Every month, Americans – both consumers and businesses -- turn to the Bureau of Labor Statistics (BLS) to understand the health of the US economy and what to expect next. How can businesses and Congress support and improve the work of this agency?
Join Steve Odland and guest Erica Groshen, former commissioner of BLS from 2013 to 2017, to find out how BLS conducts surveys, how monthly revisions work, and what's limiting the agency from being even more effective.
For more from The Conference Board:
Join experts from The Conference Board as they share Trusted Insights for What’s Ahead®
Former Commissioner
US Bureau of Labor Statistics
C-Suite Perspectives is a series hosted by our President & CEO, Steve Odland. This weekly conversation takes an objective, data-driven look at a range of business topics aimed at executives. Listeners will come away with what The Conference Board does best: Trusted Insights for What’s Ahead®.
C-Suite Perspectives provides unique insights for C-Suite executives on timely topics that matter most to businesses as selected by The Conference Board. If you would like to suggest a guest for the podcast series, please email csuite.perspectives@conference-board.org. Note: As a non-profit organization under 501(c)(3) of the IRS Code, The Conference Board cannot promote or offer marketing opportunities to for-profit entities.
Steve Odland: Welcome to C-Suite Perspectives, a signature series by The Conference Board. I'm Steve Odland from The Conference Board and the host of this podcast series, and joining me today is Erica Groshen.
Erica is a labor economist and was the 14th commissioner of the US Bureau of Labor Statistics from 2013 to 2017, and she's here today to discuss the role of the BLS, the Bureau of Labor Statistics, and how business leadership is important for the US statistical system. Erica, welcome.
Erica Groshen: Thank you for having me. I'm glad to be with you.
Steve Odland: So Erica, I'mpretty certain that few of our listeners are as expert on BLS and everything that goes on there as you are. So maybe you could just start with an overview of what is the BLS, what kind of data do they produce, what's the role of its commissioner?
Erica Groshen: Great. SoI'm going to read you the official mission of the BLS cause that's actually, they worked hard on getting this right. So BLS measures labor market activity, working conditions, price changes—so that's inflation, right—and productivity in the US economy. All of this to support public and private decision making.
And so some of the background is it's an agency of the Department of Labor, and it's really the principal fact-finding agency in the broad field of labor economics and statistics. And it serves as a part of the US federal statistical system, which is a decentralized system. So this is a big component of that system, but it is separate within that system.
It collects, on a day-to-day basis, it collects data, it calculates, it analyzes the data, it publishes this data, all of which are essential to the public, to employers, to researchers, and to government organizations.
Steve Odland: So lots of customers, policy makers at all levels. Not just the federal government, but all governments. And also the private sector, because private sector leaders need also to understand what's going on and how it impacts their business. So, maybe, just a little more granularity on the type of data that are produced. You talked about jobs and productivity.
Erica Groshen: Yeah. So in labor market activity, we're talking about things like the unemployment rate and the payroll numbers, the participation rate, things like that. Working conditions, we're talking about illnesses and injuries and fatalities, but also environmental differences and things like that. And wages and benefits, right? Price changes. That's all about the various measures of inflation for consumers, for businesses, importers, exporters. And productivity, which is the key to economic growth and to wealth creation in our economy.
Steve Odland: Yeah, so, really important, and a lot of those kinds of data are collected or produced by any individual company or organization, but you're aggregating it at a national level.
Erica Groshen: That's right. BLS has access to some data that no private sector does because it's got access to some government administrative data, and that allows it to really give the best estimates for the nation as a whole.
Steve Odland: So, if BLS didn't exist, how would the private sector gain access to these kinds of data?
Erica Groshen: Well, they'd have to pay for it individually, right? So you might have people stepping up and trying to make estimates like this. Some companies saying, "OK, there's no payroll number out there. We're going to try and do that." They would have a harder time doing it because they wouldn't have access to, say, the underlying UI data, unemployment insurance system data. But they could certainly try to do that. And in order to make it worth their while, they would probably charge a fairly high amount of money for getting access to that, which means that some people would have it and other people wouldn't.
Steve Odland: Yeah. So it's needed, it's necessary to run a business, and certainly to direct an economy. So it has to be done somewhere. Now you mentioned the BLS is part of the Department of Labor. Department of Labor is a government organization. But the BLS is not a political organization, per se. But are there political appointees?
Erica Groshen: Yeah, so the BLS is an agency. It had about 2,500 people when I was there. It'sprobably down to less than 2,000 now. And they're all civil servants, with one exception: The commissioner. The commissioner is appointed for a fixed four-year term, which makes, even though that's a presidential appointee, there's a difference between the people being appointed for a fixed term and serving at the pleasure of the president.
And that's, it's actually a big deal in Washington. Most people outside don't recognize that difference. But the difference is that if you're pointed for a fixed-year term, it's because you are a professional and with technical skills. And you're not there advancing the agenda of any particular administration. You're there administering some agency that Congress decided should be done right? And soit's like the scientific heads of the scientific agencies and things like that. So basically, you're working on producing public goods that we've all decided are important to have.
Steve Odland: So the four-year fixed term aligns with a presidential term, and so therefore it's a presidential-appointed position, one position, the head, which is the commissioner, and it's a Senate-approved position, too. So this, it's a really important position that you held, and really very impressive. And congratulations. and thanks for all that you did.
Erica Groshen: Oh, you're welcome. Yeah, it was like the best job ever. Anybody ever offers you that job, take it. It's great.
Steve Odland: Yeah. Well, they'll never offer it to me cause I'm not an economist. But you are.
Erica Groshen: I want to correct one thing you said though. This is important. The four-year fixed term starts when you're sworn in. So most BLS commissioners start during one president's term and then serve through to part of the next president's term. So it doesn't line up.
Mine did, but it was unusual because the Senate held up my confirmation. At that time, they were hoping that, the Republicans in the Senate thought, "Well, we're close enough to the end of the term. We're just going to hold off until the election, and then decide whether or not we're going to confirm."
Steve Odland: So the fact that yours aligned with serendipitous. OK, that'svery helpful.
Erica Groshen: That's unusual, yeah.
Steve Odland: OK, great. That's great. So how does the BLS provide its data? I mean, it'sessentially just issued on a website, any other ways?
Erica Groshen: Yeah, I mean, that's the way almost all the access happens, right? It used to be a lot of paper, but now it'spretty much all on the website. Although I should say that the BLS has a lot of other tables it produces regularly. And if it turns out you want one of those tables, you can call them up, and they'll send it to you regularly. It's not worth it to put every single table they make regularly up on the website. Sothere's more than even what's on the website.
But anyway, the concept is equal access to all. And there are very few exceptions to that.
Steve Odland: And hence, it's free.
Erica Groshen: And free.
Steve Odland: And therefore funded by the taxpayers in order to make that equal access.
Erica Groshen: That's right. The one exception to that part is that anybody can ask the BLS to do reimbursable work. And then the BLS charges you for doing that reimbursable work. And the one stipulation is that after BLS does it, everybody still gets access to it. So you get a role in determining what it will be, but not in who gets access to it.
So an example of that is that the aerospace industry actually pays BLS to produce an employment cost estimate just for the aerospace industry, because they really need that because it takes 10 years to make an airplane after you order that. And they index the cost, eventual cost of airlines, to airplanes, to the EC, the employment cost index for aerospace workers.
Steve Odland: Wow. Yeah, so that's great. Soyou're able to leverage the resources that are there, but then not burden every other user with the same cost if there's discrete work that's done. That's really interesting. So how does the BLS gather its data? I mean it, is it a survey that is sent out? And at what frequency?
Erica Groshen: So BLS has about 25 programs, and each one is designed for a particular measurement objective. So, I mean, the short answer is it depends, but the longer answer is BLS is still very heavily survey-centric. So when, when the statistical system started then, there was no concept of a survey yet. The data gatherers just went out and gathered whatever information they could, and that was in 1884.
By the beginning of the 20th century, survey methodology started to be designed, and so we had this development of modern survey methodology. And the statistical agencies really led the development of this. And I mean, it was a great innovation because then you could really you knew what size sample you needed. You could do causal inferences, you could put, have standard errors.
You had an idea how good the information was, and that was great. Especially since at the beginning, survey response rates were really high because it was novel, it was new. Everybody thought this was cool. We're in a different world now, right? So now, there's a lot of—survey fatigue is the technical term for it. There's so many surveys out there, some of which are not really surveys, but just pretending to be surveys, and it'sactually much more difficult to collect it. But I’ll get back to your answer.
So in addition to surveys, how else does BLS collect data? Government administrative data is a big source. So for instance, the payroll employment numbers start out by an annual exercise that the BLS does, looking at the list of all the employers who have unemployment insurance accounts with states across the country. And that forms the business register that the BLS uses to draw a sample from, to choose which employers it's going to ask, "How many jobs do you have on a regular basis?"
Steve Odland: So Erica, are these paper surveys that are sent out, or is it all done electronically?
Erica Groshen: It's almost all electronic these days.
Steve Odland: Oh, OK.
Erica Groshen: And, so household surveys, will be a phone or electronic. The employer surveys are not opinion surveys. There's no opinions in them. It's all asking how many people do you have, what prices are you charging? Soit's not opinions. And the companies or employers have a choice of different modes depending on what's easiest for them. So really small employers may fax in the number, but the bigger ones are sending it through some big electronic collection system.
Steve Odland: So having run a couple public companies, the lawyers always say to you, don't respond, don't share any of this data because it's material nonpublic information. You can't do it, you know, yada, yada, yada. So how do you ensure respondees that their data are anonymized and that they won't run afoul with anything? In other words, to try to improve the response rate.
Erica Groshen: Yeah, so there's a whole section of the BLS website that goes through all of the protections for confidentiality and privacy. So, legal protections and then operational protections and all of that. And the BLS is, and the other statistical agencies, are roach motels. Information comes in, and it doesn't go out, other than in statistical aggregates. And that's pretty much, that's the rule for the statistical agencies. But there's a key point that I want to bring out, which is that BLS, almost every one of its surveys is voluntary. There are a few exceptions, but.
Steve Odland: It's voluntary, it's safe. There's no repercussions. Nobody's going to come back at you. You don't share it with the IRS. So the IRS isn't going to ring you up and say, you know, da, da, da, right?
Erica Groshen: Or wage-and-hour division, or any of those things. That's right. They have no, OSHA has no access to it whatever. Yeah.
Steve Odland: Yeah. And so this is something, I think, that our listeners need to understand because we do need a better response rate because you need the data. And that the better, the more data you have, the better the output, and the more statistically significant, and hence the better decision-making that can happen.
Erica Groshen: Yeah. Oh, absolutely. And with falling response rates, what's happening? Well, the national data, you know, the top-line number, there's almost no change in the standard error for that. But you're still getting three bad things as the response rates so far. I mean, if they fall low enough, then yes. But right now, what's happening is that costs are rising because, if people don't respond, BLS spends time trying to convince them to respond, more people. So it just raises costs, and sometimes you just increase the sample in order to make up for a smaller response.
Steve Odland: Right. Because you're trying to get a certain number for the data to be statistically significant.
Erica Groshen: Right. So it raises your costs. There's more risk, then, of undetected bias. So if the people who aren't responding have something in common with each other that matters for what, what you're trying to measure, then you may have more bias in there that creeps in. BLS tries to correct for biases that it detects. So if there's an industry bias or geographic bias, then it can reweight things. But if for example, you had some economic influence out there that was depressing, some group of employers, and those were the ones that weren't responding. Then you'd be off, right? You'd have this undetected bias.
So the lower the response rate, the more you have this undetected bias. And then the final thing that I think is really important for businesses is that where you do see the impact is loss of granularity, less ability for BLS to accurately say what's going on at the state and local level for different demographics, for different industries, for different products or occupations, depending on the kind of data it is. And companies really need that granularity because they're usually interested in, they're a part of the state or they're a part of the industry, or the occupations that they hire, that sort of thing. So you lose visibility for firms to target their questions and to target their activities.
Steve Odland: We're talking about the role of the BLS system and how business leadership's important. We're going to take a short break, be right back.
Welcome back to C-Suite Perspectives. I'm your host, Steve Odland, from The Conference Board, and I'm joined today by Erica Groshen, the 14th commissioner of the US Bureau of Labor Statistics. So, Erica, before the break, we were talking about how the BLS gathers the data, the survey responses, and all of that. Recently, BLS has been in the press probably more than I've seen it in some time because of revisions to previous months data. Talk about why are they revised, I guess, even subsequently a couple months down the road. What causes that? It's got to be some responses coming in after the cutoff date, but two months? So explain why.
Erica Groshen: Yeah, so this is one particular program called the Current Employment Statistics Program, or also often called the payroll survey. And the design of the survey says, well, we know people want really accurate information, and they want it to be really timely. And at a certain point, you have to choose timely or accurate, you don't get both. And so what BLS has done is given people a choice. You can use the most timely data, or you can use the most accurate data. And that's up to the user to decide.
So here's how it goes: At the end of every month, the BLS reaches out to the companies that are in the sample at the time and ask them to report in how many people they had on their payrolls for the pay period that contains the 12th of the month. And so those companies ship that to BLS, and some of them at the end of the month haven't finished their payroll for the pay period that contains the 12th of the month. Because they pay monthly, or maybe just because of the way the calendar is going. Or maybe they're in some turmoil. For some reason they haven't gotten around to it, or maybe they finished it, but the person who sends it in was sick that day or something. So it just doesn't come in. So it comes in for about two-thirds. But a third don't ship it in.
Well, BLS then essentially takes that information where people don't report and they say, well, it's a missing value.We're going to replace it when we get it, but we're going to give you the estimate based on everybody who did report. And that's the first preliminary measure. Now, the next month, of course, they ask about what happened in the most recent month, but they also ask the companies to go back and say—so right now, we are in August, so around now BLS is going to start collecting the information for August.
But they're also going to say, for July and for June, how many people did you have on your payrolls for the pay period that contains the 12th of the month? And so the first report for June came in at the end of June. The second report for June came in at the end of July, and the third one is going to come in at the end of August.
And you start out with about a third of the people. And by the time the companies have had three times to report, your response rates are up to 95%. The first estimate uses only the two-thirds that reported first, the second uses in between, and the third uses 95%.
Steve Odland: Yeah. And so normally you don't see big variances. You just see better data quality.
Erica Groshen: Yeah. Usually they offset each other, so there's not a big change because some of them report higher than the BLS imputed, and this imputation is mostly kind of passive. It's a missing value, so essentially, you're attributing to the people who didn't report that they were more or less like the ones who did report. But there are certain times when you're more likely to get a correlation between not reporting and actually job loss. And that's around—
Steve Odland: So there's some underlying reason that drives the delta between the first, second, and third.
Erica Groshen: That's right. So if they're in turmoil, if demand is falling. Yeah, things like that.
Steve Odland: But recently there have been some big variations and revisions, and accusations that the revisions were politically motivated, which, because BLS is not a political organ, and the people are civil servants, as we've discussed, that's unlikely. But it's open to interpretation, and the press goes wild. So, as you look at this, are there any suggestions or thoughts that you have on revising methodology so that it's either tighter or there's less variation in the revisions so that you don't get into these accusations, however unfounded or founded they may be?
Erica Groshen: Yeah. So there are a number of different answers to that. First thing is that users of the data have to educate themselves enough so that they understand how the revision's happening happen, right? It's not a bunch of people looking at it saying, "Oh, should we change this?" Each time, it's the same computer program, just taking in new numbers. All of this estimation takes place over a very shortperiod of time with programs that are tightly choreographed.
This is much more like a manufacturing process than you might think. The data comes in, and the tables are spit out, so there's actually no way to go in and manipulate them. So how can you avoid revisions? Well, if you are a data user, and it bothers you to have the revisions, then you can do two things. One is you just look at the first number, and don't look at the other ones. Or you can say, well, it'sworth, I'm going to wait until the third month. All of that's there. You're free to ignore or use the ones that are most useful to you now.
Then there's a bigger long-run question about, can we do, can we come up with a better program than this? Can we go to one where the data are perhaps being fed on a flow basis, and you've got something that's both more granular and more timely and more accurate. And the answer is yes. There are ways.
Australia right now has a system where their version of the IRS gets payroll reports from companies at the end of every pay period, and they turn those over to the statistical system, and that's covers basically the entire country. And the statistical system puts out very timely and very accurate numbers on a flow basis. And we could do something like that, too, if we had the funding to design it and the willing participation of companies and the government.
Steve Odland: So what I hear you saying is yes, there are some revisions that could be made to the process. It would have to be funded and so forth. So that's one thing. Second thing is, the users themselves could be timely and respond, which would make the data better. So get off the stick and respond on time. But the third thing is if you don't like the variations, you could choose to only accept one in the first, second, or third month. So there are a number of ways that you could proceed on this.
Erica Groshen: Yeah, absolutely. Employers throughout the economy can choose to participate in the surveys more than they are doing now. They can get this in on time, and that means monitoring this as an expected part of people's jobs, telling them to prioritize it, rewarding participation. Too many of them aren't doing that. And why would you do that? Civic responsibility, enlightened self-interest. And just doing that would reduce these revisions.
Steve Odland: Yeah. So it sounds like business leadership is really important here. I mean, yes, it's a government agency. OK. But business leadership is necessary to reinforce the participation in the survey on a timely basis, but also for funding. Talk about that, and how should business leaders engage?
Erica Groshen: Yeah, so BLS has lost close to 20% in real terms, so inflation-adjusted terms, it's lost 20% of its funding in the last 15 years. And at the moment, it's also down 20% of its staff because of the administration's policies on staff reductions throughout government. So that's not directed at BLS. It's collateral damage, but it is collateral damage, nonetheless. So what can business do from an advocacy point of view? Well, businesses are some of the most effective advocates in the government right now. And they could expand their advocacy agenda to invest in improving and modernizing BLS' methods.
BLS, because of the funding issues, BLS just hasn't been able to innovate in the way it should be because there is amazing expansion of data throughout the economy, right? We've digitized all of our transactions, and we have the internet and all of that. And BLS is using more and more of those for the programs designed as is, but it needs to have the opportunity to totally redesign some of these programs to take advantage of these sources. And this means partnerships with companies. It means standardizing data and technologies so that the burden on companies is much lower.
Steve Odland: So what should our listeners do? Write their congressmen?
Erica Groshen: Your listeners are, a lot of them are C-Suite employers, right? So what would I say? First of all, we talked about survey participation. It shouldn't be an afterthought. It shouldn't be left to the attorneys who don't know anything about the benefits and only see possible costs. So it should be a C-Suite decision, not a risk-averse attorney decision. And it shouldn't be, employees shouldn't feel like they could never get promoted if they do a good job filling out government surveys.
There are partnership opportunities with the statistical agencies. So if a statistical agency like the BLS approaches a company and says, "Look, we'd like to do an experiment to find out if there's a different way to design a program," then entering into some of these partnerships could be really huge.
And if the BLS is a customer of a company producing data, they can not try to gouge the BLS, because that has happened, too. BLS, if it starts using data from a company, then it builds an entire system around it. And then that's an opportunity for that company to try and gouge BLS. And I saw that happen a couple of times when I was at BLS.
Steve Odland: OK, so our listeners, you need to engage here, and make sure that the data are being submitted, the surveys are being answered ina timely basis, that will improve data quality, and then engage in partnerships. Very helpful.
Erica Groshen: I have two more suggestions.
Steve Odland: Good
Erica Groshen: One is public statements. Business leaders are influential. They can attest to the importance and the reliability of federal statistics. If they rely upon them, they should be willing to say so and tell others that they should rely upon them. They should encourage other people's participation, households and employers, their business peers. And then they should decry political interference and support the independence of the statistical agencies, so that's using their public mouth.
But then there's also advocacy, and we talked about businesses being some of the most effective advocates in the government today. So they need to step up at their advocacy game to protect the agencies from political interference. And that means, now I'm going to get into hairy weeds, but there is a proposal out there that would convert many civil service positions into very close to political positions, and it's called schedule policy career. The statistical agency employees should be exempt from this. You really do not want that happening. And also traditionally, fixed-term appointees were only removable by for cause. And that seems to no longer be the understanding of the way the law was written. Well, Congress needs to think about do they need to rewrite the protections for fixed-term employees, particularly for statistical agencies.
And then, Congress can relax some of the data-sharing restrictions so that, for example, BLS could get access to IRS data. That would be a huge increase in efficiency for BLS and for the users of data because you no longer get disagreements between the Census Bureau and the BLS data on number of employees in various industries. And the administration can just lift the hiring freeze for statistical agencies. BLS, as we've talked about, is down about 20%, and they can't do anything to address that.
Steve Odland: All right, great suggestions. Erica Groshen, 14th commissioner of the US Bureau of Labor Statistics. Thanks for being with us today.
Erica Groshen: Thank you for having me, and thanks to all of the listeners.
Steve Odland: Yes, and thanks to all of you who listen to C-Suite Perspectives. I'm Steve Odland, and this series has been brought to you by The Conference Board.
C-Suite Perspectives / 02 Sep 2025
Find out how BLS conducts surveys, how monthly revisions work, and what's limiting the agency from being even more effective.
C-Suite Perspectives / 28 Aug 2025
Electricity demand is expected to increase 25% by 2050. How can renewable energy sources in the US help fill the gap?
C-Suite Perspectives / 26 Aug 2025
Consumer confidence mostly held steady in August – what's driving people’s current mood and spending habits?
C-Suite Perspectives / 25 Aug 2025
How successful has the US been with tariffs, and what other aspects of trade could be negotiated next?
C-Suite Perspectives / 21 Aug 2025
How can multinational companies operating in China evaluate the risks and opportunities—including gray swans and even black swans?
C-Suite Perspectives / 18 Aug 2025
Find out how Synchrony is improving financial literacy, uses machine learning, AI, and reset its culture through active listening and transparency.