tv Discussion Focuses on Government Statistics CSPAN March 2, 2017 11:03pm-12:17am EST
be in the profession that we are in, to be journalists. you are the best of the best. thank you so much. [applause] >> thank you. >> tomorrow on c-span after washington journal, a discussion about the u.s.-russia relationship at george washington university. and in the afternoon, an interview with the chief advisor with the president of afghanistan live from the center for strategic international studies on c-span two. rebecca blank is a former acting commerce secretary from the obama administration. she and martin feldstein, former chair of the -- chair in the reagan administration, discussed government collected data. this runs a little over one
hour. [crowd murmuring] >> good afternoon. i am president of the american enterprise institute and i am delighted to welcome you to this session, the vital role of government statistics. i've been president for the last 8 1/2 years. i love this place for a lot of reasons. the number one reason today is because only here is a conference on government statistics more popular than the rolling stones. [laughter] that is a great thing. this is a topic that doesn't normally get headlines. in a time when the phrase "alternative facts" has entered the american
lexicon, is anything more important than what we're going to talk about today? at a.e.i., we're completely dedicated to the proposition that rigorous, rock solid, empirical data are essential for evaluating the success and failure of public policy. and for ensuring that we're fulfilling the true purpose of our work, which is to help more people live a better life. we're delighted that a.e.i.'s partnering with our neighbors, the brookings institution, and their hamilton project, to examine this critical subject and we're co-releasing a new paper on the topic and this is the kickoff event for that co-issued paper. i want to introduce our panel today and then after that, our keynote speaker, or my co-introducer, at least for this day. our panel today starts with martin frontline -- feldstein of harvard. rebecca blank, the chancellor of the university of wisconsin-madison. ellen davis from the national retail federation. david lenhart from the new york times.
and diane whitmore schanzenbach , a senior fellow at brookings. finally, let me introduce robert ruben. robert is the founder of the hamilton project and the leading voice on many economic issues. he's also the co-chairman of the council on foreign relations and a member of the board of -- trustees of a lot of institutions. mount sinai health system, and many others. previously, as you all know, he set aside a private sector career to serve our country during the clinton administration. first as the first ever director of the national economic asncil, the nec and then 70th secretary of the treasury. ladies and gentlemen, please welcome robert ruben. [applause] mr. rubin: thank you. for those of you frightened by the notion that i am going to be
keynote speaker, i'm not going to be a keynote speaker. i had about 30 minutes of comments and i thought everybody would find them fascinating but he did not seem to think so. we're going to pass on that. but i am going to make a few comments. let me welcome you and my colleagues at the hamilton project. let me make three comments. firstly, a.e.i. and hamilton project most likely have different views on a lot of subjects. but this collaboration recognizes what i think is an absolute imperative, which is that across political and substantive divides, there must be a commitment to the integrity, to the intellectual integrity, around facts, data, projections and the like. and there must be a commitment to rejecting political, ideological and tactical influence. another point with respect to a.e.i. and hamilton project working together, it is a collaboration of two
organizations with rather different perspectives on a lot of issues, but i think it makes a larger point, which is the only way our political system can work is if people on different sides of these divides work together. so i think it's a nice model. albeit a rather small example of such a model, of what's what it's going to take to make our political system work again. secondly, this event reminded me of the meeting we had with president-elect clinton. this was during the transition, before the first term. and what he said to all of us as we sat there was that he was willing to fight all day long with whoever wanted to fight with him about policy, but he never wanted to be in a position where the validity of his facts could be challenged. he lived by this dictum through his entire eight years. and while there certainly were monumental policy struggles, the fact is that there was never a serious attack on the validity or intellectual integrity, if you will, of his facts.
and that, by the way, was important not only with respect to decisions we made, but also because the markets and business community recognized that we were grounded in facts that were themselves approached with intellectual integrity. it contributed to the credibility of our decisions in both the markets and business and i think that had a lot to do with the success of our policies. finally, if we lose the absolute integrity, the absolute intellectual integrity of our facts, projections or analyses or whatever it may be, then we don't have any more reliable moorings for policy, for business or investment. i've been involved in decision making for a long -- decades. first in the private sector, and then later in the public sector. and i think the absolute fundamental of all the decision making i had the opportunity to participate in was to make sure you had your facts right.
then you could argue about them. as i said a moment ago. as president clinton said. you can argue about your policy, you can argue about whatever else it may be, but make sure you have your facts right. because if your facts are influenced by politics, by ideology, by tactics, then your decisions will be commensurately faulty. senator daniel patrick moynihan almost certainly said this best, "everyone is entitled to his own facts." but not his own that seems to me is the basis on which all of us should be involved in everything we do when it comes to decision making. let me end by saying, and arthur i think referred to this briefly, we unfortunately are in an environment today, i'd say more than unfortunately, i think actually dangerously, where there's a serious risk that political, ideological, tactical factors are going to influence data.
by data, i mean raw data, methodologies, projections and the like. and it seems to me that it is the obligation of appointed officials and elected officials to protect the intellectual integrity of our data, whatever their views on policy and the like might be. i also think there's a responsibility of the media, of financial analysts, and of policy organizations to hold our political system accountable for maintaining the intellectual integrity of our data and that is exactly the point of the event that we have today. with that, i will turn the podium over to diane. [applause] diane: thanks for coming.
the modern economy is heavily reliant on data. business, governments and families must navigate the complexities of a world made possible by new technologies and innovative business practices. it's vital to have a reliable information about the economic and social environment to make informed choices. government statistical agencies provide an important source of this information. a desire to collect data for the common good dates back to our founding fathers, when james madison argued that reliable data on agriculture, commercial and manufacturing interests would allow congress to represent the interests of its citizens more effectively. hard numbers would be useful to congressional debaters, as you can read in the madison quote on the screen, in order that they might rest their arguments on facts instead of conjectures. it's perhaps not as well understood by the public as it should be, the care and rigor with which government data are collected. stacked overwhelmingly by dedicated career civil servants, the principle statistical
agencies spend less than 1/5 of 1% of the federal budget. what they do with this is extraordinary. their data collection is rigorous. operations of the statistical agencies are conducted in accord with rigorous and binding procedures. they operate in a transparent, public fashion. mistakes are promptly corrected. and revisions are made to ensure that the highest quality data are available. agencies frequently consult with outside subject matter experts to refine and improve their measures. the statistical agencies are politically independent. political appointees at the agencies are few in number and they have very limited access to the data before its publication. the entire culture of the statistical agencies is organized around protecting the security and the confidentiality of data. data are only handled by staff on a need-to-know basis with criminal penalties in place for misuse of data. statistical agencies promise confidentiality and survey data may not be used for enforcement
actions against individual respondents. our joint document that we're releasing today highlights a portion of the breadth and importance of government statistics to the economy and to public policy. the public good provided by the statistical agencies is useful to businesses, policymakers and families. first, to the business community. in this era of big data, businesses collect and analyze vast quantities of their own internal data to forecast sales, predict needs and weigh all sorts of decisions. a firm's own data are often not enough to paint a comprehensive picture of the market and there's great value when businesses' own data are complimented with a wide range of data that are collected by the government. federal data are comprehensive, covering the entire u.s. and as a result are useful for supplementaling businesses' own data. they are also consistent with many data series spanning
decades. in particular, data from the large american community survey , or a.c.s., provides fine-grained local information about a range of issues. from demographics and economic activity and housing information. retailers and merchants make ex tensive use of these data when deciding when and where to open stores and distribution centers. target and kroger, in particular, report that they use these data to tailor product mixes and advertising appropriately across locations. a.c.s. data can be useful to understand where particular workers can be found. for example, this map shows the number of engineers by local area. i'll run through the highlights of other graphs in the document quickly. we also see the share of the within an that is
local area. -- that is elderly a local area. just one more example of the many ways the data can be cut to provide detailed information across local areas. data also collected on the distribution of visas, and data from the bureau of economic analysis, track trade, as shown in this graph, demonstrating the increase in total imports and total exports over time. we can break this down by sector. here we show the u.s. energy imports and exports over time. -- overtime, including future projections. until recently, demand was such that the u.s. relied on large net imports of energy. over the past decade, domestic production of energy shot up. as new technologies allowed extraction of previously unavailable deposits. as a result, exports rose sharply during the 2000's. the energy information administration projects that exports will exceed imports by the year 2026. in addition, our highly efficient agriculture sector has seen sharp increases in exports from grains to livestock since the year 2000. policymakers negotiating international trade agreements will benefit from a clear understanding of the economic impact of trade by sector. federal data are also a great use to policymakers. timely collection of data allows us to track important trends and evaluate public policy.
uh-oh. is there a backup way to move this? michael? [laughter] all right. so when it moves to the nest, -- next, it will be a lovely picture of the unemployment rate. [laughter] you can follow along in your books. the variety of measures of labor market health collected by the bureau of labor statistics. including the payroll jobs count, the unemployment rate, and the labor force participation rate, and more. the unemployment rate, here, i'm going to try it one more time . there we go. all right. the unemployment rate is the number of people without jobs who are currently seeking work, divided by the number of people in the labor force, rises during recessions and falls during recoveries. but it also produces a number of alternative measures to provide a more comprehensive picture on the labor mark. one of these measures is the u-6
rate, which adds discouraged workers as well as those who have part-time work but would prefer full-time work. as shown here, this more comprehensive measure is substantially higher than the traditional unemployment rate. b.l.s. data are also used to track the vitally important recent decline in labor force participation rate. which was not the next slide. also the increasing share of children living in households without married parents. critically important trend. researchers have used the data to understand the impact of pro-work policies, such as the expansion of the earned income tax credit and welfare reform on women's employment and children's well-being. they also have used it to understand the increase in participation in the social security disability insurance program. these have sparked conversations about ways to strengthen and reform the program. families gain insight from the data, too, especially around major life events such as choosing a college and choosing
a major field of study. so information on expected earnings by college major here or across occupations can be useful when making decisions about college investments. to be sure, there are important shortcomings of federal data collection. we should absolutely strengthen the data collection by addressing limitations. for example, a major limitation is that we do not collect systemic evidence on a share of population with criminal records. we also do a poor job of collecting information on the gig economy, along other shortcomings. also of great concern is declining participation in surveys and increasing rates of refusal to answer questions and underreporting of certain behaviors. proposals to augment survey data withadministrative data, appropriate confidentiality safeguards, would likely improve this, increasing data synchronization across federal agencies will likely also improve data quality and also the efficiency of the agencies. so continuing to invest in federal data collection is important to business, to policymakers and to all americans.
americans across the philosophical spectrum should recognize the importance of protecting the intellectual integrity of the data collected by the federal government. so please feel free to join in the conversation on twitter, with our #govdata, and also join us for our twitter chat at 4:00 this afternoon, which should be a lot of fun. now i'd like to invite the rest of the panel to the stage. [applause] david: good afternoon, everybody. i'm david with "the new york times." this is a topic that is a favorite of mine. i was a math major in college. and like the old joke, i raised the statistical abilities of both mathematicians and journalists when i switched from math to journalism. thank you for those opening
remarks. thank you also to michael, one of our hosts here. and congratulations to a.e.i. on this gorgeous new building. we are going to talk a little bit about this topic. we have a really nice mix of folks up here to talk to you about it. in thing we want to to invite -- we want to invite you to participate. my understanding is that at some point we will have card to circulating. please write your questions down, include your name if you're willing. i want to ask each of our panelists to give a short opening remark. i'm going to start with becky. then we'll come back tolan and go down the line. what's nice here is that in becky and marty, we have two people both from an economic background and also have been deeply involved in government. in ellen and torsten, we have people who can talk to us about the important of these data for private businesses and how it's used out there. in your opening remarks, what i
would love to hear each of you talk about is some combination of why this data is good. why we should have faith in it. also, how it's used. >> thank you. very few things that i would rather talk about than the value of statistics. i'm delighted to be part of this panel. i'm delighted with a.e.i. and brookings for putting this paper together. the united states for many years has been the gold standard for government data, and we can't take that for granted. but there are reasons why that's true. let me mention four things that we have really worked hard on in this country. the first is accuracy of the data. accuracy of the data's not about at one point in time figuring how to do it right. it's about keeping your methodology and data collection methods up to date and revising and doing research and one of the things that people often don't appreciate about the data agencies is the extent to which they're constantly involved in that type of methodlogical research, beyond just the actual collection and putting together of the data. secondly, the data has to be credible.
and credible includes political independence, of the sort that diane mentioned. but credible also includes not appearing to benefit any particular group. so, for instance, when we release things like g.d.p. data or unemployment data, that moves markets. there's certain sections of the get thatould love to data ahead of everyone else and making sure that data isn't released ahead of everyone else is a major undertaking. and i can tell you, having been at the department, the work that goes into making sure that reporters get this and request and -- and can write
their stories ahead of time but no one can file their story ahead of anyone else, it's both a difficult project and a highly important one and one that people focused on every month when the data came out. thirdly, in addition to accuracy credibility, public availability -- this is part of credibility. but it's more than that. having people be able to get their hands on the data. so we don't just release census data by county. you can go in and look at the census data in a very real way and do different tabulations with it. it's a way to cross check it. people regularly come back and say, something looks funny in this data. and we go back and look then and the government agencies and figure out, well, maybe there was a mistake, maybe something was released wrong. that type of public availability matters. and then lastly, timing matters as well. this data is going to be useful, it has to be useful in some sort of realtime. timeliness, of course, fights against accuracy. there's real tradeoffs to that. one reason we do three releases of g.d.p. and the first one we say right up front is less accurate than the last one, because it comes out faster. but it was because we needed it in a quicker time period. the u.s. statistics are far from
perfect, but we've spent decades putting this together. one of the questions i would love this audience to answer for me is, how do we do a better job of communicating broadly to the public about the value of these statistics and the extent to which we work for accuracy, credibility, public availability and timeliness? let me just say one last thing. everyone knows how you use these statistics to track the economy, to guide policy, to plan. but i want to make one other comment about how these statistics are used that i think many people don't fully appreciate, and that's the extent to which these statistics drive federal and state expenditures of dollars. so, for instance, the unemployment rate is a trigger for unemployment insurance. the c.p.i. is what determines whether your mom or grandmother's social security check are going to go up this year or not, and by how much. the state per capita income dollars drive the medicaid distributions to states. i promise you, people don't want that to change only once every 10 years. if states are seeing changes in a living population, they want that updated in realtime and that requires data. similarly, housing information is driven by section 8.
so there are literally tens if not hundreds of billions of dollars driven by these data and their accuracy truly matters. people understand, i think many of the planning ways in which the data are used, but they're used in very realtime to distribute dollars written to statutory law. that's one of the things that makes keeping these data too to -- to a gold standard so important. let me stop there. david: thank you. >> while we're talking about staff, five. to give you context about retail and the reason why our industry and our association uses government data so much and so frequently, retail is the largest private sector employer in the country. consumers represent 70% of g.d.p. so consumer spending and information is critical for not only our industry but the larger economy as a whole. 96% of retailers in this country have only one location and that's important. because you might have a retailer with tons of access to their own data or their own
metrics and government data allows us to provide parity for small businesses who are trying to understand many things the same way a large business would. the final statistic is, 32% of people in this country got their first job from a retailer. for us, as a large employer, and also as a first employer for many people, understanding what's happening with employment -- i was actually in a meeting this morning about teen unemployment and looking at the difference in rates and why and what can we do in order to get more people into those first jobs which give you, you know, the platform to find another job in the future. whether it's in retail or another industry. from our perspective, really understanding the scope of government data and applying it to many different parts of the business is truly important not only for the retailers we represent, but also to try and better understand and analyze what is happening from an economic perspective. really the two different -- the two different agencies that we end up going to the most are department of commerce data, looking at retail sales data,
and department of labor statistics to understand what's happening with employment trends. but our industry broadly looks at many different types of data sets to determine things like benchmarking employee wages and benefits, consumer expenditure patterns, understanding local demographics from consumers and trends, which diane talked about earlier. and then understanding national and regional economic factors that impact product selection. we've been working on a number of initiatives over the last few years where we have used government data to help us better articulate a narrative or better understand what could be happening. for example, our chief economist has every year posed a number of data for us to forecast holiday spending and retail sales. that's critical because there's a certain amount of -- there's a lack of bias in government data
and a longevity that's really important for us that adds credibility to our forecast and helps our members and others understand what to expect over the next several months or the next year. it also helps us establish the role and the value of retail to the economy and its employer, which is an important strategic initiative for us. and it also helps us examine consumer spending behavior, understand what's happening, how -- help our industry determine changes, where shifts might be occurring, and how to move forward. there are a number of different ways we look at this, a number of ways that both large and small retailers are using government data to make better business decisions. david: thank you. >> i'm the chief economist at deutch bank. one simple and useful way of think being the importance of government statistics is, if you're an investor and you want to invest in a country or don't want to, it becomes important the quality of the statistics. i used to work at the i.m.f. down the road. when i joined it in the 1990's,
one of my colleagues went to an emerging market, well known emerging market, and sat down with the central bank governor and discussed that inflation was a little bit too high. the central bank governor took his pen out and said, what would you like it to be? [laughter] >> and i think this tells you an d illustrates not only that anecdote, but it tells he very -- tells you very importantly that if we were a hedge fund and we had $1,000 we had to invest in something, we would then ask the question, what country do we want to invest in? do we want to invest in some emerging market? there are still a lot of challenges. the world bank and the i.m.f. are helping with this. but you still have a number of issues with getting the data -- with getting the
data. you can argue that those countries that are most challenged on getting the data together are paying a premium in borrowing costs, in everything in financial markets, in the stock market, in the exchange rate. in interest rates across the board. they are paying a price for having uncertainty about what is actually going on in this country. so that a illustrates very well as an example that the u.s. has always been the gold standard and we've known for a long time that we have tremendous data in the u.s., and we should be very proud of that. but we have a lot of issues with that data collection and emerging markets and elsewhere, when you start to discuss what is the importance of this. even in greece we've had a debate about this. in a number of emerging markets still to this day, we have debates about what is the data showing. so use that as an example for why this is relevant. when you take your 401-k and decide, do i want to invest in emerging markets? one of the problems with them, there are a in the of political problems and problems with policymaking, problems about policy uncertainty.
there's certainly also the issue about, what is going on exactly? this is a very real problem for investors, a very real certain way where this has significant impact on the returns you get. you can say you trust policymaking so you trust countries. this is absolutely critical. not only in a domestic u.s. context, but a context from a global perspective. when you think about those who are not doing it right. so the final thought i want to make, this is of course not only for the rest of the world. also in comparison, we're already so good in the u.s., so what do we need to worry about? it also becomes extremely important, again, think about your own investment decision in 401 k's and whatever you invest in. all this is driven very importantly about narratives and stories about what's going on. is this company investing in a good product? does this company have the right product? are we seeing this sector grow? what is that based on? all of that is driven by what is the data showing. the bottom-line becomes for investors, it's absolutely critical that you trust data and the u.s. has probably the lowest of any premium at all in terms
of financial markets. because we do believe that everything is right. to the fact that we're opening the door so slightly to even having a conversation such as this one is from an investor perspective, this is not just wall street, this is real money, insurance companies, pension funds and hedge funds that make decisions based on data, it is quite worrisome that we're having this debate at all, unfortunately. david: thank you. >> i'm a data guy. data buta producer of i'm a consumer of data. i look every day at the beginning of the day, once it gets past 8:30 in the morning, releases are out there and available, i look at releases that come from the b.l.s. or the b.e.a. or the fed or others. so this morning we got the unemployment insurance claims for the last week, a big drop, the lowest level during the period since the economy turned up.
but in fact, according to some estimates, it's been two decades since we've seen numbers as low as this. so, i look at these to get a better sense of where the economy is going. it's a habit that i developed when i was chairman of the council of economic advisors, and needed to have an ongoing, up-to-date sense of the data to talk about not just with the president, but with other senior members of the administration. but i continue to feel the need for this kind of data, to feel the pulse of the economy. i need it because i write about economics, is a talk to clients about it, and because i teach about it and i couldn't do it without the kind of timely data that we get from the government agencies. i was president of the national bureau of economic research for 30-plus years. i think one of the most useful things that i did was to get our
staff to create online access to these daily data. so anybody in the world can sign up for this data release. we don't make the data. we simply take the data from b.l.s. or the fed or the b.e.a., and make it available as soon as it becomes available to the public each morning. also in addition to that, an ongoing inventory of previous data releases so you can not only see the current number, but you can go back and see other numbers that have been coming out. i think that makes for a much better informed group of researchers, of people in the business world who for nothing, for free, can draw upon those data. i also teach a course at harvard
called "american economic policy," and i say to this large group of undergraduates that they should understand these data. when numbers come out, i take a minute or two at the beginning of the class to say something about the number that has come to encourageand them to go to this n.v.r. website where they can find these same kind of data. but i also remind them that they have to be very careful about interpreting the data. that the data don't always mean what they seem to mean. one of the areas of the data that i've been thinking about in detail in recent years is the way we measure real growth rates. real growth of g.d.p., real growth of personal incomes, real growth of productivity. the agencies that are involved in this do, as becky said they
do, an amazing job of collecting data from a vast number of sources following a very strict protocol. i think when they go from g.d.p. or nominal output to real output, i think there are serious problems and it which cause an underestimation of that. but i'll come back to that later in our discussion. david: let's talk about some of the issues and the shortcomings and weaknesses and flaws with data that cut across multiple -- different data series. the sort of challenges. i would list a couple things that i see. i'd be interested in whether you all agree with this list and if you do, what you think is the most important on them. so diane mentioned nonresponse. that's a really serious problem. if you look at americans' willingness to respond to polls, it has plummeted. the easiest way to think about
this is, if you see a 1-888 number on your cell phone, do you answer it? no, you don't. it's not just technology, the also the general level of trust in american society and all kinds of institutions is in decline. even if you reach people, people's willingness to answer question from the census, from the private pollster has declined. that's one. another is, it seems to me that a lot of the data we're now collecting is actually being collected by the private sector rather than the public sector. which has some advantages, the private sector is good at lots of things. but that means that the private sector isn't going to release it unless it's in their interest to do. facebook and google have enormous amounts of data that would be useful for society to have but would be damaging to facebook and google's business interests. so understandably they don't release them. a third thing i would say is a sort of strain of anti-intellectualism in society. obviously that's quite clear right now. there's always been conspiracies on the right and the left, that the data is made up. but i think they've moved from the fringes more into the
mainstream. we had jack welch, a respected c.e.o., claiming that the b.l.s. just makes up its employment data. he was making this argument seriously. it's clearly false. you have a respected c.e.o. on television making this argument. and now obviously we have a president who on a relatively regular basis makes statements about data that are also simply false. whether it's about voting or elections, whether it's about the murder rate, whether it's about the unemployment rate. and it seems to me there is a general strain of anti-intellectualism. if you hear the president saying something about the murder rate, which is false, how can you trust anything? then the fourth thing is money. we have really squeezed, despite how little money this costs, we are protecting everyone over the age of 65 from anything that has to do with budget cuts. we are now protecting the military. that leaves a relatively small portion of the government that does all of this work. that would be my list of four. i'd invite any of you to disagree with one of those four, to say that you think one is clearly the most important or to
add an item that i didn't put on there. >> so, one of the stories in the weather bureau, i think this probably did happen, was one -- the head of the weather bureau was testifying in front of congress and someone leaned forward and said, i don't understand why you need to keep collecting weather data from the u.s. government. this is a waste of money. you can get all of this on the weather channel. [laughter] >> of course, the weather channel uses the weather data, for anyone that doesn't know. i also i heard similar comments occasionally about census data. i can get this online and -- online on google. why do i need anyone collecting this? and of course we all know where it comes from originally. so there's a huge lack of understanding out there about how this data permeates our lives in enormous numbers of ways. i completely agree with your list. i think the advent of big data
and big data, not just in the private sector, but in certain areas of the public sector as well, and very high frequency data. second by second if you're talking about some stuff, whether stock exchanges or weather data for that matter. and our ability to integrate that and to figure out how we make that quite as publicly accessible and accurately use it in the less frequent statistics is i think an interesting question as well. david: how much do you all worry about funding? is it so small that it's going to be -- that it's not so much like scientific research, that you think it will be safe? or do you actually worry about whether we're going to have funds to collect this? >> i think there will be a tendency to think across the board. i think every part -- every receiver of government funds thinks that, of course, they ought to be an exception. whether it's the national
endowment on the arts or the n.i.h., how can you cut those? and all of these tend to be what economists call public goods. things where we can all benefit from it, where my benefiting from it doesn't reduce your ability to benefit from it. and so there will be a tendency to lump them all together. >> i would -- one particular pleading about one data source, which is absolutely essential to everyone else, because it creates the frame that everyone else uses for their individual surveys, which is the census, and the problem with the census is that it is not flat funding over time. it ramps up to tens of millions of dollars. then it falls off. we're at a point right now where the census has to be ramping up
if we are going to run a census. we're hitting that at exactly the point we're talking about freezes in terms of personnel and big budget cuts only discretionary side. there has to be an understanding if we're serious about the census. it has to have exceptions to that. >> for many investors, obviously big data is a very big deal. it gives you proprietary information about what's going on. some of the problems with big data, as great as it all is, it's really private information. that means that someone knows something. that's ok, people are very good at knowing something. it's not inside information. you can certainly be very skilled. but one very good example is that there are a lot of people in hedge funds who are spending time on looking at satellite images of china, even of the u.s., to figure out, is the manufacturing industry in china really as strong as they say, are they moving trucks and containers around at the speeds they're saying, and use this as an indicator for saying, i think it's getting better or worse. this is proprietary information you can buy from a private provider. you can say that's a good idea or bad idea, but it becomes more pronounced that it's private. the higher the risk that this is no belong ar public good. and therefore it is no longer insider information.
it brings with it a whole host of problems. david: do you worry in retail about, with the rise of big data, there being a real separation between your larger members and your smaller members? the smaller members just can't do this? >> yes. i think this also goes back to the point about accuracy versus frequency. if you look at large retailers, especially if you look at what's happening now with the evolution of consumer spending to the web, retailers can tell you by the second how things are doing. the large ones who track. so yes, there's some disparity between what the large with the obstruction -- infrastructure can understand. and also there's some disparity between how quickly they can understand it versus how quickly we receive government data to either economyment or to negate
that -- to compliment or negate what we're understanding and give a scale. so if home depot is seeing a real huge increase in something, but the larger housing sector isn't, that is an interesting trend for them to recognize. but when there's a gap in understanding that information, you might miss a window. i think for small businesses, there is a real -- and this isn't to say that large businesses don't value government data. they do. they use it probably differently than the small businesses. but from our perspective, a lot of the small retailers can really have access to data that for them is not something they need to go pay a solution provider a bunch of money to find and can instead leverage data that exists, that's credible, that will help them kind of level the playing field between them and potentially different competitors. from our perspective, there is huge value. i think if you look at one of our big concerns, it's funding. funding for a lot of different agencies and sectors is critical if we are going to see us as a country continue to value this data. david: let's spend two minutes on nonresponse. it seems to me it's a really big deal right now. there's a chance -- we don't know for sure, but there's a chance that it explains the
polling misses in israel, in britain. the overall polling in the united states is not huge, but it was significant in states like becky's and pennsylvania and elsewhere. the national polling miss wasn't so big. it's a really big deal right now for census, for politics. in the long term, i'll confess that i'm pretty optimistic. it seems to me we're in a transition in which polling is having to move from your phone ringing at home and you picking it up, but we are all carrying around these things now. it just seems to me that in the long term, someone's going to figure out and the government's going to be able to use ways to survey people that i think will be at least as good. and i being sort of a fuzzyheaded optimist about that? >> i'll add in qukly, we're concerned about nonresponse as well. i also think there's this challenge of needing to evolve the way that the population is evolving. we need to collect information differently than we have in the
past, about different things. i think to your point about, of course we've got how many american families don't have a home phone anymore, how do you look at cell phones and how do you look at email and texting. that's something our industry's going through. but i think broadly as we look at data and how to collect it, we've got to think differently. i think sometimes that lends self to more difficult decisions of we've never done it this way. i think understanding and tackling the nonresponse issue is critical. >> i think it's also true that we've got to move away from thinking that the survey is the only way do you this. because surveys are increasingly difficult and the merging of survey data with administrative data and use of administrative data is deeply important. the main problem we have here is knowing what your total sample frame looks like. why do we use houses in the census? houses don't move. i can tell you if you got a response from every house. but phones come and go and people come and go.
you really have difficulty when you try to think creatively about how do you do this differently, of knowing what your sample looks like out of the total population. that's where administrative data can be quite useful. because you have relatively complete samples and you know who is in and out of the sample. >> one of the problems with nonresponse is that of course we want to know what the nonrespondents would have said had they said something. so what the statistical agencies have to do is answer to them. otherwise, we don't get a picture for the whole. i don't know enough about it to know how much ampletation was going on. how accurate is it, what the problems are with amputation -- imputation. i imagine most users of data don't realize that when there's
a missing answer, when the respondent doesn't want to say how many weeks they works last year because they can't remember. they're not trying to hide anything. what their income was in the previous year. something gets imputed to it. that may produce error, bias, confusion. and i think it's important for that to get more attention. david: one of the hardest things for the polling industry is there is no more randomization in many ways, right? four -- for a long time the polling industry said, you have to use phone numbers because you can randomize them. you can't randomize email addresses. but while you can randomize phone numbers, it's not that meaningful because you can't randomize who actually responds. so there are no more perfect answers. really it's got to be a mix of things, just as you were saying. >> in polling what we hear is, well, so we waited. there's so much room for errer
or to creep into that process. david: my colleague had a delightful piece where he dug into the microdata of a survey and he realized there was one african-american trump supporter in a poll that was moving the numbers. because that supporter was weighted so heavily. [laughter] before we open it up to everyone here, i want to invite each of you to pound the table a little bit. you don't all have to do this, but jump in if you want to. what is a specific statistic that you have concerns about its accuracy? i know some of you have some thoughts here. >> there are several things, obviously. you can think about, first g.d.p. and productivity and the whole national accounts. there's already a lot of uncertainty that goes into this. when we think we have a productivity slowdown or g.d.p. has been weak, revisions can go away from one minute to the other. the way we thought about the narrative was for the economy,
for the last several years, can change when revisions come around. so that's not an easy task to change how g.d.p. is done. but it's just to say that often you walk around with a certain story, not only financial markets, but elsewhere also and say, this is the story, this is what's going on. productivity is weak, growth is weak. but maybe we have been measuring this the wrong way. we have a lot of issues with how gdp is measured. there's all kinds of things that go into thoughts around what is the right way to measure. so we already have issues. even beyond this conference of how do you these things and how they're calculated. but if there's one thing in particular that's got more attention, more recently, is of course how you characterize the trade deficit. if we need to go back and say that exports are suddenly something else that needs to be calculated, subtracted from exports rather than imports, then you get a much bigger current account deficit and much bigger trade deficit. then you're opening up a door to, if we can do that, then we can basically discuss any statistic.
it is possible, not only on this election bias issue of who is the answering the phone, but to get some different calculations. if you say the sample was a little bit different than what we actually had, and it can be somewhat dramatic in some cases. but at the end of the day, we are somewhat worried about the recent noises, how those changes might come through. david: would you be ok if there was a change but it was still possible that the old series remained and it was possible to do comparisons? or do you think any kind of big change is dangerous? >> the united nations has clear standards about how do you national accounts. there are clear standards how do you everything on trade statistics. so we can't just say oh, those other countries in the world don't understand it and we can do it much better. if you end up with that, then you get back to these emerging markets of countrying saying, wow, if the u.s. doesn't have to follow these standards, why do i have to follow those standards?
>> i am concerned, as i said, in my first remarks, about what happens when we go from nominal g.d.p. to real g.d.p. and it's not about the recent numbers. i'm not concerned about the problem of why we've had a slowdown in productivity in the last few years. that is a deep problem and an important problem, and i don't have a solution to it. but what i'm concerned about is that over the long run, the way in which the government as a technical matter goes from nominal g.d.p., from nominal output from nominal real incomes, nominal incomes to real incomes and real g.d.p. i think leads to serious underestimates. i think that has a significant effect on how people view the way the economy is going and what the future holds. and it has political as well as just economic effects. what do i mean by that?
when you look at the surveys that ask people, how has your family done for the last five years or over the last decade, the general answer is, pretty well. of course not everybody. but most people would say they're better off than they were five or 10 years ago. but when you look at the official statistics, they don't say that. the official statistics say that we have had no increase in real incomes for the middle quintile over the last two decades. so when people are asked, not how did your family do, but how is the country doing, their general answer is, terribly. of course they know how well their family is doing, but they don't have a clue about how well the country as a whole is doing. all they know is what they read in the press or hear from politicians, and that reflects the official statistics.
the official statistics say we've had less than 1.5% growth of per capita income per year over the last 20 years. and if there's been a shift of some of that income toward the higher part of the income distribution, then as the economic report of the president said a couple years ago, the middle quintile didn't see any improvement at all in its real income. i don't believe that. so i have been studying what the b.l.s. and the b.e.a. do to estimate that. i think they try to do the best job they can, but they have a very faulty methodology for doing it, in two ways. one is, what do they do about quality change? when a product changes, how do they decide how much better the product is, how much more real value there is in it? and they don't have a good way of doing it. what they end up doing is
looking at the cost of product change from one year to the next. the cost, but not the value to the consumer or the user. and the basic rule is, not 100% of all products, but in general the basic rule is if it doesn't cost more, it isn't a better product. we all know that's not right. then when it comes to new products, new drugs, new technology, they don't even try to capture the benefits to the public, to users, of the new product. so i think we're underestimating that number, that roughly 1.5% per capita income growth over the last couple of decades. that number could be twice as large, could be three times as large. we don't have a good fix on that. and that's not just a u.s. problem. that i think is a problem for every industrial country.
if you look at the i.m.f. suggested statistical methods, they're basically what the b.l.s. and b.e.a. do in the united states. so we're all in this same situation of understating how well we're actually doing. i think that contributes to anti-globalization. i think it contributes to a kind of anti-business. i think it's an unfortunate thing. i don't know how to fix the problem, but i think we have to change the perception about what we're seeing in those reports. david: i agree, that's a vital issue. >> i've got a specific example that relates to something that i think is really important, both from a retail perspective and a larger economy perspective. i talked about consumer spending at 70% of g.d.p. when the department of commerce reports monthly retail sales,
you see a lot of conversation about what's happening in the industry, what's happening with the consumer, where is the consumer spending, how is that happening? and anyone who follows the retail industry or, quite frankly, anybody who shops, knows that the consumer is shopping differently than they ever have before. one of the challenges we have with department of commerce data is that they have always categorized retail sales by where the sale occurs, not the type of retailer who is selling that product. so for example, walmart, the largest retailer in the world, largest retailer certainly in the u.s. when they report their sales, they report sale by store and their ecommerce channel. when department of commerce releases its monthly sales, the store sales go into the discount retailer category. the ecommerce sales go into ecommerce. that's a challenge because there's this big narrative right now, our stores dying? are stores going away? are the retailers going to
overtake traditional brick and mortar stores? from our perspective, if you look at what's happening in retail, eight of the 10 top ecommerce retailers in the u.s. have traditional stores. wal-mart, macy's, home depot. and you see retailers investing in their ecommerce business because they know that's where their consumers are going. but if you look at department of commerce data, what you're going to see is an inaccurate reflection of how the consumer is shopping today, because what they're doing is they're taking macy's store sales and putting it into department store and they're taking macy's ecommerce sales, which are growing at a much faster rate for the company and where the company is investing, and putting it in the ecommerce category. that's a problem because if you look at department sale stores over time, without understanding the context of the industry, what you'd say is, whoa, the department store sector's really struggling. but look at ecommerce. that's where we need to invest.
go to amazon. what's really happen something macy's is saying, look, i don't care where you shop at macy's, whether it's dot-com or in our stores, as long as you shop with us. from our perspective, we understand why it was done this way in 1995 before the internet. but we need to better understand and think about how do to reflect the changing -- how to reflect the changing consumer spending behavior with what's happening today, to better understand what is happening with the economy, to better understand what's happening with the consumer and to better plan for the future. david: that's fascinating. i didn't know that. there are huge gaps in data about higher education. i think it's remarkable what we don't know about higher education. it's incredibly important to economic growth. taxpayer funds it to a great, great degree. if you advocate going to college and you want to go online and look at what is learned at any given college, good luck. even that is difficult because the way we calculate the graduation rate is so flawed. higher education is an area where we have been willing -- where it is comfortable with data, we will trust that
dory." ng is "hunky i will end with one little advertisement. one of the best data we got, they used tax returns to tell us about an economic return of who -- an economic portrait of who goes to what college. and that data we couldn't have had before. one thing that is important in this era of the big data is government allows access to data because it leads to important findings. most of the big things we have learned about have come from private researchers getting access to public data. >> which of course is about to happen. it does mean you have to match
your data and keep the individuals private. nobody can track the individuals and that stops a lot of data. that. don't trust it goes back to, how do we communicate about this? and the extent we can be careful. >> i am going to go through a bunch of these questions. here is one. which is, derek, i apologize butchering your last name, how do you improve how do you improve the delivery of data? it is absolutely dead on. should we have a centralized thing? >> at the moment, the solution to that is if they are private
companies, data stream have said it is confusing but you can get everything here and -- >> for a lot of money. >> they do it very well. the private company can fix whatever is going on. but it is incredibly confusing. you can respond, you can just google it. >> and don't give it away for nothing. maybe that is anticompetitive. but for nothing is a good price. for anyone who wants it looks for data releases and signs up and every day, they get the daily releases and have access to this list of all of the data, if you want to go back and see what is happening a month ago to some statistic and it is searchable. >> i see a lot of people from
o.m.b. and it is one of the responsibilities to release the data so there is a lot more consistency. >> to me of all the agencies, bls is the best. it is really easy to search. thank you. if you look at the way the wage data, it is easy and really wonderful. does go related questions. questions, please expand on the and the data and polymerization but of the concept of reliable data, how can we protect the integrity of data. is it new? this calls for a new defense strategy. how can reconfigure its approach
and thriving in the public sector? >> i don't think this is anything unique. we are at a moment where public institutions and the concept of all the goods have come under attack and not that people are and you see the same thing happening in the area of public institutions where they say you are not a public good. people are going to get their returns. and this level of suspicion of what used to be a value of public institutions -- i don't think you saw the data problem without changing the larger dynamic. i wish i had an answer. but we need more people in leadership positions talking about the value of public institutions and common
good-type products. you don't hear that from almost any leaders right now. >> and the answer to this question should be no. should private sector users be macroto pay for the data -- for the data? >> almost impossible to do that. if you want to make the data, almost impossible. >> are there particular statistics that you were worried to be affected by funding cuts? >> and it has been under attacks and only source of small area details democratic data -- demographic data and if we lose it, i don't know how local governments make decisions. >> what should be changed about the 2020 census? what do think people should be
adding to the senses that we do not have there now? >> it's kind of remarkable we don't ask about religion. >> there is strong opposition to having an inventory of who are muslims and baptists and the jews in the population. >> separation of church and state. >> i would go further than that. >> i see that argument more strongly today than i did six months ago. [laughter] >> >> there was talk about doing it and i think it was in the current population survey and not in the census and i was on an advisory panel and i didn't like it then. the cps is a voluntary participation online census. it was argued it is a voluntary
-- it was argued you could ask about religion but shouldn't ask about it in the census. >> and pew does phenomenal work on religion and it is publicly available. >> and it is controversial which is the questions. the race and enough -- race and ethnicity questions. those categories are incredible gross and don't categorize about how people think of themselves very well, but taking on the question of what should those categories be is one of the lyrical topics you can imagine -- one of the hot political topics you can imagine. what should those questions should be. it's easier to field those questions. and it's about how people identify themselves. particularly as the population is changing so rapidly. by ethnicityn which is a waste of money,
especially in a world where you can get all this online anyway, but we take that as a reason to do everything by hispanic and non-hispanic. >> is there any future in the just future in being able to see into the past? >> i was reporting in salt lake city and i went to the church of latter day saints and i looked up these old censuses and my wife's family and mine and they were -- and they were chilling. i'm looking at handwriting that someone who is speaking to my children's great-great-grandfather. it is really remarkable. do you think there are some prospect that wouldn't be on line? but could it be linked from parents to children that would a llow a richer history?
>> there are a number of researchers and we digitized old censuses. you don't release that data until 70 years afterwards, but we digitized the data up until that point. and there are researchers who have gone in and 1880, what are the names the children and linked them into the 1900s and look at changes in location and migration and changes in economic status. some great work in economic history and that. once the work is digitized. it's not easy, but you can do it. >> data are snapshots in time. last question. taking off of what marty does, someone asked what is your favorite public beta -- public
data? there not going to hold you -- hold you to this. what is one you like to look up? you would encourage other people to check out. >> if i were being alan greenspan i would have to say the weight of gdp. gdp.e rate of >> i would say and that's measured -- compared to many other things. unemployment falls everything -- unemployment everything about wages and tells you about how many people got a job and unemployment ration. and at the end of the day it has a wealth of information what is .he state of the business cycle
are we at full employment? should we be thinking about whether there is signs of a slowdown? you can get so many answers to so many questions. this is a pitch. it is outstanding. >> first friday of the every month, with the exception of -- what is the exception? >> he depends when the 12th of the month and it is next week rather than this coming friday. >> there is no question and release of are statistics. i take home and spend hours reading every chart and looking at the data. and this is really gives you all of the detail by different demographics and how you cut it by educational levels. what is happening with the economic well-being in the country and makes for fascinating reading. >> it is the snapshot.
if you want to know how -- [laughter] >> you can only do that if you have priced numbers to the box, and that is where we get into trouble with the gdp, real incomes if we are not adequately adjusting the price statistics improvement quality and for new goods. overtime, we don't did the comparisons right. >> i agree. here about relative changes. -- you hear about relative changes. ismarty, your least favorite present fighters. reports. >> please join me in thanking our panel. [applause]
>> tonight on c-span, jeff sessions on his recusal and contact with russian officials. president trump in newport news, virginia, where he visited a naval aircraft carrier. later, a discussion about trump's listenership with the media -- trump's relationship with the media. >> c-span's washington journal live everyday with news and policy issues that impact you. coming up, newsmax media incorporated ceo discusses donald trump's residency so far in the white house's relationship with the media. the national journal health care correspondent discusses the divisions among republicans on how best to repeal and replace the audible care act.
jonathan greenblatt assesses the ongoing threat made against jewish community centers and schools. be sure to watch c-span's washington journal getting live at 7:00 a.m. eastern, friday morning. join the discussion. in the news today, attorney general jeff sessions announced he is recusing himself from any investigations involving the 2016 campaign and the russian interference. here's his announcement. mr. sessions: thank you, peter. it is good to be with you. welcome to the department of justice. jody, thank you for being with me. he's my chief of staff and jody has been almost 20 years in the department of justice.