Examining the Partnership between Federal Statistics and National Policy
On October 20, Kenneth Prewitt (left), editor of the September 2010 Annals volume The Federal Statistical System: Its Vulnerability Matters More Than You Think, joined Census Director Robert Groves (center), and Bureau of Economic Analysis Director J. Steven Landefeld (right), for a Capitol Hill policy briefing on the importance of the nation’s statistical system. U.S. Chief Statistician Katherine Wallman moderated.
The special program exploring the science and policy behind the nation’s statistical system was held to celebrate the globe’s first World Statistics Dayinspired by the United Nations. The program, titled “National Statistics: Sound Science, Sound Policy, Strong Democracy” was designed to communicate the importance of a high quality federal statistical system. AAPSS was a cosponsor of the program, which coincided with the Academy’s recent release of the Annalsvolume on The Federal Statistical System. The other organizing cosponsors included the American Statistical Association, the Committee on National Statistics, the Council of Professional Associations on Federal Statistics, the Consortium of Social Science Organizations, the Interagency Council on Statistical Policy, and the National Institute of Statistical Sciences.
Katherine Wallman, author of the Annals article Federal Statistics: Understanding a Crucial Resource, opened the event with a message from President Barack Obama praising the value of statistical data and its far-reaching impact on the nation. The president stated in his message that statistical data “guides representation in the United States Congress; informs our economic, social service, and national security outlook; and helps determine where infrastructure like schools, hospitals, and roads should be built.”
After Wallman concluded her own remarks, she turned the floor over to Kenneth Prewitt, former director of the U.S. Census Bureau, currently the Carnegie Professor of Public Affairs at Columbia University. Prewitt explained the motivation behind The Annals volume (explored in his introductory chapter Science Starts Not after Measurement, but With Measurement ) as the great need for federal statistics to be viewed “through the lens of science.” He added, “The inability of this country to see that obvious fact is every bit as much a danger to the federal statistical system as the thing we worry about a lot, political interference”—a topic he explored in What is Political Interference in Federal Statistics?
Prewitt also addressed the growing emergence of digital data and the government’s role with regard to privacy, representation, and quality. He concluded his presentation with a call for the government to elevate the status of the U.S. Census Bureau as a scientific agency, so that it would gain increased independence enjoyed by the nation’s other major scientific agencies.
Current U.S. Census Bureau Director Robert Groves delivered the next presentation about the role statistics play in a democracy and their impact on the electoral system. He said, “Such statistical information that we supply collectively, therefore, is the source of strength of a people in a society and exercising their responsibilities.” Groves further pointed out why statistical information must remain non-political. He said, “Key to the success of this enterprise, all of us here know, is that the numbers we produce— the estimates we produce— are viewed as credible…For this reason, statistical agencies have to be vigilant to separate their activities, as Ken [Prewitt] said, from policymaking and program execution that occupies other agencies.” Groves’s contribution to the Annals volume was The Structure and Activities of the U.S. Statistical System: History and Recurrent Challenges.
The final presentation was given by J. Steven Landefeld, director of the Bureau of Economic Analysis. His talk delved into the history behind the emergence of national accounts data, in particular, the Gross Domestic Product or GDP. He described GDP as “a mosaic of data that come from all the other agencies,” including the Census Bureau, Bureau of Labor Statistics, the Internal Revenue Service’s Statistics of Income group and more than a dozen other federal statistical agencies. Landefeld drove home the point that the partnership between statistics and policy has “helped sustain economic growth and raise living standards around the world.”
The following is a transcript of the program, which may also be listened to or downloaded as a podcast.
Katherine Wallman, Chief Statistician of the U.S., Office of Management and Budget: I am Katherine Wallman. I am the Chief Statistician for the [United States] government, and I am very pleased to have the opportunity to be here to welcome you all to 20-10-2010, which in Euro-speak that is the way we do the calendar, and it is deemed by the U.N. as World Statistics Day.
The U.N. Statistical Commission actually is one of the oldest parts of the United Nations, it has been around since the beginning of the U.N. If you took your U.N. tour in high school, as I did, and walked along the upper decks there you probably did not know you might have been actually observing the work of the U.N. Statistical Commission, which has been doing yeoman’s work mostly in the area of classification and technical cooperation for more than six decades. The current director of the office there in New York thought with the coming of 20-10-2010 this would be a smart way to celebrate the work of statisticians around the world.
Our purpose today in coming to the Hill was to hopefully see some old friends here on the Hill and also to make some new ones, and to let you know informally, mostly informally, about the work that is carried out by the agencies in our decentralized Federal Statistical System. There are about fourteen, depending upon which day and how I am counting, about fourteen agencies that do statistics as their primary line of work in the government and all of those agencies are represented here today. Indeed, twelve of the fourteen are represented by the heads of the agencies and I hope you will take time to meet them if you have not already.
I would like then to finish my part of this presentation, before I turn to the real presenters, with something very special: a message that I would like to share with you. It says:
I am pleased to send greetings to all those celebrating World Statistics Day. Statistical data drives countless decisions which impact our nation. It guides representation in the United States Congress, informs our economic, social service, and national security outlook, and helps determine where infrastructure like schools, hospitals, and roads should be built. The Federal Statistical System including, among others, the United States Census Bureau, the Bureau of Labor Statistics, the Energy Information Administration, and the National Agricultural Statistic Service, has been pivotal to our progress as a nation. I commend all of the hard-working women and men who are part of this endeavor as their efforts to provide critical data have helped maintain and improve our country. As you gather to acknowledge the Global Statistical System and raise public awareness of statistics, I wish you all the best. – Barack Obama
With that, I would like to turn to the others on the program. Our first speaker we are thrilled to have here for a few minutes (he is in between something) is Ken Prewitt. Ken is currently the vice president for global something at Columbia University (sorry, he will fix that as he often does). He is known to many of you in the social science research community for positions he has held in the Social Science Research Council, the National Opinion Research Center, and a couple of foundations, and we were extremely lucky to have Ken here to conduct, or sit at the helm I should say I suppose, of Census 2000, and Ken also was the mover behind the last get-together of this sort we had actually, in May a year and a half ago or so at the National Academies. I called it a celebration of statistics, I am not sure that was its official name, but from that has come a volume, The Annalsvolume, that is available for you as a takeaway and Ken has a few words that he would like to share with us this morning.
Kenneth Prewitt: I want to just remind us of the motivation for this issue of The Annals that we assembled a large number of people to write pieces for and so on. The motivation was the attempt to state clearly, clearly, clearly that federal statistics have to be understood as part of the nation’s scientific infrastructure. They have enormous program responsibilities but the data also have deep scientific responsibilities and that it really weakens the Federal Statistical System not to see them through the lens of science, as well as the lens of programatic responsibilities. And I think in some respects the inability of this country to see that obvious fact is every bit as much a danger to the Federal Statistical System as the thing we worry about a lot, political interference, and so forth and so on. And I simply want to make that point again, this volume is organized sort of around that theme, and I want to take my two or three minutes here to say why it particularly strikes me as important today.
Here is a piece from the Financial Times of just last week, and you cannot read it, of course, but the title is Government By Search Engine, and I will just read one sentence from this article, journalistic hyperbole notwithstanding, nevertheless, “If Google is better at predicting inflation, unemployment, and influenza, it will probably be better at predicting crime, terrorism, and political unrest. What government that claims to be vigilant on these issues would fail to use Google’s data?” Now, again, journalistic hyperbole notwithstanding, there is a tremendous flood of digital data out there and I do think that within the next quarter of a century we will find the government gravitating to more and more use of that and that will, in part, come at the expense of survey/Census-based data. We already see the track we are on with respect to administrative data, and one of the things that this journal argues is that the other sources of data which sit out there fail to meet three high standards. They are much, much worse on privacy issues than certainly the statistical system is. They are much, much worse on quality issues, they do not think about data quality; we have been thinking about data quality for almost a century and perfecting sampling theory and question-order wording and all the kinds of things we worry about all the time. Administrative data, digital data do not have quality on the mind. And, finally, they are seriously worse on representation because they are not designed to represent the population. So, fine, we know that, that is not my real point. My real point is I really think we need a very strong Federal Statistical System to be the repository of the work on privacy, quality, and representational issues as the administrative data and the digital data move into our information systems. And for that I think we need to understand the Federal Statistical System is about science as well as about program. And getting that message across will be, I think, part of what we can do on days like this. I am, as many in the room know, extremely disappointed that we did not take a very small step to try to make the Census Bureau more independent than it currently is. I think [Daniel Patrick] Moynihan had it right many years ago when he said it would be very good for this society if the Census Bureau and maybe B.L.S. [Bureau of Labor Statistics] and B.E.A. [Bureau of Economic Analysis] was all mixed up and so forth. And so we know that it was more independent like the National Science Foundation, N.I.H. [National Institutes of Health], and so forth. The way to say that it is science is to give it the same kind of status that we give to our major scientific agencies in the country and the effort to sort of move slightly in that direction has thus far been blocked, and I am sorry about that. But I worry a lot, and I have to leave you all with this thought because you are the statistical system, that at least at the level of putting the Census Bureau director’s job on a five-year cycle, rather than at the pleasure of the President, is important. You are really blessed, we are really blessed with the quality of the person now leading the Census Bureau, who is a significant step forward over who did it in 2000, I know that because I’ve got real data on that, and it would be a real loss to the country and to the statistical system if there were turnover administrations in 2012 and we lost the Census Bureau who is willing to be down here for a decade and do this hard work because we had not managed to get that bill through. So I leave that with you. Do not let loose of the idea that we need to create a statistical system which has independence in the classic sense, but that is not what I am worried about. I am really worried about understanding it as a repository of the key issues that have to do with privacy, data quality, and data representation issues and unless the statistical system itself is strong, those things will be dealt with in other parts of our government, and that is not a good place to be. Thanks.
Katherine Wallman: Thank you, Ken. Our second speaker favoring us with brief remarks this morning (he told me he only has forty PowerPoints), he is the current director of the Census Bureau whose praises were just sung by the former director, Ken Prewitt. For those who do not know Bob Groves, it is my pleasure to introduce him. He is currently the director of the Census Bureau. I know his current title; his previous titles were mostly at the University of Michigan, although he also was the founding father/director of the Joint Program in Survey Methodology, which is very important to the future of our Federal Statistical System, and he did do actually a brief tour at the Census Bureau at a lower level some time back, but it is my pleasure to introduce Bob Groves to you.
Robert Groves: I hope, like me, you feel slightly silly celebrating World Statistics Day but also slightly happy about it. I am told there will not be a parade on Constitution Avenue of statisticians. Or I should say there probably will not be. And there were questions about whether we got the day off or not and I had to put the kibosh on that. But we are also here for…there is one other fun thing that is happening probably as we speak. At Carnegie Mellon they are creating a human histogram. I just cannot wait to see the photos of that to see what it looks like. But we are also here for some serious things and I want to speak on that kind of note, I guess, for a few minutes. Alexander Hamilton, in the 21st Federalist Paper, wrote the following sentence (I have left out a few words, though): “The wealth of nations depends on an infinite variety of causes,” and then he begins a list that I think has about twelve or thirteen, but they include “the nature of the government, the genius of its citizens,” and listen to this, “the degree of information they possess and the state of commerce,” and on and on and on. Statistics applied by folks who work in agencies that are represented in this room are part of that degree of information that the American public possesses. And all of us as participants in a democracy in fulfilling our civic duties have questions that that information supplies answers to. So we are all answering questions for ourselves like, “Am I better off than I was in the past?” “Is this society on a good path?” “Is my community better or worse off than another community?” “Are government actions helping or hurting the status of me and my group, the people I care about?” “Are investments that my group makes in education and welfare paying off better or worse than those of other groups?” Many of the roots of central government statistical agencies lie in notions associated with democratic forms of government. If the citizenry is to make good decisions in their electoral behavior, they need to assess the current status of the society. To the extent that things are worse, they have the obligation to vote for change. To the extent that things are better, they should support stasis. Such statistical information that we supply collectively, therefore, is the source of strength of a people in a society and exercising their responsibilities. In most societies around the world, central government agencies have arisen in democracies to provide this statistical information. And why is that? Well, since those central government agencies are controlled by the same democratic processes as the rest of the government, they can be mandated, as we are mandated, to serve the full society and that is a good thing. Key to the success of this enterprise, as all of us here know, is that the numbers we produce, the estimates we produce, are viewed as credible, they are believable by the people that see them. And I think key to that credibility is that the estimates are viewed as free from a political point of view, from diffraction through a political lens. For this reason, statistical agencies have to be vigilant to separate their activities, as Ken said, from policymaking and program execution that occupies other agencies. Now, we must admit the following observation. It is true that there is a burden connected to the collection of data from our fellow residents and, indeed, businesses and households are directly asked to produce the data that these statistics are based on. And so we have an obligation to make sure that burden is fairly spread. And just a footnote for us statisticians, that is why probability sampling is such a wonderful invention. It is a tool to ensure equity in that burden. It is appropriate that we talk from time to time about the balance of the burden of the data collection and the benefits of the data collection. This is the right thing for us to do and it is the duty of the official statistician, I think, to think of ways constantly to reduce that burden and to maximize the benefits. So, said in another way, it is not useful to think only of the benefits of the statistical information, which we tend to emphasize, or only of the burdens of the society to produce it but others sometimes emphasize; the balance of the two is critical.
So let me end by noting that there is, because of this burden, because of our role, because of the burden we impose on respondents to collect the data, we assume two obligations that are central to the core of our activities. One, we pledge, and laws enforce it, that we will never reveal the identity of respondents, we will never intervene in their lives because of their individual answers, that we do not…we in some sense are uninterested in the individual. We are interested in the characteristics of groups. And two, equally important I think, we pledge to extract the most information we can from every dataset that we extract from the population. And why do we do that? We avoid bothering people, we reduce the burden by analyzing and re-analyzing and finding new uses of data that have been supplied by the good graces of our fellow citizens. So as an informed citizenry as the engine of democracy, I think official statistics are the fuel to the word informed. We are proud I know in this room to be part of World Statistics Day, we think it is really cool, and we are prouder I know, I speak on our behalf I am confident, we are prouder of our service to this democracy. Thank you.
Katherine Wallman: Thank you, and we have one more speaker, who is Steve Landefeld, who is the director of the Bureau of Economic Analysis and has been in that role for some time. Other than me, Steve represents, I think, the longest career statistical leader in our family because his position, as mine, is a career appointment in the government and Steve has done time among other things at the Council of Economic Advisors, I think, when I first knew him. But Steve will be our final speaker. He has a little tag on that says, “Ask me about G.D.P.,” I do not know, maybe he will tell you about that now.
J. Steven Landefeld: Thank you, Kathy, that is exactly what I am going to tell you about. I want to do something a little different. We heard from Ken Prewitt sort of some future kinds of things, Bob Groves talked about some principles and practices of the system. I am going to do a little celebration of statistics by telling the story of G.D.P. and the national accounts. Now, of course, you expect me to do that because I am Director of the Bureau of Economic Analysis, but it is also because G.D.P., unlike many of the statistics that we produce in the system, is made up of a mosaic of data that come from all the other agencies, our principal suppliers being the Census Bureau, the Bureau of Labor Statistics, and I.R.S. Statistics of Income group. But we draw data from virtually every one of the other members of the Interagency Council on Statistical Policy. So this is sort of everybody’s story, if you will.
Now, the development of the national accounts and the rich set of price, employment, production, trade, and finance data that they depend on are a shining example of the development of statistics to meet policy needs. That partnership between statistics and policy have helped sustain economic growth and raise living standards around the world. Prior to the development of the national accounts in the 1930s, there was only fragmentary and sometimes conflicting data on the state of the economy. As a result, Presidents Hoover and Roosevelt and their advisors had no comprehensive information on the state of the economy with which to develop economic policy during the Great Depression, and they were left with indicators such as manufacturing production, factory employment, department store sales, and freight car loadings. In response to this critical gap in data, the Department of Commerce worked with Simon Kuznets of the National Bureau of Economic Research to develop a comprehensive and consistent measure of economic activity. This measure was based on national income, or what we now cause gross national income, in the aggregate and by industry. These national income accounts were delivered to the Congress in 1934, one year after they had requested them. Regular annual estimates were begun in 1935, and to show you the importance of these estimates by January of 1936 President Roosevelt used them in his address to the Union and they were intensively used in budgeting purposes throughout the Great Depression. World War II came along and the planning needs associated with it led to an extension of the counts to measure production in the aggregate or measure production as gross national product today. We not only had an aggregate measure but we had one by the components of G.D.P., for those of you who suffered through an undergrad Econ 101 course, the familiar C+I+G+X=M. So we had that for all of you. They were introduced in 1942 and, again, immediately used for war and post-war planning activity. Over time the national income and product counts have responded to business and policy needs to develop a rich set of integrated national, international, regional, and industry accounts. The national accounts and the associated data, in coordination with better-informed policies and institutions, including such things as deposit and unemployment insurance, have contributed to a reduction in the severity of business cycles and a post-war era of unprecedented economic growth. The last of the long string of depressions recorded in economic history was in the 1930s. During the last “Great Depression,” although many others were called great before that one, real G.D.P. dropped by almost twenty-seven percent and unemployment reached twenty-five percent. Since then the average decline in real G.D.P. during what we now call recessions, to denote their much smaller magnitude, was two percent and unemployment average six percent. Today, G.D.P., C.P.I., and a host of census indicators are key inputs into the nation’s economic decision-making. For example, B.E.A.’s data on G.D.P. and other macroeconomic data formed the baseline for the Federal Reserve Board projections and monetary policy. Data on real G.D.P. growth, inflation, wages and salaries, and other incomes are critical inputs into the federal budget. For example, according to the Office of Management and Budget and the Congressional Budget Office, a one-half of one percent lower real G.D.P. trend growth will raise the federal budget deficit by over one and a half trillion, larger than many of the policies we think about as affecting the budget. National accounts are also used in the allocation of funds to assure open and equitable distribution. For example, our state personal income per capita data are used to allocate over three hundred billion dollars in federal funds to states and localities. Virtually every state and many localities in the U.S. use our data for the allocation of funds. Some, by law, use them as spending caps and they use them for planning. Now, I should note that G.D.P. accounts are not a uniquely American phenomenon. In Canada, in the United Kingdom they were being developed in tandem with the U.S. accounts and used intensively in the war and post-war period and had developed on pace with the U.S. accounts over time and became, not only for the U.S. but for all countries, the mainstay of macroeconomic analysis. In short, the history of the national accounts illustrates the importance of government providing its people with accurate and objective statistics relevant to the decisions made by their government.
And in order to end with something more eloquent than my own words, I close with three quotes that support the central role of statistics in a democracy. First, Thomas Jefferson, “Information is the currency of democracy.” Second, Abraham Lincoln, “Let the people know the facts and the country will be safe.” And third, an important reminder from Mark Twain, “A lie can travel halfway around the world while the truth is putting on its shoes.” Thank you very much, and may each of you have a happy World Statistics Day.
National Statistics Briefing Podcast