Moody's Talks - Inside Economics - In Defense of the BLS

Episode Date: August 7, 2025

Former Bureau of Labor Statistics Commissioner Erica Groshen joins Mark, Cris, and Dante to cover a wide range of topics, including a somber discussion about the recent firing of the current BLS commissioner. Erica provides key insights into the role that BLS commissioners play in the day-to-day publication of economic data, as well as the longer-term challenges facing BLS and other federal statistical agencies. She also weighs in on the recent revisions to employment data that have garnered much attention and provides a thorough explanation of why revisions happen and the tradeoff between timeliness and accuracy. Guests : Dr. Erica Groshen, Senior Economic Advisor at Cornell University—ILR and Research Fellow at the Upjohn Institute for Employment Research and Dante DeAntonio, Senior Director of Economic Research, Moody's AnalyticsHosts: Mark Zandi – Chief Economist, Moody’s Analytics, Cris deRitis – Deputy Chief Economist, Moody’s Analytics, and Marisa DiNatale – Senior Director - Head of Global Forecasting, Moody’s AnalyticsFollow Mark Zandi on 'X' and BlueSky @MarkZandi, Cris deRitis on LinkedIn, and Marisa DiNatale on LinkedIn Questions or Comments, please email us at [e-mail]. We would love to hear from you.  To stay informed and follow the insights of Moody's Analytics economists, visit Economic View. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Welcome to Inside Economics. I'm Mark Sandy, the chief economist of Moody's Analytics, and I'm joined by two of my colleagues, my trusty co-host, Chris DeReedy's. Hey, Chris. Hey, Mark. How are you? It's early in the week. This is a Wednesday, August 6th.

This is a special podcast, and we'll let everyone in on that shortly. But it feels a little weird on a Wednesday, doesn't it? It does. It does. Yeah. And without Marissa. And she's on another plane somewhere, I think. Yeah, but she really wanted to do it.

to make this one. I know. Yeah. And we'll clue everyone on why is on that as well. And we have Dante, Dante D'Antonio. Dr. D'Antonio. How are you, doctor? Doing well, Mark. How are you? You're back from the beach, I see. I am. Yep, made it back. Yeah. No worse for the where. How was the beach? All good? It was good. Yeah, weather was good. Good time. Good, good. You're arrested? Finally, yep. Good, because we have a lot of work for you. I know. Yeah. And this is an important podcast because we have Erica Groshen on the line here with us. Erica, good to see you. Good to see you too, Mark. Yeah. Erica was the former commissioner of the BLS, the Bureau of Labor Statistics, obviously top of mind, given everything that was going on. Hey, can I, Erica, you're at Cornell now,

right? You're a professor of Cornell. I'm not professor, senior economic advisor. A senior economic advisor. That sounds better, actually. It means I don't teach on a regular because I do like little guest stints and things like that. Well, that sounds like a good gig. Yeah. I'm very happy to be there with them, yes. Although I work out of Seattle, so there we are. Oh, you're in Seattle. Is that home?

I guess that's home. Very good. Are you from Seattle, Erica? No, no. I moved here four years ago. But my kids settled here, and I have a granddaughter now, and off we went. Congratulations. I get that. I get that. And did you, when you were BLS commissioner, that was from 2012 to 2017?

Did I get that right? 2013, January 2013 to January 2017. So under President Obama? Yes. My confirmation was delayed, so my term ended up, and it's very unusual for the BLS, not really spanning presidential terms. Usually it does span terms, but I was only a commissioner during the first week of President Trump's first term. And then William Beach succeeded you as the commission.

Yes, with a gap in between, but then he succeeded me. Yeah. Right. Got it. You know, I do recall that that was to be confirmed for that spot was a bit of a, as I recall, a chore. I thought that was It took almost a year. Yeah.

I was nominated in February and sworn in in January. Right. I remember writing a recommendation letter to you to a bunch of senators saying, hey, what the heck's going on? But I'm glad that worked out.

What was the holdup? Why didn't it take so long? Do you know? They base, I think the Republicans were hoping that a that a Republican would be, would win the election.

Oh, I see. And so they wanted to hold it open to be able to appoint the BLS commissioner. Got it. So that then when President Obama won a second term, then they were willing to move ahead. And I was confirmed unanimously. Got it. Congratulations. Thank you.

That's fortunate for the BLS and for the nation that that happened. But before we, I want to get a better sense of, you know, life as a commissioner of the BLS, you know, obviously in the context of everything that's going on. And we'll come back to all that stuff as well. But before you were a commissioner, I was reading your bio. I didn't, I had forgotten you were at the Fed, the New York Fed for a long time. Yeah. I spent most of my career in the Federal Reserve system. So I started at the Cleveland Fed for seven years. And then I went to the, actually, I taught for a year at Barnard. And then I went to the New York.

Fed and I was there continuously except for a year in Basel at the Bank for International settlements. I was there until 2013 when I started to BLS. Yeah. Doing economic research, advising on monetary policy, did stint, doing regional communications work, let's see, what else, editing their economic journal. and writing papers and doing fun things. Being a Fed economist. Yeah. Being a Fed economist.

Right. Right. And so you were there for most of your career, then you went on to the BLS and then now at Cornell. Got it. Yeah. 27 years there and then Cornell, rather than BLS and then Cornell, right. Great. So can you give us a sense of the day in the life of a BLS commissioner?

Maybe to make it even more specific, you know, in the days leading up to and during the day the jobs numbers are released. I mean, BLS, Bureau of Labor Statistics. Oh, by the way, I have a good question. I have a quiz for you. Because I'm doing my research. What year was the BLS established? Do you know?

Before. Okay, very good. I could fool her. I was pretty surprised by that. That's a long time. It was the first independent statistical agency created in the federal government. I did led the way. Wow.

I didn't know that. So 150 years of history here. 140. Yeah. I mean, my round up to 150. Yeah. Closely. Closely.

Right. All right. So, and of course, the BLS does all kinds of things. The thing that's kind of front and center in the minds of most people, certainly folks like us who are on Wall Street and, you know, worried about what's going on with the economy and jobs. and Fed policy and everything else. We're focused on that jobs release that comes out the first Friday of every month.

And then the other big release that is important at different times, but certainly important now is the consumer price index, the CPI. But the BLS, it's enormous, right? The amount of statistical data that comes out of that agency on a regular basis. What it produces is enormous,

but it's actually not that large an agency in the federal government. When I was there, it was about 2,500 people. Now it's down to probably less than 2,000. About 1,500 of them in Washington and the other thousands scattered across the country, taking care of regional operations. It has about 25 programs. You've just mentioned the products of three of those programs,

the consumer price index, which is consumer inflation. And then the payroll, rather the employment situation that comes out every month is actually the product of two separate programs, the current population survey, which is the household survey measuring, how households are doing, you know, the labor market conditions from a household perspective. and then the CES, the current economic statistics program, and that's often called the payroll program or the business survey or establishment survey. It's got lots of names, but that's all the CES. And those two things, the CPS and the CES are the input to the employment situation that comes out monthly.

Got it, got it. And I mentioned before we went on the air here. that Dante was a former BLS employee. Dante, what program did you work on? I worked in the CES program, but I worked on the state and area side, so I did the state and metro employment data, not national.

Right. And you didn't overlap with Erica. You were there just right before her. That's right. Yeah, so you handed off the baton. Yeah, I was a lowly economist, but yeah, I handed the baton. Okay.

So, Erica, what's the day in the life like? What is, is it most, just a general sense of, you know, what's the job exactly? Yeah. Well, I mean, that week is a special part and it's, you know, basically, you know, about I guess in some ways a fourth of the time there because, you know, every month there's the employment week. So it's an important week, but it's actually pretty different from most of the other weeks. So it's a week where I didn't give many, I very rarely gave a,

talk or appeared outside to people during that week because I knew the numbers. And so it was both to avoid a slip or someone's thinking that maybe something had been slipped. Oh, can I ask you? So you actually knew what payroll employment, what unemployment would be almost a week before the release? No, no, a few days before. A few days before.

During that week. During that week. Because it's released on a Friday. So during that week, you would learn what the numbers were at that. Right. So usually by the end of the day Monday, I had the CPS numbers. Okay. It would be finaled by them.

And they had been, the collection is all done. For that survey, the collection is all done at census. So they collected the process, they collected the data, shipped it to the BLS. And the BLS then processed it and turned it into the tables that I would receive by the the end of the day on Monday. Mm-hmm. And did, you know, do you have any input at all into the numbers and the data? Does the commission...

On a monthly basis, not at all. Not at all. I have no access. No access to the underlying data. No access to the programs that generate it. Just get the PDFs. Right. Why do you suppose that you even know what the numbers are?

Why would, why? Because it's need to know, isn't it? I mean, I think. Especially the first few times, like, hey, I could move markets, you know. Yeah. That would make me nervous, by the way. I'd be very nervous for that information. It did make me nervous in the beginning.

Yeah. Because you could spill the beans inadvertently. That's part of the reason not to go out. Not to go out. Right. Yeah. Right. Yeah.

Right. So need to know. I mean, like the, my understanding is the chair of the Council of Economic advisors gets the data maybe the Thursday, the Thursday night, right? And I'm sure. Yeah, something like that. And then he or she would give that to the president, I assume. Brief the president, yes. Right, but no one else outside the BLS would, other than you and CEA and the president would know, I presume, or would the Fed chair know? Would the Fed get the information? I'm embarrassed to say I've forgotten, but I think the Fed chairman does. I think so. I think that that they tell the Fed around the same

time as the CEO, maybe the CEO tells them. I forgot. Right, right. It's 10 years ago now. Yeah, yeah, yeah, no, absolutely. It stands to reason, but yeah. Because I was really part of that, of transmitting that or in that conversation. Yeah, so that's what I, so I get the household number, and we'll talk about why I heard about. For one thing, it's just, you know, to let you know that, to let us know that everything's on track, right? I see. I see. And also, I was going to be briefing first the Council of Economic Advisors or some staff from the Council of Economic Advisors on Thursday, and also be briefing the Secretary of Labor from 8 to 8.3, So the half an hour before release at the Department of Labor under lockdown conditions.

So go into a room with the secretary and a few of their staff, a few of the BLS staff, and go and spend a half an hour going through the whole release. Got it, got it. So that prepares you, you look at them. You know, I'm an economist. There are things I always look for. They're trends. So I might, you know, do my own little spreadsheet about trends or this or that.

But you might do a little QC. Does this make sense? Does it not make sense? That kind of thing. Right. What explanation do I need if I'm brooding on about this? What do you feel that I'm going to need to explain? Right, right. Okay. What do I want to point out? Yeah. You, you, the way you answered the question is you don't have, my interpretation of what you said was when you get that number, you really have no input into it at all at that point in time. But it sounds like over time, you could. could have some impact on how that is calculated. That number is ultimately calculated. Is that right? Yes. So, you know, head of the agency, right? Views on what to prioritize.

Views on whose input to bring into the agency to have people thinking about what modernizations or changes are needed. So the long-term direction, yes. I mean, that's an important part of your role, looking over the horizon, what matters to emphasize and do, but not on a month-to-month basis. And when BLS makes any kind of changes in methodology, they test them, they run them by their advisory committees, they do research and write papers, they announce it's going to be changed. They, when the change comes up, they very often will do the old way. in the new way for a long time, so people can compare them and then they drop the old way. So there's a real process for being very transparent and very careful about methodological changes.

Okay. I do want to come back and obviously talk about the events of the past week, you know, with regard to the revisions to the current data and what's happened to the BLS, the current BLS commissioner who was just fired by President Trump. But before I do that, can you just give us a sense of the culture of BLS. Like, you know, from my perspective as a very careful user of BLS over decades now, I have found BLS to be highly professional. In fact, I've hired lots of folks, Dante being an example, I've hired lots of folks from the BLS. Marissa, the listeners of the podcast know, Marissa, she's another co-host, she came from the Bureau of Labor Statistics. And I find

the folks that come from BLS to be, as I said, highly professional. but, you know, very careful, you know, very thorough, very thoughtful, not willing to make grand statements without, you know, significant support for what they're saying. The temperament is, you know, very equine, what's the word, equinemias, equanimous, you know what I'm saying, very, you know, I'm not excitable. Like Dante, I've never gotten Dante to, you know, other than smile. Tate's not effusive. He's not effusive. But that's my experience. Is that, am I seeing the real BLS when I say that? Yes. I mean, there used to be a joke when I joined there. Someone said, oh, how do you tell the

extrovert at BLS? And said, oh, the extrovert and BLS is the one who looks at your shoes in the elevator instead of their own. That's great. That's great. That's a great line. These are the world's most dedicated data nerds, right? They are dedicated to accuracy, truth, the mission of the agency. It's the culture just, you know, really embedded in this sense of mission.

And that's what I saw everywhere. It was really impressive. And I would say when people asked me what I loved about the job, that was one of the three things that I always mentioned, right? That they were just like the best colleagues ever, because you could ask them any question, and they didn't take it personally. They weren't defensive.

They were totally about, oh, that's a limitation. And let me show you the work I've done on that limitation, and this is how we deal with it. It was really ingrained in them that everybody there was doing the best possible, you know, working as hard to make the best possible outputs and be as transparent as possible. And nobody thought that what they were doing was perfect and didn't need to be improved. But they also thought that they were doing the best that they could right now. And always thinking about how to make it better.

because statistics are inherently imperfect and yet really, really valuable. And that tension, some people think that something that's imperfect isn't valuable. But actually, the job is to try and illuminate things that are inherently uncertain. Got, got it. You said three things. You mentioned one. What were the other two? the other two were like the ham sandwich you could get in the cafeteria every day

it was learning something new every day uh-huh yeah and i learned something new about the economy statistics or how government works or how some other part of the economy operated or communication. Just every day. In the beginning, it was totally drinking from the fire hose. But even by the end, I was still learning something new every day. And that really made it a lot of fun. Sounds like a lot of fun.

Hey, before we move on, let me hand up. You want the third one? Oh, yeah, absolutely. I'm sorry. Yeah, I was, I thought you intentionally left out the third one, but please, I'm all for it. Okay, well, I could keep it a secret. No, don't do it. Don't do it. Let us know. What is number three? It's the pure mission.

Oh, the pure mission. Yeah. Right. Yeah. It was clear and it was pure and it just made me feel good every night when I went to bed and every morning when I got up. Right. With my colleagues, I was pursuing a really pure mission that I could feel good about every day. Right. That is a reason to like a job.

That's for sure. For sure. Mission is critical. Before we move on, let me hand the baton back to Chris and Dante. Guys, you know, given, I'm sure I missed something that you'd like to ask Erica about up to this point. So, Chris, anything I missed that you'd like to ask? Maybe just a general question in terms of, you mentioned everything you liked. Maybe what would you change?

What do you think is the biggest challenge facing a BLS commission? or the agency more broadly? Oh, wow. I only have an hour. I can only pick one. I, okay. The constraints on modernization.

Okay. Okay. So, you know, and we'll be talking about, you know, big, long-term things that the agency was facing. basically declining response rates, combined with a permanent part of the mission of a statistical agency is continual improvement. It's never one and done. The economy is changing. Technology is changing.

Everything is always changing, and the statistical agencies have to adapt to that. Needs change. and at the same time, the opportunity is there because we have this huge influx of digitized information all the way through the economy. And the statistical agencies are chomping at the bit to be able to really take advantage of that. But the funding isn't there. It requires an investment. It's not going to make anything cheaper, at least in the short run. Maybe it will in the long run.

but in the short run, you need to continue to produce what you're doing and develop a whole new approach to creating statistics. And the funding isn't there. And then there's some structural barriers we can talk about that make it harder to do that. So that was really the most frustrating and difficult part of it. How do you communicate to Congress and the people who influence Congress that this, public good that they take for granted is at risk and could be so much better, but it doesn't produce ROI in any obvious sense. I mean, we all know it.

In fact, the uproar over the firing of the BLS commissioner this week shows it's a lot of people aren't used to speaking up for statistics, but when they're about to lose them, they realize that they'd be lost without them. Yeah, yeah. Yeah, I mean, I think you can actually towed up the cost pretty easily of bad statistics. I mean, take the CPI, you know, the consumer price index. So you can correct me if my wrong. My understanding is there's been budget cuts there recently, so-called doge cuts,

so we've seen staff reductions, and therefore the BLS has not been able to send out as many folks to kind of canvas for prices for different goods and services. Therefore, the BLS has been forced to impute more of the prices that they're including in the CPI. So prior to the start of the year, say roughly 10% of the products and services had imputed prices now were up to 30%, 35%. Now, you know, that may or may not result in worse CPI statistics. My guess is it can't be good. I mean, there's no upside to that. There's nothing but downside. But if you're, you know, sitting in as a bond investor and you're seeing that, the one thing that you immediately come away with is there's more uncertainty around that estimate. For sure, there's more uncertainty

around that estimate. Therefore, if I'm buying a Treasury inflation protected security, a tip security that price and its yield is determined by that CPI measure, I will demand a higher interest rate to compensate for the risk posed by that increase uncertainty, right? Is that just, and, you know, I mean, if I'm a, just by a 30-year treasury bond, same deal. If I don't know what inflation is or I'm unsure about what inflation is, I'm going to ask for a higher interest. And, you know, it may not, it's not going to be 10 basis points, but even if it's one basis points, even it's half of one basis point, you multiply that by 30 trillion plus.

us dollars outstanding. That's real money. That's real money. No? No. Totally. Totally. But that's not it can be hard to convey, right? So even that, you know, you took, you took several paragraphs to explain. Yeah. Right. Right. The example I often used, and it's really similar, is social security benefits are indexed to the CPI. If the BLS is off by a tenth of one percent, a basis point, the federal government will overpay

or underpay beneficiaries by a billion dollars. There you go. Great example, great example. And that's just, you know, one use of the number. If people don't trust the numbers, then they won't write long-term contracts that are based on, that have an escalator using the CPI in it, right? And they won't write, they won't write those long-term contracts at all. They'll be renegotiating them every year or every two years instead of five years or something like that.

And then you have, you know, the economy has this wasted kind of a deadweight loss from, extra negotiations that wouldn't be needed otherwise. Yeah, so you say ROI, return on investment, it feels pretty obvious to me that, you know, the ROI is massive. I mean, think about just go hire a few hundred more BLS folks to go out in Canvas, please, and the savings would be enormous to taxpayers. And every American, right, think about mortgage rates, think about auto loan rates, think about credit card rates. They're all higher because we can't figure out or we're unsure about, you know, what inflation is. It's just, it's mind-boggling.

It's just, you know, highly- But if you take that to a congressperson who said, well, you know, show me how, show me how government spending is going to go down then. You know, ROI, I want to know, I want to know what, you know, how much, if you make this change, how much is your budget going to go down? That's the ROI I'm looking for. Yeah. Right.

Actually, it would be really good. I wonder if CBO would ever take that up, Congressional Budget Office, because that feels like, you know, you could really easily show, I think, that the benefit to taxpayers is quite substantive with, you know, better data anyway. Hey, Dante, before we move on, anything for Erica? Yeah, I'd like to get your thoughts on just one other sort of long run challenge that we've talked about on and off over the last few years, and that's response rates to not just, you'll be a less survey, government surveys in general. obviously response rates have fallen a lot over the last 15 years since I've been at BLS. And some of that is just general fatigue of people from filling out surveys. Some of that is certainly funding. But I'm curious what your view is, you know, is there hope to turn that trend around?

How would that happen? I think the most we can probably hope for us to stabilize them. The research that has been done on what could turn them around has not found a silver bullet. paying people, making them mandatory. Those things just don't give you much of a boost. So, but you can do things to reduce the burden. You can do things to reduce people's fear that the data is going to be misused or misplaced or whatever. The statistical agencies have a very strong, almost,

on blemished record on keeping people's data secure. You have not gotten one of these, oh, your data was released accidentally messages from a statistical agency. And there's a reason for that. But so what's the solution? Well, there are things you can do to make the burden lower. And for instance, when I was at BLS, for that period of time, I'm not going to take credit for it, but I will point it out to you, the establishment response rates weren't declining. Now, they were declining before that and they declined after that. But it was because during that time, BLS was implementing a number of different modes for responding, more and more electronic modes, that made it easier for companies to respond. And during that time, it kind of kept

the response rates stable. And the most burdensome surveys are like the consumer expenditure survey, hugely burdensome on the consumers. The CNSTAT wrote a report saying, here's how you could redesign it to take advantage of digital information that most people already have, their bank accounts and their credit cards and things like that. And the BLS is really looking to push forward with that, but has never gotten funding to do it.

So they're doing it very slowly. And now the modernization plan needs modernizing. So there are ways. I think the solution is what we were talking about before. Surveys are not going to go away, but really concentrate them on the information you can only get from the survey. Things like if someone's not working, why aren't you working? You have to ask them that.

You're not going to find that another way. But other things, there are many other things, hours you worked, things like that. Those are things that are present in other records. And the BLS needs to be actively involved in how business records, in particular are kept and how to make it easy for that information to be transmitted to the BLS so that they can be used. And there's a payoff to companies moving in that direction because then they'll be able to benchmark much better. Well, this is a good segue into the recent events. The last Friday, August 1st, there was the release of the, you know, the monthly

job numbers, both the payroll survey and the household survey. The payroll survey showed some large revisions downward. So they showed not only that the job gain in the month of July, the last data point was on the soft side, I think it was around 70, 75K, but the down revisions to the month of June and May were substantively down. And so now it changes kind of the whole picture that we had of the labor market. Before the July data, we thought average monthly job growth was somewhere between 100 to 150K. Now it feels like even given the potential for further downward revision that were somewhere no higher than 50K, we could be 2550K, which is, that's a big difference. That's a big difference. And it changes your perception, you're thinking

about not only the perception, but the reality of what's going on. That's close to stall speed for the economy economy feels like it's struggling going into recession potentially going into recession so uh thus you know obviously the firestorm that that that precipitated politically and the firing of the BLS commissioner those revisions what do you what do you make of them you know are they completely out of bounds are they consistent with historical norms you know we you know what do you make of them and what do you think's going on here what's behind those revisions So they're big. They're big.

But they're not unprecedented, right? Revisions tend to be bigger and pro-cyclical at turning points. Right. So think about how revisions come about, right? The revisions, and I've said this, I don't know how many times I've said this in the past six days. And I said it often before, but not so often. So revisions, and repeat after me, revisions are a feature, not a bug. Right.

Good point. All right. They are the intentional feature added to a data series when an agency wants to have data that are both. very timely and very accurate and can't get those at the same moment. So the first release is the 60 to 70% of the sample who got their information in on time. And this is the picture that you get from those people, those companies. And there's a lot of information in it, but it's not complete. Right.

And the BLS doesn't want to sit on that because here you go, guys. Here's the preliminary number. And they are preliminary, and BLS tells you that they're preliminary, and everybody forgets the preliminary part of it. The second month, by that month, the collection rate is much higher. And by the third month, it's up to 94%. percent or something like that. And by the way, Erica, just as a sidebar,

correct me if I'm wrong, but if I look historically, that's pretty consistent. You know, it hasn't come down all that much. No, it's a little bit lower on the first, but the 93-94, that's pretty normal. Yeah, right, okay. Fair enough. Okay.

And so when, you know, in that first month, for all of those missing values in the cells, The cell is a group of employers that's the same occupation, same size, same geographic location, right? All of the missing values mostly are just treated as missing values. So for that cell, you just take the average of the people who did report. So that is an implicit imputation to the non-reporters that they're exactly like the ones who did report. When you're not at an inflection point, then that's, that's usually pretty good.

In such a point, meaning the economy going up or down. Yeah, that you're a stable economy. Yeah, that you're not about, you're not, you're not just taking off into an expansion or heading into a recession, right? But at the inflection points, then you've got a lot of companies that are changing from adding jobs to cunning jobs. And you've got some at the beginning of that and some at the end of that. And you don't have a trend yet developed around it.

And so the indicators are pointing in a lot of different directions. That's the time when everybody wants to know as much as possible. And it's also the time when information is most confusing. It's just at its nature, right? And so that's what you see in the data. So you get this kind of serial correlation on the revisions, right, which doesn't happen normally. The correlation for the listener means that if you have a downward revision, it's more likely that the next revision will be down.

Yes. Yeah. Right. And that doesn't happen when things are fairly stable, but it does happen at these inflection points, right? Right. And all right. So now BLS has actually done some things to improve that. For very large companies that dominate cells,

it now has firm-specific imputation routines. To take, so because they're very often different from the small companies in their cell. And they're important, right? Right, right. And so BLS will estimate those separately and put them in instead of doing the implicit imputation. Got it. And that has helped to reduce revisions overall. They started doing that around the time that I was there.

They chose some of these big companies. And they said, so big companies are in the certainty part of the CES. sample. They, you know, they don't go in and out. If they're participating, they're in all the time. Because we pose them randomly, it would be 99% of the time anyway, so why, you know, do anything else? Right. And so the, because not all of them participate, but the ones that do, you know, thank you. Thank you. Thank you. Right. Don't a bottle of wine. Yep. But we can't do that. No, you can't do that. Too bad.

You can tell you who there are, although. Take them to a ballgame or something. Nothing that could be observed. They could announce it. And any company that doesn't participate, shame on you. Okay. Okay, so just to replay it back so people can digest it because this is obviously quite complex if you're not doing this on a regular basis

is you know the first release the like the July number we got on August 1st that has a 60 70 percent response rate so well collection rate because if you collection rate yeah because if you know if you refuse to ever participate yeah we're not putting you in the yeah okay collection rate And then you start getting more information as you move forward here. So we're now in the month of August. More companies are reporting. You're collecting that information. And so next month, when the bailouts comes out with this estimate, it's going to take

account for those additional responses that it got. And what you're saying is that in an economy that's stable, you know, generally the revisions are not going to be that big because the companies that are reporting late. are doing what the companies that did it early were doing. But if you're at a turning point, and you said inflection, I'll use turning point. You know, the economy's going, and in this case, South is going into recession, then the companies that are reporting late, that's real information there. They're saying, hey, things have gone south here for me in the last few days, last few weeks,

and things are worse, and that's what we're picking up in these downward revisions. So as you say, that's not a bug. That's a feature. There's real information there saying, hey, there's a problem, guys. the economy is going south. And we're, and when you're down to, you know, 35, 40, 45, 50,000 jobs a month, you're saying, we're pretty darn close to some negative numbers. And by the way, and this is now Zandi speaking, this is a, you know, my, I'm just taking one step further. The next revisions we get, this is a forecast. I bet you we start getting some negative

numbers. We started getting some negative number. Very possible we get some negative numbers here, particularly when you consider the so-called benchmark revision, which we don't even need to go into, but that's coming too, and we're going to learn more about that next month. So did I get that roughly right, Erica? Yes. Yeah, you did. I'll throw in one other little thing just for the people who are really in the weeds. Another factor.

That's us, by the way. That's this podcast because we're in the weeds, in the weeds. So feel free. Okay, I'll feed that. Feed that. Okay. Okay, very good. From the second to the third one.

that's also when you get the impact of changed seasonals. Right. So every month, BLS redos the seasonals, and that's when the seasonals change. And around this time of year is when seasonals can be changing because school schedules change and hiring for schools changes. And in fact, one of the bigger, one of the bigger industry jobs, groups that that had this, that, that saw these big revisions was state and local education. And this happened to be around the time in a direct causality, I can't prove, but this is around

the time that federal subsidies to, pandemic-related federal subsidies to state and local governments expired. So they may, so the state and local subsidies. schools that were relying on those subsidies may not have hired as many as much staff as they would have if they'd had those subsidies continued and that would be this combination of governments state government often state and local government often report late late right right and and so that's you know something that that might be in their initiative addition to a more broad spread, you know, turning down. Before we move on to the firing of the former commissioner, one question, that one kind of idea that's been out there in the atmosphere is,

well, maybe the BLS should not release, not do the first release, you know, wait a month and, you know, release. What do you think of that idea? Well, I mean, that's always an option. If you don't want to revise a number, you can wait and, you know, wait. until you think it's perfect. You know, why not wait a year or two, right? And then you nail it, right? Or you can only, you can do it once and never look at it again. Right, right.

These programs, in the CES, the wisdom in this program is to say we value both and we're going to do it. So anybody who wants to can ignore that first number. Go right ahead. Ignore it. Right. But the BLS has found by test that there's information in it that is valuable. It's just not as good as it will be in the next month or the month afterwards. But it's like, you know, if you go back to the analogy of driving a car, people often use and, you know, what's the state of your windshield?

Well, the first month is after you wipe it once. The second month is after you wipe it twice. And the third month, it's after you wipe it again. Oh, I like the metaphor. I like that metaphor. Yeah. And you see better, but do you not want to wipe it the first time? Yeah.

Or do you want to claim that it's clean when it's not, you know? I guess what you're saying is, yeah, if you know, you're not going to drive the car until you wipe it twice. Do you really want to wait to wipe it twice before you make that decision to drive the car? Yeah, I got it. Okay. Hey, Dante, anything on this? before we move forward, any other questions you'd like to ask about, you know, the nitty gritty

here? No, or the questions, I think maybe just to reiterate the fact that, yeah, there are these two pieces of the revisions, right? One piece is from new sample information, right? Additional businesses coming into the sample for month to month and then the seasonal adjustment issue, right? And I think it's probably about 20 years ago now when BLS started concurrent seasonal adjustment.

So we get these additional revisions as a result of that updated seasonal adjustment, right? So as we get new information, we got July, and as we revise June, that's going to impact the seasonal adjustment for May and June and July, right? So all of that new information is being included in the seasonal adjustment process to try to make it as good as it can be, but that also leads to an additional source for revisions every month. Yeah, good point. I think, you know, the other important thing to emphasize is that these revisions are not the result of a discussion or a conversation. right this is it's not a judgment call it's just arithmetic yeah that's right the the judgment was on how to set these uh and the judgment really judgment based on research right is how you set up these estimation routines and then what comes out every month is what you collect and then what spit out

the other end. Yeah. Let me, let me ask one more question before we move on. So what's your judgment play economist now? You're in my shoes, Dante's shoes. How do you think the economy is doing in the context of these numbers? Are you concerned about the direction of travel here? Yeah. I think that this is at least inflection, if not turning point. I mean, Inflection would mean, you know, something flat, turning point would mean actually heading into a recession, right? And somebody asked me an interesting question that I answered off the top of my head and I still agree with. That's not always true, but, you know, what's changed? Why do you think this has happened?

And ordinarily, a time of a recession, there's some glaring examples, COVID's example, but usually you don't know exactly what's causing. Afterwards, you look at it and you say, oh, okay, financial crisis or, you know, or this or that. Well, this time, what I'm looking at is the main thing that's changed is government policy, right? At the root of it, nothing else that I can point to has changed. So to the extent that the slowing is due to uncertainty or the direct impact of government policies, tariffs, immigration, federal government operations, things like that, then the good thing is it could be turned around, but not if those policies continue. Got it. Got it. Okay, let's turn to the firing of the BLS commissioner, another Erica.

And can I ask, how do you pronounce her last name? Erica. McIntoffer. McIntoffer. McIntyff. I've wondered how to say that. What do you make of that? How do you process that as a former BLS commissioner?

What's your takeaway? What is your thought process around that? I would have said that this was. wasn't possible that Congress had designed this position, like all term presidential appointee positions, as a position that spans presidential terms, where the holder of that position could only be removed for cause, where cause is dereliction of duty, malfeasance, embezzling, you know, something like that, not performing the job, right? And certainly not on a whim.

So this, the firing seems to have result from a thought that there's not much difference between a term appointment and a, and the other political appointments where the purpose of the job is to advance the policies of the administration. And those people serve at the pleasure of the president, and that makes sense. And that's fine. And they do important work. I'm not denigrating it, but it's a different sort of work. The term positions are for people who are involved in producing the public goods that the government produces. And their job is to make sure that, that the production of the public goods is done well. And so it's a little bit of a mind thing.

You know, it's not that I didn't, in some sense, fear it might happen, but when I had the job, I was secure in the fact that it couldn't happen. Right. And the reason it's damaging at this time is because then it makes people wonder about what the new commissioner will do when the new commissioner comes in. So even if that new commissioner has no intention of change. the objectivity of the Federal Reserve, the BLS, people are going to be wary until they finally get comfortable again. And that means there's going to be more uncertain than there should be about the quality of the data.

Should we be worried about the quality of the data, Erica? I mean, the timeliness, the accuracy, the quality, should we be? Or is that shown we have to wait and sit? how this all plays out. So in the very short run, you should not be. Because Bill Wyattrowski is the acting commissioner. I appointed him as acting commissioner. He has been acting commissioner twice before already. He is a career BLSER. And I don't think there will be no meddling as long, I think, as he is assistant commissioner. When the new commissioner comes in, it really will depend on what that commissioner does.

And I think there will be some grounds for worrying about it. There's another damage that we haven't talked about also, which I'll throw in, which is that during the past 10, 12 years, there have just been. an increase in baseless attacks on statistics for political gain. And that does a disservice to the agencies. It reduces use of the data. And that damages people's decisions after that. And it also probably discourages some people from participating in the surveys.

which damages the quality of the data. And so I fear that concerns about the quality of data will have that ripple effect on further depressing response rates. And I fear you can imagine a scenario where we go down the route of Argentina or Greece. Oh, goodness. And it did not serve them well. It would not serve us well.

Now, I'm not saying that will happen. I have no idea who the president is likely to appoint. And, you know, I don't know if they're going to be legal challenges about this and whether those might win. But there are things Congress could do. You know, Congress could investigate. Congress could clarify the protections intended for people who are term appointees, make that more clear than that to avoid the kind of arguments that may be bandied about that that not following.

I don't know. Somebody somewhere is arguing that it's okay. So whatever it is they're saying, figuring out their right, they're right, that doesn't work, right? Right. Right. The other thing that I wonder about and worry about is the chilling effect this might have on how other statistical agencies within the government operate. Oh, totally, totally. Yeah. I mean, this is not just the BLS.

Yeah. I mean, what does it mean if you're at the Bureau of Census or the BEA or? Mm-hmm. Right? Okay. Yeah. I mean, yeah, the education statistics. The National Center for Education Statistics has already been gutted. Its head was fired. Peggy Carr was fired. That has up to now been the most direct attack on a statistical agency. The others have been more of this collateral damage that has affected all of the agencies in BLS2.

but so this but I think because of the jobs report it's sent the BLS produces seven of the principal federal economic indicators and so it's much closer to it's visible to a much broader swath of the American public yeah for sure let me ask you we're coming to time here I just want to ask you one more question. I mean, it feels like in the context of everything that's going on that we need to start thinking more systematically, carefully about private sources of data. Yeah. And, you know, there are sources of data, like the employment, the ADP, the payroll, the human resources. Billion prices project for CPI, consumer prices. What do you think? Is that possible? Do you think the private sector step up here. And I don't know how they could fill the void, you know, created, but do you

think they could play a significant role in trying to, you know, fill the gaps that might develop here in terms of our data availability and timeliness and accuracy? Well, I mean, they could do it temporarily. Temporarily. But they actually benchmark baseline to the federal statistics. Oh, that's a great point. We knew that. We knew that. about ADP, ultimately benchmarked to the same unemployment insurance records, right? So, yeah. Yeah. So, yeah. So for a short time, but they, you know, basically, they, you know, they don't have a long run interest in doing this, right? Federal statistics are privacy protected. They're aggregated to answer the most relevant questions.

They're constructed very transparently. They're widely available to people without a prescription. This is not within the normal profit-making mandate of a private company to do all of those things. So they can be very innovative, but to produce the kind of documentation that's necessary for transparency, it's just, you know, they, just not going to spend their time doing that. They, they're not going to create a 40 year long time series if they can. They're, and they're going to want to sell some of it because that's how

they sustain their operations, right? Right, right, right. And that means restricting the information of it. So federal statistics are a very, a classic case of a public good. And they're really much more like infrastructure than most people realize, like roads and bridges, they facilitate transactions. And it's easy to take them for granted, but when they disappear, you're in trouble. So that said, it's great to have all of those new sources of information. They provide a lot to our understanding, and they help the statistical agencies, and they help people who are trying to understand what the statistical agencies put out. The future I imagine is one where the federal government works out relationships with the companies

that do this so that some of their information is fed to the statistical agencies to create these new products that have the features that are necessary for federal statistics, but take in some of the information that right now they don't have access to. And that will be, there will be governance questions and privacy questions. And there's going to be a big need for data standards and technology standards to accomplish that. It's all doable. We need a structure for our statistical system and the funding to get there. And then we can have data that is more,

timely, more granular, and covers more of the economy than anything we've got now, because the information is out there, but we're not getting the network effects yet. All right. Well, let's end the conversation with this. So you're in our shoes and trying to understand what's going on with the economy. What indicator or indicators are you watching to kind of send. Because as you said, you called it an inflection point. Actually, now that's thinking about it, that's the right word.

I use turning point, but you're right. I think inflection point is the right word. We're at an inflection point. What are you looking at to determine whether it's going to turn into a turning point? What's your favorite indicator or two? Okay. So I realize that I'm a labor economist. So I'm going to go full labor here on you.

Full labor on. Fire away. Temporary help services. Okay. If you are adding or shrinking, then you are either not hiring, well, temporary help services are often leading indicators. You put people in temporarily and then permanently. And when you're shrinking, you get rid of your temps first, right? So temp help services.

I look at them. I do like the diffusion index to get a sense of how widespread any downturn is. Right. That is looking at the number of industries covered by the BLS that are adding to payrolls. What percent are adding to payrolls, right? So if that's low, if that's like below 50 percent, less than half are adding to payrolls, that's a sign that we've got a problem. Yeah. Got it. The lower that gets, the more widespread this impact is. Right. Not waiting. That's just saying, you know, people doing different things. Are they all being affected the same way or not? And by the way, that's down. If you look at the July data, I think that three-month diffusion was like, what was it, Dante, like 46 or something, 46 percent? Something like that. I think it's moved under 50 in recent months. It's under 50 for sure.

Yeah. Anyway, that's a good one. Great. Yeah. I look at hours. Hours worked. Okay. Yeah, hours worked because you can cut hours before you cut people.

Right. Right. Right. And I do look. at initial claims and exhaustion of claims. Right. Right. How many people have exhausted their claims?

It's kind of, you know, ideas similar looking at number of people long-term unemployed. Right. So those are some of the things I look at. Yeah, those are good. I mean, if you look at them, they send a little bit of a mixed picture, right? I mean, diffusion index, negative. Job, I mean, hours work, negative, temp help. That's actually stabilized over the last six, 12 months.

It's down a lot from where it was a couple, three years ago. So maybe. Yeah, it's been doing weird things. Weird things, yeah. So I'd say that's a wash. And then UI claims are low. I mean, you know, the initial claims. So I guess that's what you're saying.

We're not, we're at an inflection point. So we've got to watch those things to gauge, you know, what's going on? They're all going to come from the same direction or not. Yeah. And, you know, there are some numbers there that we know are a little misleading. The federal government numbers we know are an underestimate, right? Because all these people on administrative leave being paid who have already lost their jobs. That's going to be in the data soon, isn't it?

I mean, the first few months. Next few months, yeah. Yeah, yeah. So we know that's coming. Okay. All right, very good. Dante, one last thing for the former commissioner. Anything else? I don't think I have a maybe that last question. Can I at least throw out a, you know, for all the people that work at BLS? You know, a lot of the attention has gone to the commissioner being fired. But I feel bad. I have friends that still work there. And, you know, firing commissioner also calls into question the integrity of everyone that works there, right? I mean, that's essentially what you're doing by by making that statement. And so I feel bad for the

people that are still there. I don't envy the position that they're in, but I still trust them to do good work. Well said, Dante. They're just the most world's most skilled and most dedicated data nerds. They are, you know, near and dear to my heart still. Yeah, same. Well, Erica, thank you so much for coming on. I know you've been very, very busy, you know, answering lots of questions, and we really appreciate that was very crystal clear. And I think, you know, I think it'll be very helpful. for people to hear this podcast because, you know, provide a lot of good information. So thank you and appreciate it. And thank you for all your service over the years to the nation.

So, you know, very important, you know, jobs. You're welcome. You know what? Can I take a second to make a plug? Yeah, sure. All there are many, many companies out there that do lobbying. And I don't think that many of them. them include speaking up for the support of the statistical agencies on their lobbying agenda.

And I think it's time. It's time. So I think government relations offices should be thinking about how to do that and encouraging all of the public officials that they talk to, to pay attention to this issue and do what they can because talk about an inflection point. We're at an inflection point for the statistical system too. Here, here, the nice way to end. I really, and I absolutely positively agree with that, you know, sentiment and view. I mean, companies have to step up here.

This is the time to do it. Yeah. Okay. Well, with that, dear listener, with that kind of sobering podcast, really, you know, sad to see, but here we are. I want to thank you for listening, and we'll talk to you next week. Take care now.