Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Inside the BLS: William Beach on Trust, Data, and the Future of Federal Statistics
332 Plays4 hours ago
In this episode, I sit down with William Beach, former Commissioner of the Bureau of Labor Statistics, to dig into how America’s most important economic data are produced. We talk about the nuts and bolts of how the BLS jobs numbers are collected, processed, and released, as well as why revisions happen and what they really mean. Bill shares his perspective on the commissioner’s role, the challenges of falling survey response rates, and how statistical agencies can rebuild public trust in their work. We also touch on his experiences working across two administrations and his ideas for the future of federal data. This conversation sheds light on a system that is often misunderstood, yet vital for understanding the economy.
Subscribe to the PolicyViz Podcast wherever you get your podcasts.
Become a patron of the PolicyViz Podcast for as little as a buck a month
Follow me on Instagram, LinkedIn, Substack, X, Website, YouTube
Email: jon@policyviz.com
Recommended
Transcript
Introduction to Season 12
00:00:12
Speaker
Welcome back to the Policy Biz Podcast. i'm your host, John Schwabisch. This is the first official episode of Season 12, and I'm glad you can join me. And I'm glad you'll be able to listen to my interview with a very special guest, Bill Beach, one of the former commissioners of the Bureau of Labor Statistics, one of the preeminent data and statistical agencies in the U.S. federal government.
The Controversy of Firing a Commissioner
00:00:36
Speaker
I reached out to Bill to see if he would join me on the show after President Trump fired the current BLS commissioner, Erica McIntarfer, on August 1st, explicitly because he accused her of rigging and faking the job numbers that were released that day.
00:00:53
Speaker
ah Those job numbers ah came in probably lower than the administration expected and also included some downward revisions to the job numbers for the previous couple of months.
00:01:04
Speaker
And of course, trusting the job numbers, ah trusting the inflation numbers, trusting the official numbers that come out of the federal statistical agency ecosystem is crucially important to governments, to businesses and to people.
Understanding BLS's Role and Trust Issues
00:01:19
Speaker
Now I really have two goals in this conversation with Bill. The first is to help people understand in more detail how BLS collects processes and communicates the data that they are responsible for doing. So this includes job numbers and includes employment numbers and includes inflation, lots of other data. We're going to focus on jobs numbers in this discussion, but I wanted to help explain to people how these numbers are collected in process and and communicated out to the world. Because in my reading of the news after commissioner McIntyre for was fired, I really felt like there was just a total misunderstanding of how these data are processed.
00:02:01
Speaker
So that was the first goal in this conversation. The second goal is to help people understand what the commissioner's role is. So the president accused Commissioner McIntyre of faking and rigging the job numbers. And as you're going to hear in this ah interview with one of the former commissioners, that's not really something that can happen.
00:02:21
Speaker
That is not an action the commissioner can take. It is not an action that is really even possible as you're going to hear. um And there's a lot of people involved in the process of collecting and processing and analyzing these data from the BLS.
00:02:36
Speaker
So we talk about those in length. This isn't to say that there isn't room for improvement at the BLS and and other federal agencies. I've sat on several different advisory groups and advisory boards for the BLS. So they are clearly aware of many of the issues. This is not easy work to do.
00:02:55
Speaker
Many of the issues that they can improve upon, or they they can change or that they can update. But what you're going to hear in this interview.
The Importance of Reliable Data
00:03:03
Speaker
I think is a real cautionary tale about the dangers that are facing our data collection and statistical agencies. If we as a country, including individuals, including businesses, including governments at all levels, if we can't trust the data that are being collected and produced,
00:03:23
Speaker
We are not going to be as productive. We are not going to be as accurate. We are not going to be able to project or forecast or estimate future behavior, future business purchases, future hiring decisions, future firing decisions.
00:03:37
Speaker
And so this conversation and several more that are coming up this season are crucially important. not just for these broader issues, but also for how we as a data visualization community are going to produce our work and produce our visuals. Because if we can't trust the data that we're using, our visuals are not going to be useful in any way, because everything that we produce when it comes to visualizing data sits on a bedrock of clear, honest, and objective data.
00:04:06
Speaker
So I think you're really going to enjoy this discussion with Bill Beach. I think you're going to learn a lot. I hope
Insights from Former BLS Commissioner Bill Beach
00:04:12
Speaker
you enjoy it. And so I'm going to turn it right over to my conversation. Here's my interview with former BLS Commissioner Bill Beach.
00:04:21
Speaker
Well, hi, Bill. Good to see you. John, it's great to be in the program. I mean, thanks so much for coming on. Your phone must be still ringing off the hook. ah Yeah, I think I should get a rebate from the phone company. um As I've mentioned to others,
00:04:35
Speaker
This is truly a remarkable um ah remarkable episode. I've been in Washington a long time. I've spoken with a lot of reporters. I've done a lot of things on the edges of you commenting on on matters in front of the public. like I have never seen a an explosion of press interest on what would I think arguably be called a very narrow, otherwise a very narrow event an agency. Most people don't know about where the agency head is is replaced suddenly becomes world
Media Attention and Trusted Statistics
00:05:05
Speaker
news. I mean, I have reporters from other countries, so it's it's remarkable. I'm very pleased by it because I think the issues raised in this in this particular episode go well beyond a personnel
00:05:18
Speaker
ah issue They go to the whole question of whether our statistics are going to be trusted and how we measure the economy. So it is is it's very important, and I'm very pleased by the attention that it's getting.
00:05:31
Speaker
Yeah. So I'd like to start by sort of kind of zooming in and then zooming back out, because I think the one thing that I've heard a lot of let's say, ah misunderstanding about is how these numbers are collected and created. That's correct right. Job numbers, inflation numbers, and then the commissioner's role in getting those numbers out the door.
00:05:54
Speaker
So I was hoping as someone who's done this, you know, did this for a while, You know, can you explain that and and pick, you know, whatever, whatever one sort of makes sense, you know, the nor either either any of the data that BLS produces, which is a lot, um you know, what is the process like? How many people work on it? How long does it take? And then what is your role to, you know, get those things out the door in the monthly release?
00:06:17
Speaker
Uh, so, so as noted, I was commissioner from 2019, March until 2023, March. So I i had a, a very rich period of time to be leading the Bureau. and what, what I mean by that is that I got to see a lot of changes to the way we collect data under circumstances, which I'd hoped were not to be repeated. So very stressful circumstances. So I saw the system stressed, which is really important for me to say, because a very stressed system survived.
00:06:46
Speaker
And we are in a period of time where there are new stresses on BLS, but I think to answer sort of an implied question, it's always out there. Will BLS do well? I think they'll, they'll do fine.
00:06:57
Speaker
Okay. So what are, what are we talking about when we say collecting data? Why are we doing that? We're collecting what are called sample data. We're not asking everybody in the country or every business in the country, a question.
00:07:13
Speaker
So there are hundreds, hundreds of millions of people and really about 112, 115 million ah establishments that are in the database of the BLS. So you can't ask all of them every month.
00:07:28
Speaker
So what we do is we. Create a representative sample. This is like I like you take 2% of all the fish in the in the lake, but you you you don't take just the fish you can see you say well there's this there's minnows here and there's bass here and your northern pike here.
00:07:45
Speaker
David And so you take a representative sample in that sample. for the survey that's been very controversially disputed here recently. This is the establishment survey, the survey of businesses to find out how many people are working this month as opposed to that month, the gain or the loss.
00:08:04
Speaker
That is 120,000 firms and cover, you know, sometimes up to 400,000, sometimes up 500,000 work sites. that's kind of important number. ah thousands sometimes up to five hundred thousand worksites okay that's kind of kind of an important So what we do is we ask these firms, would you like to participate in the survey?
00:08:26
Speaker
It's a voluntary survey. That's another thing people should know. It isn't required by law that they do so. those Though there are 15 states that do require the businesses and people in those states to respond to federal surveys. We nevertheless don't ask people to do this and say, well, under penalty of law.
00:08:43
Speaker
yeah So if they say yes, then they get a ah form that is filled out electronically every month and sent to an electronic collection data center.
00:08:58
Speaker
There are two of them, one in Chicago and one at Fort Walton Beach, Florida. ah So depending upon where they're assigned, they send this form in. Now they they're looking at their business and what we ask them to do is look at their business for about a two week period in the middle of the month.
00:09:17
Speaker
that includes the 12th of the month. Now we do that because we have other surveys on the on the labor force that also say, let's look at your employment is a situation who's working in your house over a period of time that includes the 12th of the month.
00:09:34
Speaker
So BLS
BLS Data Collection and Release Process
00:09:35
Speaker
has designed their surveys of the labor force to include that that kind of that lodestone date. So we're centered on that date. ah The surveys come in to these electronic data centers. They're filled out.
00:09:49
Speaker
We're asking them things like how many people are working this month? ah ah whats What's your industry code? Where you located? it By the way, these are confidential data collections.
00:10:01
Speaker
they They're protected under the Privacy Act and the Confidential Information and Privacy ah Security Act. so They come into these data centers and then from the data centers, they're sent to Washington, DC, to the BLS headquarters, which is currently at Suitland, Maryland at the Suitland federal building.
00:10:21
Speaker
They're about, um I would say of 180 people are working on some slice of those data that have come in.
00:10:32
Speaker
And when they have finished processing, their slice and they only have a couple of days at the end of the month, of suit they usually Monday and Tuesday of ah the last week. Then about 40 people at BLS, sometimes a little bit less, bring all of that together and create the employment situation report or that portion of it dealing with establishments.
00:10:56
Speaker
Um, the, the people who finally see all pieces of that final number, maybe by Wednesday of the, of, of that week are about 18 people in BLS. BLS has 3000 employees, see that.
00:11:10
Speaker
about eighteen people see that Okay, so there's, there are some, there's mathematics and statistics that have to be used. Remember, it's a sample, very small sample.
00:11:21
Speaker
And we have to then use some ah formula to say, well, this, this is how much we got here in this industry. And then we blow it up to the whole country. And that's where the national numbers come from.
00:11:34
Speaker
You'll note that the commissioner didn't have a single yeah thing to do in all of this. Not once that I mentioned the commissioner is out there collecting sample or or processing it. ah The commissioner is the only presidentially appointed Senate confirmed person at the bureau.
00:11:51
Speaker
I was the commissioner, so I'm a political. ah And I would see the data at 11 o'clock in the morning on Wednesday of the final week of the month. That's the first time i would have seen the data. By that time, the data is completely done.
00:12:08
Speaker
It's written up in a report in a draft. It's loaded on computers in the United States, ah in data centers and in data centers around the world. is right We have 16 distribution sites worldwide and they simultaneously release will release in about less than 48 hours from from when I see the data, the commissioner needs the data.
00:12:31
Speaker
I have, there's nothing the commissioner can do to influence the data at that point. The only thing I could have done is if I had been informed that there was widespread fraud or interference, I could have not approved the publication of the employment situation report. That's the only thing I could, I could stop it, but then I'd have to, I'd have every financial entity in the world i'm on my back. But yeah yeah yeah that's how it that's basically how it's collected and then how it's processed.
00:13:04
Speaker
And so throughout that process, how are checks made at each of those different steps or each of those different locations to ensure that you know, the seasonal adjustments being done correctly and, you know, and the data are being processed correctly. Like, is there a a fact checking procedure that, you know, there's big teams here are like, how are all these numbers sort of fact checked and, ah you know, quality assurance?
00:13:34
Speaker
Sure. So there are about 460 people at the national headquarters that work for the, um, associate commissioner for employment and unemployment statistics.
00:13:46
Speaker
And they do all these things that you just mentioned. They are ah adjusting the data for seasonal seasonality, which you'd want to do. you know ah every Every summer, ah ah bunch of people start looking for jobs. These are called managers, winning summer jobs.
00:14:02
Speaker
And if you didn't seasonally adjust that number, you would say, oh my gosh, look at the unemployment rate's way up. because We ask the question, are you working or looking for work in the past four weeks? And a person could say, yeah I just got out of high school and I'm looking for work.
00:14:16
Speaker
ah So we have a seasonal adjustment to kind of average through that normal increase. ah There are people who are looking at ah duplicate returns. Sometimes a firm will send in their returns two or three times because these are machine loaded and machine transmitted data.
00:14:37
Speaker
And sometimes the transmission is replicated for whatever strange reasons happen. ah We're looking for incompleteness. And when we have a situation where a firm, particularly a large one, let's say, just imagine any of your large box retailing operations, when their form doesn't get in, then someone will be on the phone. Some people, maybe many will be on the phone trying to find out if they're just late in sending it in or they're not going to send it in or it's in the mail. yeah
00:15:09
Speaker
So there's, you know, normal sort of things are going on. Um, I would say is, as I mentioned about 120 people out of the 460 or so are engaged in that monthly, monthly profits. It's a, it's a, it's a big operation.
00:15:23
Speaker
It's not so big at the field level because at the field level is electronic. Uh, the data are collected rapidly. Um, there's another survey that's reported in the employment situation report called the household survey.
00:15:36
Speaker
And that's, is a a big field operation. but it's entirely conducted by the Census Bureau and paid for by a BLS. So right this is where people call households and interview them.
00:15:48
Speaker
Right. So hopefully people listening now have a better sense of how this process works. A lot of people working a long time and the commissioner sees it about, what, 48 hours before it actually yeah hits our screens and sort of already locked in.
00:16:06
Speaker
So I think hopefully that will that will help people understand that piece of the puzzle. I think the other piece of the puzzle that... I think is often confusing for folks is why the numbers get revised. Right. And this was sort of the reason why we've, why we're having this conversation, right. Is that we have these big revisions to the couple previous months.
00:16:26
Speaker
And so given everything you've just told us, maybe you could tell us why these numbers end up getting revised. We've got this sample of 120 some odd thousand establishments or with a lot of people collecting the data and processing it.
00:16:38
Speaker
And so. I think just instinctively, it makes sense that there'd revisions, but I'm go to let the former commissioner tell us like ha why these numbers are revised
Why Job Numbers Are Revised
00:16:48
Speaker
and then how that process sort of works its way through.
00:16:51
Speaker
Sure. ah Well, if all 120,000 firms reported on time, there would be no revisions. Yeah. So they don't all report on time.
00:17:04
Speaker
ah Unlike the household survey, where 100% of the sample is surveyed every month without fail, ah even though our response rate not 100%, we know the non-responders have just said we don't want to. We've contacted them. Everybody's contacted each month.
00:17:20
Speaker
With the business survey and they're voluntary turning it in we give them two more months to get that survey in. So let's go back to the lake with the minnows and the bass and the Northern pike and all that stuff in the lake.
00:17:36
Speaker
We've done this survey and we've sent out the surveys to the minnows and to the bass and to the Northern pike. And um about 70% of the fish have have have reported.
00:17:49
Speaker
So on the basis of that, we say, well, the other 30% must be like the 70%. And so we'll use our mathematics to say, here's the we'll blow up the sample and we'll say, here's the total number of fish by type in the lake.
00:18:04
Speaker
So if you get a pretty good first response, 70, 80%, so seventy eighty percent then, you know, your revisions, if from data coming in on the second month and the third month are not going to be very big. And that's, that's been the history.
00:18:18
Speaker
um So the problem we're having now is the response rate has dropped down to around 50%.
00:18:27
Speaker
That's the first month response rate. Now, by the time we get to the third month, we get 93, 94% of the returns in. So it's still an excellent, excellent sample, but there's now fewer returns in the first month. So when we say, here's here's our estimate for the number of types of fish and the types of of changes in this in the in the lake from last month to this month, our first month estimate could be significantly off If the second month comes in with information that disconfirms our assumption that what came in the first month is like what's going to come in in the in the second month.
00:19:06
Speaker
And that is that's the nature of it. There's no conspiracy here. There's no rigging of the data. There is a ah ah much more much more awesome problem. It's much more severe.
00:19:18
Speaker
And that is that we're losing our ability to estimate accurately the final number because the first month number is falling in and responsory is falling.
00:19:28
Speaker
Well, just go out and crack the whip, right? Just say, you've got to get it in, require it. Well, we know from other countries and we know from other surveys that response rates are not easily reversed, no matter what you do. Pay people, ah change the survey, make it easier, you know, make it two questions as opposed to 18.
00:19:48
Speaker
um Almost nothing will change that. So there are solutions, hope we can talk about them, that to this problem. But if you say, Bill, tell me about the that the revisions. Is it because we're doing a worse job?
00:20:00
Speaker
Well, okay. yeah My first answer is no, we're not doing a worse job. We have we have this this problem. The second thing is that revisions do tend to be bigger, even with good response.
00:20:14
Speaker
When the economy is coming out of slow times or when the economy is slipping into slow times, we appear to be slipping into slower economic activity. ah Maybe I'm, I hope it's just temporary, right?
00:20:26
Speaker
It's in periods of slow down that our first estimate will overestimate the number. The second month comes in maybe smaller businesses, right? Maybe the small businesses were underrepresented in the first one. And they they're the ones that are having the economic trouble.
00:20:44
Speaker
Well, how we get this better information? And so we revise downward. Downward revisions, upward revisions tend to follow the business cycle. Right. So in terms of the sample, are the smaller businesses, they're trying to represent the overall, the aggregate sort of makeup of of the economy. That's right. um But the small businesses are more likely to be making changes as the economy sort of ebbs and flows.
00:21:12
Speaker
Yeah. so So what happens then for these small businesses that are making larger changes, or or I guess really more of the question is the creation and destruction of small businesses.
00:21:28
Speaker
How does the BLS then address that either because of the revisions or they're trying to find, you know, Joe's burgers that no longer exists, right? How do they go in and sort of try to fill in some of the gaps of like,
00:21:42
Speaker
re I guess, resampling the establishments. So we have a real challenge here in doing our survey work because when we first draw our sample, it is drawn from all of the data on all of the businesses as of March of the year in which we draw the sample.
00:22:01
Speaker
I say that because that is a particularly important month. The quarterly census of employment and wages is the sampling frame. That is, that's the entire lake. And we're drawing it.
00:22:13
Speaker
We're drawing our sample based on that one month. We're saying this is representative of the economy as of March. Well, now now you're in November. Okay. yeah And the economy has softened.
00:22:25
Speaker
Well, what we have is a model. it's and it's ah It's a statistical model called the birth-death model. We run it every month to estimate based on overall economic conditions, other things happening in the economy, GDP, doubt about ah all all that stuff.
00:22:41
Speaker
How many firms are being formed that were not in the March sample? How many firms are dying that were in the March sample, but now we think are not? Because we don't really know, right?
00:22:53
Speaker
We just don't know. And so the birth death model is very important to the monthly estimates. It's not, it'll add maybe 15% change one way or the other.
00:23:04
Speaker
Uh, but it is also highly sensitive to the business cycle. So when the economy particularly is decreasing, we have a hard time getting that death rate or that decrease in infirm rate correct. So one area where we would like to make improvements at BLS is in the birth death model. I think there are ways to do that.
00:23:26
Speaker
yeah um We'd also like to increase the the the kinds of data that are used each month to estimate the number of jobs. We can take, have survey data like we've talked about and blend that with data from other other sources. Perhaps we can blend in the the data from ah ADP, which looks at private sector employment each month, or from indeed.com, or from the other websites that follow employment.
00:23:54
Speaker
And by having a blended set, one survey that we're controlling totally, one set of what are called internet data or or publicly available data,
00:24:05
Speaker
we can get a better fix on the direction and magnitude of the employment change month to month. Yeah. I wanted to ask quickly ah before moving on, on the response rates. I think at least I'm sort of familiar with falling response rates at on the household surveys, right? You know, people are tired of answering surveys. There's landline move from landline on mobile phones and and more migration, all these sorts of things.
00:24:29
Speaker
What is the state of the art understanding of why response rates at the establishment level has changed? Is it is it this is kind of the same reason that there's still a person at the end of the day that needs to fill out or submit a form?
00:24:41
Speaker
Or is there something different about establishments from individual householders? There seems to be two things at work here. um First off, a systematic ah change. when And that is that all firms, large and small, are less willing to prioritize the form as opposed to their business.
00:25:02
Speaker
I mean, that's shocking, isn't it? yeah And so oh so that, that is part and parcel with the general overall trend in people saying, I'm going to be less responsive to what government's asking me to do.
00:25:15
Speaker
So we do think that that's part of it. Another one, frankly, John, is just, there is just a, an almost a completely episodic and unexplainable month to month reason, you know, one,
00:25:26
Speaker
One, one time I remember the the employment people at the, at the Wednesday briefing that I've referred to said, oh, commissioner, we have a very serious problem, um, with the numbers we didn't get to have the big box retailers submitted. And so, well, that is a big deal. And so why didn't they do it? Well, they both were doing inventory checking and inventory year end inventory checking.
00:25:48
Speaker
And, uh, they were just too damn busy. Okay. Well, that happened. Um, it's, it's, you have peculiar circumstances come up in the end is a solid survey. Cause you get 95, 93% of 91%. That's very good from yeah survey statistics.
00:26:05
Speaker
It's that first month. That's the issue. Right. That's the weakness. We don't throw the baby out with the bathwater because it's performing well, the survey is performing well overall, but not in that first month.
00:26:17
Speaker
Right. So I'm hoping that the purpose of this conversation that people sort of understand this process a little bit better. But I'm sure there are some people listening to this saying, okay, so the commissioner sees this report or these reports, right, for all these different data series a couple of days before they go out.
00:26:35
Speaker
There's no conspiracy. They can't change anything. It's locked in. So what does the commissioner actually do then? Like, what like what is the what is the job? So, yeah. So what is what is the day-to-day role of the of the commissioner?
00:26:50
Speaker
So the commissioner has a big job, um and that is to lead change at the Bureau. Because the economy is constantly changing, and we've actually mentioned several things that are happening in the economy right now that are affecting our data.
00:27:04
Speaker
The commissioner has to be a good and effective leader. ah There are continuous problems. So you're busy making the big long-term changes. I'll illustrate it from my own term.
00:27:17
Speaker
I came in with three objectives. ah to kind of redo the way we think about retail, retail employment, because we were not counting in the retail category online retail, which was really strange.
00:27:28
Speaker
So our our employment numbers were like with all those people were in we warehousing or something, which is completely bizarre. Yeah. The second thing was to improve our productivity estimates because we were not doing a good job there in counting service productivity in the service sector. But the biggest thing I did was to work on the CPI and make major changes to the CPI, which hadn't been updated ah since 2012.
00:27:52
Speaker
31 of 37 recommendations that came from the National Committee on Statistics, we recommend we we implemented those. All right. So that's that's how you affect the day-to-day activities by affecting the long-term picture. We've made investments in ah automation research. we We looked at AI.
00:28:11
Speaker
You know, we, we kind of said, here's where we're going to be. And I spent a lot of time doing things like interacting with the the two administrations, Trump and Biden, getting their policy priorities, um, kind of reflected in the product we were doing, certainly the annual publications.
00:28:27
Speaker
Um, you're, you're continuously busy with, with other smaller things, visiting the regional offices, uh, boosting morale. Um, I had a lot of work I had to do to get the, uh,
00:28:39
Speaker
BLS ready to move from the Postal Square building and but next to Union Station in Washington DC to a suburban location. Yeah, that's the kind of thing you do. You do not manage on a day to day basis. There are good managers there. You do not collect the data.
00:28:55
Speaker
these are These are product ah efforts done by experts. And there's no role for you to play there other than to be the pretty face in front of the camera. I want to come back to the firing of the of the commissioner, but I want to sort of like go forward and come back.
00:29:10
Speaker
The president has nominated ah someone to to take over. There's been lots of other names sort of floated without sort of getting into individual right experiences.
00:29:21
Speaker
From your perspective, if you were nominating someone, what are you looking for? Are you looking for government experience, being an economist, ah you know we having worked in government agencies before? like what do you you know you You are now in charge. you get the You get the honor. I'm putting you in charge now to get to nominate the next ah commissioner. What are you looking for in that person?
00:29:45
Speaker
ah So this particular nomination cycle is unique. ah The job is a much bigger job today than it was on July 31st, the day before the firing.
00:29:56
Speaker
know You have so much more you have to do as commissioner, not only lead to change, but you've got to do an immense amount of immediate work ah triaging trust, ah trust in the statistics.
00:30:08
Speaker
um All right. So what am I looking for? Well, I'm not looking for a person who understands what's happening, have happens in BLS. And people are surprised by that because you don't know on the outside what's happening on the inside.
00:30:22
Speaker
This is a statistical agency that is handling extremely sensitive data. And as a consequence, it is not open to the public for inspection.
00:30:33
Speaker
It's like a security agency. Yeah. So what you want the nominees to have had yeah He or she should be familiar with all the products that BLS has demonstrated a an interest in the statistical system, perhaps had some service or some interaction. you a lot of private economists are very active in the statistical system, uh, through their associations and through their own work.
00:30:59
Speaker
I think the major thing you want is you want someone who has demonstrated leadership in a situation of change.
Leadership Qualities for Future Commissioners
00:31:07
Speaker
ah Managing knowledge workers is extremely important.
00:31:10
Speaker
um You have to have experience there. Otherwise, you'll be overwhelmed by all the things you don't know about how to manage people's agenda and how how to work with their but their views.
00:31:22
Speaker
But the main thing is you have to really have been through a life experience of successfully producing change in an or organization. And I think that's why good commissioners tend to be mid to late career ah people, ah not because they're wiser, but because they're just more more banged up.
00:31:39
Speaker
ah they've they've They've kind of gone through things. it is It's not an easy job. um And it's a job that is term limited. You have four years and basically you could you could be reappointed, but the expectation is you only have four years.
00:31:54
Speaker
right So so that's what I'm looking for. I'm looking for I think that a PhD in economics is important because you're dealing with the academy and they're kind of touchy about that. um I think you should know statistics pretty well because you're going to get a lot of that language and in meetings and know you you can't ask what a standard deviation is. and there big um so um So that's that's what i'm I'm looking for. But I'm definitely not looking for a person who knows how to solve the problem on the outside.
00:32:24
Speaker
ah They need to be open. But once they're on the inside, that's their job. Problem solving, you know getting to those big changes, leading leading the organization. Yeah. So now stepping back to to August 1st, the president fired the commissioner, Erica McIntarfer, who I know very briefly through my service to a couple of BLS commissions.
00:32:46
Speaker
and And I just want to get in your head a little bit when you heard that she had been fired. ah Explicitly for, you know, I think he the president sort of, you know, tweeted out or whatever, you know, that she faked the the job numbers. what was your What was your immediate reaction? Like, what was your immediate thought? were you First off, were you like, I'm glad I'm not commissioner anymore? Like, I'm guessing that was your first thought.
00:33:08
Speaker
but I actually was. ah It is a good story. I was at lunch with a good friend of mine when we were talking about revitalizing a kind of an institution in Washington.
00:33:21
Speaker
And so I went out to the parking lot and I drive a pickup truck and I was sitting in my pickup truck and, um you know, so I thought, well, I'll just check my emails because you don't do that during lunch and text and see what's going on, you know?
00:33:34
Speaker
And in my phone was blowing up. It was just blowing up. And I sat there with the engine, eight the AC running, it's pretty hot day, for 45 minutes, I think, um partly on a conference call. I'm the co-chair of the Friends of BLS.
00:33:50
Speaker
and it's an advocacy group. And we were doing our response to the firing. And I knew I was stunned. i I've known Erica for a long time. She's maybe the best qualified person in a long time to to have that position, ah just a distinguished career in statistics and in management.
00:34:12
Speaker
So the accusation was, was amazing. and ah And so we issued this the statement and then thereafter I've been really busy on media. ah I want to back up and say something about President Trump and his and what he did.
00:34:29
Speaker
and he He certainly had the right to do what he did. i'm i'm i'm Maybe I'm in a minority, but I believe that the president does have the right over the executive branch appointees to remove any ah any appointee that he that he wants.
00:34:45
Speaker
ah Some argue that that's not the case, and if it if it shouldn't be the case, then Congress should act so specifically in legislation to exempt that that person or that position from such an action.
00:34:56
Speaker
But this was not an exempted position. The second thing I'd like to say is it was extremely ill-advised. um ah I don't blame him for not knowing how BLS operates on the inside. I made this point a couple of times in our conversation today that you really don't know unless you're there and or unless someone tells you.
00:35:15
Speaker
That said, John, there are people in the White House that I know know how this system works. They know. yeah And i if they were unable to get to the president to advise him otherwise, okay, fine. They're off the hook.
00:35:30
Speaker
But shame on them if they said this is okay thing to say because it has done damage to the BLS. It's done damage to economic statistics. um I was contacted once, if I have time to tell you, I was contacted once by a banker in Dusseldorf.
00:35:47
Speaker
Um, that's in Germany. And, uh, this is for another reason, but during the course of the conversation, he said, do you know that the exchanges here in Dusseldorf shut down for 10 minutes prior to every first Friday's labor labor report?
00:36:01
Speaker
And of course I didn't know that, but that's how important it is to decisions made in remote areas, remote from the United States. ah It is a crucial ah monthly set point for financial decision and other kinds of decision making and to impugn it without any shred of foundation.
00:36:25
Speaker
is is i'll use the term ill-advised but you could use a stronger term so so i and i i hope the president now and i think he has sort of backed walk back walked that a little bit ah let it kind of subside uh because it's so it's damaging to the economy you have to be able to have statistics to see the economy the economy does not grow on trees gdp is nowhere to be dug up
Restoring Trust in Federal Statistics
00:36:48
Speaker
like a root vegetable right um it um It's only defined in the numbers and the statistics that our statistical agencies produce.
00:36:56
Speaker
I do want to come back to that, and I do want to touch on the the solutions that you mentioned earlier. i did just want to ask, um you were appointed by President Trump. You served into the Biden administration. Right. And so you sort of had both of those. And McIntyre was appointed by Biden, served, obviously, just for a few months in the Trump administration.
00:37:17
Speaker
Right. Was your job, your role as commissioner any different as you stretched between those two presidencies or was the work you just, you know, you know, November was the same as March is just two different people sitting in the white house.
00:37:35
Speaker
Well, there were stark differences. um And I think that's the that's kind of a unique perspective. There wasn't, if you, let me answer it this way. If you were talking to a senior staff member inside BLS, they would not have noticed any difference. Their job didn't change from one administration to the next.
00:37:56
Speaker
My job changed a lot. um In the Trump administration, we were looking at a lot of issues like re-employment, What happens to unemployed people when they leave unemployment? How do we trace them?
00:38:08
Speaker
We were looking at local wages. we were we were we were very much interested in the immigration issues. um And so BLS was through me was getting special requests, tabulations, activity reports and so forth.
00:38:21
Speaker
I was on a lot of panels. I was doing a lot of national security work. I was in skiffs um and we and we were 24 seven for 18 months on the pandemic.
00:38:32
Speaker
When the Biden people came in, I then pivoted to looking at how do we create statistics that are more inclusive? Uh, how do we, how do we define workers by other kinds of characteristics and, and make them visible in our labor force reports?
00:38:49
Speaker
Um, I think the most, the biggest example of that was, but I initiated a monthly, not seasonally adjusted report on the unemployment rate of indigenous peoples. And that was, that was very reflective of, of President Biden's priorities.
00:39:04
Speaker
I did not sit in any national security issues that that that dropped off. And of course we were in the period of inflation, um, where's the end of my term. So there was quite a bit of work to be done on the price price side.
00:39:16
Speaker
So elections matter, John, they, they, they really do. And priorities are embodied in, especially when the parties change their new priorities. And, and, um, as a presidentially, uh, responsible person,
00:39:30
Speaker
because I was, while I was appointed by Trump, the Biden people liked me fine enough. I had to make sure that I was ah responsive to their requests and ah that was appropriate.
00:39:41
Speaker
At the same time, want to reemphasize no difference whatsoever in the daily working of BLS. right the you know The top management worked with me on the special requests, but everyone else did their work. Yeah.
00:39:55
Speaker
yeah So looking forward, this is gonna be a very broad question. So we can, you can take it however you like, but but you've mentioned ah trust in the federal data collection, statistical agencies, and also solutions to some of the lots of issues. I mean, it's data collection, it's analysis. There are a lot of issues that pop up big and small that that have solutions.
00:40:17
Speaker
And so this sort of a two part question, I guess is What is the solution to some of these detailed issues you've mentioned? Maybe we just pick a couple. And what is the solution to the broader issue of maybe this sort of cracks in the trust federal data that are being collected and being ah published? Like, what do you think needs to happen, both at kind of a micro level and a macro level?
00:40:47
Speaker
Well, there are problems that need to be addressed. I think restoring trust in any product starts with your recognition that there's a problem and you're working on it.
00:41:00
Speaker
ah People are not going to be more trusting if you just ignore their criticisms. And I think that this particular thing on the revisions, um yeah Yeah, initially I wasn't convinced it was a problem. I thought it was just episodic.
00:41:17
Speaker
um But I do think that there's a systematic element that needs to be addressed. So let's do blended data. let's let's But Congress needs to step forward with some some money. I think the whole ah salvation of all of our surveys, they're called probability surveys.
00:41:31
Speaker
ah The probability surveys can be restored or at least augmented with other data sets for it in BLS's case for just about $40 million over a two year period.
00:41:44
Speaker
the i heard I've heard an estimate for system-wide that is all within all the 13 recognized statistical agencies of 160 million. I don't know if that number is correct at all, but there is a number less than what Congress spends per day on, um I don't know, flights back to its constituency or something. it um It is not a large number in terms of the numbers that the federal government deals with.
00:42:13
Speaker
And i I'll just real because I was just looking this up. I think the BLS's budget, right, is something like $600 million dollars a year. So we're talking about, and that's fallen by, I think my numbers are like $150 million over the last decade. So yeah they're getting less money, but the numbers you just quoted are a fraction of that money.
00:42:32
Speaker
Fraction of that. And the reason why we don't have large numbers to deal with in BLS's case is that we've been spending time ah without asking for additional funding, studying this problem and devising and testing solutions. So there are things out of the, just really out of the out of the bank we can we can pull and say, well, let's let's let's now test this in real time.
00:42:56
Speaker
It takes about a year minimal of testing to make sure the new system is working like the old system and producing statistically sound results. Okay. Yeah.
00:43:07
Speaker
so The second thing on trust is to understand that part of the reason why people don't trust is that they're not getting what they want. ah it's It's a kind of a funny connection between trust and utility.
00:43:22
Speaker
um If you are continuously told, oh, we can't do this, we can't do this, you just have to live with this product that you don't really like, then you begin to like it less and less.
00:43:32
Speaker
um We learned during COVID that the public particularly businesses, want higher frequency so official statistics. And we began to produce them during the COVID years using the Household Pulse Survey and the Business Pulse Survey, which was a weekly survey with weekly results.
00:43:51
Speaker
And everybody thought were great. um We didn't ask many questions, but those questions were pertinent to the times we were in. So i I think we should find ways of doing high frequency official statistics.
00:44:05
Speaker
And that would mean we'd roll out new products. We would we would amaze people by the kind of higher utility the statistical system has in their daily lives. um And the frequency may be as high as as daily.
00:44:19
Speaker
ah Many banks now, um as as well as well as Federal Reserve banks, as well as universities, do these daily forecasts of gross domestic product of, of demand.
00:44:33
Speaker
It's called go forecasting. It's, uh, it's increasingly accurate. And I think the statistics could do something like that. What, what if you got a daily report of job changes in the economy?
00:44:45
Speaker
Um, uh, by industry, by large industry sectors. I mean, it's, it would all be modeling, but, but it it would be useful for a lot of firms and then Lee and, and, uh, maybe bi-monthly.
00:44:58
Speaker
ah So, uh, I think restoring trust is also doing a better job. overall of being more responsive. We're we're in a 20th century mold, post-World War II mold of doing survey work, probability surveys.
00:45:12
Speaker
That was the big thing in the 1950s. That was the cutting edge. And it has been very useful. But now we have other ways we can use the internet, AI, and lots of business data to see the world more clearly. And we need to do that.
00:45:28
Speaker
Yeah. Well, ah Bill, thanks so much for coming on the show. I really appreciate it. I'm sure your phone is still blowing up even a few weeks later. And I hope people will find this value. Hopefully they understand a little bit more what goes into this process and the commissioner's role. So um thanks so much for coming on the show. i hope I hope I get to see you again soon, maybe in that pickup truck. Hey, man. Thank you very much for having me.
00:45:54
Speaker
Thanks. All right. Take care. Thanks everyone for tuning into this week's episode of the show. i hope you enjoyed that. I hope you learned a lot. I hope you have a keener sense of what the BLS does, how it processes its data and the commissioner's role in that entire process and in the agency more generally.
00:46:13
Speaker
have a moment, please rate, review the show wherever you get it on the website, on Zencaster, iTunes, Spotify, wherever you get their show, even if you're watching it right now on YouTube. So thanks so much for tuning in.
00:46:24
Speaker
This has been the PolicyViz podcast. Thanks so much for listening.