[Senate Hearing 105-170]
[From the U.S. Government Publishing Office]


                                                        S. Hrg. 105-170


 
 ROLE OF THE DEPARTMENT OF COMMERCE IN FEDERAL STATISTICAL GATHERING, 
     ANALYSIS AND DISSEMINATION, AND OPPORTUNITIES FOR REFORM AND 
                             CONSOLIDATION

=======================================================================

                                HEARING

                               before the

                      SUBCOMMITTEE ON OVERSIGHT OF
                 GOVERNMENT MANAGEMENT, RESTRUCTURING,
                      AND THE DISTRICT OF COLUMBIA

                                 of the

                              COMMITTEE ON
                          GOVERNMENTAL AFFAIRS
                          UNITED STATES SENATE

                       ONE HUNDRED FIFTH CONGRESS

                             FIRST SESSION

                               __________

                             APRIL 9, 1997

                               __________

      Printed for the use of the Committee on Governmental Affairs


                               


                        U.S. GOVERNMENT PRINTING OFFICE
 40-030 cc                     WASHINGTON : 1997
_______________________________________________________________________
             For sale by the U.S. Government Printing Office, 
 Superintendent of Documents, Congressional Sales Office, Washington, DC 20402


                   COMMITTEE ON GOVERNMENTAL AFFAIRS

                   FRED THOMPSON, Tennessee, Chairman
WILLIAM V. ROTH, Jr., Delaware       JOHN GLENN, Ohio
TED STEVENS, Alaska                  CARL LEVIN, Michigan
SUSAN M. COLLINS, Maine              JOSEPH I. LIEBERMAN, Connecticut
SAM BROWNBACK, Kansas                DANIEL K. AKAKA, Hawaii
PETE V. DOMENICI, New Mexico         RICHARD J. DURBIN, Illinois
THAD COCHRAN, Mississippi            ROBERT G. TORRICELLI, New Jersey
DON NICKLES, Oklahoma                MAX CLELAND, Georgia
ARLEN SPECTER, Pennsylvania
             Hannah S. Sistare, Staff Director and Counsel
                 Leonard Weiss, Minority Staff Director
                    Michal Sue Prosser, Chief Clerk

                                 ------                                

SUBCOMMITTEE ON OVERSIGHT OF GOVERNMENT MANAGEMENT, RESTRUCTURING, AND 
                        THE DISTRICT OF COLUMBIA

                    SAM BROWNBACK, Kansas, Chairman
WILLIAM V. ROTH, Jr., Delaware       JOSEPH I. LIEBERMAN, Connecticut
ARLEN SPECTER, Pennsylvania          MAX CLELAND, Georgia
                        Ron Utt, Staff Director
      Laurie Rubenstein, Minority Staff Director and Chief Counsel
                      Esmeralda Amos, Chief Clerk



                            C O N T E N T S

                                 ------                                
Opening statement:
                                                                   Page
    Senator Brownback............................................     1

                               WITNESSES
                        Wednesday, April 9, 1997

Hon. Daniel Patrick Moynihan, a U.S. Senator from the State of 
  New York.......................................................     2
Hon. Stephen Horn, a Representative in Congress from the State of 
  California.....................................................     7
Vincent P. Barabba, General Motors Corporate Strategy and 
  Knowledge Department...........................................    12
Janet Norwood, Senior Fellow, The Urban Institute................    21
Maurine A. Haver, Chair, Statistics Committee, National 
  Association of Business Economists.............................    29
Leonard Nakamura, Economic Adviser, Federal Reserve Bank of 
  Philadelphia...................................................    35
L. Nye Stevens, Director, Federal Management and Workforce 
  Issues, General Government Division, U.S. General Accounting 
  Office.........................................................    78

                     Alphabetical List of Witnesses

Barabba, Vincent P.:
    Testimony....................................................    12
    Prepared statement...........................................    13
Haver, Maurine A.:
    Testimony....................................................    29
    Prepared statement...........................................    31
Horn, Hon. Stephen:
    Testimony....................................................     7
    Prepared statement...........................................     8
Moynihan, Hon. Daniel Patrick:
    Testimony....................................................     2
    Prepared statement...........................................     5
Nakamura, Leonard:
    Testimony....................................................    35
    Prepared statement with attachments..........................    38
Norwood, Janet:
    Testimony....................................................    21
    Prepared statement...........................................    24
Stevens, L. Nye:
    Testimony....................................................    78
    Prepared statement...........................................    80

                   ADDITIONAL INSERTS FOR THE RECORD

Letter to Senators Moynihan and Kerrey from former Chairmen of 
  the Council of Economic Advisers, dated September 23, 1996.....     4
Articles by Leonard Nakamura:
    ``Measuring Inflation In a High-Tech Age,'' Business Review, 
      November/December 1995.....................................    53
    ``Is the U.S. Economy Really Growing too Slowly? Maybe We're 
      Measuring Growth Wrong,'' Business Review, March/April 1997    66



 ROLE OF THE DEPARTMENT OF COMMERCE IN FEDERAL STATISTICAL GATHERING, 
     ANALYSIS AND DISSEMINATION, AND OPPORTUNITIES FOR REFORM AND 
                             CONSOLIDATION

                              ----------                              


                        WEDNESDAY, APRIL 9, 1997

                                       U.S. Senate,
         Oversight of Government Management, Restructuring,
                      and the District of Columbia Subcommittee,   
                        of the Committee on Governmental Affairs,  
                                                    Washington, DC.
    The Subcommittee met, pursuant to notice, at 1:45 p.m., in 
room SD-342, Dirksen Senate Office Building, Hon. Sam 
Brownback, Chairman of the Subcommittee, presiding.
    Present: Senator Brownback.

             OPENING STATEMENT OF SENATOR BROWNBACK

    Senator Brownback. This Subcommittee hearing of the 
Governmental Affairs Committee will be called to order.
    We are holding our third in a series of hearings on the 
Department of Commerce, and this one in particular will look at 
the statistical gathering interest of the United States and our 
abilities in these area, or lack thereof, our lack of 
coordination. It has been interesting to me, in looking at this 
issue, that we have some 89 different organizations involved in 
statistics in the Federal Government, and yet with that and the 
amount of budget we spend, which I believe is about $3 billion 
annually, we are ranked seventh amongst the leading 
industrialized countries for the quality of our statistics.
    Well, that seems to me to be a bit odd, and we have in 
front of us in the first panel a couple of gentlemen who have 
been around this issue and who have either produced statistics 
or used statistics, either of which they are very knowledgeable 
about it. Let's just get started on this with the two of you 
gentlemen, looking at this area of the statistical functions of 
the Federal Government, what needs to be done differently. And 
we are trying to build a set amount of information as to what 
should be done differently in these areas. We have several 
panels today.
    Senator Moynihan, thank you very much for joining us. I 
apologize for being late. We were in a major discussion on the 
chemical weapons convention, which is an issue of some 
important interest to a few, so I was following on that.
    Thank you for joining us. Please feel free to illuminate us 
on what we should know about the statistical gathering from 
your background, either in the agency, in the entity, or in the 
U.S. Senate. The floor is yours.

TESTIMONY OF HON. DANIEL PATRICK MOYNIHAN, A U.S. SENATOR FROM 
                     THE STATE OF NEW YORK

    Senator Moynihan. Thank you, Mr. Chairman. It is a special 
honor to appear here with my old White House colleague, Steve 
Horn. I can tell you it was a Republican White House, if that 
is of any reassurance.
    Senator Brownback. Those were the days.
    Senator Moynihan. And you will soon be hearing from Janet 
Norwood, who is a former Commissioner of Labor Statistics, and 
Vince Barabba, a very distinguished head of the Bureau of the 
Census.
    I have a statement which I would like to place in the 
record.
    Senator Brownback. Without objection.
    Senator Moynihan. And to make a very brief statement, which 
I would hope you might Representative Horn just join in. I 
think this is--it would appear this is an idea whose time has 
come. In the last Congress, Senator Kerrey and I introduced--
and we have done it once again--a proposal to set up a Federal 
commission to think about consolidating and rationalizing these 
agencies. Representative Horn put in a bill that would actually 
do so. I think you had the Census, the BLS----
    Mr. Horn. And the Bureau of Economic Analysis.
    Senator Moynihan. And the Bureau of Economic Analysis, the 
three most important ones, into one institution. I gather 
that--and he can speak for himself--that Mr. Horn thinks that 
maybe we could get a commission to look at the whole subject 
first before deciding. The Heritage Foundation has come up with 
a proposal for a national statistical office.
    There is a certain simple point here. We built statistics 
into our Constitution when we required a decennial census to 
apportion the House of Representatives. And so we have always 
had a powerful statistical basis in the Bureau of the Census--
which I guess was started formally about 1860?
    Mr. Horn. A little later.
    Senator Moynihan. And the Bureau of Labor Statistics about 
1880?
    Ms. Norwood. In 1884, and the Census was slightly after 
that.
    Senator Moynihan. The Census was slightly after. In 1884, 
the Bureau of Labor Statistics was established, and then as 
different departments of government are established, Mr. 
Chairman, almost invariably, if not from the outset, sooner or 
later they get their own statistical agency. And I have had my 
share in that wrongdoing. In the Intermodal Surface 
Transportation Efficiency Act of 1991, we created a Bureau of 
Statistics in the Department of Transportation, which needs 
data, but it does not need necessarily its own. And that is 
what other countries are finding.
    In Canada, almost by a historical accident, they have had 
one agency, Statistics Canada, since 1923. They found 
themselves--in the First World War, their status as a country 
was indeterminate with regard to Great Britain, but Great 
Britain would ask them how much can you do in the way of 
munitions and what can you send us in the way of wheat and how 
many troops do you think you can provide. And the answer was 
nobody in Ottawa knew. So they created a central agency, and it 
has been that way since, and very effective, partly because it 
gives the head of that agency--the chief statistician of the 
dominion--is on equal status with the other senior civil 
servants in the other departments of government.
    I just learned from Janet Norwood that the British, who 
have had a system not unlike ours, are beginning to bring it 
together. They are taking it sort of one step at a time, but 
there is now in the United Kingdom a national statistical 
service, and with more merging yet to come.
    The need in our case is twofold. One is that the 
mathematics of statistics gathering--index number theory--
continues to advance. We surprise ourselves perhaps to learn 
how recent some of these advances in theory are. For example, 
the first unemployment rate published by the Department of 
Labor using current monthly survey methodology was 1948. We had 
to learn sampling, the kind of sampling that every member, in 
the White House does every night to find out what people think 
about the chemical treaty. They did not know how to do that. It 
was learned in the 1930's and gradually brought in. We used to 
take the unemployment rate in the census by the process of 
counting everybody and seeing how many were unemployed. We took 
it in April of 1930 and then April of 1940, and there was no 
depression. It just never appears.
    I was once, in the Kennedy administration, I was Assistant 
Secretary of Labor with a nominal responsibility for the BLS. I 
can tell you that when the unemployment rate would come out 
people were fascinated with it because it was still new. And 
immediately it would be an issue of controversy. It is too 
high, said the Chamber of Commerce; too low, said the AFL-CIO. 
So we would have our meeting--but gradually over 30 years I 
think that number has achieved an acceptance that is rarely 
questioned.
    On the other hand, you can see recently how much 
controversy and doubt is raised over the issue of how we adjust 
various benefits for the cost-of-living changes, and, indeed, 
we have indexed the Internal Revenue Code as well. All of which 
argues that, you know, let's take a look at this. And the view 
that it ought to have come about is very widespread, and I 
would like just to close my remarks by asking that you place in 
the record a copy of a letter from what I believe is every 
living former Chairman of the Council of Economic Advisers, a 
letter sent to me and Senator Kerrey just last year, saying--we 
write to support the basic objectives and approach of your bill 
to establish the commission to study the Federal statistical 
system.
    [The letter follows:]

  LETTER TO SENATORS MOYNIHAN AND KERREY FROM FORMER CHAIRMEN OF THE 
                      COUNCIL OF ECONOMIC ADVISERS
                                                 September 23, 1996
The Hon. Daniel P. Moynihan
The Hon. J. Robert Kerrey
U.S. Senate
Washington, DC
    Dear Senators Moynihan and Kerrey: All of us are former Chairmen of 
the Council of Economic Advisers. We write to support the basic 
objectives and approach of your Bill to establish the Commission to 
Study the Federal Statistical System.
    The United States possesses a first-class statistical system. All 
of us have in the past relied heavily upon the availability of 
reasonably accurate and timely Federal statistics on the national 
economy. Similarly, our professional training leads us to recognize how 
important a good system of statistical information is for the efficient 
operations of our complex private economy. But we are also painfully 
aware that important problems of bureaucratic organization and 
methodology need to be examined and dealt with if the Federal 
statistical system is to continue to meet essential public and private 
needs.
    All of us have particular reason to remember the problems which 
periodically arise under the current system of widely scattered 
responsibilities. Instead of reflecting a balance among the relative 
priorities of one statistical collection effort against others, 
statistical priorities are set in a system within which individual 
Cabinet secretaries recommend budgetary tradeoffs between their own 
substantive programs and the statistical operations which their 
departments, sometimes by historical accident, are responsible for 
collecting. Moreover, long range planning of improvements in the 
Federal statistical system to meet the changing nature and needs of the 
economy is hard to organize in the present framework. The Office of 
Management and Budget and the Council of Economic Advisers put a lot of 
effort into trying to coordinate the system, often with success, but 
often swimming upstream against the system.
    We are also aware, as of course are you, of a number of 
longstanding substantive and methodological difficulties with which the 
current system is grappling. These include the increasing importance in 
the national economy of the service sector, whose output and 
productivity are especially hard to measure, and the pervasive effect 
both on measures of national output and income and on the Federal 
budget of the accuracy (or inaccuracy) with which our measures of 
prices capture changes in the quality of the goods and services we buy.
    Without at all prejudging the appropriate measures to deal with 
these difficult problems, we believe that a thoroughgoing review by a 
highly qualified and bipartisan Commission as provided in your Bill has 
great promise of showing the way to major improvements.

            Sincerely,

                       Professor Michael J. Boskin,
                                        Stanford University

                              Dr. Martin Feldstein,
                       National Bureau of Economic Research

                                     Alan Greenspan

                       Professor Paul W. McCracken,
                                     University of Michigan

                                Raymond J. Saulnier

                               Charles L. Schultze,
                                  The Brookings Institution

                                  Beryl W. Sprinkel

                                     Herbert Stein,
                              American Enterprise Institute

                       Professor Murray Weidenbaum,
                  Center for the Study of American Business
    Senator Moynihan. It is signed by Michael Boskin, Martin 
Feldstein, Alan Greenspan, Paul McCracken, Raymond Saulnier, 
Charles Schultze, Beryl Sprinkel, Herbert Stein, and Murray 
Weidenbaum. I do not want to add an ideological tint to this 
because there is none, but I do somewhat regret to say that of 
all those illustrious names, only one of them is a Democrat. 
[Laughter]
    But this is the view of the persons who had to advise the 
President under the legislation, the Employment Act of 1946, 
have had to advise the President about the state of the 
economy, the utilization of resources including manpower, and 
are the people who use data in the White House. They find they 
have to get it from too many places, and often it is simply 
inconsistent and sometimes incompatible, and so good management 
and good government suggest we would take a look at our present 
arrangements.
    I thank you, sir.
    [The prepared statement of Senator Moynihan follows:]

                 PREPARED STATEMENT BY SENATOR MOYNIHAN
    Mr. Chairman: Statistics are part of our constitutional 
arrangement, which provides for a decennial census that, among other 
purposes, is the basis for apportionment of membership in the House of 
Representatives. Article I, Section I provides that:

        .  . . enumeration shall be made within three Years after the 
        first meeting of the Congress of the United States, and within 
        every subsequent Term of ten Years, in such Manner as they 
        shall by Law direct.

President Kennedy once said:

        Democracy is a difficult kind of government. It requires the 
        highest qualities of self-discipline, restraint, a willingness 
        to make commitments and sacrifices for the general interest, 
        and also it requires knowledge.

That knowledge often comes from accurate statistics. You cannot begin 
to solve a problem until you can measure it.
    As you know, Senator Bob Kerrey and I have introduced S. 144, a 
bill to establish a commission to study our Federal statistical system. 
The Commission would consist of 13 Presidential and Congressional 
appointees with expertise in fields such as actuarial science, finance, 
and economics. Its members would conduct a thorough review of the U.S. 
statistical system and issue a report including recommendations on 
whether statistical agencies should be consolidated.
    We, of course, have an example of a consolidated statistical agency 
just across our northern border. Statistics Canada, the most 
centralized statistical agency among OECD countries, was established in 
November, 1918 as a reaction to a familiar problem. At that time, the 
Canadian Minister of Industry was trying to obtain an estimate of the 
manpower resources that Canada could commit to the war effort. And he 
got widely different estimates from statistical agencies scattered 
throughout the government. Consolidation seemed the way to solve this 
problem, and so it happened--as it can in a parliamentary government--
rather quickly just as World War I ended.
    A member of my staff has just returned from a working meeting in 
Ottowa with the Assistant Chief Statistician responsible for Business 
and Trade Statistics, including price statistics. We learned that 
Statistics Canada is doing quite well. Decisions with respect to the 
allocation of resources among statistical functions are made at the 
highest levels of government, since the Chief Statistician of 
Statistics Canada holds a position equivalent to Deputy Cabinet 
Minister. He communicates directly with Deputy Ministers in other 
Cabinet Departments. In contrast, in the United States, statistical 
agencies are buried several levels below the Cabinet Secretaries, so it 
is difficult for the heads of these statistical agencies to bring 
issues to the attention of high-ranking Administration officials and 
Congress.
    Our bill, S. 144, would require the Commission to focus 
particularly on the agencies that produce data as their primary 
product--agencies such as the Bureau of Economic Analysis (BEA) and the 
Bureau of Labor Statistics (BLS).
    I have here a letter from nine former Chairmen of the Council of 
Economic Advisers (CEA) endorsing this legislation--virtually every 
living chair of the CEA. While acknowledging that the United States 
``possesses a first-class statistical system,'' these former Chairmen 
wrote:

        Without at all prejudging the appropriate measures to deal with 
        these difficult problems, we believe that a thoroughgoing 
        review by a highly qualified and bipartisan Commission as 
        provided in your Bill has great promise of showing the way to 
        major improvements.

    The letter is signed by Michael J. Boskin, Martin Feldstein, Alan 
Greenspan, Paul W. McCracken, Raymond J. Saulnier, Charles L. Schultze, 
Beryl W. Sprinkel, Herbert Stein, and Murray Weidenbaum. I ask that the 
full text of this letter be printed in the record.
    It happens that this Senator's association with the statistical 
system in the Executive Branch began over three decades ago. I was 
Assistant Secretary of Labor for Policy Planning and Research in the 
administration of President Kennedy. This was a new position in which I 
was nominally responsible for, among other things, the Bureau of Labor 
Statistics. The then-Commissioner of the BLS, Ewan Clague, could not 
have been more friendly and supportive. And so were the statisticians, 
who undertook to teach me to the extent I was teachable. And so it was 
that I came to have some familiarity with the field.
    In 1961 we received a report on ``The Price Statistics of the 
Federal Government'' from a committee led by George J. Stigler, who 
later won a Nobel prize in economics. The Committee stressed the 
importance of accurate and timely statistics, noting that:

        The periodic revision of price indexes, and the almost 
        continuous alterations in details of their calculation, are 
        essential if the indexes are to serve their primary function of 
        measuring the average movements of prices.

    More recently, in December of 1996, the Final Report of the 
Advisory Commission To Study the Consumer Price Index (The Boskin 
Commission) also addressed itself to the effectiveness of Federal 
statistical programs. Although the Boskin Commission focused primarily 
on the extent to which changes in the CPI overstate inflation, the 
Commission also recommended that:

        Congress should enact the legislation necessary for the 
        Department of Commerce and Labor to share information in the 
        interest of improving accuracy and timeliness of economic 
        statistics and to reduce the resources consumed in their 
        development and production.

    There is some momentum for a more centralized statistical system, 
as you will shortly hear from your other witnesses. I would like to 
commend Rep. Steve Horn on the bill he introduced in the 104th 
Congress, H.R. 2521, which would establish a Federal Statistical 
Service by merging the Bureau of the Census, the BLS, and the BEA.
    Janet L. Norwood, former Commissioner of the BLS, wrote in her book 
Organizing to Count:

        The U.S. system has neither the advantages that come from 
        centralization nor the efficiency that comes from strong 
        coordination in decentralization. As presently organized, 
        therefore, the country's statistical system will be hard 
        pressed to meet the demands of a technologically advanced, 
        increasingly internationalized world in which the demand for 
        objective data of high quality is steadily rising.

    And in a recent report on Balancing America's Budget, the Heritage 
Foundation recommends merging nine core statistical programs into a 
single, independent Bureau of National Statistics (BNS) headed by a 
Chief Statistician and modeled after the Canadian system.
    While I share many of the objectives of those who want to move 
quickly, I believe we must approach this subject somewhat more 
cautiously than some have suggested. Issues such as privacy must be 
carefully evaluated by a Commission. For example, earlier this week, 
reports in the media indicated that it may be possible to gain 
unauthorized access to confidential Social Security Personal Earnings 
and Benefits Estimate Statements (PEBES) via the Internet. Along with 
Finance Chairman Roth and others, I sent a letter this morning to the 
Social Security Administration requesting that these statements via the 
Internet be suspended until the Social Security Administration is able 
to convene and receive recommendations from an independent panel of 
computer and privacy experts on what additional safeguards are needed 
to protect the confidentiality of Social Security records.
    In an era in which people are fascinated by technology, but are not 
always aware of its consequences, it is important that a Commission 
take a comprehensive look at how our statistical infrastructure 
protects the privacy of individuals and their personal records.
    Our bill is only a first step, but an essential one. The Commission 
will provide Congress with a blueprint for reform. It will be up to us 
to finally take action after nearly a century of inattention to this 
very important issue.

    Senator Brownback. Thank you, sir, Senator Moynihan, whom I 
have followed for a long period of time and respect deeply your 
opinions, and I have quoted you often over the years as well, 
when you were with Republicans or Democrats, either way. I very 
much appreciate your being here, and we will look forward to a 
discussion.
    Congressman Horn has joined us. He has a bill, and what I 
want to get to eventually is a discussion between the two of 
you of an approach to take now, whether it should be a 
commission or a bill arrangement. I want to make sure we get to 
that. Congressman Horn, thank you for coming across the Hill, 
and the floor is yours.

 TESTIMONY OF HON. STEPHEN HORN, A REPRESENTATIVE IN CONGRESS 
                  FROM THE STATE OF CALIFORNIA

    Mr. Horn. Well, thank you very much. It is a great pleasure 
and delight to be with Senator Moynihan, who certainly has 
influenced all of us on many aspects of public policy.
    Let me just say I agree with the commission approach. 
Perhaps we both came out of the academic background and we 
found it is better to try and get a consensus by getting all 
the different views around the table and then pass legislation 
based on that. So I do not have any problem with the commission 
route.
    What got me into the issue was in my role as chairman of 
the Subcommittee on Government Management, Information, and 
Technology of the House Committee on Government Reform and 
Oversight. I took a look at this, and have a strong belief that 
in the organization of the government, the controversial issues 
should be in Cabinet Departments. That is why we pay Cabinet 
officers and Presidents of the United States. Hopefully, the 
non-controversial issues--and I realize the Consumer Price 
Index is very controversial would be handled appropriately. My 
own experience was not as lofty a position in the Department of 
Labor as my colleague here in the Senate, I was assistant to 
the Secretary of Labor under President Eisenhower, James P. 
Mitchell. And I know from that experience that we had high 
respect and never interfered with the Bureau of Labor 
Statistics within the Department.
    I feel very strongly that both the Census and the Bureau of 
Labor of Statistics--less I know about the Bureau of Economic 
Analysis--deserve the credibility of what I would call an 
independent agency. The National Science Foundation gets along 
quite well as an independent agency. It goes on about its 
business. To my knowledge, there has never really been a 
scandal involving the National Science Foundation. And we ought 
to get those agencies that need the credibility and should not 
be mixed up even by innuendo in partisan politics, we need to 
let those agencies function on their own in an independent 
agency status.
    So I am looking at, suggesting here as I did last year, a 
Federal statistical service. Former Commissioner Norwood was of 
great help as a principal witness, and I am glad to see that 
she is one of your principal witnesses. But I think we are not 
talking about merging all statistical services. I think there 
are certain advantages to incrementalism and not having 
everything come out of one service. For example, Agriculture 
will have some of its own statistics. Any Cabinet Department 
has some of this. And OMB, the Office of Management and Budget, 
is there to supervise the degree to which you will burden the 
taxpayers and others with this or that survey.
    But there are historically two agencies--Census and BLS--
that are looked to not simply by the Department in which they 
happen to be located, but they are looked to by millions to 
affect various decisions. Not only legislators at the local, 
State, and national level, but corporations, labor unions, non-
profits, all of those groups rely on the data that is reflected 
in the 10-year census, as well as in the monthly series in the 
case of BLS or the half-year series. We look to those for 
credibility, and my feeling is get them out of the Cabinet 
Departments where, even if the critic is mistaken, they would 
seem to be under political influence. And that is what has 
motivated me here, not to get everybody in one tent, but to get 
the two major ones plus the Bureau of Economic Analysis in one 
tent and have it assured that there is no political intrusion. 
We could have a professional advisory committee which could be, 
for terms certain, made up of leading statisticians. You could 
analyze that by category, and they serve between 
administrations and give policy guidance if needed to the 
director involved. That is just the way I would come at this, 
having looked at the successes or non-successes in government 
over the years.
    So I would like, Mr. Chairman, to file the statement which 
gets into detail really on the bill, but I just wanted you to 
know where I am coming from in terms of suggesting this.
    [The prepared statement of Mr. Horn follows:]

               PREPARED STATEMENT OF REPRESENTATIVE HORN
    During the last Congress, I introduced two pieces of legislation 
relating to Government statistics.
    The first, the Statistical Consolidation Act of 1995 (H.R. 2521), 
would have consolidated the three primary economic statistics agencies 
of the Federal Government--the Bureau of the Census, the Bureau of 
Labor Statistics and the Bureau of Economic Analysis--into a new, 
independent Federal Statistical Service. The second, H.R. 3924, would 
have permitted data sharing among eight government statistical 
agencies.
    The economic statistics gathered and analyzed by the Federal 
Government are integral to public and private decision making. The 
financial markets rise and fall based on the data provided by these 
agencies; Federal aid is determined and distributed using this 
information. Businesses make a variety of decisions with reference to 
these statistics, ranging from the development of new products to the 
opening and closing of factories. Although sound statistics and 
analysis do not automatically produce sound public policy, they do 
provide a necessary foundation from which to identify problems, to 
evaluate options, and to monitor results.
    Historically, most Federal statistical programs were established to 
serve the information needs of the particular department or agency in 
which they were based. However, the increasingly interconnected nature 
of major economic and social issues far exceeds the bounds of any 
single agency's work program. The information needs of Congress and the 
President transcend the data and statistics compiled by any single 
statistical agency.
    A new Federal Statistical Service would streamline and improve the 
production of key economic data. The three principal economic 
statistical agencies affected by this legislation cannot continue to 
operate in isolation.
    Additionally, the independence of the consolidated Federal 
Statistical Service will ensure freedom from partisan influences. I 
cannot emphasize strongly enough the need to protect the integrity of 
the data during the collection and interpretation stages of the 
process.
    The Director of the Census and the Commissioner of Labor Statistics 
must demonstrate an absolute commitment to the integrity of their 
statistics and be willing to take decisive action if he or she believes 
the statistics are not being protected from external political 
pressures.
    Confidentiality issues in statistical policy are also vitally 
important. More uniform confidentiality standards could permit the 
sharing of data for statistical purposes with Federal and State 
agencies. This would be integral to the success of the consolidation.
    As for those concerned that in the current atmosphere of reducing 
government's size and cost, we anticipate that the creation of a 
consolidated Federal statistics agency would result in significant 
savings through the streamlining of personnel and field and procurement 
offices. Further savings will be achieved from the consolidation of 
population list management operations.
    The other bill I introduced in the 104th Congress, H.R. 2521, would 
have taken existing resources and reallocated them to meet future 
statistical needs while also improving the accuracy of existing 
statistical measures.
    The time has long since passed to bridge the schism between the 
responsibilities given the statistical agencies and their capabilities. 
Consolidation is a long awaited and logical extension of the decades 
old collaborative relationship between these three agencies.
    Before introducing new legislation in the 105th Congress, I look 
forward to working with interested legislators, such as Chairman 
Brownback and Senator Moynihan, to develop bi-partisan legislation that 
could be introduced simultaneously in both bodies. Working together, we 
can develop legislation along these lines with important provisions 
that have been recommended again and again by those most familiar with 
government statistics.

    Senator Brownback. Good. I appreciate that.
    A bit of discussion here. Senator Moynihan, you believe 
that the approach we should go is the commission route. That is 
the way to get everybody headed the same way at roughly the 
same time.
    Senator Moynihan. Yes, sir, and I think Chairman Horn is of 
the same view, and the former heads of the Council of Economic 
Advisers say, yes, it is time to take a look at this.
    I would make the point, if I may, that we must not think of 
statistics as a sort of given thing, as something you know how 
to do and that is it. It is evolving all the time. An example, 
if I may, I was once Director of the Joint Center for Urban 
Studies at MIT and Harvard, and in the mid-1960s we held a 
conference down here with Census on what do you do about the 
undercount. And it was a very elegant arrangement. We had an 
academic--for each topic, a paper was prepared by an academic 
and a counterpart in the Bureau of the Census. And it was 
agreed that there was a very large undercount, that it was not 
evenly distributed. It was in central cities. It had racial 
components. It affected the distribution of seats in the House 
of Representatives.
    But they also agreed there was not much you could do about 
it because, in theory, a sample would be better, but they did 
not know that much about sampling.
    Thirty years go by, Mr. Chairman, and the Bureau of the 
Census now says, yes, we are ready, we can sample. But that is 
a learning process, and I think an independent agency might 
encourage that kind of creativity in a way that does not 
necessarily happen when you are way down at the bottom of an 
agency that has political activity going on all the time and 
the Secretary is not thinking about you. He is thinking about 
that other Secretary or what the Vice President says and what 
Chairman Brownback thinks.
    Senator Brownback. Let me ask you, would either or both of 
you be willing to hazard your opinion on what a consolidated 
agency or independent entity should look like, or at a minimum, 
which of the current functions it should contain?
    Mr. Horn. Well, I suspect some economies could be made in a 
merger between what Census does and what BLS does. But I would 
think that is the kind of thing that might well come out before 
a commission where you have the chance and the time it takes to 
get into the----
    Senator Moynihan. And the Bureau of Economic Analysis.
    Mr. Horn. Right. And see if there is a way you can do a 
certain series where all three of those agencies contribute 
something to the series. What I am trying to do is get strong 
professional guidance throughout the profession, be it the 
American Statistical Association, American Economics 
Association. From all of these I would put representation on an 
advisory board.
    I have had the experience of doing that when the late 
Robert Cutack and I drafted the legislation for the National 
Institute of Corrections, which was Chief Justice Burger's 
idea. We needed to do something about the States and 
localities. Well, we put together an agency within Justice that 
has a professional board that actually recommends the director 
to the Attorney General, and what we did was establish various 
categories of representation. That has worked well, so I am 
used to a system like that where you have got a strong advisory 
board; they make their recommendations to the director, and if 
there is a vacancy there, they make the recommendations to the 
Attorney General about who should be the director. It has 
worked well for over 20 years. I am sort of thinking of that in 
this type of a consolidation.
    Senator Brownback. You are saying, Senator Moynihan, at a 
minimum you would look at Census, BLS, BEA, as being a 
consolidated--at least a core of that that you would be pulling 
these together.
    Senator Moynihan. That I think, sir, is your basic economic 
data system. Now, you know, the FBI crime statistics are going 
to be over there whether we like it or not because it is the 
FBI. But your economic data, population data, resource data 
would come together, as Chairman Horn proposed last year. That 
need not be the only way to do it, but certainly it would be 
one of the first things you would look at.
    Senator Brownback. OK. I have to admit, too, to my 
participation in this game previously when I was Secretary of 
Agriculture of Kansas. We had an arm that was counting 
different statistical items which the Federal Government was 
counting as well. Would you bring in something like that or 
not? You know, it is a very costly operation overall. I do not 
know what it is in USDA's budget, but it is a substantial 
dollar investment that is put in there.
    You mentioned, Senator Moynihan, if I could, that you think 
maybe this is an idea whose time has finally come. The fortune 
of a good staff, they have looked into some of the background 
of this, of which I am sure you are familiar as well. Seventeen 
years ago, Senator Ribicoff introduced the Statistical Policy 
Act. Even before that, the now-defunct Bureau of Efficiency in 
1922 issued a report urging centralization of Federal 
Government statistical agencies. There have been other 
initiatives.
    Why now do you think we could pull this together and 
actually move it forward and get something constructive done, 
when obviously this has been identified at different times over 
history as being an area of some problem for us, or if not 
problem, at least an area where gained efficiencies should be 
able to take place? Why could we get it done now?
    Senator Moynihan. May I say I had some involvement with 
Senator Ribicoff's proposal. Why now? I will tell you why now. 
Because a bunch of crazy Republicans have taken over Congress 
and they think it is time to change some things.
    Senator Brownback. Good enough for me. [Laughter]
    Mr. Horn. I might add, instead of saying why, why not, I 
guess is where I am coming from on this.
    Senator Brownback. I guess I am asking for the detractors. 
Where are we going to get into the fights before you start into 
that?
    Senator Moynihan. You have touched the BLS, and you have 
touched the AFL-CIO. You have touched the Bureau of Economic 
Analysis. You have touched the Chamber of Commerce.
    Mr. Horn. And we want to touch them all.
    Senator Moynihan. All at once. Or I think they will get you 
one more time.
    Mr. Horn. I think the Senator is absolutely correct on 
that. That commission will smoke out where people are on these 
various areas in public, and I think that is important. I think 
we should be concerned about what the ultimate end of the 
public is as a whole as we try to legislate for America as a 
whole, not just the specific interest groups. But as the 
Senator says, every group in this town has their little bureau, 
and often, I remember when I went to the Department of Labor, 
it was filled with retirees from certain special interests. And 
I am sure the Senator ran into that in his role as Assistant 
Secretary.
    So I just think if we start, I think we could do this in a 
year between the commission and the idea of moving ahead on 
legislation, or do it during this Congress.
    Senator Brownback. And that is a reasonable time frame, you 
think as well, Senator Moynihan?
    Senator Moynihan. That would be my judgment.
    Senator Brownback. Good. Gentlemen, thank you very much. I 
look forward to working with both of you as we push this on 
forward and catch some of those sacred cows. Thank you very 
much.
    The next panel will consist of Janet Norwood, who served as 
Commissioner of the Bureau of Labor Statistics from 1979 to 
1991, is now a senior fellow at the Urban Institute; and Vince 
Barabba--I hope I said that correctly.
    Mr. Barabba. You did, sir.
    Senator Brownback. Good--twice Director of the Census 
Bureau, past president of the American Statistical Association, 
currently general manager of General Motors Corporate Strategy 
and Knowledge Development.
    Both of you have a distinguished past and current 
occupations as well and, I am sure, distinguished futures. I 
look forward to your presentations and your thoughts. We can do 
this just as the last panel if you desire, or if you do want to 
read your statement in the record, we can do that as well. I 
think you can gather what I am after is what sort of proposals 
can we move forward on and in what sort of time frames, bottom 
line, for me and for this Subcommittee.
    Ms. Norwood, if you would like to take the podium or the 
mike first, we would love to hear from you.
    Ms. Norwood. Mr. Chairman, since Vince Barabba has a plane 
to catch, perhaps we might start with him.
    Senator Brownback. In the interest of reverse chivalry, Mr. 
Barabba, please.
    Thank you for joining us.

   TESTIMONY OF VINCENT P. BARABBA, GENERAL MOTORS CORPORATE 
               STRATEGY AND KNOWLEDGE DEPARTMENT

    Mr. Barabba. You should be aware that you have two people 
in front of you who have served in statistical agencies under 
the directorship of two different political parties, and I 
think that----
    Ms. Norwood. That is, both Republicans and Democrats, each 
of us have served.
    Senator Brownback. Good.
    Mr. Barabba. And so I think it is clear that we sit here 
without any political point of view in mind.
    I will just submit my testimony. I would say that Senator 
Moynihan and Congressman Horn did a very good job of 
articulating what I had in mind, and I learned a long time ago 
not to try to enhance what Senator Moynihan has said.
    But let me address your question of ``Why now?'' One of the 
reasons of why now is that it is no longer a question of 
whether our society is changed. It is a question of the extent 
to which it has changed. And it is no longer a question of 
whether we are able to predict what that change is. It is clear 
that we are not capable of predicting what the future will be.
    We could spend a lot of time trying to do it, but what we 
find out is the more we make assumptions, underlying 
assumptions about what our forecast is, the less likely you are 
to be correct because we are not very good at being correct 
about these assumptions because society is changing so fast.
    So it becomes quite necessary to move from a predict-and-
prepare mentality--that is, I can anticipate what the world is 
going to be like for some period of time, therefore I will 
prepare a set of actions--to much more of a sense-and-respond 
mentality, which says I have got to be prepared and I have to 
be sufficiently capable of adapting to changes that I did not 
expect.
    That is no longer a question. Industry is coming to 
recognize that. I am not sure that government has, but the 
traditional ways of doing business just do not hold anymore. So 
that is another reason to change because the current 
statistical system was designed around a predict-and-prepare 
mentality. And I do not think we can handle that anymore.
    The other point I think I could make is, having had the 
experience of working in both a large government agency and a 
reasonably large company, the problems are fundamentally the 
same. Up until recently, inside of General Motors you would 
find yourself dealing with the different units of the company, 
virtually being silos of intense competence, but not willing to 
share that competence across the activities, and that poses a 
problem for a customer because a customer looks at the whole, 
not at any one of the entities. And if you come together in 
front of the customer and you are not consistent, you find 
yourself in trouble, as we did just recently.
    It has taken a fundamental change of how we understand 
customers and how we have to look across our activities to put 
us in the position where we have improved our situation from 
the situation we found ourselves in in the late 1980's. We had 
to make some really hard choices inside the company, and we 
actually had to change how we gathered information and 
consolidated the activity. But there is a trade-off there. You 
do not want to consolidate information gathering so that it 
becomes an entity in itself and it collects what it wants to 
collect. You have to design a balance so that those who use the 
information are in a position to affect the relevancy of what 
is collected.
    The other aspect of it is that those who are responsible 
for the collection have to be responsive to the issue of 
relevancy, but then they have to have in place a set of 
procedures, agreed to by all involved, that this is how we will 
collect the information and this is how we will present it.
    I would say that the difference between the public and the 
private on this issue is not very great. In fact, there are 
more similarities between big agencies and big companies than 
there is between public and private.
    The third point I would make is relative to the commission, 
which I would support, because there is a lot of sorting out to 
do before you would move forward at any level of consolidation. 
Based primarily on the comments I just made, the worst thing to 
do is to take a set of agencies that were designed for one era 
and mush them together without understanding what their new 
assignment is. So the notion of a commission to really sort out 
the needs of society, not just the needs of government, to 
design a system based on those needs would be very important.
    This issue of finding out where the barriers generated by 
the vested interests becomes a really important item, and so 
the makeup of the commission I think has to be really carefully 
thought through, because I think you want the people with 
specific interests involved, but you want them to show up at 
the meeting with the interests of society at the forefront, not 
the interests of their vested interest group.
    I think I could stop there, Mr. Chairman, and let Janet go 
ahead.
    [The prepared statement of Mr. Barabba follows:]

                PREPARED STATEMENT OF VINCENT P. BARABBA
    Let me, at the outset, suggest that determining whether the Census 
Bureau be housed in the Department of Commerce or elsewhere, although 
an important topic, will not, by itself, significantly improve the role 
or performance of the Census Bureau in particular, nor Federal 
statistics in general. The issue is far more complex, and so this 
afternoon I will attempt to provide a framework for addressing both the 
challenges faced by the Census Bureau and those faced by Federal 
statistics overall.
    More specifically, I suggest the development of a broader 
information system (not necessarily a large central organization) 
within which the agencies of the Federal Government can adequately meet 
the ever-changing information needs of our society.
    As currently operated, few statistical agencies are either equipped 
or authorized to help determine what society needs to know to 
meaningfully improve our well being. Doing so would require a cross-
societal discussion process, which our current system does not support. 
An ideal process would allow interaction between those who determine 
what citizens need to know and those who collect the data. These 
parties would be encouraged to discuss all aspects of the information, 
including its form, accuracy and cost (both time and money), and would 
go far toward ensuring that the right information was collected, and 
was then used as effectively as possible.
    This issue is important for at least two reasons:

    1. LIt is no longer sufficient to address the issues of society 
from the perspective of limited purpose functional organizations (i.e., 
labor, commerce, health, education, etc.).
    2. LA government that merely ``predicts and prepares'' is no longer 
viable. Our society faces an increasing rate of change and increasing 
complexity, and therefore requires a government with the ability to 
``sense and respond'' in an iterative, interactive learning process.

    For the duration of my testimony, I will elaborate on these two 
points.
A Systemic Approach
    First, it is no longer sufficient to address the issues of society 
from the perspective of limited purpose functional organizations (i.e., 
labor, commerce, health, education, etc.).
    Each of us sees the world through a unique set of lenses, 
determined by our personal experiences, our responsibilities and our 
singular interests. As a result, none of us sees reality in its 
entirety. This is sometimes referred to as the silo or smokestack 
problem--a problem of particularism versus globalism.
    In any large organization, individuals have particular tasks and 
responsibilities, which tend to be organized into narrow, vertically 
structured functions. Those who manage the business, on the other hand, 
have broader interests, as well as more general tasks and 
responsibilities.
    I have encountered this problem throughout my career in both the 
public and private sector. Interestingly, my first encounter with the 
silo problem occurred during my senior year as an undergraduate 
student. One of my professors had developed a business simulation in 
which students were organized into teams that competed in making and 
selling a product. One year, instead of assigning students to a 
particular team at random, the professor organized the students 
according to major. This led to strikingly different outcomes.
    The marketing majors spent most of their time and money on sales 
and promotion. They acquired an impressive share of the total market, 
but at high cost, and were bankrupt before the game ended. The 
accounting majors aimed at maximizing profits by minimizing investments 
in products and promotion. With no new products and only meager 
promotion of existing ones, the eyeshade brigade lost market share and 
slipped by degrees into bankruptcy.
    At the other extreme, the production majors spent all their money 
on product development and manufacturing processes. They ended up with 
great products at the right prices, but with no money to tell customers 
about them, they too went out of business.
    To the consternation of all concerned, the personnel majors won. 
The marketing majors ran out of money, the accountants ran out of 
products, and the production majors ran out of customers. The personnel 
types occupied themselves with endless changes to the organization 
chart. Having spent no money, they simply ran out of time and won the 
game by default.
    Unfortunately, a similar scenario plays itself out in our 
government as well. Our functional policy makers often fail to bring 
together diverse government departments to address the cross-functional 
requirements of society. More often than not, information acquired by 
different departments falls into functional information depositories, 
where the data does little to improve government's understanding of the 
constituencies it serves. Department A knows what it knows; Department 
B knows what it knows, and so forth. By themselves, the isolated bits 
of information are less useful than they might be if combined with 
information from other sources and placed in the context of a decision 
making process.
    But working cross-functionally is not the answer on its own either. 
In addition to tapping into cross-functional networks, a new system 
should also include methods for accessing deeper knowledge within the 
individual functions. A strong statistics system must support both.
    One of Federal statistics most knowledgeable observers and 
constructive critics, James Bonnen, pointed out the need to organize 
statistics around knowledge needs, rather than points of data 
collection during a debate over whether the Census of Agriculture 
should be collected at the Census Bureau or the Department of 
Agriculture. While almost everyone else focused on who should collect 
the information, Jim reminded the group of how the data would be used. 
He pointed out that society needed to understand the specifics of each 
aspect of the system, as well as the interactions of the entire 
system--a system that started in the farmer's field and ended on the 
consumer's dinner table.
    This meant we needed to integrate the data and information 
collected from numerous sources, ranging from the input supplies (seed, 
fertilizer, machinery, etc.) to agricultural production, commodity 
assembly, processing, manufacturing, wholesaling, retailing, 
transportation and eventual consumer consumption.
    With that system view in mind, Jim suggested that the Census Bureau 
commit resources to identifying and integrating the various food sector 
statistics scattered throughout several economic censuses and surveys, 
then relate those statistics to the agricultural census. In essence, 
Jim knew it would be best to align our statistical practices around the 
user's needs and not around the existing organization structure.
    This is not only a government problem. Many large organizations--
both public and private--have failed in similar ways. The communities 
or markets of interest they wish to serve have changed, while the 
organizations themselves have remained much as they were at the turn of 
the century. Many public and private enterprises remain relics of the 
industrial age and a long-vanished society in which governments and 
companies offered services or products to customers and constituents 
who bought or accepted what they were offered. If government and 
corporate leaders do not drive the necessary changes, their 
enterprises, like endangered species, will be replaced by better, more 
adaptive competitors.
    Faced with an uncertain future, we require structures and processes 
that can adapt quickly and flexibly to change, because while we can 
guarantee that change will occur, we can not predict what such changes 
will actually look like. Creating such structure and processes demands 
a systemic approach--an approach that takes its cues from the realm of 
systems thinking.
    Russell Ackoff describes a system as ``any entity, conceptual or 
physical, which consists of interdependent parts.'' At the same time, 
``a system is a whole that cannot be divided into independent parts.'' 
\1\ Each element of the organization must rely on and interact with the 
rest if the organization as a whole if it hopes to succeed. Today, 
problems are best solved not by breaking them into functional bits, but 
by carrying them into the next larger system and solving them through 
integrative mechanisms.
---------------------------------------------------------------------------
    \1\ Russell L. Ackoff, The Democratic Corporation: A Radical 
Prescription for Recreating Corporate America and Rediscovering Success 
(New York: Oxford University Press, 1994), 21.
---------------------------------------------------------------------------
    Systems thinking allows us to see the structure that underlies 
complex situations, while still seeing the whole. It provides a 
framework for seeking interrelationships rather than things, as well as 
a framework for seeking patterns of change rather than static 
snapshots. In essence, it encourages us to create a whole whose value 
is greater than the sum of its parts.
    Currently, the totality of Federal statistics is not more valuable 
to society than the sum of its parts. I make this assertion recognizing 
that each agency individually makes valuable contributions, but as a 
statistical system designed to contribute to a broader system of 
democratic governance, the current combination of functionally aligned 
statistical agencies falls considerably short of its potential value. 
By looking at the problem from a systems thinking framework, the 
solution clearly does not lie in fixing any one or some combination of 
the parts (i.e., the statistical agencies). The solution lies in fixing 
the way the agencies interact--not only among themselves but also with 
those for whom they provide information.
    Solving the problem, of course, has become more difficult in that 
the combination of parts is increasing. For example, just the transfer 
of responsibility for many social programs from the Federal Government 
to the states and localities has dramatically increased the number of 
customers served by Federal statistical agencies. Yet, with some 
notable exceptions, the attention of many Federal statistical agencies 
is directed to the needs of Federal functional departments within which 
they are located.
    The problem presented by this lack of a systemic approach was aptly 
described by a participant in a conference I attended in 1991, at which 
we discussed the barriers to making more data readily available for use 
by State and local agency staff. As my colleague observed, ``It's 
nobody's job to see that data are integrated across programmatic areas, 
that standards are established, or that needed data are collected and 
made available.''
Moving Toward Sense-and-Respond
    My second point builds on the first, and requires that we 
acknowledge the changes in society overall, and act on what that means 
for government in particular. A government that merely ``predicts and 
prepares'' is no longer viable. Our society faces an increasing rate of 
change and increasing complexity, and therefore requires a government 
with the ability to ``sense and respond'' in an iterative, interactive 
learning process.
    Peter Drucker recognized this fact nearly ten years ago. In 1988, 
Drucker observed the following:

          ``We are entering a third period of change: a shift from the 
        command-and-control organization, the organization of 
        departments and divisions, to the information-based 
        organization, the organization of knowledge specialists . . . 
        But the job of actually building the information-based 
        organization is still ahead of us--it is the managerial 
        challenge of the future.'' \2\
---------------------------------------------------------------------------
    \2\ Peter F. Drucker, ``The Coming of the New Organization,'' 
Harvard Business Review, 66, no. 1 January-February 1988: 53.

    Drucker has also pointed out that knowledge must serve a purpose to 
have real value. The power of knowledge lies in its role as a basis for 
action. For the public or private enterprise, knowledge must serve as a 
basis for decision making and the allocation of resources that follow 
from those decisions.
    While it may seem painfully obvious that using knowledge 
effectively makes sense, organizations consistently create barriers to 
doing just that. Key among these self-imposed barriers is the 
intervention of ``vested interests'' and ``information handlers.''
Vested Interests
    To remain relevant and useful, the information produced by Federal 
statistical agencies depends on linkage and communication between data 
providers and data users--those who produce information and those who 
make laws and policies. Most data providers understand the importance 
of this linkage, but generally prefer arm's-length transactions with 
policy makers and political officials, in order to guard against the 
manipulation of data collection, production and dissemination. There is 
some historic evidence for their concerns; nevertheless, closer 
encounters must occur.
    To ensure society's investment in relevant knowledge creation stays 
relevant to national goals and needs, the data provider communities 
require the user communities provide a clear statement on overall, 
societal information needs. To ensure all this is accomplished 
efficiently, someone must coordinate efforts and make sure that only 
the minimum allocation of resources (time, money and respondent burden) 
are expended to gather all that is needed.
    Without such a systemic approach, the budgeting process will be 
increasingly dominated by special interests. This will occur due to the 
nature of the budget process and the multiple client users of 
statistical agencies. The individual agencies, in budget terms, link to 
their departments first, then OMB, and then the congressional 
appropriation cycle. If the budget process is neglected or becomes too 
decentralized, attempts to allocate the limited statistical resources 
of government to needed activities that cross departmental boundaries, 
at the expense of department budgets, will face the difficult political 
reality of what is sometimes referred to as the ``iron triangle''--the 
affected departments, special interest groups associated with those 
departments, and related congressional committees.
Information Handlers
    ``Information handlers,'' those competent department employees who 
collect, package and interpret information and construct databases, 
also intervene between data collection and its useful application. 
Typically, these individuals are attached to specific functional areas 
and operate on behalf of their respective functions. They transmit 
information within the function's hierarchy, offer advice, and act as 
keepers of the function's accumulated data, information and knowledge. 
These information handlers are often individuals who, by virtue of long 
tenure, have a unique grasp of the information used by others within 
their functional area.
    Information handlers, no matter what their titles, have one thing 
in common: their position and status in the organization is closely 
bound up with the control of information. Information and its methods 
of acquisition are the coin of their realm, and they guard it 
carefully. This is neither unnatural nor entirely bad. The most adept 
information handlers are generally those who recognize the dangers of 
allowing people who lack training in research methodologies to get 
involved with data collection, because these individuals often draw 
inferences that cannot be supported by the statistical requirements of 
professional research.
    In this context, information handlers are justifiably concerned 
that ``insights'' be objectively and genuinely determined. They are 
aware that information gained from listening can influence decisions 
involving large sums of money and other resources. As a result, 
information handlers prefer to be the singular channel for listening, 
the intermediary between the outer environment and the many individuals 
in their function who have an interest in probing that environment.
    But information handlers can create problems within their own 
functions when they become more focused on the methods of analysis than 
the use of analysis. The consequence of this behavior is described by 
Russell Ackoff in a recent interview regarding operations research (OR) 
and management science (MS) specialists.
    He points out that although OR and MS ``started out as the 
application of science to the problems of managers of large-scale 
operations,'' eventually ``the researchers became enamored of the 
techniques, which eventually became ends in themselves, taking the 
focus off management and its problems.''
    The role of information handlers was most appropriate during the 
Industrial Age when we viewed the enterprise as a simple machine of 
interrelated and replaceable parts. Its limitations are more obvious 
today, particularly as we require greater cross-functional activities 
to address the problems facing us from a total systems perspective. 
Since most information handlers are bound to particular functions of 
the enterprise, they inadvertently maintain barriers to cross-
functional information sharing. Their higher order mission has been to 
collect, order and analyze information for the particular uses of their 
parent functions, not to make it available or meaningful to others 
outside their domain. There is also evidence that simply being part of 
a function skews listening toward those things the function is keen to 
hear.
Potential Solutions
    As external pressures force us to move from the familiar ``predict 
and prepare'' mentality, and toward a ``sense-and-respond'' capability, 
we need to design a Federal information system that is positioned and 
empowered to do several things. It must:

      1. LBalance needs of society from a total system perspective with 
the natural desire to create narrow departmental specific budgets, 
championed by vested interests.
      2. LBe motivated by an incentive system that rewards functional 
statistical agencies and the users of their information for focusing 
their attention on the requirements of the total system, even at the 
expense of functional interests.

    How does this relate to the concern of this Committee regarding the 
location of the Census Bureau? The Census Bureau, as currently 
constituted and perceived by most observers of Federal statistics, is a 
general purpose statistical agency--often referred to as ``the fact 
finder for the nation.'' This designation has occurred mostly by 
practice, because of the central and integrative role of the decennial 
and economic censuses, and not by organizational design.
    If we are to achieve the advantages of a truly Federal statistical 
system that also serves the needs of local governments, there is no 
question the Census Bureau must be an integral part of that system. If 
that direction is taken, as others have clearly stated, the Bureau of 
the Census will require a broader perspective on the needs of our 
society than is found in any current department of government. The same 
could be said for the Bureau of Labor Statistics, Bureau of Economic 
Analysis and other major departmental statistical agencies.
    The Census Bureau, in this case, must be empowered to anticipate 
the needs of local as well as Federal Government. Using methods as 
simple as geocoding, we could then use local administrative records 
regarding specific populations in tandem with Census Bureau small area 
data to study, in a more consistent and inexpensive manner, any number 
of local public policy, business and health issues.
    As I was preparing this statement, it occurred to me I would be 
participating in yet another plea to Congress to address this issue 
during the 20th century. Throughout this century--a century of 
incredible change--we have not been able to fundamentally change our 
approach to measuring and understanding ourselves, although endless 
reports and time have been expended to try to get us to do the job 
better. Given that continued change is inevitable, I believe we will 
change. The question is whether that change will be designed in 
anticipation of future needs, or forced on us as we react to breakdowns 
in our system.
    As the century comes to a close, several valuable suggestions have 
been presented: the well-reasoned and thoughtful account by my 
colleague Janet Norwood published in Organizing to Count; H.R. 2521, 
submitted by Congressman Horn; and the comprehensive review, Improving 
the Federal Statistical System: Issues and Options prepared in 1981 by 
the President's Reorganization Project for the Federal Statistical 
System.
    A skeptic reviewing this material would conclude that ``never has 
so much been said and so little done.'' As a practical optimist, 
however, my hope is that the needs of society have reached a point 
where something must be done. Past experience has shown that our 
elected and appointed officials, when confronted by a pressing need 
such as this, are capable of coming together to solve the problem.
    I am convinced that the pressing need for a reform of how our 
statistical agencies are coordinated is about to burst upon us because 
of the complex problems we are facing--problems that require more 
relevant, accurate, timely, integrated, easy to understand and cost 
effective information than our current approach to information 
gathering is capable of achieving.
    In commenting on the inability of the government to coordinate the 
Federal statistical agencies, Jim Bonnen referenced the sage comments 
of Sir Claus Moser, at that time the distinguished director of the 
Statistical Service of the United Kingdom.

        Sir Claus Moser once observed to a conference that 
        ``statisticians must suffer disasters as a hazard of their 
        profession. But, they should never allow disgraces to occur.'' 
        He paused at the puzzled expressions of his audience and added 
        ``You know what a disgrace is? . . . It is a disaster that is 
        allowed to continue.'' We now have such a disgrace.

    Mr. Chairman, you and your congressional colleagues have the 
opportunity and responsibility to keep this disaster from becoming a 
disgrace. The solutions, painful as they might be to some of us, are 
available and clearly conceived.
    From personal experience, I can attest to both the pain and the 
benefits of taking the type of action necessary to avoid a disgrace. In 
the early 90s General Motors found itself in very difficult 
circumstances. Our Chairman, Jack Smith, faced the daunting task of 
returning the company to profitability after years of disasters that 
brought the company to the brink of bankruptcy.
    Among the steps he took was one similar to what I am suggesting 
today. In June 1994, he announced the formation of the Strategic 
Decision Center (now called Corporate Strategy and Knowledge 
Development) to support management in the integration of market and 
business knowledge, the management of information systems, and the 
development of a global direction on core business and strategic 
intent. He directed the center to create a knowledge- sharing network 
that would support GM's efforts to meet several goals:

      1. LBetter align strategic and operational business plans.
      2. LImprove management's understanding of complexity, uncertainty 
and opportunity in the market.
      3. LDetermine the required resources for knowledge development 
and clarify roles and responsibilities.
      4. LEffectively capture ideas for innovative products and 
services.
      5. LDevelop organizational learning as a system.

    This direction led to the consolidation and coordination of the 
market research function throughout the entire corporation. Today, our 
market research budgeting process, the manner in which we determine our 
information needs, and the manner in which they are incorporated into 
our decision making process all contribute to our understanding of 
General Motors as a system within the context of the environment in 
which we do business--on a global basis. I believe my colleagues within 
GM, including some of those who had their roles and positions within 
their functions severely altered, would agree that GM is better off 
today as a result.
    Mr. Chairman, I find it interesting that I am at this hearing 
offering advice to the government, based on my experience, much of 
which was gained as an employee of the General Motors Corporation. I 
say that somewhat concerned that many people familiar with the often 
repeated quotation: ``What's good for General Motors is good for the 
country,'' might be so bothered by the statement that they would ignore 
the transferability of the GM experience to the situation at hand. 
After reviewing the events leading up to the reported quotation, 
however, I believe a little historical clarification would be of value.
    During his Senate confirmation hearing to become Secretary of 
Defense, former GM President, Charles E. Wilson, was asked, ``If a 
situation did arise where you had to make a decision which was 
extremely adverse to the interests of your stock and General Motors 
Corporation . . ., in the interest of the United States Government, 
could you make that decision?'' Wilson replied, ``Yes, sir; I could. I 
cannot conceive of [a conflict] because for years I thought what was 
good for our country was good for GM and vice versa.'' \3\
---------------------------------------------------------------------------
    \3\ U.S. Senate Nomination Hearing, Charles E. Wilson, January 15, 
1953 Washington, D.C. GPO: 1953.
---------------------------------------------------------------------------
    It is unfortunate that the news reports of his testimony 
interpreted the ``vice versa'' to mean ``What's good for the GM is good 
for the country'' and published only this interpretation as a direct 
quote. Other accounts attribute Mr. Wilson confiding to GM's general 
counsel that what he had meant by his ``vice versa'' was ``and what is 
bad for the country is bad for GM.'' \4\
---------------------------------------------------------------------------
    \4\ See Robert A. Nitschke, The General Motors Legal Staff 1920-
1947 (Detroit, Mich. 1989) p. 41.
---------------------------------------------------------------------------
    Mr. Chairman, If you and your congressional colleagues and the 
Executive Branch can find a way to bring together the appropriate 
Federal Government statistical agencies into a system designed to 
provide the information that Federal and State governments need to make 
appropriate policy and legislative decisions, I can say without 
hesitation that what is good for our country will also be good for all 
elements of our society--including businesses such as General Motors. I 
also contend that taking no action and allowing the current disaster to 
turn into a disgrace would harm all elements of our society.
    Let's optimistically hope that at the beginning of the 21st century 
we will be celebrating, among other things, the establishment of a 
Federal Statistical System that better serves the needs of our society 
as we move into the uncertainty and opportunity of the next century.

    Senator Brownback. Because he has a plane, will you mind if 
quiz Mr. Barabba just a couple of times?
    Ms. Norwood. Yes, go right ahead.
    Senator Brownback. Why are our numbers not considered that 
good by consumers of statistical data when they compare us to 
other industrialized countries?
    Mr. Barabba. I am not familiar with who made that 
evaluation, but it kind of startled me. I do not know who said 
that, but in some areas I cannot imagine that anybody would be 
doing it any better--in some areas. In some areas, we are not 
quite good. Maybe Janet has more--I have been out of the 
business for a while on the government side, so maybe Janet 
could respond.
    Ms. Norwood. I think that summation was done by looking at 
the number of times there were revisions, particularly in the 
national accounts. And the number of times that there are 
revisions is not a demonstration of the quality of the data. It 
is a question of how frequently, how quickly you get data out. 
In this country, because of the uses of data, there has been a 
push to get data out very quickly with incomplete information 
available to people with the understanding they would be 
revised when more complete data became available. And the 
article which ranked countries found that agencies which 
delayed putting data out, of course, revised them less 
frequently.
    Generally I think people recognize that data produced by 
the statistical agencies of this country, particularly the most 
important ones, are really still, compared to those of other 
countries, quite good. The issue is not whether the quality of 
our data are adequate now. The issue is where we are going to 
go in the future as the world becomes more complex and, more 
importantly, where other countries are supporting statistical 
infrastructure, making it more efficient, and paying a great 
deal more attention to it. The European Community, for example, 
is integrating data across the European Union, and other 
countries are moving to improve their data systems at a much 
greater rate than we are in this country.
    Mr. Barabba. I would just add to that, at General Motors we 
have businesses across the globe, and the information we get 
from the U.S. Federal Statistical System, and even some of the 
States, is as good as anything we see anyplace else we go.
    Senator Brownback. Good. Mr. Barabba, you have been past 
president of the American Statistical Association. Would that 
group generally support a consolidation of some of these 
functions? Or can you speak with any sort of certainty----
    Mr. Barabba. Janet also was a president of this 
association. She can probably add to this. But I would be 
surprised--well, first of all, I have never found anything upon 
which the American Statistical Association had unanimous 
agreement.
    Senator Brownback. I am not asking unanimous agreement. If 
we could get 50 percent plus 1, we will----
    Mr. Barabba. You would find a lot of support within the ASA 
for this kind of activity, but there would be some people who 
would be concerned because consolidation to them means too much 
central control. And this is an issue, I think, that both 
Senator Moynihan and Congressman Horn referenced the issue and 
that Janet also alluded to it. This is not just the statistical 
activity. This is an activity that is generating information in 
a form that policymakers can make use of. And some 
statisticians--not all, but some statisticians really focus on 
the statistic, sometimes at the expense of its use. So though I 
would certainly think it would be important to have the support 
of the American Statistical Association as well as the American 
Demographics and American Economics Association, those also are 
vested interests. So I would not let that be the sole guidance 
of the direction of this commission.
    Senator Brownback. Ms. Norwood, on that question?
    Ms. Norwood. I think I will leave it at that.
    Senator Brownback. OK.
    Mr. Barabba. She will let me deal with our colleagues now 
that I said it. [Laughter]
    Ms. Norwood. I would say that I think there are many forces 
and many different ideas within the statistical community, as 
there are in the economics community, and in some of the 
population and social data groups. But everybody wants better 
data. Everybody wants data that are more relevant, and I think 
those issues could be worked out.
    Senator Brownback. Mr. Barabba, one final question before I 
let you catch your plane. You heard Senator Moynihan talk about 
at least consolidating the economic population and resource 
gathering devices and dissemination groups. Would you generally 
agree with that point of view, or do you not want to be heard 
yet on where you would pull things together?
    Mr. Barabba. I would think that--and I agree with what 
Janet expressed in her book. I think that is a very good start. 
I think there are some very, what we would refer to in private 
enterprise as some low-hanging fruit out there that we could 
really pick off and take advantage of. But how you go about 
doing that requires a lot of very interesting, thoughtful 
thinking because there may be other agencies that should be 
incorporated into that as well, and so you would have to think 
about that a lot. But my biggest concern would be not focusing 
on the organization structure, but focusing on what we expect 
the outcome of this consolidation to be, and that requires a 
pretty good understanding and agreement that the role of these 
agencies would be to provide information about the system of 
government and the system of society, and not just functional 
information that allows one part of it to get better 
information than another part.
    Senator Brownback. I am gathering from your statement 
earlier you are saying it is not enough just to put these 
together for cost efficiencies and to gather the same sort of 
data and disseminate it. We need to step to the next wave of 
statistical type of information that can be more usable in 
looking forward rather than just historical. Do I understand 
that point of view?
    Mr. Barabba. Absolutely. If I could draw a comparison on 
the private side, if, say, in a company that produces vehicles 
one hand of the company would say we are going to make a claim 
that we are going to satisfy you no matter what it takes, and 
then another part of the company says we want to reduce costs 
and let's say service is part of cost. So you make a claim that 
we will satisfy you, no matter what. A person purchases the 
vehicle. Something goes wrong. They bring it in, and they say I 
am not satisfied. The person who did not make the claim but who 
is held accountable for reducing the cost says, ``But we do not 
fix that.'' At that point two independent people, both meeting 
the requirements of their superiors in their functional area, 
are doing their job, but the customer is not satisfied.
    If you took someone on welfare and the same kind of 
situation, and then you asked them about their needs for 
health, education, and other services, they would say, gee, I 
find agencies out there asking me to do different things, some 
of which are contradictory.
    Now, if the information design is on the old system, then 
you are going to create strong functional silos of information 
rather than an information system that lets you address the 
needs of that person who needs help.
    Senator Brownback. That is a good way to put it. Thank you 
very much for joining us. If you have further statements----
    Mr. Barabba. I have some more time. I will just stay until 
the end of Janet's presentation, if it is OK with you, Mr. 
Chairman.
    Senator Brownback. That would be fine by me.
    Ms. Norwood, we look forward to your presentation. You have 
the floor.

 TESTIMONY OF JANET NORWOOD, SENIOR FELLOW, THE URBAN INSTITUTE

    Ms. Norwood. Thank you, Mr. Chairman. You have my 
testimony. Let me just summarize a few points.
    First of all, I think we would all agree--and both 
Congressman Horn and Senator Moynihan made the important point, 
which is we need professionalism, we need objectivity, we need 
freedom from political control. As we meet today, there are a 
number of controversial issues. Certainly you are aware, I 
know, of the most important ones. One is the decennial census 
and the use of sampling. There are a number of critical issues 
involved in that, and there are several studies by the 
Committee on National Statistics. I am a member of that 
committee, and I served on one of those panels, and I also 
served on a so-called blue-ribbon panel of the American 
Statistical Association, which decided that sampling was 
indeed, as Senator Moynihan said, a useful tool for the census 
as well as for other products.
    The Consumer Price Index is an issue as well, which we have 
been hearing a great deal about. Part of the problem, I think, 
is that the American public is not very understanding or very 
knowledgeable about statistics, in part because of the 
separation and the compartmentalization of these agencies. They 
do not understand why the Census Bureau cannot count. It seems 
very clear that it ought to be able to count. Why is it, 
therefore, that we undercount? They do not understand the 
difference between a cost-of-living index and a price index, 
and they do not see what all this controversy is about. Too 
many of the people in the country, sometimes even public 
officials in the Executive Branch and the Congress, tend to use 
data and look at data in ways that suggest that if they want 
them to go up and they do, they are good. If they want them to 
go down and they go down, they are good. So it depends. You 
cannot please everybody. I learned that a long time ago. People 
who pay always wanted the index to go down, and people who 
received payments always wanted the index to go up.
    We have problems of that sort, and I think that suggests 
that it is useful to have a place in the sun, to have an agency 
that has respect, that is professional and objective, that can 
pull together the kinds of issues that need development.
    I have looked at this a great deal. I was Commissioner of 
Labor Statistics for 13.5 years, and while I was there, I 
thought a great deal about whether the Bureau was better off in 
the Labor Department or whether it should be elsewhere. I had 
mixed feelings about that, mainly because I wanted to be 
certain that the data that we produce are relevant to policy 
issues.
    Since that time, I have had a chance to look at the rest of 
the system and recognize that most of the other agencies have 
very little relationship to the departments in which they are 
located--very little relationship, that is, in terms of the 
data needs of those programs. They are not sufficiently high in 
the structure to be able to know what the programs are or what 
the program needs should be.
    In my view, a statistical system should be able to define 
the problems, not to develop policy solutions but to identify 
the issues which need policy direction--or policy solutions. 
And to do that, we need to have a system that is large enough 
and strong enough to use state-of-the-art techniques, to have 
the professional respect that is needed in this country, to 
have the opportunity to educate people on these needs, but 
also, I think, a system which can be related to policy needs.
    Last year, I was at a European conference on statistics in 
Bologna, Italy, and I had the opportunity to spend some time 
with the head of the new British Statistical Service. One of 
the issues that we discussed over dinner was how the British 
were trying to look at just that issue, could some of their 
people sit at and work with the Department of Labor and with 
the other Ministries and still be a part of the central 
statistical organization.
    I think all that needs to be worked out. It is for that 
reason that we should not move so rapidly that we ignore the 
needs that exist. I do believe that it is time for us to move. 
I believe that it is time for us to move toward a gradual 
integration.
    Why did I pick out Census, BLS, and BEA? For several 
reasons, one being that if you tried to move the entire system 
together, it would fall of its own weight. It would be so 
enormous an instrument, so large a group in government, that I 
think it would not work very well.
    Second, I do not think we should underestimate the 
stakeholders involved. I have dealt with many of them. I think 
that the Senator and the Congressman suggested that a 
commission might help to develop public support. That is 
certainly probably a very good idea.
    The Bureau of Economic Analysis is frequently not discussed 
very much at all, and yet when you think about it, it is one of 
the smallest agencies of all, but it is one with probably the 
most difficult job to do and the broadest job to do. It has to 
measure the entire economy, and in order to that, it has got to 
rely on data that are produced in other agencies. And it is for 
that reason that it seemed to me that if we are going to move 
toward a gradual centralization, we ought to bring those three 
agencies together.
    I would also argue, as I testified before Congressman Horn 
on his bill, that the new agency must really be responsible for 
statistical quality standards, for definitional standards, for 
representing the system abroad. And that means that much of the 
work that is done very well by the small group under the chief 
statistician at OMB would have to be moved with it.
    I should say, Mr. Chairman, that no organizational change 
will solve the particular problems of the Census. Changing the 
location of the Census Bureau is not going to solve the problem 
of the undercount or how we deal with it. Changing the locus of 
BLS is not going to affect what is done on the CPI. In fact, if 
we move very rapidly without thinking through what we do, we 
could derail some of the work that is going on in those two 
agencies to improve the data system, and I think we should be 
very careful about that.
    In this country, we have a habit of looking at something, 
dealing with it by jumping into it, and sometimes that is a 
very good thing. In this case, my fear is that we may 
reorganize the statistical system because of a desire to do 
some other kind of reorganizational structuring in some other 
agency, or even worse, because we believe that the way to do 
this is to reduce budget. We want only to get a lot of money 
out of this.
    I would argue that we can have savings in the long run, but 
that it will take many years to work them out rather carefully. 
I have some ideas about how that can be done, but it is not 
going to happen overnight. It is not going to happen in the 
first few years because a great deal of research needs to be 
done.
    So all in all, I believe that the time has come for us to 
move toward a gradual organizational change. We should be 
careful how we do it. And we should make sure that we examine 
all facets of it.
    [The prepared statement of Ms. Norwood follows:]

                 PREPARED STATEMENT OF JANET L. NORWOOD

 (Any opinions expressed herein are solely the author's and should not 
    be attributed to the Urban Institute, its officers or funders.)

    Mr. Chairman and Members of the Subcommittee: I appreciate this 
opportunity to comment on the state of our Federal statistical system 
and on methods to improve it. As you know, I spent much of my 
professional life at the Bureau of Labor Statistics. I served three 
terms as Commissioner with appointments from both Democratic and 
Republican Presidents. I have worked closely with all of the Federal 
statistical agencies, and I know the importance of their work.
    I approach these issues with several strong convictions. First, I 
am convinced that the effective operation of democracy requires that 
our citizens have access to an accurate and objective data base of the 
highest possible quality, one that is clearly relevant to the policy 
issues that confront them. Second, those responsible for producing the 
nation's statistical data base must be professionally competent and 
completely free from political interference. They must work in an open 
environment in which all methodological changes are fully explained and 
freely discussed. Third, an effective statistical system must be 
grounded in an institutional and legal framework which provides the 
authority and public credibility to permit the setting of priorities, 
the protection of confidentiality, and the flexibility to conduct 
research for improvement. And, finally, our statistical series must be 
based on sound concepts, tested methods, and state-of-the art 
statistical techniques.
The Current Environment for Statistics
    The data produced by government statistical agencies affect many 
critical policy decisions, and it is important that they be produced 
efficiently, accurately, and objectively. In recent months, programs of 
the two largest, general purpose statistical agencies in the system--
the Bureau of the Census (Census) and the Bureau of Labor Statistics 
(BLS)--have been the subject of controversy in the Congress and the 
press. While public discussion of statistical issues is always useful, 
the current criticism of the Consumer Price Index and of the planning 
for the 2000 Census demonstrates the difficulties government agencies 
face when they produce statistics which affect the lives and incomes of 
a large part of our population. People do not understand differences in 
concept--between a cost-of-living index and a price index--and they 
become impatient with a Census Bureau which finds it hard to count all 
of the people in the country with complete accuracy. These criticisms 
are symptomatic of the challenges that face all of the official 
statistical agencies This is a time when we must reinforce the 
objectivity and the professionalism of the agencies as we support their 
search for new survey and compilation methods. But it is also a time 
when we must find ways to improve their credibility with the public and 
their operational efficiency.
    We live in a period of great change, when survey operations are 
becoming increasingly more complex. The environment in which Federal 
data producers operate today is very different from the past, in part 
because our citizens have begun a fundamental rethinking of the role of 
government in our society and the need for information for decision-
making. The policy implications of statistical information have become 
more significant even as statistical agencies have found it more and 
more difficult to introduce state-of-the-art techniques to ensure the 
quality and relevance of the data they compile. And yet, the need to 
seize the opportunity for use of new statistical techniques and 
technological innovation has never been greater. The use of federally 
produced data by both the private and the public sectors has increased 
exponentially; they drive many public policy decisions and they affect 
family life. Our statistical data have become increasingly complex, 
much harder to collect, and much more expensive than in the past. Even 
more difficult is the need to keep up with a world which is constantly 
in a state of change. The Administration, the Congress, and the public 
all play an oversight role, but the risk is that those responsible for 
public policy may insist on techniques which produce data that are 
higher or lower--depending on their policy views--rather than those 
which will provide the best estimates of that which we are trying to 
measure.
Statistical Agencies in the Department of Commerce
    As this hearing takes place, the Bureau of the Census is hard at 
work planning and testing methods to improve the 2000 Census of the 
population. A number of Congressional committees have been reviewing 
this work, and several panels of the National Academy of Science's 
Committee on National Statistics have been advising the Bureau. I am a 
member of the Committee and served on one of those panels. I also was a 
member of the American Statistical Association's Census 2000 Blue 
Ribbon Panel which pointed out that sampling ``. . . can be an 
appropriate part of the methodology for conducting censuses.'' (ASA 
1996). Like all previous Censuses, the 2000 Census will incorporate 
new, more modern collection techniques and may, therefore, provoke 
considerable public debate.
    The other statistical agency in the Department of Commerce, the 
Bureau of Economic Analysis (BEA), one of the smallest agencies in the 
Federal system, has perhaps the largest and most difficult job of all--
compilation of the national accounts. Since our Gross Domestic Product 
must be built up from data produced by a large number of agencies, BEA 
must, of necessity, rely on data produced by many of the other parts of 
the Federal statistical system.
Current Statistical System
    The two agencies in the Commerce Department are part of a 
decentralized system consisting of more than 11 separate agencies 
located in 9 different executive government departments; and some 70 
other agencies of the government produce statistical output as a part 
of their programmatic responsibilities. We have a statistical system 
that is more decentralized than that of any other large country. 
Moreover, the group at OMB which coordinates the system is one of the 
smallest in the world.
    In recent months, Katherine Wallman, OMB's Chief Statistician, has 
been successful in spearheading completion of a new North American 
industrial classification system, and in sponsoring monthly meetings of 
the statistical agency heads. OMB has also developed a legislative 
initiative to standardize confidentiality and to permit the exchange of 
data for statistical purposes among the major Federal statistical 
agencies. Passage of that legislation would help considerably to 
strengthen the system. In addition, the OMB Director has requested 
development of a statistical budget so that resources devoted to 
statistics can be looked at across the entire system.
The Statistical System Problem
    In spite of this progress, however, we must ask whether the current 
structure of our Federal statistical system is efficient enough to 
provide the kind of information base needed for the social and economic 
challenges of a democracy. Will our present system, as currently 
organized, be able to develop the kind of integrated data base required 
to solve the complex, multifaceted policy issues we face? When we 
compare our Federal statistical system to those of other countries, we 
find that our system, although much larger in size and scope, 
nevertheless, has more problems in operational efficiency and in public 
trust than they do. Our system is one that seems disjointed, with 
multiple data bases and little integration. Although we continue to 
produce data of relatively high quality, the risk is that statistical 
information will suffer as agency heads try to deal with the demands of 
their parent departments in a period of serious budget constraint.
    Why do we have these problems? Is it because the environment in 
which the data producers operate has changed so dramatically? Or are 
there problems inherent in the system itself which need repair? I 
believe that both of these forces are at work. They must be addressed 
if we are to succeed in the development of the objective system of 
information that is so crucial to democracy in our country.
A Fresh Look at the Problem
    It is time for us to take a fresh look at the organization of our 
Federal statistical system. We heave multiple statistical agency heads 
with no one having sufficient authority, resources, power, and public 
prestige to make the entire statistical system function as efficiently 
as it should. Those heading statistical groups at OMB and other Cabinet 
agencies are dedicated and competent, but they work in a variety of 
different situations and at different levels within their own 
departments. Several of our laws, especially individual agency 
confidentiality legislation, make it difficult to engage in cooperative 
research and to benefit from economies of scale. We have insufficient 
long-range planning and budgeting of data products and of data 
production.
    The system is too slow to adjust data to changing economic and 
social conditions, in part because investment in data is generated only 
when the statistics reflect deteriorating conditions; once improvement 
occurs, the public's interest in the production of data of high quality 
tends to disappear. And there is insufficient investment in 
coordination in a system that is heavily decentralized. Data priorities 
are all too often determined almost entirely among programs within each 
sponsoring agency instead of across the statistical system itself. This 
compartmentalization continues into the Congress as statistical 
activities are spread among a large number of different Congressional 
committees.
    None of these conditions is new. The Pratt Commission was the first 
to study the need for statistics in 1844. In the century and a half 
since then, we have had some 15 more committees or commissions to study 
the nation's statistical system. Although each commission differed in 
emphasis, every one of them struggled with the same critical questions: 
centralization vs. decentralization, location and power of the 
coordination authority, protection of confidentiality and reduction of 
respondent burden, as well as with problems of information 
dissemination and data integration. Unfortunately, the interest which 
generated appointment of each commission was generally short-lived. By 
the time the reports were issued, the commission was forgotten, and 
there was no sustained interest to bring about change. Thus, in spite 
of a long series of studies, the Federal statistical system remains 
relatively unchanged either because of a lack of popular support for 
statistics, because of bureaucratic inertia, or because of the 
unwillingness to upset stake-holders with a particular interest in 
retaining the status quo.
Steps Toward Centralization
    The United States has neither the benefits that come from strong 
centralization of a statistical system nor the efficiencies that come 
with strong and effective coordination of a decentralized system. Our 
existing system will find it increasingly difficult to meet the demands 
for data from an increasingly technologically advanced and globalized 
world. We should move carefully and gradually toward greater 
centralization of the system. In a recent book, Organizing to Count: 
Change in the Federal Statistical System (Norwood 1995), I have 
outlined a plan for a new statistical agency that would house the two 
large multi-purpose statistical agencies--the Bureau of the Census and 
the Bureau of Labor Statistics--as well as two smaller groups--the 
Bureau of Economic Analysis and OMB's Statistical Policy Branch. This 
new agency would collect, compile, analyze, and disseminate statistical 
information and, at the same time, set quality and classification 
standards and provide oversight over statistical work done elsewhere in 
the government. Census and BLS would make up the core of the new 
agency, an arrangement that would permit development of a comprehensive 
effort to evaluate existing data sets and engage in research for 
efficient survey design aimed at the elimination of duplication and 
overlap. This work must be done with great care because much of the 
data produced are among the most sensitive and critical of all the data 
produced by the government. A new confidentiality protection law would 
be part of the package creating the new agency, so that protection of 
data collected from respondents with a pledge of confidentiality would 
be uniform across the system. I have provided a more detailed 
explanation of the organization and functioning of this Central 
Statistical Board in my recent book. What I suggest is a first step 
toward the consolidation of the Federal statistical system into a 
coherent and efficient agency of our government. This is the route that 
the British have taken, and they have done it with great success. The 
National Statistical Service of the United Kingdom is by now a well-
established group. It started with the coordinating authority and the 
national accounts, then added labor statistics, and then demographic 
and census operations. We can learn from their experience.
Conclusion
    In conclusion, let me emphasize several important points. First, 
the recent criticisms of the plans for the 2000 Census and of the 
Consumer Price Index are useful issues for discussion but they must not 
be allowed to result in politicization of the process of compiling 
these important data programs. Second, the nation's statistical system 
needs to operate more efficiently and more effectively. Third, we must 
consider very carefully how any proposed changes would affect the data 
systems upon which the whole nation depends. Fast re-engineering of the 
nation's statistical system without sufficient research and thought, 
carried out simply as a by-product of other government reorganization 
or only as a means to reduce statistical budgets, will surely damage 
the quality and the relevance of many of our most important statistical 
series.

                           List of References

    American Statistical Association, Report from the Census 2000 Blue 
Ribbon Panel. 1996.
    Committee on National Statistics, Modernizing the U.S Census, Barry 
Edmonston and Charles Schultze, eds. National Research Council. 
National Academy Press. 1995.
    ------, Counting People in the Information Age, D.L. Steffey and 
N.M. Bradburn, eds. National Research Council. National Academy Press. 
1994.
    Norwood, Janet L., Organizing to Count: Change in the Federal 
Statistical System. Urban Institute Press. 1995.

    Senator Brownback. Thank you, Ms. Norwood, for your 
presentation, and your testimony. Your background speaks 
volumes about your ability to articulate and speak with 
knowledge on these issues.
    Let me ask you something about the speed of change, because 
both of you, as I gather, endorse change, endorse substantial 
change. Don't let me misstate your positions.
    Ms. Norwood. No. That is correct.
    Mr. Barabba. We agree.
    Senator Brownback. But you question how fast we move at 
this. It has been my observation, Mr. Barabba, that your 
statement about big government and big business being pretty 
similar is pretty accurate. Both move as big entities and 
frequently move slowly. But when big business went through most 
of its reorganization in this country, it moved quite rapidly 
in moving forward under the theory, at least from folks I have 
talked to, that if you move slowly you are going to get stopped 
because of either the inertia of the bureaucracies or the 
special interest groups that support the current operation. The 
key was to move wisely but rapidly. What do you think about 
that in this setting here? Because it does bear upon how you go 
at this reorganization.
    Mr. Barabba. If you put wisely before rapidly, I would 
agree with that. Take, for example, all the movement that was 
made on--the buzz word--re-engineering and reinventions of 
companies. Most of those activities by any analysis says that 
most of the endeavors did not achieve what they set out to do 
and because everybody wanted to make a fast decision.
    The thing that we have found is that it is not how fast you 
make the decision to do something, it is how fast you implement 
it. And I think what we are saying in this case is that we 
would be able to implement significant change if, in fact, we 
spent a little time making sure what we wanted that change to 
accomplish. It would be--and I know in our own case, in our own 
company, we for a long time made very fast decisions. But then 
as soon as you found out the ramifications, after a little bit 
of analysis and attempt at implementation, you went back to the 
decision time after time and that kept delaying the 
implementation even longer.
    Our experience has been is if we walk a little bit, we find 
out we can run a lot faster.
    Senator Brownback. Ms. Norwood, what about the idea that 
you can do this within a time frame, if you establish a 
commission, it reports out, and you could actually implement or 
pass implementation legislation this Congress if we move 
forward expeditiously, as Senator Moynihan was suggesting in 
the earlier panel? Is that too fast?
    Ms. Norwood. I think it would depend on what the commission 
came up with, obviously. I would not want to prejudge that. It 
might not be too fast. But I think we should be clear about the 
complexity of the task. It is not just a matter of taking a few 
agencies and saying we will pull you from here and from there 
and put you together. The question is how you put them 
together. The question is how you integrate programs. A great 
deal of work needs to be done looking at data integration.
    There is another question, if I may say so with due 
deference, of the way in which the Congress is organized in 
order to provide oversight to statistical agencies. I once 
spent some time trying to count up the number of congressional 
committees with oversight over the Federal statistical system. 
And I will tell you that there are an enormous number of them, 
and each of them looks at a little piece of the system but not 
at the total system.
    If we move toward an integration, even a gradual 
integration, of the statistical system, we have got to look at 
that. My experience has been that changing the Congress takes 
even more time than for the Congress to decide to change the 
Executive Branch.
    So I think there are a lot of complex issues that we need 
to think about. If this is to be an independent agency, exactly 
how will it relate to the Congress? How will it relate to the 
Executive Branch? How will its budget and its personnel be 
handled? These are details, but they are very important 
details. And I would hope that a commission, if one were 
appointed, would have people who are knowledgeable enough to 
look at things of that sort.
    Senator Brownback. And you would support an independent 
agency, independent entity with a civil servant head? Or are 
you willing to make that type of statement?
    Ms. Norwood. Yes, I would like to see someone with a long 
fixed term of office heading it who was qualified. I do not 
know what you mean by a civil servant. I was a civil servant, 
but I was a Presidential appointee with the consent of the 
Senate. I believe that this is an important enough position to 
be a Presidential appointment with the consent of the Senate.
    Senator Brownback. What I think most people look at is 
whether you have a long enough term and a qualification 
requirement. You mentioned how much of government looks at 
these statistics and depends upon actions based upon these and 
how much of the private side, markets are driven. I mean, my 
little neck of the woods that I know about, as far as the 
agricultural sector, you release those reports and the markets 
move.
    Ms. Norwood. Yes, I know that.
    Senator Brownback. And my dad complained about it every 
time we put one out, saying, ``I wish you guys would just quit 
putting out those dang numbers.'' Because he thought we were 
counting cattle by how many legs they had instead of how many 
head of cattle were there and driving his market down.
    Ms. Norwood. Well, you know, I am currently serving on a 
board of a very large international bank, and I am chairing the 
board of directors committee on risk assessment. And what we 
are doing is essentially looking at data in order to determine 
the decisions that we make or that the bank officers make in 
their business judgments. So I recognize that this is certainly 
very multifaceted.
    I think that we need to be careful, and that we let the 
improvement of the CPI and of the Census go along as planned. I 
would not want to see those derailed because suddenly we moved 
the agencies.
    Second, it is fine to say these agencies should be put 
together, but how do you do that? Are you still going to have 
the Census Bureau out in Suitland and BLS over here just a few 
blocks away and BEA somewhere else in the city? Or are you 
going to put them together? And how are you going to relate 
this agency to OMB and to the Congress?
    Those are issues which really need some considerable 
thought. There is the GAO example of a 12- or 14-year term for 
the Comptroller General, and that is a semi-independent agency. 
There are other examples in government.
    The important thing is that we need to move, I think with 
some deliberate speed, but I emphasize the word ``deliberate,'' 
and with some wisdom and in a non-political manner.
    Senator Brownback. You remind me of the old Russian 
proverb: The slower I go, the further I get. I do not know that 
that applies to government. I am still mixed in my opinion on 
that. But thank you both very much. You have a tremendous 
amount of expertise and background and have obviously thought 
about this a great deal. If you have additional thoughts, 
either flying back or at another time, please let us know as we 
consider that in moving forward.
    Thank you very much.
    Ms. Norwood. Thank you.
    Mr. Barabba. Thank you.
    Senator Brownback. The next panel will be Maurine Haver, 
the past president of the National Association of Business 
Economists; Dr. Leonard Nakamura, Economic Adviser for Federal 
Reserve Bank of Philadelphia; and Nye Stevens, the Director of 
Federal Management and Workforce Issues, General Accounting 
Office.
    I do not know if the three of you have discussed who should 
go first or if we have particular problems. If not, I am going 
to go with Ms. Haver first on the list, unless you all--does 
anybody have any scheduling difficulties? If not, then, Ms. 
Haver, the mike is yours.

  TESTIMONY OF MAURINE A. HAVER, CHAIR, STATISTICS COMMITTEE, 
          NATIONAL ASSOCIATION OF BUSINESS ECONOMISTS

    Ms. Haver. Thank you. I would like to have my written 
statement put in the record, and I will summarize some of my 
thoughts.
    Senator Brownback. Without objection.
    Ms. Haver. First of all, I am representing the National 
Association of Business Economists, and we have been supporting 
a single agency for at least 3 or 4 years now. We came to the 
recognition that we needed consolidation because we have 
realized that, despite the fact that the U.S. statistics really 
remain among the best in the world, we have been lacking the 
investment in our statistical infrastructure to keep our 
statistics up to date. And this is what Janet was alluding to 
when she talked about the European Community investing in their 
statistics. They now have industry statistics that actually 
represent the growth industries of today, whereas the United 
States is truly lacking detail in services and high-tech 
industries, the industries of the information age.
    In fact, on some of the statistical tables that we look at, 
the most important item is the one listed as ``all other.'' 
This, I think, is due in large part to the fact that our 
budgets have simply been growing at a rate that perhaps covers 
wages and rent escalations. But if you look at the past budget 
increases of BLS, BEA, and Census, taking out for a moment the 
decennial census and periodic programs, they have been very 
flat for a number of years. And these have not allowed for the 
research and development that is sorely needed to expand the 
scope and to improve the quality of our statistics.
    So as we look at it, we have to make the system as 
efficient as possible, and we feel that will be achieved 
through consolidation.
    Now, we agree with both Senator Moynihan and Representative 
Horn that we should start with BLS, BEA, and Census. These are 
the core statistical, economic statistical functions. They also 
are general purpose agencies that do not have the strong 
special interest groups such as agriculture. I gave a talk on 
consolidation in Omaha, and I walked away from that talk 
realizing that we must start small if we are going to 
accomplish something. And I feel we can accomplish a great deal 
for those economic data that drive our financial markets, the 
data on which businesses make many decisions, if we restrict 
ourselves to these three agencies.
    NABE sent a letter on February 24 to the President, to many 
of you on the Hill, stating our position and our 
recommendation. We recommend that the new agency be an 
independent one, that it report to the Office of the President, 
and that it be headed by a Statistician General who would serve 
for 7 years. We think it is very important to get the 
statistical agency outside of any of the Departments of Labor 
or Treasury or Commerce.
    One of the problems in improving our system has been a 
problem of response rates from businesses and individuals. And 
certainly I think businesses are more concerned with 
confidentiality and the existence of statistical agencies in 
organizations that also have enforcement areas. Whereas if the 
statistical agency is separate, I think you would find in 
business a great willingness to provide the data.
    Also, within a single agency, we would envision a single 
reporting form so that you would not often have to report the 
same data twice. Right now, because of the confidentiality 
between BLS, BEA, and Census, oftentimes they do not have the 
right to see the data that are collected perhaps by Census.
    In a meeting at BLS recently, we learned HHS is starting a 
survey on health benefit costs of companies. Well, the BLS 
already goes to companies as part of their employment cost 
index program and requests information on health benefits. I 
think that consolidation could also mean much less respondent 
burden, and that would be important to our business members.
    Finally, just one last point, and I believe it is a very 
important one. Consolidation without data sharing does not 
matter. If we do not have data sharing, there are really very 
few benefits to consolidation. Now, there is a bill proposed by 
the executive, I think, from OMB for data sharing among the 
statistical agencies BEA, BLS, and Census, and I believe the 
remaining 12 statistical agencies, and also with the Treasury. 
This is very, very important. And I know the response NABE got 
back from the White House about our proposal on consolidation 
was that data sharing would solve the problems that we have 
now.
    We really feel data sharing is important. It is a first 
step, but data sharing is not going to realize all the 
efficiencies and all the advantages of one unified consolidated 
system.
    Thank you.
    [The prepared statement of Ms. Haver follows:]

                 PREPARED STATEMENT OF MAURINE A. HAVER
    I am Maurine Haver. Today I am speaking in my capacity as the chair 
of the Statistics Committee of the National Association of Business 
Economists. (NABE).
Economic statistics are important to every American.
    Statistics produced by our Federal statistical system are vital to 
the functioning of our market economy. Businesses make decisions about 
where to locate a plant, how much to produce and how much to pay their 
workers based on data provided by the Bureau of Labor Statistics (BLS), 
the Bureau of Economic Analysis (BEA) and the Bureau of the Census. 
Participants in financial markets make investment decisions which in 
turn affect interest rates, the stock market and the value of the 
dollar. These data also serve as critical inputs into the formulation 
of monetary, fiscal and trade policy. In short, the quality of our 
economic statistics impacts the lives of every American.
Is the present system producing what we need?
    While U.S. economic statistics remain among the best in the world, 
lack of investment in our statistical infrastructure has left us with a 
system that does a better job of measuring the industrial economy of 
the past than the information economy of the present. Services and high 
tech industries are not well covered. The most important industry in 
some statistical tables is now the one labeled ``all other''. Budgets 
for statistical agencies barely cover mandated wage escalations. Funds 
for research and development are sorely needed to expand the scope and 
improve the quality of our statistics so they remain relevant in a 
rapidly changing economy.
    The European Union is devoting considerable resources to the 
development of statistics for the 21st century. Its industry statistics 
have been reorganized to reflect new industries. U.S. agencies are only 
beginning a comparable effort. Adoption of the new system--the North 
American Industrial Classification System (NAICS)--is long overdue. 
Funding for the NAICS is provided for in the President's budget, but 
unfortunately these resources may not be adequate to develop comparable 
history for business analysis.
    NABE members have been very concerned for many years about the 
quality and availability of economic data for business decisionmaking 
(see attached recent member survey results). But only one economic 
statistics program--the consumer price index--has received attention 
among Members of Congress. While measurement error in the CPI clearly 
impacts the Federal Government budget, mismeasurement of other 
important indicators can affect the actions of the Federal Reserve and 
businesses causing the economy to perform below potential. This 
underperformance also negatively impacts the Federal budget and the 
welfare of all Americans.
    Some statistical programs have been terminated altogether. Many of 
these programs measured economic activity in local areas and were 
especially important to small and medium-size businesses without a 
national presence. Nonresidential building permits by State and 
locality were terminated by Census last year. Monthly retail sales by 
State and metro area were discontinued in January.
    Other terminated programs were important for policymaking. For 
example, information on business spending for pollution abatement is no 
longer collected. If Congress considers any changes to clean air or 
water regulations in coming years, current data will not be available 
on business expenditures required to meet existing regulations.
NABE recommendations.
    The time has come to organize our economic statistical system so it 
can operate as efficiently as possible and so that decisions affecting 
statistics are made within the context of the complete economic 
monitoring system.
    NABE recommends that our major economic statistical agencies--BLS, 
BEA and Census--be consolidated into a high-level statistical office 
similar to those in other industrialized countries like Canada and the 
United Kingdom. This new agency should report to a Statistician General 
of the United States who would report to the President. The 
Statistician General would serve for seven years and be removable only 
for cause. He or she would have clout in the fight for resources and 
could resolve questions of statistical policy free of pressure from 
political and special interests. The Statistician General would be 
responsible for setting system standards and the consolidated agency 
could provide technical assistance to special interest statistical 
groups within other government agencies.
    Statistical agency consolidation alone will not realize the 
efficiencies that must be achieved, and we would caution supporters 
consolidation not to expect immediate budget savings. Consolidation is 
important because it starts the process of functional integration and 
will bring decisions on program tradeoffs together so that more 
rational decisions are made.
    Data sharing among these agencies is critical and would solve some 
of the problems that arise from our fragmented system. NABE strongly 
supports pending legislation which would provide for data sharing among 
statistical agencies and with the Treasury which is now often 
prohibited by law. However, we do not believe this legislation goes far 
enough. Consolidation is needed.
    The business community and financial markets derive significant 
benefits from the collection and dissemination of economic data. 
Complaints of respondent burden are often misinterpreted. Many 
businesses are ready to provide data but object to rigid reporting 
requirements which preclude the submission of computer-generated 
reports. Some progress has been made in allowing for computer input but 
no single standard exists among the agencies. Duplication of requests 
due to confidentiality barriers which cause different agencies to 
request identical or similar data is another familiar complaint. We 
hope consolidation would free resources for the creation of one 
automated reporting standard which we believe would lower costs and 
improve response rates.
    NABE recognizes the importance of balancing the Federal budget. Our 
members have consistently supported a balanced budget since we began 
policy surveys more than 20 years ago. However, we also recognize the 
costs of incomplete and inaccurate information which are well in excess 
of the combined budgets of our major statistical agencies (BLS, BEA, 
Census).
    As Michael Boskin stated in a recent article in NABE's journal 
Business Economics, ``Virtually every major private firm in the world 
is spending heavily on information technology--hardware, software, and 
human capital--and we should not expect better statistics from our 
government agencies without a corresponding investment.
    Problems of data quality or lack of economic information pose heavy 
costs on our society. We must begin a renewed effort to improve our 
statistical system so it can provide us with the information we need to 
make appropriate decisions for the 21st century.

                   NABE Economic Statistics Questions

    From the February 1997 NABE Outlook Survey (36 profession economic 
forecasters responding)

    (A) Budget constraints have led to cutbacks in Federal statistical 
programs. Have these cutbacks impaired the availability and quality of 
information that you use in your work?

          Yes  78%            No  14%            No Response 8%

    As a general matter, are you satisfied with the scope and quality 
of measurement and reporting on the U.S. economy?

          Yes  25%            No  69%            No Response 6%

    (B) Should President Clinton create a blue-ribbon commission to 
review the content of economic indicators as well as the organization, 
management and funding of the agencies that assemble the data?

          Yes  67%            No  19%            No Response 14%

    (C) Would consolidation of Federal statistical activities into a 
new Central Statistical Office headed by a high-level, nonpartisan 
Chief Statistician of the United States help to improve the scope and 
quality of economic measurement and reporting in the U.S.?

          Yes  67%            No  17%            No Response 16%

    From the February 1997 Economic Policy Survey (228 business 
economists responding)

    A. How adequate are the quality, timeliness, and coverage of 
current economic statistics produced by the Federal Government agencies 
for your needs as a business economist? Please circle the appropriate 
number below where 5=very adequate and 1=not at all adequate.

      5-5%      4-39%      3-35%      2-15%      1-4%      No response 
2%

    B. Do you feel the quality of your analysis and decision-making is 
adversely affected by the lack of quality economic data from the 
Federal Government?

          (a) no, the quality of my analysis and decision-making is not 
        adversely affected by the lack of quality economic data       
        29%

          (b) yes, the quality of my analysis and decision-making is 
        somewhat adversely affected by the lack of quality economic 
        data       62%

          (c) yes, the quality of my analysis and decision-making is 
        very adversely affected by the lack of quality economic data     
          9%

        No response       10%

    C. What Federal Government data series (e.g. GDP, retail sales, 
etc.) is the most important for you to have improved?

            % of responses
          32%      Inflation
          22%      GDP
          10%      Employment
          17%      Regional data
          17%      Productivity
           4%      Service sector data
          14%      Retail Sales
          14%      Other

    NABE Membership Survey for a Statistics Action Plan--August 1995

    A questionnaire on economic data quality and possible approaches to 
the improvement of government statistics was mailed to NABE members in 
late July. 394 questionnaire had been returned by the cutoff date of 
August 22. This was a 13% response rate for U.S. members. The 
tabulation below reflects these questionnaires. Over 400 questionnaires 
were received in total.

    1. Which data are most important to you in your work?

          45.9%      U.S. Macro Indicators
          23.9%      Regional
          13.2%      Industry except Agriculture and Energy
           1.3%      Agriculture
          14.1%      Energy
           7.4%      Demographic

    2. How do you view the quality of data you ranked #1?


                                             Total    Macro    Regional  Industry    Agr     Energy  Demographic

              --for accuracy?
Excellent.................................     14.0     14.4       10.6       9.6      0.0     25.0        24.1
Good......................................     44.2     46.4       40.4      38.5     20.0     56.3        55.2
Acceptable................................     33.0     33.7       34.0      38.5     80.0     18.8        17.2
Poor......................................      7.4      5.0       12.8       9.6      0.0      0.0         3.4
             --for timeliness?
Excellent.................................     11.9     19.3        2.1       9.6     20.0     12.5         6.9
Good......................................     35.0     46.4       19.1      25.0      0.0     31.3        34.5
Acceptable................................     35.5     27.6       45.7      44.2     60.0     50.0        44.8
Poor......................................     17.0      6.1       31.9      21.2     20.0      6.3        13.8
              --for coverage?
Excellent.................................     15.0     16.0        9.6       5.8      0.0     25.0        17.2
Good......................................     40.4     44.8       35.1      25.0     40.0     50.0        41.4
Acceptable................................     30.7     28.2       38.3      51.9     40.0     18.8        31.0
Poor......................................     12.9      9.9       14.9      17.3     20.0      6.3        10.3


    3. How do you view the trend?


                                             Total    Macro    Regional  Industry    Agr     Energy  Demographic

              --in accuracy?
Improving.................................     16.8     14.9       16.0      15.4      0.0     25.0        17.2
Stable....................................     53.0     49.7       52.1      61.5     40.0     56.3        55.2
Deteriorating.............................     28.4     34.8       27.7      23.1     60.0     18.8        24.1
             --in timeliness?
Improving.................................     12.9      9.4       13.8      15.4      0.0     25.0        13.8
Stable....................................     64.7     74.6       56.4      61.5     40.0     68.8        55.2
Deteriorating.............................     20.6     14.9       25.5      23.1     60.0      6.3        27.6
              --in coverage?
Improving.................................     14.0      8.8       12.8      11.5      0.0     31.3        10.3
Stable....................................     55.3     56.4       61.7      55.8     20.0     62.5        62.1
Deteriorating.............................     28.7     33.1       21.3      32.7     80.0      6.3        24.1


    4a. Most important step to improve quality of government data:


                                                          Rank 1  Rank 2

  Create a single statistical agency....................    27.9    17.3
  Eliminate obstacles to data sharing...................    24.6    28.2
  Increase funding targeted for R&D and technology......    20.8    12.7
  Make surveys mandatory................................     9.4     9.6
  Increase R&D with current budgets by eliminating
   programs.............................................     1.5     4.1


    4b. Greatest obstacles to accurate and timely government data:

  Budget cuts..............................................       70.1%
  Lack of coordination among agencies......................       67.0%
  Poor response rate on surveys............................       47.7%
  Inability of agencies to share data due to
   confidentiality agreements..............................       45.4%
  Lack of R&D effort.......................................       38.1%


    5a. LDo you support the consolidation of the major 12 statistical 
agencies into one STAT-USA similar to Statistics Canada?


                                                                      Total     Academic   Government   Industry

Yes...............................................................       59.1       50.0         54.4       62.2
No................................................................       12.9        9.4         16.2       11.7
Undecided.........................................................       27.7       40.6         27.9       26.1


    5b. LIf no or undecided, would you support consolidation of only 
BEA, BLS, Census and the Statistical Policy Office of OMB?


                                                                      Total     Academic   Government   Industry

  Yes.............................................................       22.3       31.3         19.1       22.3
  No..............................................................        5.6        0.0          5.9        4.9
  Undecided.......................................................       11.9       18.8         16.2       10.2


    6. What would be the advantages of a single statistics agency?

      75.1%  data sharing could eliminate duplication of surveys
      74.6%  higher productivity--expertise would not be duplicated
      65.5%  ease of locating data
      54.8%  investment in technology could have higher payoffs
      43.9%  data gathering would be separate from regulation
      38.6%  unified budget could be reallocated based on priorities
      34.8%  would have more clout in budget negotiations

    7. What would be the disadvantages of a single agency?

      49.5%  might be less responsive to user needs
      45.4%  priorities might not coincide with needs of business 
economists
      45.4%  easier to politicize data in a single agency
      18.8%  agency with all data would have too much power
      11.7%  data gathering should be in same agency as regulatory work
       6.6%  Lcongressional committees would be less able to control 
allocation of resources

    8. LWould you be willing to pay higher user fees if the money 
collected could be applied directly to data improvement?


                                                                      Total     Academic   Government   Industry

  Yes.............................................................       75.1       90.6         69.1       75.6
  No..............................................................       16.8        3.1         13.2       19.8


    9. LWould your company agree to extend existing confidentiality 
agreements to all parts of a consolidated agency?


                                                                      Total     Academic   Government   Industry

  Yes.............................................................       32.0       18.8         30.9       34.3
  No..............................................................        4.6        0.0          8.8        4.2
  Don't know......................................................       51.5       53.1         39.7       55.5



    Senator Brownback. Thank you, Ms. Haver, for your concise 
and very good comments.
    Dr. Leonard Nakamura is the economic adviser for the 
Federal Reserve Bank of Philadelphia. We very much appreciate 
your joining us today, and the mike is yours.

  TESTIMONY OF LEONARD I. NAKAMURA, ECONOMIC ADVISER, FEDERAL 
                  RESERVE BANK OF PHILADELPHIA

    Mr. Nakamura. Thank you very much for the opportunity to 
testify. I have submitted a written statement and two articles 
on the quality of our economic measures. What I am about to say 
represents my own views and not those of the Federal Reserve 
System or the Federal Reserve Bank of Philadelphia.
    In my testimony, I would like to make the following five 
major points: First is that the 1978 CPI methodology revision, 
the largest revision in the history of the CPI, now appears, 
with hindsight, to have worsened our measures of inflation 
rather than improved them. The consequence is that over the 
past 3 years the U.S. economy was probably growing twice as 
fast as our real GDP growth rate figures indicate. The U.S. 
inflation rate may well be zero rather than 2 to 3 percent.
    One of my studies shows that the rate at which consumers 
have been changing their spending patterns away from 
necessities like food and towards luxuries imply that 
mismeasurement of inflation overall worsened by 2 percentage 
points beginning in the late 1970's. One of the implications of 
this is that it may well be the case that the productivity 
slowdown we have been worrying about for the past 20 years may 
be a statistical figment.
    I present some details on the CPI for air fares, food at 
home, and medical care. I estimate that air fare inflation 
since 1978 has been overstated by nearly 6 percentage points a 
year; food at home inflation appears to have been overstated by 
1.5 percentage points annually; and medical care inflation 
appears to have been overstated by 4 percentage points 
annually.
    Known fixed statistical procedures cannot currently measure 
the inflation rate accurately. It will likely take a major 
ongoing economic research effort to ameliorate this problem.
    Let me elaborate a little bit about the 1978 revision. The 
methodology that we used seemed sensible at the time. I myself 
was very active in measurement at that point. I was a member of 
the Rees--I was a consultant to the Rees Commission on 
productivity statistics, and I was among the many economists 
who, in effect, vetted the new BLS methodology. I thought it 
was a solid step in the right direction.
    Well, we were wrong. The new methodology happens to lean on 
an economic principle called the law of one price, and the new 
methodology was implemented just as deregulation and 
computerization made changes in retailing that basically 
repealed the law of one price. And I will go into that more a 
little bit later.
    Our current economic statistics do not give an accurate 
picture of what is happening in the U.S. economy, but this is 
not the fault of the agencies that collect and publish the 
statistics. It is primarily due to the extremely rapid rate of 
change of the economy itself to which we have already had 
allusions in this testimony.
    One point that I would like to make is that it is sometimes 
argued that even if the data are biased by a large amount on a 
trend basis, they are accurate on a short-term basis. This may 
be so, but the little evidence that we do have is actually to 
the contrary. One of my exhibits shows the Bureau of Labor 
Statistics estimates on the year-to-year changes in the source 
of inflation bias that is easiest to measure, and in a number 
of years, the year-to-year inflation bias doubles or falls in 
half in a very striking way. So the year-to-year changes in CPI 
inflation may reflect changes in bias and may not reflect true 
changes in inflation.
    I would like to give an example which I think is 
particularly striking, and the example is one of air traveling. 
In 1978, when our current methodology for the CPI was put in 
place, there was only one round-trip coach fare on most routes 
because fares were regulated by the Civil Aeronautics Board. 
Now, as we all know, dozens of different fares are available 
with a variety of restrictions on every route, and the fare 
structure changes by the minute.
    Between 1978 and 1996, if we asked the question, What did 
passengers actually pay per mile to travel on airlines? that 
price per mile grew at a 2.7 percent annual rate. The CPI for 
air fares, however, grew at an 8.3 percent annual rate during 
that same period, a difference of 5.6 percentage points. Now, 
if we take that CPI and use it to deflate airline passenger 
revenues for that period, we find that the real output of 
airline passenger travel fell from 1978 to 1996. If, on the 
other hand, you look at the actual miles that passengers flew, 
passenger miles on airlines more than doubled, from 100 to over 
240.
    How can such a substantial gap have been sustained for so 
long? The reason is precisely this dispersion of fares. Full 
fare for unrestricted travel has risen very rapidly. It has 
risen at nearly a 9 percent annual rate. The CPI for air fares 
has basically tracked the full fare. However, the average 
restricted discount fare has increased at only 2 percent a 
year. As a result of the accumulation of those differences, the 
average unrestricted fare is now more than 3 times as great as 
the average restricted fare. You often hear about air travelers 
buying two round-trip restricted tickets instead of one 
restricted fare--in order to get around these restrictions, and 
you can see why. The unrestricted fares themselves cost three 
times as much.
    Only 7 percent of passenger miles are flown at full fare. 
So almost all passengers are, in fact, flying at the restricted 
fares but we are tracking the full fare. Nevertheless, full 
fares account for 20 percent of passenger revenues because the 
gap between the prices is so large.
    Now, you can argue and I believe that this constellation of 
fares is, in fact, highly efficient. In essence, airlines 
divide the customers into two broad groups: business travelers 
and vacation travelers. Business travelers care most about 
saving time. Their time is extremely valuable, and they want to 
have the maximum possible flights to a wide variety of 
destinations, and they are willing to pay more for that. 
Vacation travelers, on the other hand, care most about savings 
money and are often flexible about exactly when they fly. The 
airline accommodate both types of travelers efficiently, 
providing a multitude of flights for business travelers, 
filling the seats with vacation travelers. The typical 
restriction on a discount flight, the Saturday night overnight 
stay, which vacation travelers can usually easily accommodate, 
is used to separate the two.
    It is the deregulation and computerization that made this 
possible, and those date from the late 1970's, precisely the 
time when we were instituting this new methodology.
    Now, that proliferation of fares and restrictions is not 
without cost. It is highly efficient, but it also has costs. 
The costs are that consumers have to be careful shoppers; they 
have to plan their trips in advance; they have to guess when to 
lock in their fares. Changing plans becomes a lot more costly 
for them.
    Now, there are ways to estimate the relative costs and 
benefits of the proliferation of fares, but that is not easy. 
It involves ongoing research.
    Now, let me tell you very briefly what forces this upon us 
and why the CPI revision was wrong in the first place. 
Basically, the message is that our system was set up for 
catching the one price that we believed was the long-run 
competitive price for that good. And yet that long-run 
competitive price has disappeared.
    In the late 19th century, when economics became a modern 
science, the principle on which that science was based was the 
law of one price. The law of one price basically says that in 
any market at one moment there cannot be two prices for the 
same kind of article. What that means for economists is that, 
as long as there is competition among suppliers, in the long 
run that unique price is a good measure of the resource cost 
for producing that good and service. Consumers for their part 
are only going to buy the good if the exchange of dollars for 
the good benefits them, which makes price a good measure of 
consumer utility, and that was the central synthesis of the 
laws of supply and demand which join in the law of one price.
    When that law holds, we can use a cost-of-living index that 
reflects consumers' welfare to help us obtain real output 
measures for productivity growth because productivity growth is 
about resource costs, and consumers is about the utility and 
efficiency with which we use that. If that law of one price 
does not hold, that logic falls apart, and our statistics 
create the kind of problems that we see in the airline fares 
example, that a model of utility based on the law of one price 
says something like, well, if a full fare is 3 times as 
expensive as a restricted fare, then that must be 3 times as 
valuable for all users. That law, that falls apart.
    Senator Brownback. Mr. Nakamura, I want to make sure that 
you do focus on whether we should consolidate these entities. I 
appreciate the points you are making here because it is a good 
point and it is one I was not familiar with. But do you see 
consolidation points, issues? Do you have trouble with doing 
that? I do not know if you have that in your written testimony 
as well, because I sure would like to make sure we do not get 
buzzed for a vote or anything before I get you to that point.
    Mr. Nakamura. OK. The main thing that I have to say about 
this is that the known fixed statistical procedures cannot 
measure the inflation rate accurately. It is going to take a 
major ongoing research effort to solve this problem. I think 
that we can agree on research principles that will measure 
price reasonably, but in the rapidly changing economy in which 
we live, we will not be able to define fixed procedures that 
are going to accurately measure inflation.
    The economics profession has been developing the necessary 
theoretical and statistical apparatus for measuring price in a 
rapidly changing world, and we have made tremendous strides. 
But we do not have the data, the funding, or the authority to 
use this knowledge to create useful statistics. And if we as a 
Nation are to agree on our inflation measures, we must figure 
out how to create an ongoing research effort in economic 
measuring.
    [The prepared statement of Mr. Nakamura with attachments 
follow:]

                 PREPARED STATEMENT OF LEONARD NAKAMURA
    (This testimony reflects the personal views of the author and not 
those of the Federal Reserve Bank of Philadelphia or of the Federal 
Reserve System.)
                               __________
Contents:
      I. LIntroduction and overview
     II. LMore on the quality of price and output statistics: details 
on food and medical expenditures.
     III. LThe rate of change of consumer expenditures and its 
implications for growth mismeasurement: Engel's Law and estimates of 
real growth
    IV. LMeasuring of inflation and economic growth requires on-going 
research, not just a change in procedures. There is no quick fix.

    Thank you very much for the opportunity to submit testimony. What I 
am about to say represents my own views and not those of the Federal 
Reserve System or the Federal Reserve Bank of Philadelphia.
    In my testimony, I will make and support the following five major 
points.
    1. LThe 1978 CPI methodology revision appears, with hindsight, to 
have worsened our measures of inflation rather than improved them.
    2. LOver the past 3 years, the U.S. economy may have been growing 
twice as fast as our real GDP growth rate figures indicate. The U.S. 
inflation rate may be zero, rather than 2 to 3 percent.
    3. LKnown fixed statistical procedures cannot currently measure the 
inflation rate accurately; it will likely take a major on-going 
economics research effort to ameliorate this problem.
    4. LDetails on the CPI for airfares, food at home, and medical care 
are presented. I estimate that airfare inflation since 1978 has been 
overstated by nearly 6 percent. Food-at-home inflation appears to have 
been overstated by 1\1/2\ percent. Medical care inflation appears to 
have been overstated by 4 percent.
    5. LThe rate of change at which consumers have been changing their 
nominal spending patterns implies that mismeasurement of inflation 
overall worsened by 2 percentage points beginning in the 1970s.
I. LIntroduction and overview
    A. The 1978 CPI methodology revision appears, with hindsight, to 
have worsened our measures of inflation rather than improved them. In 
1978 the Bureau of Labor Statistics undertook the most fundamental 
revision to its methodology that has ever occurred. The methodology 
seemed sensible at the time; I myself was among the many economists who 
vetted it and thought it was a solid step in the right direction. We 
were wrong. The new methodology
    was implemented just as deregulation and computerization began 
changes in retailing that made it inaccurate.
    B. Known fixed statistical procedures cannot currently measure the 
inflation rate accurately it will likely take a major on-going 
economics research effort to ameliorate this problem. Our current 
economic statistics do not give an accurate picture of what is 
happening in the U.S. economy. But this is not the fault of the 
agencies that collect and publish the statistics. It is primarily due 
to the extremely rapid rate of change of the economy itself. A major 
research effort in economics may be necessary to ameliorate this 
problem.
    Currently, measured inflation is approximately 2 to 3 percent. I 
argue that a much more reasonable estimate of the true inflation rate 
is zero and that the mismeasurement of inflation has been accelerating. 
This upward bias in inflation has downwardly biased our measures of 
growth and productivity.
    I believe that over the past 3 years the U.S. economy has been 
growing perhaps 5 percent a year instead of 2 or 3 percent. When we 
look back on this period of time 10 or 15 years hence, we will likely 
acknowledge that this was a period of unprecedented change and growth 
for the U.S. economy.
    Our current statistics say that the U.S. economy is growing 
unprecedentedly slowly; our statistics argue that there has been no 
technological progress, despite 20 years of computerization, 
automation, deregulation, tax cutting, and downsizing. Exhibit 1 
presents data that say that the efficiency with which the U.S. economy 
uses its capital and labor rose roughly 1.7 percent a year from 1947 to 
1978 and that since then there has been no gain in efficiency. Our 
current statistics say that the average U.S. worker has been losing 
ground since 1965 (Exhibit 2). Every time we compare our paychecks to 
the consumer price index, we think we are falling further behind.
    It is sometimes argued that even if the data are biased, they are 
informative on a short-term basis. This may be so, but the little 
evidence we have is to the contrary. Exhibit 3 shows the BLS estimates 
of the easiest-to-measure source of inflation bias, the inflation bias 
that arises from not updating the weights of major product groups in 
our market basket. It accounts for one-fifth of the bias in the CPI 
according to the Boskin Commission, and less than one-tenth by my 
reckoning. Year-to-year changes in CPI inflation may reflect changes in 
bias as much as true changes in inflation.
    An example: Air travel. In 1978, when our current methodology for 
the CPI was put in place, there was only one round-trip coach fare on 
most routes, as fares were regulated by the Civil Aeronautics Board. 
Now, as we all know, dozens of different fares are available with a 
variety of restrictions on every route, and the fare structure changes 
by the minute.
    Between 1978 and 1996, the average price paid per mile by 
passengers grew at a 2.7 percent annual rate (Exhibit 4). The CPI-U for 
airfares grew at an 8.3 percent annual rate, a difference of 5.6 
percentage points. If we use the CPI to deflate airline revenues from 
passenger travel, we find that ``real'' airline passenger travel output 
fell from 1978 to 1996. But, in fact, passenger miles on airlines more 
than doubled.
    How can such a substantial gap have been sustained for so long? The 
reason is the dispersion of fares. Full fare for unrestricted travel 
has risen at nearly a 9 percent annual rate, and the CPI for airfares 
has basically tracked the full fare. The average restricted (discount) 
fare has increased at only 2 percent a year. The average domestic 
unrestricted fare is now 3.1 times as much as the average restricted 
fare. But only 7 percent of passenger miles are flown at full fare 
(full fares account for 20 percent of passenger revenues).
    It can be argued that this constellation of fares is highly 
efficient. In essence, airlines have divided their customers into two 
broad groups: business travelers and vacation travelers. Business 
travelers, who care most about saving time, want to have the maximum 
possible flights to a wide variety of destinations. They are willing to 
pay more for this privilege. Vacation travelers, on the other hand, 
care most about saving money and are often flexible about exactly when 
they fly. The airlines accommodate both types of travelers, providing a 
multitude of flights for business travelers and filling the seats with 
vacation travelers. The typical restriction on discount flights--a 
Saturday overnight stay, which vacation travelers can often easily 
accommodate--is used to separate business travelers from vacation 
travelers. Note that deregulation and computerization has made this 
possible.
    The proliferation of fares and restrictions also has many costs. 
Consumers have to be careful shoppers, planning their trips in advance 
and having to guess when to lock in a nonrefundable fare. Changing 
plans has become more costly. There are ways to estimate the relative 
costs and benefits of the proliferation of fares, but they are not 
simple. One approach is to directly measure the cost of the 
restrictions to the flier. Another approach is time diary measures, 
which track shopping times (the amount of time consumers spend 
searching for the lowest fares) and compare them to estimated waiting 
times (the amount of time consumers save by having more flights 
available). A complementary approach is to measure the elasticities of 
demand for different types of travelers, which gives us estimates of 
the travelers' tradeoff between time and price. A third approach is to 
estimate the market structure of airline competition, to see where 
airlines have market power and where competition reigns. A fourth 
approach is to analyze and measure the role of the various contributors 
to airline output: airplane manufacturers, travel agents, airlines, 
airports. None of these approaches alone will do the job.
    The law of one price. The proliferation of prices for air travel is 
the product of computerization--it also violates the most fundamental 
principle in economics, the law of one price. Jevons's law of one price 
is the principle that ``in the same open market, at any one moment, 
there cannot be two prices for the same kind of article (Jevons, 
1879)''.
    As long as there is competition among suppliers, in the long run 
this unique price will reflect the resource costs of producing the good 
or service. Consumers, for their part, will only buy goods if the 
exchange of dollars for the good benefits them; this makes price a good 
measure of consumer utility. Thus the law of one price is that prices 
are a useful measure of resource costs and of utility. When the law 
holds, we can use a cost of living index (based on measuring consumer 
utility with prices) to deflate nominal output to obtain real output 
for productivity growth measures (productivity is our ability to reduce 
resource costs). If the law of one price does not hold this logic falls 
apart. In the absence of the law of one price, the meaning of prices is 
difficult to know without thorough study of supply and demand 
conditions and the exact nature of the price variations.
    The phenomenon that we have been examining, the breakdown of the 
law of one price, appears throughout our economy. The computerization 
of transactions and recordkeeping has made it possible for managers to 
know the impact of price changes on profit in item-by-item detail, and 
has also made possible daily variation in prices. Meanwhile, we 
shoppers have all learned to be careful shoppers, and to shop at large 
malls, which are designed to help us compare as many goods as possible 
in the shortest period of time.
II. LMore on the quality of price measurement: details on food and 
        medical expenditures. These two categories account for roughly 
        one-fourth of consumer expenditures and one-sixth of the 
        consumer price index.
    A. Food. Among all consumer prices, economists are most experienced 
at collecting food prices. The following extended example shows that 
our measures of food prices (narrowly defined here as food purchased 
for consumption at home) went dramatically awry beginning in 1978.\1\ 
The argument is a reductio ad absurdum: our official statistics imply 
that the real output of retail services at supermarkets fell 
dramatically, but direct measures of supermarket services rose 
substantially over this period.
---------------------------------------------------------------------------
    \1\ This is drawn from my papers, Nakamura (1997b).
---------------------------------------------------------------------------
    1. The Bureau of Labor Statistics has been collecting monthly data 
on food prices since World War I, when the CPI, then called the cost of 
living index, was institutionalized. Prior to 1978, the prices 
collected were for the same goods and services across all the cities 
surveyed. Price inspectors throughout the country would collect prices 
for ``milk, delivered, glass bottles,'' or ``bacon, first quality, hand 
sliced.'' Uniformity poses some problems. Over long periods of time the 
quality of these goods might well vary, and indeed the products might 
well disappear altogether. Milk might be rich or watered or sour; first 
quality bacon in one city might be second quality in another. And 
delivered milk has become a rare commodity in most cities.
    2. In 1978, when the new methodology came in, this uniform 
specification of products was replaced by decentralized specification 
of products. Price inspectors were asked to define detailed product 
specifications in the field. The price inspectors were given broad 
product definitions, such as flour and prepared flour mixes, and a 
store location based on a nationwide survey called the Consumer Point 
of Purchase Survey. For example, the Survey and the randomization 
process might result in the choice of the Acme supermarket at 
Germantown and Sedgwick in Philadelphia. Then the price inspector, with 
the help of store personnel, chooses several possible items, and 
randomly picks one, say, Betty Crocker Chocolate Fudge cake mix. For 
the next 5 years the item priced by the price inspector would be this 
particular item at this particular store (unless the store stopped 
carrying that item or closed).
    3. The BLS also collects and publishes average price (AP) data on a 
selected group of foods. This is a separate series that prices products 
(such as flour; white, all-purpose) that are relatively broadly defined 
when compared to the very narrow product-store combinations used in the 
CPI. The AP series gives the average price for that product per pound. 
The prices are weighted by the relative sales of the outlets at which 
they are collected. The AP series is apparently piggybacked on the CPI 
data, in the sense that the basic data in the AP series are, to the 
extent possible, taken from the CPI collections.
    The AP series, it should be pointed out, is essentially what 
economists have typically collected historically. The AP series (except 
for a break from 1978 to 1980) is available going back to 1890 for nine 
foods.
    4. Before the introduction of random sampling by the price 
inspector in 1978, the CPI series and the AP series showed no 
systematic tendency to diverge. An economist at the BLS, Marshall 
Reinsdorf, published an article in 1993 that has been one of the 
seminal articles in the area of CPI price mismeasurement. He discovered 
that from 1980 to 1990, the CPI and AP series for comparable products 
diverge by roughly 2 percent a year, with the CPI series rising faster 
than the AP series. As can be seen in Exhibit 5, the divergence over a 
recent 6 year period is quite substantial for many of the products--and 
the divergence is almost universally in the same direction. And as seen 
in Exhibit 6, the roughly 2 percent a year divergence between the two 
series continues to January 1996.
    5. In principle, there are two reasons the CPI and the AP series 
might diverge. One is that customers may be switching to lower quality 
goods within each product category. The other is that customers may be 
switching to less costly outlets for goods. And there is an additional 
technical reason: the method that the BLS used to reweigh goods when it 
updated its sample was biased in the absence of the law of one price. 
This so-called ``formula bias,'' which apparently accounted for \1/2\ 
percentage point a year of the 2 percentage point annual divergence, 
was corrected in January 1995.
    One possible reason for the CPI to rise more rapidly than average 
prices is if consumers were shifting to lower quality foods. We would 
have evidence of a switch to lower quality goods if the CPI rate of 
increase were mirrored by an increase in the PPI for comparable goods. 
It is not. The CPI series for food at home grows 1.4 percent faster 
from 1977 to 1992 than does the PPI series for consumer food (Exhibit 
7).
    6. Another possibility is that supermarkets' retail services could 
be declining rapidly, if, for example, variety were decreasing or 
service personnel were declining or stores became more cramped. This is 
also not the case. There has been some switch to discount warehouse 
type stores, as shown in Exhibit 8, but the greater switch has been to 
the superstore format, in which the supermarket sells extensive 
additional lines of goods, such as drugs, and provides additional 
services, such as a deli counter, fresh fish, flowers, and even 
banking.
    In this enlarged format, supermarkets are larger (Exhibit 9), stock 
more items (Exhibit 10), and have more employees (Exhibit 11). While 
some of the growth in number of products is due to a shift toward more 
drugs and other nonfood products, most of it appears to be due to an 
increase in variety of food products.
    Consider the following. We can use the CPI for food commodities to 
deflate food store sales for 1992 to measure the real value of food 
products and retail services delivered to consumers. Similarly, we can 
use the PPI for finished consumer foods to deflate 1992 food store 
goods purchases to get a measure of the real value of products farms 
and manufacturers delivered to food stores. The difference should be 
real retail services added by the food stores: the economic 
contribution of supermarkets. This calculation is shown in Exhibit 12, 
when we use this so-called ``double-deflation'' methodology to estimate 
the real contribution of supermarket output. The implication of our 
official statistics is that food store output has been declining at a 
7.7 percent annual rate. This is absurd, because as I have shown along 
a variety of dimensions, food store output has been increasing.
    In short, the CPI attributes declining real output to a retail 
segment that by every conceivable measure has been rapidly providing an 
ever greater abundance of value added services. This unreasonable 
result is the outcome of the clash between the methodology put in place 
in 1978, and the fact that foods do not obey the law of one price in 
our current retail environment.
    B. Medical expenditures. Between 1984 and 1994, measured per capita 
real expenditures on medical care rose at a 2.2 percent annual rate; 
real spending on drugs, a subcomponent, rose at a 2.3 percent annual 
rate. But most of the components of medical care are priced by inputs, 
such as office visits, procedures, and hospital room charges, rather 
than by outputs such as conditions treated successfully. Since 1974, 
the proportion of nominal expenditures on medical care in personal 
consumption expenditures has more than doubled and medical expenditures 
now account for one dollar in six in consumer spending. Note that this 
figure includes health insurance benefits paid by corporations, and 
Medicare and Medicaid payments as well.
    An argument can be made that the true rate of growth of real 
spending on the drug subcomponent was over 8 percent annually in the 
1980s. This would include an adjustment of 3 percent to the inflation 
rate from late introduction of products into the survey (Berndt et al, 
1993), a 2 percent adjustment from brand names and generics and 
conversions from prescription drugs to over-the-counter status (Fisher 
and Griliches, 1995, and Temin, 1992), and a 1 percent adjustment from 
uncaptured consumer surplus from the introduction of new varieties of 
drugs at prices substantially below reservation levels.
    Since the value of medical care depends on the ability to treat, 
and treatment to a large extent consists of prescribing drugs and 
performing surgery, the real rate of increase in drug supply would be a 
useful proxy for treatment success. Thus the quality improvement in 
medical services might be usefully proxied by the growth of real 
expenditures on drugs.
    Cutler (1995) has examined the growth of spending in medical 
services. In a detailed study of heart attack treatments from 1984 to 
1991, he measures the inflation rate on a Paasche basis as 3.5 percent 
annually. During that time the GDP deflator rose at a 3.7 percent 
annual rate. By contrast, the PCE deflator for medical services rose at 
a 7.5 percent annual rate. This deflator prices inputs, such as the 
price of an hour of a surgeon's time, and not treatments. Even Cutler's 
measure does not impute any benefit for the improvement in techniques 
for treating heart attacks, particularly angioplasty as an alternative 
to bypass surgery. Cutler's measure does capture accurately the 
declining price of the angioplasty, which fell at a 2.3 percent annual 
rate. The difference between Cutler's heart attack price series and the 
PCE deflator is roughly 4 percentage points per year. This understates 
the quality improvement associated with hospital services, as it does 
not include any imputation for the availability of angioplasty or 
improvements in techniques in performing these operations. 
(Angioplasties are less expensive than bypasses, as well as being less 
invasive and therefore requiring less patient time for recovery.) 
Shapiro and Wilcox (1996) show similar price mismeasurements in an 
analysis of cataracts.
    Cutler et al (1996) revisits these issues, and includes an analysis 
of ultimate benefits in terms of quality of life years (QOLY) saved. 
Taking a quite conservative measure of the value of a QOLY, $20,000, 
they confirm that conventional measures understate the growth of value 
produced by at least 4 percentage points a year.
    Changing medical technology includes improved diagnostic machinery 
(Trajtenberg, 1990) and surgery. The development of noninvasive or less 
invasive surgical procedures such as ultrasound treatment of kidney 
stones and arthroscopic surgery as well as improved understanding of 
the healing process has led to shorter hospital stays. As a 
consequence, the 6 percent annual rate of inflation in medical care 
could be entirely a result of mismeasurement of quality. Even with a 
conservative estimate of bias, the inflation rate of medical care 
services is likely being overstated by 4 percentage points.
III. LThe rate of change of consumer expenditures and its implications 
        for growth mismeasurement: Engel's Law and estimates of real 
        growth.\2\ Here I argue that the rate of change of American 
        spending suggests that growth is being undermeasured by two 
        percentage points a year, compared to how growth was measured 
        previously.
---------------------------------------------------------------------------
    \2\ This is drawn from my papers, Nakamura (1997a, c.)
---------------------------------------------------------------------------
    A systematic way of testing for the presence of economic growth is 
to examine the rate at which basic economic necessities, such as food 
and clothing and household operations, are shrinking as a proportion of 
total expenditures. The basic empirical principle in this regard is 
Engel's Law: As real income per person rises, the proportion spent on 
food declines. The eminent Harvard economist Hendrik Houthakker has 
said, ``Of all empirical regularities observed in economic data, 
Engel's Law is probably the best established; indeed it holds not only 
in the cross-section data where it was first observed, but has often 
been confirmed in time-series analysis as well.'' \3\
---------------------------------------------------------------------------
    \3\ Hendrik S. Houthakker, ``Engel's Law,'' in John Eatwell, Murray 
Milgate, and Peter Newman, eds., The New Palgrave: A Dictionary of 
Economics. Volume 2 (Macmillan, 1987), pp. 143-44.
---------------------------------------------------------------------------
    Exhibit 13 illustrates the basic idea. Suppose at time 0 real 
income is 1000, of which 60 percent is spent on food and other 
necessities, while the other 40 percent is spent on luxuries. Now 
suppose that real income grew 20 percent, to 1,200. Demand for food 
doesn't increase as much as demand for luxuries, so although food 
purchases increase, they shrink as a percent of expenditures. Suppose 
that real income grows another 20 percent. Food purchases continue to 
rise, but less rapidly than total income and spending. The share spent 
on food declines over time. Moreover, equal percent increases in real 
income lead to equal changes in the share of nominal expenditure on 
food and in the share of nominal expenditure on luxury: in both 
periods, each share changes 5 percentage points, food down and luxury 
up.
    This formulation of Engel's Law is based on work by Angus Deaton 
and John Muellbauer.\4\ It implies that equal percent increases in real 
incomes per person should lead to equal percentage point changes in 
shares of expenditure.
---------------------------------------------------------------------------
    \4\ ``An Almost-Ideal Demand System,'' American Economic Review 70 
(June 1980), pp. 312-16. Their system formally says that holding 
relative prices constant, equal changes in the logarithm of real income 
lead to equal changes in shares in nominal expenditures. Here we 
discuss the system in terms of percent changes as we assume most 
readers are more familiar with that terminology.
---------------------------------------------------------------------------
    How do we apply their formulation to U.S. data? From 1959 to 1974, 
according to the official statistics, real income per person grew 45 
percent. In the longer period from 1974 to 1994, real income per person 
grew 39 percent. If these numbers are accurate, one would expect that 
the share of necessities in total expenditures should have shrunk by 
about the same amount in the two periods (or perhaps a bit less in the 
second period). In fact, the proportion of the average budget spent on 
food fell from 27.3 percent in 1959 to 23.1 percent in 1974, or 4.2 
percentage points, but fell substantially more--7.1 percentage points--
from 1974 to 1994 (Exhibit 14).
    The proportion of household budgets spent on other necessities, 
such as clothing and home heating, also almost uniformly contracted by 
more in the period 1974 to 1994 than in the earlier period 1959 to 1974 
(Exhibit 15). In contrast, the share spent on luxuries, such as medical 
care, personal business services, recreation, education, and foreign 
travel, generally rose more in the later period than in the earlier 
one.\5\ This faster shift away from necessities as a proportion of 
budgets in the second period suggests that real income per person grew 
more in the second period than in the first, not less as the official 
statistics say.
---------------------------------------------------------------------------
    \5\ What is a necessity and what is a luxury is not always easy to 
determine. Food is the clearest example of a necessity. Goods and 
services whose consumption declines over long periods of time when 
incomes are rising are defined as necessities here; the consumption of 
luxuries rises over the same time periods.
---------------------------------------------------------------------------
    How much more? To answer this question, calculate the average 
absolute change in shares for all consumption categories over each 
period, that is, take the average without considering whether each 
change is up or down. In this way, a decline of 2 percent for a 
necessity like food and a rise of 2 percent for a luxury like travel 
both correspond to rising real income. The nine consumption categories 
in Exhibit 15 changed absolutely by 1.50 percentage points, on average, 
in the period 1959 to 1974, while they changed 2.88 percentage points, 
on average, from 1974 to 1994. If we use Deaton and Muellbauer's 
application of Engel's Law, the fact that the average shift in spending 
shares (away from necessities and toward luxuries) was almost twice as 
big in the second period as in the first--2.88 to 1.50 percentage 
points--implies that the true rise in real income in the second period 
was about twice as large as that in the first (so long as prices of 
luxuries did not rise at a substantially different rate than prices of 
necessities). If real income rose 45 percent from 1959 to 1974 as the 
official data show, the change in spending shares from 1974 to 1994 
suggests that real income rose just over 100 percent during those 20 
years, not 39 percent as reported in the official statistics.\6\ Over 
1974 to 1994, this represents a per-person annual growth rate of 3.7 
percent, not 1.7 percent--a difference of 2.0 percentage points per 
year.
---------------------------------------------------------------------------
    \6\ According to the Deaton and Muellbauer model, the ratio of the 
change in the log of real income between 1974 and 1994 to the change in 
the log of real income from 1959 to 1974 equals the ratio of the 
absolute change in expenditure shares between the periods 1959 to 1974 
and 1974 to 1994. The change in the log of real income from 1959 to 
1974 is 0.375. We multiply this by the ratio between the percent 
changes in shares, 0.375 x (2.88/1.50) = 0.720. The antilog of .720 is 
2.05, suggesting that real per capita income in 1994 was 2.05 times 
real per capita income in 1974.
---------------------------------------------------------------------------
    Now let's reexamine the productivity slowdown that began around the 
mid-1970s. That slowdown is reflected in the official data in Exhibit 
16 in that more real growth per person took place from 1959 to 1974 
than in the longer period from 1974 to 1994. But the slowdown is not 
consistent with the changes in the consumption expenditure shares. The 
implication of the calculations reported above is that growth in real 
income per person was mismeasured by 2.0 percentage points annually 
from 1974 to 1994--slightly more than the measured slowdown in 
productivity growth in the official statistics of 1.7 percentage points 
annually. That is, it is possible that the entire productivity slowdown 
of the past two decades revealed by the official statistics is the 
result of mismeasurement! Put another way, the shifts in composition of 
expenditures from 1959 to 1974 and from 1974 to 1994 are consistent 
with the view that productivity growth was the same in both periods. 
Households are spending in a pattern that is inconsistent with the 
official statistics on real output and price; that is, the average 
household has expanded the proportion of luxuries it buys as if its 
real income had doubled over the last 20 years, while the official data 
report that its real income rose by less than half.
IV. LMeasuring inflation and economic growth requires on-going 
        research, not just a change in procedures. There is no quick 
        fix.
    A. Why measuring inflation appears to be easy but is really quite 
difficult. We live in a society that measures physical units with the 
accuracy of angstroms. Moreover, a manager in Omaha can know what we 
had for breakfast and what we paid for it. How could we not know what 
the inflation rate is?
    At first glance, price appears to be as clear as day. The 
arithmetic problems we give first graders are about prices: if an apple 
costs ten cents, what do two apples cost? And we are all consumers, so 
each of us is an expert on price. In every shopping decision we make, 
we are using our knowledge of prices and inflation.
    But as the airfare example shows, we have lost the simple 
connection between prices and products. In 1978 we all paid the same 
fare. Since then, the full fare has risen 8.8 percent a year, the CPI 
for airfares 8.3 percent, the average fare 2.7 percent, and discount 
fares 2 percent a year. Almost all the time we fly at the discount 
fare. Is the rate of inflation in airfares 9 percent, 8 percent, 3 
percent or 2 percent?
    We buy most goods on sale. We have all become careful shoppers. 
Markups and discounts have always been with us, but they have exploded 
since the computerization of retailing in the 1970s and early 1980s. In 
1965, the average department store discount was 6 percent, almost 
unchanged from the 5 percent rate of 1955. But by 1986, large markdowns 
were endemic. A study by Peter Pashigian, a professor at the University 
of Chicago, of men's white dress shirts in 1986 revealed that two-
thirds were sold at discount, with the average discount varying from 33 
percent to 50 percent. In this environment, it becomes harder to know 
what the prices of products mean and to assess the rate of inflation.
    B. Measuring inflation requires ongoing research, not a fixed 
procedure. The fundamental message of my testimony is simple: the 
ongoing measurement of inflation and growth is a research issue and not 
a procedural issue. In the economy in which we live, there is no fixed 
procedure that will accurately measure price forever. I believe that we 
can come to agree on a dynamic set of research principles that will 
measure price reasonably but for the forseeable future we will not be 
able to define fixed procedures that will do so.
    Ideally, the measurement of price would be accomplished by the 
economics profession as part of its regular research program. But 
measuring price is extremely expensive by academic standards.
    The economics profession has been developing the theoretical and 
statistical apparatus for analyzing and measuring price, and has made 
tremendous strides. But the profession lacks the data, the funding, and 
the authority to use this knowledge to create useful statistics. If we 
as a nation are to agree on our inflation measures, we must figure out 
how to create an on-going research effort in economic measurement.
    [Exhibits and two articles from Business Review follow:]
    [GRAPHIC] [TIFF OMITTED] TH170.001
    
    [GRAPHIC] [TIFF OMITTED] TH170.002
    
    [GRAPHIC] [TIFF OMITTED] TH170.003
    
    [GRAPHIC] [TIFF OMITTED] TH170.004
    
    [GRAPHIC] [TIFF OMITTED] TH170.005
    
    [GRAPHIC] [TIFF OMITTED] TH170.006
    
    [GRAPHIC] [TIFF OMITTED] TH170.007
    
    [GRAPHIC] [TIFF OMITTED] TH170.008
    
    [GRAPHIC] [TIFF OMITTED] TH170.009
    
    [GRAPHIC] [TIFF OMITTED] TH170.010
    
    [GRAPHIC] [TIFF OMITTED] TH170.011
    
    [GRAPHIC] [TIFF OMITTED] TH170.012
    
    [GRAPHIC] [TIFF OMITTED] TH170.013
    
    [GRAPHIC] [TIFF OMITTED] TH170.014
    
    [GRAPHIC] [TIFF OMITTED] TH170.015
    
    [GRAPHIC] [TIFF OMITTED] TH170.016
    
    [GRAPHIC] [TIFF OMITTED] TH170.017
    
    [GRAPHIC] [TIFF OMITTED] TH170.018
    
    [GRAPHIC] [TIFF OMITTED] TH170.019
    
    [GRAPHIC] [TIFF OMITTED] TH170.020
    
    [GRAPHIC] [TIFF OMITTED] TH170.021
    
    [GRAPHIC] [TIFF OMITTED] TH170.022
    
    [GRAPHIC] [TIFF OMITTED] TH170.023
    
    [GRAPHIC] [TIFF OMITTED] TH170.024
    
    [GRAPHIC] [TIFF OMITTED] TH170.025
    
    [GRAPHIC] [TIFF OMITTED] TH170.026
    
    [GRAPHIC] [TIFF OMITTED] TH170.027
    
    [GRAPHIC] [TIFF OMITTED] TH170.028
    
    [GRAPHIC] [TIFF OMITTED] TH170.029
    
    [GRAPHIC] [TIFF OMITTED] TH170.030
    
    [GRAPHIC] [TIFF OMITTED] TH170.031
    
    [GRAPHIC] [TIFF OMITTED] TH170.032
    
    [GRAPHIC] [TIFF OMITTED] TH170.033
    
    Senator Brownback. Good. And I do not mean to put you on 
the spot on anything at all, and if anybody is uncomfortable 
with any question, just say I am not interested or I do not 
feel like I can. But I do appreciate that comment and your 
thoughts.
    Mr. Nye Stevens is the Director of Federal Management and 
Workforce Issues for the General Accounting Office. Mr. 
Stevens, thank you for joining us, and the mike is yours.

 TESTIMONY OF L. NYE STEVENS, DIRECTOR, FEDERAL MANAGEMENT AND 
  WORKFORCE ISSUES, GENERAL GOVERNMENT DIVISION, U.S. GENERAL 
                       ACCOUNTING OFFICE

    Mr. Stevens. Thank you, Mr. Chairman. I will give a very 
brief summary, if I may, and submit the statement for the 
record. It is based on a body of work that we have been doing 
on statistical issues in statistical agencies over the past 4 
or 5 years, and the list of those products is at the end of my 
statement. We have applied those to four specific questions 
that you asked us in preparation for this hearing.
    First, you asked for an assessment of the quality of 
statistical data that is produced by the agencies of the 
Federal statistical system, which is very extended. We believe 
that the Federal statistical agencies have done some things 
very well. In fact, we applied to the major agencies--BEA and 
Census and BLS among them--the principles for good professional 
standards by the National Academy of Sciences and found that 
generally they conform very well to those. The one area that 
raised some difficulty was that of data sharing, where there is 
a statutory restriction on them.
    Nevertheless, it is also true that a number of studies of 
the Federal statistical system based on the assessments of 
government officials and users, like the two we have heard from 
today, do identify a number of concerns about the quality of 
statistical data. I think the view that Ms. Norwood and Mr. 
Barabba presented would have some people disagreeing with that 
in the statistical community. We note in particular that there 
are problems with the international transactions of the U.S. 
economy and the way they are measured. We would expect that 
probably there are criticisms in the area of investment and 
savings, and a number of other economic areas.
    We are finding, as we get into how agencies are responding 
to the requirements of the Government Performance and Results 
Act, which was developed by this Committee, that the importance 
of data to measure the outcomes of their efforts is more 
important than ever. I should also point out that the economic 
statistics initiative, sometimes called the Boskin Initiative, 
identified 38 major problems with the collection and 
dissemination of statistics. We did a review of the 
implementation of those 4 years afterwards and found that only 
about half had had any implementation associated with them. The 
others still remained on the ``to do'' or ``to be funded'' 
block.
    Finally, our current work on the decennial census reveals 
that there has been some deterioration in the quality of the 
data. There are many causes for that, of course, but we are 
worried enough to have put the conduct of the 2000 Census on 
GAO's high-risk list just 6 weeks ago. So we do believe that 
there are major problems there.
    You also asked us how the decentralized structure of the 
system affects these quality problems, and while our work does 
not really show the precise relationship between 
decentralization and quality, it does show that the 
decentralized nature of the system contributes largely to a 
number of other problems which may have quality implications, 
such as inefficiency, the lack of an ability to set priorities 
among the kinds of data that you collect, the burden of the 
collection on data providers and also the burden on users of 
having multiple resources for it, and restrictions on the 
exchange of data among statistical agencies. These all are 
contributors to quality problems and direct consequences of the 
decentralized nature of the system.
    Some of the data quality problems that were identified by 
the Boskin Initiative, for example, have yet to be corrected 
because the corrective action involves steps by more than one 
agency, and in some cases we find that an agency would receive 
funding for its part of a corrective action and another agency 
involved in the same problem would not receive funding. The 
result was no real improvement in quality.
    The lack of an effective mechanism for setting national 
priorities is also a problem. Each statistical agency has its 
own budget, its own trade-offs within the department where it 
finds itself. There is a similar dispersion of congressional 
attention to these agencies that results in an uneven quality 
of determination of what is most important. The burden on 
providers is another consequence, and this stems in large part, 
really, from the inability of statistical agencies to share 
data with one another because of legislated confidentiality 
restrictions.
    You asked about the potential of consolidation to provide a 
more streamlined and effective system. Here we turned 
immediately to the Canadian system on which we have done a good 
deal of comparative work with our system. We did find a great 
deal more consolidation, and I would have to say a number of 
advantages from their consolidation that we do not have and 
that would potentially result in changes here. For example, the 
use of common data collection methods and development of more 
efficient survey designs, cost savings and reduced burden on 
data providers to be achieved through greater sharing of data, 
avoidance of duplication. I think Ms. Haver is certainly right, 
however, that if there were a consolidation that did not 
involve breaking down some of these barriers that now exist in 
data sharing, it would not do any good.
    And then finally you asked whether the Census Bureau and 
the Bureau of Economic Analysis benefit from their present 
location within the Commerce Department. We have testified 
before the full Committee, in fact, on the legislation 2 years 
ago to dismantle the Commerce Department, and we described that 
Department then as essentially a holding company for a great 
many disparate programs, subject to almost constant 
organizational change over its 84-year history when it broke 
off from what was then the Department of Commerce and Labor. It 
has historically not been managed on the basis of any unifying 
mission or shared goals among its components. The components 
are overseen by a number of committees within Congress, none of 
which has jurisdiction over the entire Department.
    I can say as an aside that, partly for that reason, GAO 
does not have a coherent overview of the Department as well 
because no committee in Congress has asked us to do a general 
management review there, for example.
    It is also true that Commerce has decentralized its key 
administrative functions so that the major Commerce components 
have been granted the authority and the responsibility by 
Commerce for meeting most of their administrative requirements. 
They do this through a number of cross-servicing arrangements, 
but those are readily transferable. They get the money. They 
pay for services from Commerce, in some cases from other 
bureaus within Commerce. But they could also buy those from 
other agencies. The relative independence of the major 
components minimizes the disruption that would occur if one or 
more were broken off from Commerce, as has frequently happened 
in the past, and we note that neither the Census Bureau nor BEA 
is physically housed in the Commerce headquarters building.
    We are not aware of any reasons that would prevent Census 
and BEA from performing their missions if they were not 
components of the Commerce Department. This is not, however, to 
say that the relationship is meaningless. It does have some 
implications. We, in fact, have, in our concern about the 
Census, called upon the administration to become more involved 
in the planning for the Census rather than leave it completely 
to the Bureau. Commerce officials have noted, for example, that 
the absence of regulatory programs in that Department is one 
positive argument for having it there, or at least the 
avoidance of some objections that might come from its being 
elsewhere.
    It is also a fact that because Census and BEA are in 
Commerce they are competing for funding with functions as 
disparate as fisheries management and coastal zones, Weather 
Service, examination of patents and a whole host of unrelated 
functions, which have nothing to do with statistical data 
collection. But it is also true that another department would 
involve trade-offs, but just different kinds of trade-offs.
    That is a brief summary of what we have done, Mr. Chairman, 
and I will respond to any questions you have.
    [The prepared statement of Mr. Stevens follows:]

                  PREPARED STATEMENT OF L. NYE STEVENS

         Statistical Agencies: Consolidation and Quality Issues

    GAO's statement applies its considerable body of work on 
statistical issues to four questions the Subcommittee asked on data 
quality and the decentralized U.S. statistical system.
    While the principal statistical agencies GAO has reviewed have 
generally adhered to applicable professional standards, there are 
reasons to be concerned about the quality of statistical data. Public 
and private sector experts have said that the current system needs a 
more coherent approach to measurement of investment, productivity, and 
services. Measurement problems, such as those concerning consumer 
prices, can affect budget and economic policymaking. GAO's work has 
also demonstrated a deterioration in the quality of the decennial 
census, which GAO designated as a high-risk area in February 1997.
Effects of the Decentralized Structure
    Although GAO's work does not indicate the extent to which the 
decentralized structure is a major cause of the quality problems, it 
does show that the decentralization contributes largely to other 
problems, such as inefficiency, the lack of national priorities for 
allocation of resources, burden on data users and providers, and 
restrictions on the exchange of data among statistical agencies. For 
example, in part because of the inability to share data, both Census 
and the Bureau of Labor Statistics have compiled and maintained their 
own lists of businesses.
Potential Effects of Consolidation
    GAO has compared the dispersed U.S. system with Canada's 
centralized system. The head of Statistics Canada has a higher level 
position than that of the U.S. Chief Statistician, can set and change 
priorities and shift resources easily, has access to all of the 
government's administrative records, and can share survey data 
internally under strict and uniform privy requirements. Potential 
disadvantages associated with consolidation would include possibly 
responsiveness to the needs of former parent departments and possible 
objections to the concentration of data in a single agency.
Benefits from Location in the Commerce Department
    Commerce historically has not been managed on the basis of a 
unifying mission or shared goals and has decentralized its key 
administrative functions. While the Commerce relationship is not 
meaningless, GAO is not aware of any reasons that would prevent Census 
and the Bureau of Economic Analysis from performing their missions as 
part of another department
                               __________
    Mr. Chairman and Members of the Subcommittee: We are pleased to be 
here today to discuss the Federal statistical system. Over the years, 
we have developed a considerable body of work on statistical issues. 
The related products list that follows my statement contains our most 
recent products. As you requested, our testimony today brings this body 
of work to bear on four issues you asked us to address: (1) the quality 
of Federal statistics, (2) how the Federal statistical system's 
decentralized structure affects statistical quality, (3) whether 
consolidating the statistical functions currently housed in the 
Department of Commerce with those of other Federal agencies could 
provide a more streamlined and effective Federal statistical system, 
and (4) whether or not the Bureau of the Census and the Bureau of 
Economic Analysis benefit from being housed in the Department of 
Commerce.
Background
    Statistical activities are dispersed throughout the Federal 
Government. The Office of Management and Budget (OMB) has identified 70 
Federal agencies that each spend at least $500,000 annually on 
statistical activities. Together, these agencies requested over $2.75 
billion for fiscal year 1997 for statistical activities. Of the 70 
agencies, 11 are considered to be the principal statistical agencies 
because they collect, produce, and disseminate statistical information 
as their primary mission. These 11 agencies together spend 
approximately $1.2 billion annually on statistical activities. Two 
Commerce agencies--the Bureau of the Census and the Bureau of Economic 
Analysis (BEA)--and the Department of Labor's Bureau of Labor 
Statistics (BLS) account for about $825 million of this total.\1\
---------------------------------------------------------------------------
    \1\ The other eight principal statistical agencies are the National 
Center for Health Statistics (in the Department of Health and Human 
Services), Energy Information Administration (in the Department of 
Energy), National Agricultural Statistics Service and the Economic 
Research Service (both in the Department of Agriculture), Statistics of 
Income Division (in the Internal Revenue Service in the Department of 
the Treasury), Bureau of Justice Statistics (in the Department of 
Justice), the Bureau of Transportation Statistics (in the Department of 
Transportation), and the National Center for Education Statistics (in 
the Department of Education).
---------------------------------------------------------------------------
    The missions of the principal statistical agencies are to ensure 
that the statistical information they collect, produce, and disseminate 
is accurate, reliable, and free from political interference and impose 
the least possible burden on individuals, businesses, and others 
responding to requests for data. Most of the other agencies that 
produce and disseminate statistical data do so as an ancillary part of 
their missions.
Quality of Statistical Data
    The principal statistical agencies have done many things well. For 
example, in August 1995, we reported that four statistical agencies we 
reviewed--Census, BEA, BLS, and the National Center for Health 
Statistics--generally adhered to applicable professional standards.\2\ 
Nevertheless, a series of studies of the Federal statistical system, 
going back several decades, have identified concerns over the quality 
of statistical data. One of the concerns is that economic statistics 
have not kept pace with changes in the economy. This has led some 
experts to question whether current statistics adequately reflect the 
importance of international transactions to the economy, or whether 
current productivity measures are adequate given the increase in 
importance of service industries. Experts who have worked in the 
Federal statistical system have also said that the current system needs 
to update its approach to measuring savings and investment. We are 
finding that agencies are devoting more attention than ever to the 
quality and coverage of statistical data series as they search for 
appropriate outcome-based performance measures in their efforts to 
comply with the Government Performance and Results Act that originated 
with this Committee.
---------------------------------------------------------------------------
    \2\ Statistical Agencies: Adherence to Guidelines and Coordination 
of Budgets (GAO/GGD-95-65, Aug. 9, 1995).
---------------------------------------------------------------------------
    In 1991, the Economic Statistics Initiative, which was led by 
Michael Boskin who chaired the Council of Economic Advisers under 
President Bush, made 38 recommendations to address well-known problems 
in economic statistics for which action was feasible in the near term. 
Among the recommended actions were (1) accelerating improvements in 
estimates of international trade in services, including financial 
services; (2) better measuring service sector production and prices; 
(3) separating quality and inflationary changes in prices; and (4) 
making it easier for statistical agencies to share data for statistical 
purposes. In reviewing the status of these recommendations, we found 
that only about half of the recommendations were funded and that the 
funding levels varied considerably among the different agencies 
producing economic statistics, thereby hampering improvement efforts? 
\3\
---------------------------------------------------------------------------
    \3\ Economic Statistics: Status Report on the Initiative to Improve 
Economic Statistics (GAO/GGD-95-98, July 7, 1995).
---------------------------------------------------------------------------
    We reported in 1995 that measurement problems can affect budget and 
economic policymaking.\4\ In that report, we pointed out that many of 
the studies we reviewed indicated that technical problems associated 
with the development of the Consumer Price Index could cause it to 
overstate inflation. We also pointed out that measures of economic 
output and productivity failed to account for the increasing importance 
of the service sector to the nation's economy.
---------------------------------------------------------------------------
    \4\ Economic Statistics: Measurement Problems Can Affect the Budget 
and Economic Policymaking (GAO/GGD-95-99, May 2, 1995).
---------------------------------------------------------------------------
    In February 1997, the Nation Association of Business Economists 
(NABE) reported that nearly 70 percent of its members who responded to 
its survey were dissatisfied with the scope and quality of economic 
data in the United States. NABE said that the current system does a 
better job of measuring manufacturing than it does of measuring 
services and the information technology aspects of the economy.
    Our work has also demonstrated a deterioration in the quality of 
the decennial census, which provides a baseline for countless other 
statistical programs. The 1990 Census, though it was the most expensive 
in history, for the first time produced results that were less accurate 
than those of the preceding census.\5\ Almost 10 million persons were 
missed in that census, although the net effect of this was somewhat 
masked by the counting of about 6 million persons twice. These 16 
million gross errors represent a minimum tally, since they do not 
include such errors as persons erroneously included or assigned to the 
wrong locations. In February 1997, we designated the 2000 Decennial 
Census as being at high risk of producing unsatisfactory results.\6\
---------------------------------------------------------------------------
    \5\ Decennial Census: 1990 Results Show Need for Fundamental Reform 
(GAO/GGD-92-94, June 9, 1992).
    \6\ High-Risk Series (GAO/HR-97-2, Feb. 1997).
---------------------------------------------------------------------------
How the Decentralized Structure of the Federal Statistical System 
        Affects Statistical Quality
    Over the years, a number of problems with the quality of 
statistical data have been associated with the organizational structure 
of the Federal statistical system. Although our work does not indicate 
the extent to which a decentralized structure is a major cause of the 
quality problems, it does indicate that not all of the quality problems 
that exist stem from the decentralized structure of the statistical 
system. For example, the deteriorating quality of decennial census data 
relates largely to limitations in the basic processes used to collect 
census data, not to the decentralized structure of the statistical 
system. On the other hand, our work as well as that of others has shown 
that the decentralized structure of the system contributes largely to 
other problems, such as inefficiency, the lack of national priorities 
for allocation of resources, burden on data users and providers, and 
restrictions on the exchange of data among statistical agencies.
    Clearly, our decentralized statistical system has sometimes 
affected the quality of statistical data produced by the system. For 
example, in estimating the National Income and Product Accounts (NIPA), 
which includes the estimate of gross domestic product, BEA relies on 
data collected by other agencies. Frequently, those data are collected 
for other purposes, and according to a BEA official, much of the data 
are not in the form that BEA needs to calculate NIPA. In some cases, 
gaps exist in the data, and these gaps, in turn, affect the NIPA 
estimates. As another example, some of the data quality problems that 
were identified by the Economic Statistics Initiative have yet to be 
corrected because the corrective action requires steps by more than one 
agency. In some cases, one agency received funding to correct its data 
problems, but another agency, which may contribute source data, did not 
get funds to address the issue.
    Many experts have expressed concern about inefficiencies in the 
statistical system due to its decentralized structure. The experts 
often cite duplicative or overlapping data collection activities or 
system infrastructure, such as field structures; computer systems; or 
administrative, technical, and program personnel as sources of 
potential cost savings. Those who have studied the systems, however, 
often disagree on how much could be saved through consolidation. In 
this regard, we have noted that many agencies have used reimbursable 
agreements with other agencies, such as the Census Bureau, to handle 
their data collection activities, thereby avoiding having to establish 
and maintain their own systems and structure for these purposes.\7\ 
These types of arrangements would tend to limit the savings that could 
come from consolidation. Further, we are not aware of any savings 
estimates that have been verified by an independent party.
---------------------------------------------------------------------------
    \7\ Federal Statistics: Principal Statistical Agencies' Missions 
and Funding (GAO/GGD-96-107, July 1, 1996).
---------------------------------------------------------------------------
    The lack of an effective mechanism for setting national priorities 
for the Federal statistical system has been another concern expressed 
over the years about the system's decentralized structure. Our work as 
well as work done by others has shown that the United States has lacked 
an effective apparatus for setting national priorities for use of the 
statistical agency resources. For example, in August 1995, we reported 
on limitations on OMB's ability to coordinate the budgets of 
statistical agencies.\8\ A number of factors contribute to the lack of 
clear national priorities for the U.S. statistical system. One of these 
factors is the nature of the budget formulation process, in which each 
statistical agency has its own budget which has been generally 
determined in the context of the competing needs and priorities of 
other components within its home agency or department, as opposed to 
the needs and priorities of the overall Federal statistical system. 
Another related factor is the dispersion of responsibility among 
multiple congressional committees and subcommittees for authorizing, 
funding, and overseeing the statistical agencies.
---------------------------------------------------------------------------
    \8\ Statistical Agencies: Adherence to Guidelines and Coordination 
of Budgets GAO/GGD-95-65, Aug. 9, 1995.
---------------------------------------------------------------------------
    Another problem arising from decentralization is the increased 
burden on data providers as a result of duplicative data collection 
efforts. For example, Janet Norwood, a former Commissioner of Labor 
Statistics, has identified surveys that she believes could be 
consolidated. She believes that the consolidation of such surveys would 
reduce cost as well as burden to survey respondents while improving the 
possibility for integrating the data collected. At least to some 
extent, overlap in the types of information asked for in surveys has 
occurred because of the decentralized structure of the statistical 
system.
    Another related factor that contributes to the overlap problem is 
the inability of statistical agencies to share data with one another 
because of legislated confidentiality restrictions. Federal statistical 
agencies generally operate under a number of laws, policies, or 
regulations that govern the collection, use, and confidentiality of the 
statistical information for which these agencies are responsible. Some 
of these laws, policies, and regulations apply only to a specific 
agency. The legal framework for the Federal statistical system also 
limits the extent of data sharing among agencies because statutes exist 
to protect the confidentiality of data providers and, in many 
instances, allow only the agency collecting the data to have access to 
them. For example, in part because of the inability to share data, both 
Census and BLS have compiled and maintained their own lists of 
businesses.
Potential of Consolidation to Provide a More Streamlined and Effective 
        System
    You asked whether consolidating the statistical functions currently 
housed in the Department of Commerce with those of other agencies could 
provide a more streamlined and effective Federal statistical system. To 
respond to your question, we drew on our work comparing the 
decentralized U.S. system with Canada's centralized system. The 
Canadian statistical system is often used as a reference point for 
considering proposed consolidations in the United States and is highly 
regarded in the international statistics community. However, there are 
some differences between the United States and Canada that need to be 
considered when making such a comparison. Also, there may be 
disadvantages associated with a consolidation, and there are 
alternative approaches to making the system more streamlined and 
effective.
    A consolidated agency could help streamline and improve 
effectiveness in a number of ways. For example, better quality data 
could be achieved by bringing together the expertise needed to address 
important issues, such as the use of common data collection methods and 
more efficient survey designs, so that data that are produced are based 
on similar concepts, time periods, and classification structures. Cost 
savings and reduced burden on data providers may be achieved through a 
greater sharing of data and agency resources in a consolidated agency, 
thereby avoiding duplication and enabling greater integration. 
Consolidation could also resolve the issue of setting national 
priorities and achieving greater coordination for the system to the 
extent that a head of the proposed consolidated agency would be able to 
set priorities for the use of its funds and require its components to 
cooperate with one another.
    Our August 1996 report comparing the Canadian statistical system 
with the U.S. system offers some insights on consolidation.\9\ While we 
did not evaluate the effectiveness of the Canadian system, we did 
identify several clear differences between the Canadian and U.S. 
systems in our review. The Canadian system is much more centralized, 
with Statistics Canada containing many of the activities currently 
divided among the principal U.S. statistical agencies and being 
responsible for the majority of the government's statistical 
information. The head of Statistics Canada has a higher level position 
than that of the U.S. Chief Statistician, has direct control over the 
agency's budget request, and can set and change priorities and shift 
resources easily. Statistics Canada also (1) has access to all of the 
government's administrative records, (2) can share survey and other 
data among its components and other government agencies and 
nongovernmental organizations, (3) has consolidated technical and 
administrative support functions, and (4) is subject to strict and 
uniform privacy requirements. According to Statistics Canada officials, 
these privacy requirements also help cure a high voluntary response 
rate to data collection efforts.
---------------------------------------------------------------------------
    \9\ Statistical Agencies: A Comparison of the U.S. and Canadian 
Statistical Systems (GAO/GGD-96-142, Aug. 1, 1996).
---------------------------------------------------------------------------
    While Canada's centralized system may appear to offer several 
advantages over the U.S. system, several factors need to be considered 
as part of the comparison. Canada's parliamentary system of government 
may lead to a clearer definition of government policy and priorities 
and the ensuing needs for statistical information than our system, 
which institutionalizes tension between different branches of 
government. The United States is a much larger nation and has a larger 
and more complex economy than Canada. Also, the Canadian statistical 
system is much smaller than the U.S. system. For example, the fiscal 
year 1997 budget for Statistics Canada was about $246 million (in U.S. 
dollars) compared with the nearly $1.2 billion budget for the U.S. 
principal statistical agencies. Finally, the Canadian public has 
accepted that a government agency will have broad access to all 
government records for statistical purposes.
    On the other hand, disadvantages may also be associated with a 
consolidation. For example, the consolidated agencies could be less 
responsive to the needs of their parent departments from which they 
came and their constituencies. Another potential disadvantage is the 
potential for abuse, such as breaches of confidentiality, that could 
occur when so much information about individuals and businesses is 
concentrated in one agency. Finally, some of the benefits expected from 
consolidation are unlikely to materialize unless the components of the 
consolidated statistical agency are authorized to share data and if 
legislative responsibility for the consolidated agency continues to be 
dispersed among multiple congressional committees. In addition, the 
extent to which benefits of a consolidation could be realized would 
depend on how comprehensive the consolidation is. If significant 
statistical activities remain outside the consolidated agency, some of 
the problems of inefficiency and priority setting in the statistical 
system could persist.
    Given the potential drawbacks of consolidation, the Subcommittee 
may also want to consider alternative approaches for improving 
statistical data collection and analysis. One option would be to 
consider alternatives to the dominant paradigm of having Federal 
employees collect, analyze, and disseminate information through the use 
of appropriated funds. Alternatives might be privatizing at least some 
aspects of data collection, analysis, or dissemination; additional 
contracting out; or the imposition of user fees. We have not explored 
such alternatives for the Federal statistical system and are therefore 
not in a position to elaborate on them.
    Concerning data sharing, one step could entail enacting legislation 
that allows statistical agencies to share data and information with 
appropriate safeguards to protect against breaches of confidentiality. 
Proposals to enable greater data sharing among statistical agencies 
have been made in the past; both the Economic Statistics initiative 
under President Bush and the National Performance Review under 
President Clinton have recommended such actions. These proposals were 
not adopted, in part because of general concerns that greater data 
sharing might endanger the privacy of individuals. In 1996, OMB and the 
Department of the Treasury sent to Congress proposed legislation that 
would permit limited sharing of data among designated statistical 
agencies for statistical purposes subject to procedural safeguards 
contained in the proposals. Although Congress did not enact the 
legislative proposals, OMB officials have told us that the 
administration plans to submit these data sharing proposals in 1997. We 
as well as others who have studied or are knowledgeable about the 
Federal statistical system believe that the inability of statistical 
agencies to share data is one of the most significant issues facing the 
statistical system and one of the major factors affecting the quality 
of data, the efficiency of the system, and the amount of burden placed 
on those who provide information to the agencies. Since 1979, we have 
recommended changes to existing statutes that would enable statistical 
agencies to share data.\10\
---------------------------------------------------------------------------
    \10\ After Six Years, Legal Obstacles Continue to Restrict 
Government Use of the Standard Statistical Establishment List (GAO/GGD-
79-17, May 25, 1979).
---------------------------------------------------------------------------
    Another approach to improve the current system would be to 
strengthen OMB's ability to set priorities for use of the agencies' 
funds and provide mechanisms that would enable agencies more easily to 
shift resources, including staff. The appropriations process constrains 
OMB's ability to independently make such resource shifts, and we, as 
well as others, have reported on limitations on OMB's ability to set 
priorities for allocation of funding among statistical agencies.\11\ In 
recognition of this concern, OMB launched an initiative during 
preparation of the administration's fiscal year 1998 budget in which 
some priorities were set for statistical agency funding. The effect of 
OMB's efforts, however, will not be known until after Congress 
completes the appropriations process.
---------------------------------------------------------------------------
    \11\ Statistical Agencies: Adherence to Guidelines and Coordination 
of Budgets (GAO/GGD-95-65, Aug. 9, 1995).
---------------------------------------------------------------------------
    Greater coordination among statistical agencies is another way to 
improve their effectiveness and streamline operations. In this regard, 
it should be noted that some consolidation already has taken place and 
additional efforts are underway. For example, statistical agencies have 
already acted to reduce duplication and inefficiency by collection 
information for one another. An illustration of this is the decennial 
census long form questionnaire. Ten of the principal statistical 
agencies and many other Federal agencies use information collected 
through the form as source of data for their own statistical 
activities. We reported in July 1996 that if agencies had to collect or 
arrange for the collection of these data on their own the total cost 
would exceed the cost of having Census collect these data.\12\
---------------------------------------------------------------------------
    \12\ Federal Statistics: Principal Statistical Agencies' Missions 
and Funding (GAO/GGD-96-107, July 1, 1996).
---------------------------------------------------------------------------
    OMB also has a number of coordinative efforts under way through the 
Interagency Statistical Policy Council, which OMB chairs. The council 
consists of the heads of the principal statistical agencies as well as 
representatives from the National Science Foundation and the Social 
Security Administration, and exists to foster greater coordination 
among statistical agencies. One such effort has been the development of 
the ``one-stop shopping'' service for users of Federal statistical 
data. This effort entails establishing an electronic link to all 
Federal statistical agencies through the Internet. OMB plans to have 
this service fully operational in 1997. With this system, a user should 
be able to go to one source that will identify the types of data 
available and will electronically link the user to the data maintained 
by the appropriate agency. Although OMB's coordination efforts appear 
promising, it is unclear at this point how effective they will be in 
resolving problems that result from the decentralized structure of the 
system.
Do the Census Bureau and the Bureau of Economic Analysis Benefit From 
        Location in the Commerce Department?
    In testimony before the full Governmental Affairs Committee on July 
25, 1995,\13\ we described the Commerce Department as essentially a 
holding company for many disparate programs, and subject to almost 
constant organizational changes in its 84-year history. Because of the 
wide diversity of its functions, Commerce historically has not been 
managed on the basis of a unifying mission or shared goals. Its 
components are overseen and authorized by several committees in 
Congress, none of which has jurisdiction over the entire department. 
Within Commerce, Census and BEA together account for less than 10.5 
percent of departmental obligations and less than 20 percent of 
departmental staff.
---------------------------------------------------------------------------
    \13\ Government Reorganization: Observations on the Department of 
Commerce (GAO/T-GGD/RCED/AIMD-95-248, July 25, 1995).
---------------------------------------------------------------------------
    Commerce has decentralized its key administrative functions. Major 
Commerce components--including the National Oceanic and Atmospheric 
Administration, the Patent and Trademark Office, and the Economics and 
Statistics Administration which comprises both Census and BEA--have 
been granted the authority and responsibility by Commerce for meeting 
most of their own administrative needs. Thus, Commerce headquarters 
provides some services but primarily sets policy and provides overall 
direction and oversight. In some cases, the major components pay for 
the services provided by headquarters through a working capital fund. 
Census and BEA receive their legal services this way, for instance. In 
addition, BEA purchases most of its administrative services from other 
components of Commerce through a series of cross-servicing 
arrangements. Commerce's decentralized approach to providing 
administrative services is a result of its response to significant 
budget reductions that occurred in the early 1980s. The relative 
independence of the major components minimizes the disruption that 
would occur if one or more were separated in a reorganization. Neither 
the Census Bureau nor BEA is physically housed in the Commerce 
headquarters building.
    We are not aware of any reasons that would prevent Census and BEA 
from performing their missions if they were not components of the 
Commerce Department. This is not to say, however, that the Commerce 
relationship is meaningless. In fact, Commerce officials have argued 
that the absence of regulatory programs within the department has been 
a factor in preserving the reputation for independence of its two 
statistical agencies. Because they are located in Commerce, Census and 
BEA must compete for attention and resources with other functions of 
that department, functions as disparate as weather service 
modernization, fisheries preservation, technological innovation, and 
trade sponsorship.
    The department's superior stature, resources, and access to the 
highest policy levels within the administration have at times been of 
value to Census and BEA; for example, our high-risk report on the 2000 
Census recognized that the Bureau itself was not capable of securing 
all the stakeholder decisions it needs to proceed with plans, tests, 
and commitments, and that attention from the administration was needed. 
The value of attachment to a Cabinet-level department to promote an 
agency's interests at the highest policy-making levels is well 
established in organizational theory and practice. Statistics Canada, 
for example, takes pride in its independence but it is, nevertheless, a 
component of the Department of Industry Canada. Granting the value of 
departmental affiliation, it does not necessarily follow that the 
Commerce Department is the only organization to provide it.
    Mr. Chairman, that concludes my prepared statement. I would be 
pleased to respond to questions on it or on aspects of our statistical 
policy work that I have not covered.

    Senator Brownback. Good. Thank you, Mr. Stevens. I 
appreciate that.
    Ms. Haver and Mr. Stevens, you both agree in commenting 
that we need a common form for companies to fill out. That is 
at least a minimal, least burdensome way to go on this. Mr. 
Stevens, you talk about the Canadian system as having 
significant advantages. Now, you say that based upon studying 
the Canadian system?
    Mr. Stevens. Yes, we have done a report that compares it 
with our decentralized system. We spent a good deal of time in 
Canada and compared the funding mechanisms, their oversight, 
their confidentiality provisions, six or eight different 
facets.
    Senator Brownback. And you are confident of that system 
being better organized than the U.S. statistical gathering 
system?
    Mr. Stevens. Better organized, I would say yes. But we also 
noted there are a number of differences between Canada and the 
United States. We have a much more complex economy and 
statistical system. It is more expensive. And we have a 
different political structure that is probably not as 
comfortable with having large amounts of data and data 
collection capabilities in one centralized place. So there are 
some differences.
    Senator Brownback. But let me ask you about that last 
statement. Why shouldn't this political system be comfortable 
with having numbers in one place? If we are confident with the 
accuracy of the numbers, why would we be discomforted by them 
being at one place or many?
    Mr. Stevens. I do not have a reason for that. It is a 
matter of values, and I think that is a value that our 
political system has put up against the data collection system. 
The Internal Revenue Service, for example, has got an awful lot 
of good data, but there are a lot of suspicions about it among 
Americans as to what it does with it and what controls should 
be imposed upon it. It is not an area without controversy.
    Senator Brownback. So as a professionals, you have no 
problem, and you would think if people look at this as 
professional, they would have no problem with consolidation of 
numbers coming from one point.
    Mr. Stevens. No, professionally, not at all, and the 
Canadians pull it off without any controversy at all, I 
believe.
    Senator Brownback. Ms. Haver, what about you? How do you 
look at that as a private professional?
    Ms. Haver. Well, actually, my company uses a great deal of 
Canadian data. We find their standards are very high, and 
essentially their programs are very impressive.
    From the point of view of one place for our data, I think 
if that place is one that is not in an organization that has 
regulators, most companies would be much more comfortable with 
it than having it as it is right now in the Department of 
Labor, for example, with BLS. But I do not think the whole 
issue is as important as I hear government people telling me it 
is, very frankly.
    Senator Brownback. That BLS collects the numbers and also 
regulates?
    Ms. Haver. Well, BLS does not regulate. They----
    Senator Brownback. But I mean the Labor Department 
regulates.
    Ms. Haver. Right.
    Senator Brownback. Are companies deeply concerned about 
that?
    Ms. Haver. I personally--I run a company, and I am not 
concerned about that at all, and I think most--at least small 
businesses actually do not know the difference between BLS, 
Census, or BEA, very frankly.
    Senator Brownback. And they really do not care. It is just 
somebody hassling them for numbers. I mean, we used to have 
that problem when I ran the Agriculture Department on the State 
level. People are tired of giving you information all the time. 
They do not get much of anything out of it that they see 
specifically. I mean, they get the general number, but----
    Ms. Haver. Well, I think a lot of businesses use these 
numbers, and they are very valuable. But it is true the people 
that use them and the people that give them are often not one 
and the same.
    I also believe--and this is just my ideal--that if we had a 
consolidated agency, we could start looking at all of these 
surveys we do of companies and figure out what we need. We 
could say to companies: Here it is, can you provide this and 
you program your computers to produce this form, we are not 
going to change it for 5 years. Because once that is done and 
it just comes out of your system, it is not as troublesome as 
getting a form where you are supposed to write these things 
down. And companies still do get such forms.
    Senator Brownback. Good. I thank you all very much. Did 
anybody have anything else to add?
    [No response.]
    Thank you all very much. I appreciate this. If you have any 
additional thoughts, please feel free to send them on in to us.
    Without further ado, we are adjourned.
    [Whereupon, at 3:17 p.m., the Subcommittee was adjourned.]