue as it works through this legislative 
vehicle.
    I think it is important to keep in mind as we pursue these 
coordination efforts, including consolidation, that the goal 
has to be a better statistical product. I think you will find 
that the cash savings are extraordinarily low, that the level 
of purely duplicative work is very low, but we do have the 
opportunity through greater coordination--if not outright 
consolidation--to pursue a variety of goals. We have the 
ability to set national statistical priorities.
    Ms. Katzen, in the earlier panel, talked about the problem 
of agencies wanting to have model or sectoral data to answer 
their own policy problems. Well, a consolidated or much more 
highly coordinated system would give us the opportunity to talk 
about whether that data was really a national statistical 
priority or if it should be funded by the agency out of its own 
resources. Consolidation would also allow far greater 
competition among the different modal types of statistical 
gathering because congressional patrons and departmental 
champions and constituents wouldn't be able to segment the 
system in its appropriation. The consolidated agency would also 
allow us to realize new opportunities. Mr. Hakes talked about 
some of those.
    One that we were trying to pursue when I left Commerce was 
getting Census and BLS to get the same data set for retail 
sales and for consumer price information using bar codes. That 
had a variety of difficulties, some of which would be solved by 
statistical data sharing legislation, but some require simply a 
far higher level of management attention that is unclear to me 
that the current system can provide.
    I am concerned that we are moving toward a consolidated 
agency that would be independent in the administrative sense of 
the word. That is, we put economists together and send them out 
to Rockville, Springfield, Morgantown, Suitland, or wherever 
and tell them to come back when they have the problem solved. 
That is not what happens in Canada, although there is confusion 
about that point.
    I am worried that an independent agency in the classic 
sense would be a political orphan when it came time to fight 
for budgetary resources or to keep agencies accountable. 
Accountability is very important in our agencies. It is what 
stops, for example, the career professionals at the Census 
Bureau from writing articles or producing product about why 
there is poverty in the United States as opposed to measuring 
it, or preventing the professionals at BLA from talking about 
what ought to be done in the business cycle at a certain stage 
as opposed to measuring the business cycle.
    For those reasons, I think we need to always think about a 
Cabinet-level steward for any consolidated statistical agency.
    The problem, Mr. Chairman, is that the obvious and ideal 
choice, to my thinking, is the Commerce Department, much as the 
central statistical entity in Canada reports to Industry 
Canada. I think we are engaged still in the remnants of a snipe 
hunt about the Commerce Department here in the Congress, and we 
need to get over that so that we can start to make decisions 
about the statistical system that are governed not by the 
desire to dismantle Commerce, or to adjust the CPI, or to 
perpetuate the undercount in the decennial census, but rather 
decisions that are made with the best interests of the 
statistical users and our Nation in mind.
    I appreciate the opportunity to edit the submitted remarks 
I have given the committee. Now that I am on my own, my only 
opportunity to perfect the product has been reduced.
    Mr. Horn. We thank you for that statement of clarity.
    [The prepared statement of Mr. Ehrlich follows:]



    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
    
    Mr. Horn. Mark Wilson, Rebecca Lukens fellow in labor 
policy, Heritage Foundation. Welcome.
    Mr. Wilson. Mr. Chairman, thank you for inviting me to 
testify.
    My name is Mark Wilson. I am the Rebecca Lukens fellow in 
labor policy at the Heritage Foundation. I would like to lend 
my voice to the discussion today on how consolidating Federal 
statistical agencies would free up additional resources and 
enhance the production of individual privacy and improve data 
quality. Please accept my written testimony and enter it in the 
record.
    I must also emphasize my remarks are my own opinions and 
should not be construed as representing any official position 
of the Heritage Foundation.
    The American statistical system, as everyone realizes, is 
one of the most decentralized data producing systems in the 
world. Although other countries have moved toward centralizing 
their statistical work within a single agency, the United 
States has moved in the opposite direction, creating more and 
more separate statistical agencies throughout government, more 
often than not with separate confidentiality provisions and 
requirements and mandates on each one. The result has been a 
patchwork of statistical agencies and confidentiality 
provisions with little or no data sharing requirements or 
mandates or provisions amongst or between them.
    Despite spending almost $2.7 billion per year, the Federal 
statistical system is in somewhat of a crisis these days. The 
country's decentralized system hinders improvements and 
squanders resources on, at times, duplicitous bureaucratic 
overhead. As a result, the quality of the Nation's economic and 
social statistics has deteriorated over time. Poor data, in 
turn, has a damaging effect on the Federal budget, a 
detrimental effect on the public policy debate, and disastrous 
implications for business decisions, points that the General 
Accounting Office acknowledged in a July 1995 report.
    Over the years, numerous improvements that have been cited 
by the experts as necessary for ensuring the quality of U.S. 
statistics have not been implemented. The decentralized 
fragmented Federal statistical system means no single agency or 
official is really answerable for the modernization and 
improvement projects that cut and sweep across agencies, such 
as improving measures of the service sector of our economy. I 
am not sure whether a virtual agency would have a clearly 
defined, singularly answerable, and identifiable person in this 
fashion that could do this.
    The chief statistician can assert leadership and attempt to 
encourage such action but currently is very limited in terms of 
what ability they have to ensure the accountability that the 
agencies that currently conduct updates to our statistical 
system carry out those improvements.
    The fragmentation and confusion of the current system has 
left many key areas of our society unmeasured, while resources 
are expended on collecting data of what I feel are limited 
public policy interest. For example, we have quarterly data on 
the number of goats that are lost to predators going back a 
number of years, but we have precious little data on the role 
of religion in creating stable and well-adjusted families.
    The topic that has been discussed here at great length has 
been tangentially the decline in the public trust of Federal 
surveys and the Federal statistical system as a whole. 
Protection of the confidentiality of data collected for 
statistical purposes is basic to the development of high-
quality data in any statistical system. Unless respondents can 
be assured that the data that they provide to the government 
for statistical purposes will not be used for regulation or 
enforcement, they will either not respond or report inaccurate 
information.
    The protection of confidentiality, again, as these 
gentleman have pointed out in previous testimony, is not 
uniform in the current Federal statistical system because the 
individual agencies have been created at different times for 
different legislative reasons. As a result, the system operates 
currently under a complex set of regulations, Executive orders, 
and laws that differ in application among the statistical 
agencies.
    Although OMB's new confidentiality order is a step in the 
right direction, I believe that legislation is necessary to 
correct this patchwork of confidentiality requirements that we 
have. Currently, the U.S. system has neither the advantages 
that come from centralization nor the efficiency that comes 
from strong coordination. While centralization alone is not a 
sufficient measure to solve all the problems facing the system, 
significant improvement, I believe, cannot occur without it.
    According to former Commissioner of the Bureau of Labor 
Statistics, Janet Norwood, consolidating the fragmented and 
decentralized Federal statistical system is one of the most 
effective solutions to the problems it is currently under. It 
would provide better data at a lower cost. It would create a 
single statistical agency that would facilitate the creation of 
a coherent national research strategy and development of better 
statistics. It would also have greater independence and improve 
the confidentiality and public trust in our statistics.
    As the 105th Congress begins its debate over the Federal 
statistical system, it should bear in mind four important 
principles to ensure the taxpayers and data users receive the 
greatest benefit from any reform.
    Combine as many agencies as possible. Although 
consolidating the four largest statistical agencies would 
eliminate some duplication, the largest budget savings and 
benefits from economies of scale will occur and come from 
integrating as many of the smaller agencies as possible.
    Improve privacy and confidentiality. The confidentiality 
protection laws established piecemeal among the different 
statistical agencies should be replaced with uniform privacy 
provisions that would permit the exchange of confidential 
information for statistical purposes only and ensure the 
independence and objectivity. Two of the most important 
attributes of an objective statistical agency are the longevity 
of leadership and the independence from political pressure.
    You should also strengthen coordination by giving a 
consolidated agency the authority and the management structure 
to enable it to develop an overall statistical research and 
development agenda and to implement modernization and 
improvement projects across agencies.
    Thank you for your time. I would be happy to answer any 
questions you might have.
    Mr. Horn. We thank you for your time.
    [The prepared statement of Mr. Wilson follows:]


    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
    
    Mr. Horn. Let me go to our last witness before we ask 
questions. Mary Susan Vickers, research director, Interstate 
Conference of Employment Security Agencies, Inc.
    Ms. Vickers. Thank you, Mr. Chairman, for this opportunity 
to testify on behalf of the Interstate Conference of Employment 
Security Agencies.
    My name is Mary Susan Vickers, and I am director of labor 
market information for the Interstate Conference, or ICESA. 
ICESA is the national organization of State officials who 
administer the Nation's employment and training services, 
unemployment insurance laws, and labor market information 
programs. The State Labor Market Information Divisions, or LMI 
units, within employment security agencies produce, analyze, 
and distribute labor statistics to improve economic 
decisionmaking. These statistics include employment, 
unemployment, and wage information produced primarily through 
cooperative Federal-State statistical programs with the Bureau 
of Labor Statistics.
    The BLS programs are housed in State employment security 
agencies because their existence depends directly on their 
connection to unemployment insurance administrative data. BLS's 
contracts in each State rely on access to and use of 
confidential administrative records collected by the States for 
the administration of unemployment insurance programs. The 
collection of unemployment insurance data is authorized by 
State law and conducted according to State policies and 
regulations. The disclosure of unemployment information is also 
governed by State statute and policy.
    These Federal-State statistical programs are a fully 
integrated component within State employment security agency 
functions. Within the unemployment insurance programs, for 
example, they are used to set unemployment compensation 
benefits and to determine tax rates for employers. For job 
training and employment programs, they are used to allocate 
resources to sub-state areas. These statistical programs, using 
supplemental resources from the U.S. Department of Labor's 
Employment and Training Administration, are key to developing 
knowledge about where there are current jobs, what they pay, 
and the background and education job seekers need to obtain 
them.
    Within the context of the Federal-State cooperative 
programs with the Bureau of Labor Statistics, the States 
perform two critical functions: First, they are the producers 
of the statistical data. Second, they provide the analysis 
requested by State and local users to create that data into 
information, labor market information.
    Labor market information is a key driving factor in the 
planning and delivery of State work force development systems 
and major State welfare reform initiatives. State staff are 
responsible for keeping employers informed of the confidential 
statistical use of the data and have a vested interest in data 
quality and timeliness. States are also strong advocates for 
reducing employer reporting burdens; for addressing State 
confidentiality concerns in State data sharing proposals; and 
to ensure that the States, as producers of labor market 
information, achieve equal status within the Federal-State 
statistical system.
    Federal sharing of State data represents a transfer of 
authority from the States to the Federal Government. This 
transfer of authority means that the States require assurance 
that Federal practice does not violate State statute.
    If, as a result of consolidation, for example, confidential 
data from State unemployment insurance records provided to the 
Bureau of Labor Statistics were to be shared with another 
Federal agency, statutes in several States would have to be 
changed. State legislatures might agree to some of these 
changes only if qualifiers were stipulated. These qualifiers 
might include that States be advised of the nature of the use 
of the data and the State would be reimbursed for the cost of 
providing the data to the additional Federal agency or the 
Federal reciprocal agency would abide by State disclosure 
rules.
    It is also our position that the States should have access 
to data held by Federal agencies for statistical purposes. Data 
sharing agreements should be reciprocal. An agency receiving 
data and then sharing that data should also provide access to 
the original collecting agency, which may be a State.
    Finally, we believe that a reformed system should ensure a 
State rule that encourages the Federal statistical system to 
evolve as the needs of our mutual customers evolve. In other 
words, consolidation for greater efficiency is important, but 
consolidation will not be effective if it is not responsive to 
our customers' needs or if it does not recognize the 
interdependence between Federal and State needs.
    The Federal statistical agencies have direct customers, 
such as the Federal Reserve Bank, while States' customers are 
employers planning business expansions, job seekers, and 
economic developers. The needs of all of these customers are 
important and must be met.
    As you develop your reform proposal, ICESA is prepared to 
assist in additional study and planning for change. Thank you 
for the opportunity to share our comments.
    Mr. Horn. Well, thank you. I appreciate your testimony.
    [The prepared statement of Ms. Vickers follows:]


    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
    
    Mr. Horn. Let's ask a few questions, and then I think we 
will wrap it up.
    Just in general are the Federal Government statistical 
agencies focusing too much of our statistical resources on the 
more traditional parts of the economy and not enough on the 
rapidly growing information, technology, and service sector?
    Just to pick one, pick biotechnology, which is the most 
rapidly growing industry in California. Any feelings on that? I 
realize some of you are specialists in a particular area.
    Mr. Ehrlich. Let me speak to that.
    Mr. Horn. What do you hear?
    Mr. Ehrlich. The short answer is yes. The problem is not as 
profound as it was several years ago.
    As you understand, Mr. Chairman, 2\1/2\ years ago we had 
the first comprehensive strategic review of the Nation's 
economic statistics in 40 years, held at the Chamber of 
Commerce downtown; and started to get rid of old programs. For 
example, regional economic projections, sub-national retail 
sales: the leading economic indicators were farmed out to the 
private sector so that we could free up resources for exactly 
those kinds of measurements. In the 1997 economic quinquennial, 
I think we will have real balance between the service and non-
service sectors.
    On the other hand, the service sectors pose special 
problems about their conceptual being. How do you measure a 
unit of software? How do you measure a unit of insurance being 
provided or a financial derivative? Who sold what to whom? That 
requires more resources so we understand their quality and 
understand their contributions to economic growth.
    With that said, if I may move to my third hand. I believe 
it is not the structure of the system that obstructs our 
ability to answer those questions right now. It is the level of 
resources provided to it, and, to some extent, there are 
questions of management.
    Mr. Horn. What sort of questions of management come to your 
mind?
    Mr. Ehrlich. Are you willing to establish priorities and 
enforce them within the individual agencies as to what is more 
important than the next thing? What we did at BEA and Census 
was establish that measuring the quality and concept of output 
and therefore prices was our most important priority. And while 
we regretted very deeply cutting such programs as regional 
economic projections or nonresidential building permits that 
had real value to real users, they weren't as important as the 
central question--are we getting inflation and growth right? 
Therefore, they had to go.
    Mr. Horn. On that very point, the regional Federal Reserve 
officers usually have a chief economist and a number of 
economists on the staff and generally try to watch some of that 
local economic data. Have you found them wanting? Does that 
make sense that they have that role? I realize the two are good 
for each other, competition and all that.
    Mr. Ehrlich. They have that role, but what they provide is 
not a substitute. The beige book, which is the summary of their 
findings, is really qualitative in nature. It is an essay as 
opposed to a data series that can be used for business planning 
and the like.
    I think they use their anecdotal reservoir as a substitute, 
to some extent, for data that they used to provide, and when 
they provided it, we used it as a complement.
    Mr. Horn. OK. Are the Federal Government agencies taking 
advantage of the improved methods that businesses use to track 
information in order to improve the Federal data collection 
methodologies? For example--and this is a rather interesting 
one--do Federal statistical agencies incorporate the universe 
America product code information tracked by scanners into 
surveys of economic data? What is the answer on that one?
    Mr. Ehrlich. Not yet. We are designing programs to do so. 
The product code isn't a full universe of products.
    The obstacle to using bar code data in a statistical system 
is to find a way to bridge between those products that are bar 
coded and those establishments that are bar coded, even though 
they are the great preponderance of products and 
establishments, and those that don't, so you avoid double 
counting and the like, and you can seamlessly capture the 
entire household goods sector.
    I referred in my testimony, in the statement I gave to the 
subcommittee, to our ability, once we have that, to get both 
price and quantity data from it. I think that the price and 
quantity data found in bar codes could be very useful for 
experiments in measuring the changing quality of goods and, 
therefore, what the real impact of inflation has been. But we 
are not yet at a point where we can use them as the basis for 
all of the GDP calculations.
    Census is also working on software that a respondent firm 
could load into their own accounting systems and that would 
automatically report to us. At first, it sounds almost 
nightmarish that you would allow that, but, in fact, it is the 
same level of confidentiality and the same level of assurance 
that over-the-phone or pencil and paper reporting provides. It 
is simply allowing it to happen automatically at lower cost. 
That is another example of the kinds of technological 
opportunities they are pursuing.
    Mr. Horn. As you talk, I have been thinking of interesting 
studies; and, of course, they probably all would run into the 
Hawthorne effect.
    For example, if you had a card that you are picked in a 
random sample nationally by BLS, BEA, Census, or one of the 
statistical agencies, you use that card whenever you make a 
purchase. That is registered, obviously. You can pull all of 
that together.
    Now, the mere fact you are designated to be on such a 
select panel, Hawthorne told us it didn't matter what you do. 
The more somebody cared about people, they would increase 
productivity. Would they go out eating hamburgers and grease 
when they think that is the normal thing to do, when the rest 
of us are out eating vegetables to avoid hamburgers and grease?
    Mr. Horn. Dr. Sondik.
    Dr. Sondik. Actually, we are using that technology in the 
National Health and Nutrition Examination Survey which we are 
in the process of implementing. We use it not so much in 
identifying products that someone is using but use the 
technology within the centers that we have to track the samples 
that are derived from the individual and track individuals as 
they move within our trailer examination centers.
    We use it in terms of the home interviews that are done; 
and we use it, in fact, in nutrition monitoring, in which we 
found that the Hawthorne effect is perhaps not as strong as we 
might hope that it would be in terms of what this country is 
eating.
    So I think in that survey, and I would like to think in 
some of the other things we are doing, particularly in 
information dissemination, we are trying to make as much use as 
we can of technology that is really not behind the wave but 
really is ahead of the wave, and we are certainly doing it with 
this survey.
    Mr. Horn. Anybody else?
    Mr. Hakes. I think all of us favor electronic reporting. We 
have used some version of it now for 3 and 4 years, and it 
keeps getting better. I think we have to be sensitive to our 
respondent base. If even as many as 10 percent of our 
respondents have difficulty reporting electronically, that may 
be a barrier to us using it for the whole universe.
    Another thing we noticed is once you have a system in 
place, the reporting entity is sometimes reluctant to change 
it. We have suggested removing certain questions to reduce 
respondent burden; and the respondent says, don't take out the 
question; we don't want to change the system.
    It gets complicated, but I think there are tremendous 
potentials in electronic reporting. A person can get a screen 
of what they reported last time, simply change the data, or 
maybe have the data go in in an automated fashion from their 
accounting system. That improves the quality at both ends, and 
I think that there will be a lot of movement forward in this 
area.
    The difficulty will come mostly in the small business area, 
where that will be a more difficult transition. But that will 
come, also.
    Mr. Horn. Would those of you that operate energy, health, 
any of the data banks before us, do you draw on BLS and Census 
data in any of your statistical operations? Do you need to use 
their duplications, their series, whatever you want to describe 
it as?
    Dr. Sondik. Yes. They are actually vital to what we do. But 
we have limitations in what we can do now, because we can't 
share data.
    For example, the Census Bureau actually implements the 
health interview study, which is the core of the Health and 
Human Services survey integration activities. In fact, we are 
trying to build surveys around that in a process we call survey 
integration. But when we do that, we actually have to derive 
the sample from that, and they implement it for us. We could 
save considerably if we were actually able to use their 
sampling frame.
    The same goes for BLS, in which we have ongoing 
discussions, but, for example, provider surveys--the Secretary 
tells me she hates that word provider--but for those who 
deliver or are involved in health care, we need to be involved 
more with them in understanding exactly what they are doing. 
The frames that we could derive from BLS and from Census would 
be enormously useful in that.
    There are frames that I am not sure we even know about that 
could be very useful in health at this point. For example, in 
energy, in helping us understand the implications of various 
forces.
    When we study health, the issue is really the interaction 
of different forces. Studying health is not only looking at 
particular genes, because, in fact, what happens to a person is 
not just a function of your gene structure. It is a function of 
all of the other interplays--the social ones, the ones that 
have to do with food intake, for example, and actually the care 
that is delivered. All of those are derived from information 
from different areas and this could be handled much more 
efficiently if, in fact, we were able to share information with 
confidentiality, which now, with the statistical 
confidentiality act, we would be able to do.
    So it would be of enormous benefit to us, not only in 
saving money but from the standpoint of how much it would open 
the kinds of studies that we are able to do.
    Mr. Horn. I would think your colleague next to you, you 
have got energy surveys, probably home heating oil, New 
England, the health relationships there, when the price goes up 
or the energy deliveries go down in either case.
    Dr. Sondik. As he was speaking, it certainly occurred to me 
that was an area that would be important to us, particularly if 
we could segment and look at the impact on us and do it on the 
basis of income. It would be very important.
    Mr. Horn. Things generally happen incrementally in Congress 
and in the executive branch, too, for that matter. What I am 
interested in, since we did have a bill in the last Congress 
and will probably have it in again, in terms of Bureau of 
Economic Analysis, Census, Bureau of Labor Statistics, are 
those agencies, the ones we could start with first, that 
consolidation might help you more than it does now?
    I would just be curious if there is any reaction on this. I 
want to go down the line and see what the views are.
    Dr. Sondik. It would certainly help us. But I would much 
rather see data sharing expanded so that the other agencies 
could also be involved.
    Agriculture is another area in which the study, for 
example, of the environment that the farmer is in, pesticides 
and so forth, would be much easier, and it is a very important 
area. These studies would certainly be enhanced with that 
freedom.
    Mr. Hakes. From our perspective, I don't think the issue is 
consolidation. It is confidentiality.
    On the residential survey, for instance, we use the Census 
data down to about the level of 50 households, and we can go 
that far and not have the confidentiality problem.
    Then we get the logical thing at that point would be just 
to continue right on with the Census data down to the 
individual household, but we cannot do that. From that point 
on, we are on our own.
    Another point of interaction with Census is we actually 
contract with them to do the Manufacturing Energy Consumption 
Survey. We don't really deal with the individual respondent 
data. It does create somewhat of an awkward situation because I 
think analysts having access to that data is important from a 
quality standpoint. But we sort of cobble together things now, 
using the Census as much as we can, but we sort of reach a 
point where we have to stop.
    Mr. Horn. We could solve the confidential situation and 
then have some things centralized, others decentralized with 
access because of changing the confidentiality access situation 
that exists now. So that is another model we might think about.
    Canada has presumably--what--a centralized statistical 
agency? They include all departmental statistics in their 
particular operation?
    Dr. Sondik. I believe they do.
    Mr. Hakes. The answer is yes, but I would caveat that in 
several ways. For instance, most of what is called--of the 
energy mining activities that occur in Canada occur in the 
Province of Alberta, and actually the statistics there are 
collected by the provisional government, not by the Federal 
Government, so it is certainly more centralized than the United 
States system, but it is not a totally centralized system.
    Mr. Horn. Mr. Ehrlich, how about it?
    Mr. Ehrlich. Mr. Hakes makes a good point. The provinces 
are more important in Canada. They do bear a much greater 
portion of the burden, and they also create a burden of their 
own, insofar as the Canadian system is asked to calculate GDP 
by province on a quarterly, if not monthly basis. It is an 
ungodly task.
    Were you to consolidate, I would advise you out of my own 
experience to start with the five agencies, not the three you 
mentioned, and Energy and Agriculture.
    You say first why in terms of what I've left out. You have 
four agencies in NCES, NCHS, the Bureau of Transportation 
Statistics, and Justice Statistics. They have very important 
functions, and it provides important analytic and policy 
direction to the data-gathering processes over which they 
preside.
    The operations of those agencies share two 
characterizations. One is that the Census is used, as you've 
heard, to be the wedge to the system. Census has a world class 
field force. It's well organized. They know how to go out on a 
low-cost basis and get work done. In that sense, the system is 
already centralized at the point of production. The other thing 
they share is that they are a conduit for money to States. And 
the States then go out and report back to those four agencies. 
So there are problems. Their issues are different regarding 
consolidation or higher levels of coordination. The five 
agencies, I think, would allow you to sit down with the 
managers' most formidable weapon, a clean piece of paper, and 
start reorganizing in a way that some past efforts--Dr. 
Norwood's, for example--have not.
    I think simply taking BLS, BEA, and Census, and creating an 
organization chart that has them as three operating divisions 
misses all of the nuances that are possible in centralizing 
functions related to national income, to personal and household 
incomes, to industry studies, to labor market studies, to 
demographic-based surveys, and the like. But we do have the 
opportunity to put common things together and arrive at common 
methodological approaches and to relieve ourselves of duplicate 
work where it exists. I think we start at those levels.
    Mr. Horn. Interesting.
    Mr. Wilson.
    Mr. Wilson. I agree with Mr. Ehrlich. I think it's 
important to--again, to be considering consolidation, which I 
hope that you do, that you consolidate along the lines, I 
think, functionally rather than just organizationally as it is 
now with just being separate divisions, of BLS being one 
division, and Census the other, and BEA being another.
    I think it's also important to consider that the 
statistical system of a Nation generally provides two forms of 
statistics, public statistics--which are of wide interest and 
are used by the general public and data users--and the 
administrative statistics--which are of narrow interest and are 
used primarily by the Federal or State Governments for 
policymaking purposes, regulatory purposes, and others.
    In that regard, it may be useful to take a look at how the 
United Kingdom, Great Britain, has consolidated their 
statistical system and how they've set up a public statistics 
service as well as the government statistics service and see 
what we can learn from that and perhaps integrate in the United 
States.
    Mr. Horn. Ms. Vickers.
    Ms. Vickers. Yes, in terms of a consolidation of Federal 
statistical programs, I think the point we would like to make 
is just that the States are definitely involved in the system 
and interested in what will happen about it. The national 
system is a little bit different, in our mind, than a Federal 
system.
    Mr. Horn. Sure.
    Ms. Vickers. Building a system from a local level up would 
be a system that would be helpful to our customers.
    Mr. Horn. Just for the record, the agencies you represent, 
the employment agencies.
    Ms. Vickers. Yes.
    Mr. Horn. The Federal-State partnerships since, what, 1934 
or so, somewhere around there.
    Ms. Vickers. Our organization has existed since 1937.
    Mr. Horn. 1937. They are providing unemployment data in 
particular, aren't they----
    Ms. Vickers. Yes.
    Mr. Horn [continuing]. Through those offices?
    What else are they providing?
    Ms. Vickers. They are collectors for the Bureau of Labor 
Statistics, statistical programs on wages, occupations, current 
employment on a monthly basis. The CES programs have a 90, as 
well as the unemployment, and employment statistics.
    Mr. Horn. Interesting.
    Ms. Vickers. Mass layoffs is another area that the States 
collect for the Bureau.
    Mr. Horn. While we're on some of this question, Mr. 
Ehrlich, you're a very eloquent person. I knew that from the 
first hearing with you. Let me read one of your eloquent 
statements you once made in the Chicago Tribune.
    Mr. Ehrlich. Oh.
    Mr. Horn. No, it's OK. Don't worry.
    Mr. Ehrlich. OK.
    Mr. Horn. Quote: If we can't maintain a pace of improvement 
in collecting economic data as rapid as change in the economy, 
we're going to have the world's most advanced economy with a 
statistical system worthy of a middle tier nation. We need to 
start looking at economic data as being as much a part of our 
infrastructure as our roads, our ports, and our bridges. 
They're part of what makes the economy go.
    Is that a correct quote?
    Mr. Ehrlich. Yes, it is.
    Mr. Horn. Could you elaborate on your comment about having 
a statistical system worthy of a middle tier nation.
    Mr. Ehrlich. This goes to a remark that has been made in 
various places in the panel. The economy is changing very 
rapidly. When we think about, for example, the problems that we 
associate with the statistical system, the measurement of 
services, our questions about whether or not we're overstating 
inflation and understating growth, they're all various 
dimensions of the problem of economic change.
    We have the most innovative, advanced, and technologically 
progressive economy in the world, probably the most that the 
world can imagine. And that means that we're continually 
creating new kinds of goods and services and that old ones are 
changing in nature and scope. And the statistical system has to 
accommodate those as they occur.
    I've often, Mr. Chairman, used the metaphor of the tailor, 
that we think of the statistical system being like a tailor 
measuring someone for a suit of clothes. And you hold the ruler 
or the tape measure up to them, and you read off the numbers, 
and you write them down in books. That makes sense only if we 
accept the fact that the person that we're measuring is 
sprinting as fast as they can, which means the only way we can 
do our jobs--and I'll allow myself to dignify myself by still 
saying ``we''--is by running as fast as the person we measure 
and being twice as agile. It's a formidable challenge.
    To some extent, moving toward consolidation would help us 
meet that challenge in the various ways that I mentioned. But 
there are other dimensions of the problem, and they have to do, 
as I mentioned earlier, with the level of resources we're 
provided and with management will.
    The economy has grown by around 40 percent in the last 15 
to 20 years, and the number of establishments--number of places 
where businesses conducted per unit of GDP--has grown by around 
30 percent. And yet in real terms, the resources we're 
providing are still what they were 15, 20 years ago.
    Mr. Horn. Yes, Mr. Hakes.
    Mr. Hakes. I of course can speak most authoritatively about 
energy statistics. I believe that the United States has by far 
the highest quality energy statistics of any in the world, and 
I think most countries in the world would agree to that. When 
the trade press, which is, I think, the most frequent and 
eligible user of our data in many instances or refers to our 
data, the word ``authoritative'' is used as an adjective on 
many, many occasions.
    Last week, one of the trade press referred to our 
statistics as the ``Bible of Energy Statistics.'' Although I 
think we need to be aware of opportunities to improve, I think 
we certainly have within our system some statistics that are 
considered the finest in the world. And we advise many 
governments around the world on how to upgrade their 
statistics.
    I think as we look for opportunities to improve, which I 
think all of us are very committed to, I think we have to look 
at the centralization versus decentralization question.
    I, like others in the room, used to teach political 
science. But I started in government 20 years ago, and one 
thing that struck me over and over again is--after I worked in 
several different agencies--there is a tendency if an agency is 
highly centralized, to believe that decentralization will 
substantially improve that agency's performance, and if the 
agency is decentralized, it can be substantially improved by 
centralization.
    Now, any change may cause progress, but any change like 
that also has very high transaction costs. And let me give you 
an example. Right now, in energy, there's a lot of changes 
going on in the energy industries. The deregulation of the 
electric industry that is going on at the State level may get 
some boost at the Federal level. Those industries will look 
very different a few years from now than they look now.
    Now, if energy has to go through four more layers to get 
approval to fix the system, that may cause problems. So that 
there are some potentials of integration. There are also some 
dangers of integration in not being able to turn the battleship 
always quickly when things need to be changed.
    I would just mention one more thing from our experience. 
EIA itself is a unified agency combined out of several previous 
agencies. It was formed in 1977 out of the Bureau of the Mines, 
the Federal Energy Administration, and other agencies that were 
brought in. When I arrived in 1993, and even to some extent 
today, you can see those operating independently within our 
operation. Whether you can tell whether they came from the 
Bureau of Mines, this is essentially a management project. It 
is difficult even within our statistical agencies to do the 
amount of integration that needs to be done. And so I think we 
have to see that there are some potentials in a decentralized 
system, some in a centralized, and hopefully we can find the 
advantages of both.
    Mr. Horn. That's a very helpful comment.
    We have a vote on the floor, and I don't want to have to 
recess and hold you here. But is there any other comment 
members of the panel would like to make that we haven't asked 
you the right question where you should make it?
    Dr. Sondik. Just in terms of consolidation, and perhaps 
this is from the health point of view, but I would think it 
applies across the board--the statistical agencies are not only 
archivists, if you will, but they're involved in providing the 
information that we need for making decisions. And I think in 
order to do that, you have to be close to the communities that 
you're serving or at least have very strong links to those 
communities.
    And I see in the health-related statistical agencies, the 
involvement of each of those with their communities is 
absolutely invaluable to what it is they bring to the table in 
knowledge concerning where the country is going. In our case, I 
think it's even worse than the tailor running after the person 
who is sprinting. I'm not sure we have an idea of any direction 
that the person is going in at this point and really need to be 
as close to that as we can be, at the same time that we don't 
lose the past thread, so to speak.
    Mr. Horn. Any other comments to be made? Ms. Vickers. Mr. 
Wilson.
    Mr. Wilson. No.
    Mr. Horn. Mr. Ehrlich. Mr. Hakes.
    Dr. Sondik. Thank you.
    Mr. Horn. Well, let me just thank each of you. I think the 
country can take great pleasure in the fact that we've got such 
able people running some of our key statistical agencies. I've 
learned a lot from this, and we're going to be consulting you, 
because I think that the point that you've made in our 
legislation, we're just trying to keep it very simple, but 
there ought to be some goals and aims in there.
    And when you mention the national economic data and the 
different series that one might think about, not as definitive 
but simply as illustrative, we would simply welcome your 
comments in the months ahead, and both my colleagues and the 
staff, I think, would want to be interacting with you and some 
of your other counterparts.
    I'm familiar with the Bureau of Justice Statistics, because 
I spent, I think--well, maybe 15 years on the National 
Institute of Corrections after helping found it. And you are 
right about the linkages with the community you serve, because 
I remember when we wanted to make suggestions to the FBI 
uniform crime survey to get women and some of their problems 
into it, it took quite a battle almost. And I think some of 
those days are in our past, but there are questions that ought 
to be asked, that if you aren't close to the people that are 
affected, you are not going to really think about them very 
much.
    So I thank you all for coming. It has been an immensely 
interesting hearing for me. And with that, we are adjourned 
exactly at 4 p.m.
    [Whereupon, at 4 p.m., the subcommittee was adjourned.]

                                   - 
