WEBVTT - Some of America's Most Important Economic Data Is Decaying 0:00:02.480 --> 0:00:14.000 Bloomberg Audio Studios, Podcasts, Radio News. 0:00:18.079 --> 0:00:21.599 Hello and welcome to another episode of the Odd Lots Podcast. 0:00:21.680 --> 0:00:24.040 I'm Joe Wisenthal and I'm Tracy Alloway. 0:00:24.360 --> 0:00:26.880 Tracy, I've said this a bunch of times, and we've 0:00:26.920 --> 0:00:29.600 sort of danced around this issue on the podcast before, 0:00:29.640 --> 0:00:31.080 but I've said this a bunch of times. I'm sort 0:00:31.080 --> 0:00:33.640 of fascinated by the degree to which we sort of 0:00:33.720 --> 0:00:36.800 take numbers on the screen. For granted, Like you know, 0:00:36.920 --> 0:00:38.879 we look at a price of a stock and it 0:00:38.960 --> 0:00:41.160 exists on the screen, it's like, where did that come from? 0:00:41.159 --> 0:00:42.080 How did that get there? 0:00:42.520 --> 0:00:45.000 Or like the first Friday of the month, the job's 0:00:45.080 --> 0:00:47.680 data flashes on the screen and we just start talking 0:00:47.920 --> 0:00:50.920 about what the data said, what the unemployment rate was, 0:00:51.000 --> 0:00:53.360 et cetera. But we don't really talk enough about like 0:00:53.400 --> 0:00:56.520 what had to happen behind the scenes to get that 0:00:56.640 --> 0:00:57.880 number to the screen, right. 0:00:58.000 --> 0:01:01.080 I find this really interesting as well, because obviously there's 0:01:01.120 --> 0:01:04.040 the data collection portion of this, Like, yeah, you have 0:01:04.120 --> 0:01:07.640 to go out and talk to people for certain surveys, 0:01:07.680 --> 0:01:10.960 certain data points, but also there are so many qualitative 0:01:11.120 --> 0:01:14.840 and subjective adjustments that you can make to that data. So, 0:01:15.160 --> 0:01:19.560 for instance, with CPI I didn't know that CPI waitings 0:01:19.600 --> 0:01:23.000 are like different depending on what city you're in. So, 0:01:23.040 --> 0:01:26.039 for instance, like food at home could matter a lot 0:01:26.120 --> 0:01:30.600 more in I don't know, Minneapolis compared to Chicago. It's 0:01:30.680 --> 0:01:33.920 really interesting. And there's also qualitative adjustments. Yeah, so if 0:01:33.959 --> 0:01:38.400 your fridge gets Wi Fi connected, then that gets incorporated 0:01:38.480 --> 0:01:41.600 into the price as well. So it's really interesting. 0:01:41.720 --> 0:01:44.880 It's super interesting, and like I have a you know, 0:01:45.240 --> 0:01:47.920 this is the issue. We don't talk about data collection, 0:01:48.480 --> 0:01:50.960 but it's obviously some of the most interesting stuff there 0:01:51.080 --> 0:01:53.200 is because of course, you know, things like food at 0:01:53.200 --> 0:01:55.360 home and one city should be different than another city 0:01:55.400 --> 0:01:57.920 if you really want to collect a basket, and so 0:01:58.120 --> 0:02:00.360 means that you have is some of the most interesting 0:02:00.440 --> 0:02:05.080 economics work being done anywhere in a realm that virtually 0:02:05.120 --> 0:02:07.880 gets no attention. And when we talk about data collection, 0:02:08.280 --> 0:02:10.560 you know, you hear it a lot in politics, right, 0:02:10.919 --> 0:02:13.680 surveys have gotten you know, no one answers the phone. 0:02:14.480 --> 0:02:16.560 It's you know, it's harder and harder to do high 0:02:16.680 --> 0:02:19.560 quality surveys. I think you've written about this, like response 0:02:19.639 --> 0:02:22.280 rates and just the actual cost of collection of all 0:02:22.280 --> 0:02:23.560 this data is on the rise. 0:02:23.680 --> 0:02:25.880 Yeah, so I have a bunch of thoughts on this. 0:02:26.080 --> 0:02:29.040 I'll just say talking about the economic data might be 0:02:29.200 --> 0:02:32.960 timely as well, because we know that Trump doesn't really 0:02:33.360 --> 0:02:36.560 respect I guess a lot of official economic data. And 0:02:36.600 --> 0:02:39.520 he's also trying to cut back on costs or funding 0:02:39.600 --> 0:02:42.880 of a bunch of different government programs. So it's clear 0:02:43.040 --> 0:02:46.279 that you know, entities like the Bureau of Labor Statistics 0:02:46.280 --> 0:02:49.320 could potentially be in the crosshairs here. But you're right, 0:02:49.600 --> 0:02:53.480 response rates lower, response rates have been happening for a 0:02:53.520 --> 0:02:56.800 long time, and it's really easy to look up like 0:02:57.040 --> 0:02:59.400 just how bad things have gotten. If you look at 0:02:59.400 --> 0:03:03.080 the BLS website, for instance, look at response rates on 0:03:03.480 --> 0:03:07.399 the housing portion of CPI that's gone from about seventy 0:03:07.440 --> 0:03:12.200 percent back in twenty fifteen to just fifty seven percent. Now, wow, 0:03:12.480 --> 0:03:15.520 response rates on jolts have gone from I think sixty 0:03:15.560 --> 0:03:20.399 seven percent to just thirty percent, So that's pretty amazing. 0:03:20.480 --> 0:03:23.440 And the other interesting thing here is it's not just 0:03:23.480 --> 0:03:27.280 a US problem, right, Like there's no US exceptionalism here. 0:03:27.560 --> 0:03:30.240 You see the same pattern in other economies like the 0:03:30.320 --> 0:03:33.880 UK and New Zealand, and actually in the UK just 0:03:34.000 --> 0:03:37.320 last month, the Office of National Statistics said it wasn't 0:03:37.360 --> 0:03:40.960 going to publish PPI because of a data quality issue. 0:03:41.160 --> 0:03:44.000 And also trade data had a problem because of errors 0:03:44.080 --> 0:03:47.520 in the data provided by HM Revenue and Customs. And 0:03:47.560 --> 0:03:49.800 now there's a task force to look at all of 0:03:49.840 --> 0:03:53.640 these mistakes and data problems at the ONS. So something 0:03:53.640 --> 0:03:55.640 that goes beyond the US here. 0:03:55.960 --> 0:03:58.000 This is a really big deal, especially when you just 0:03:58.080 --> 0:04:02.400 consider how much economic activity is based on being able 0:04:02.480 --> 0:04:05.520 to look at high quality data and make decisions from it, 0:04:05.880 --> 0:04:07.920 obviously in the market, but also in just sort of 0:04:07.920 --> 0:04:11.520 the quote real economy, et cetera. Anyway, without further ado, 0:04:11.680 --> 0:04:15.360 we really do have the perfect guest, someone who knows 0:04:15.400 --> 0:04:18.440 a lot about how the sausage is made and why 0:04:18.480 --> 0:04:21.880 making the sausage is getting more expensive. We are going 0:04:21.920 --> 0:04:24.440 to be speaking to Bill Beach. He was most recently 0:04:24.480 --> 0:04:28.440 the commissioner of the BLS, and he knows all about this. 0:04:28.520 --> 0:04:31.599 He was the fifteenth commissioner at the BLS, does research 0:04:31.920 --> 0:04:35.680 affiliated with the Economic Policy Innovation Center, and he's going 0:04:35.720 --> 0:04:37.640 to talk to us about all of these issues. 0:04:37.680 --> 0:04:39.719 Bill, thank you so much for coming on odd Lats. 0:04:40.520 --> 0:04:42.159 Oh Man, it's just it's great to be with you 0:04:42.200 --> 0:04:44.559 and Tracy. Thank you very much. It's a great topic. 0:04:44.800 --> 0:04:45.640 It's a great topic. 0:04:46.160 --> 0:04:48.839 I don't know like when the process begins, but it's like, 0:04:49.279 --> 0:04:52.200 I get this number. It says what the jobs report, 0:04:52.360 --> 0:04:55.680 it says how many jobs were created that month? How 0:04:55.680 --> 0:04:59.120 did that number get onto my screen or ontobls dot gov? 0:05:00.040 --> 0:05:01.760 And it gets there every month, Yeah. 0:05:01.720 --> 0:05:04.640 Every month, every month. It's never missed except maybe once 0:05:04.680 --> 0:05:07.719 send like a hurricane or so. Anyway, So to get. 0:05:07.560 --> 0:05:11.760 That first, yeah, the first Friday report, the Jobs Report, 0:05:11.920 --> 0:05:14.599 which comes out at eight thirty Eastern time on the 0:05:14.640 --> 0:05:18.000 first Friday. It consists of two surveys. So let me 0:05:18.000 --> 0:05:19.640 talk about the first one, which is the one you 0:05:19.760 --> 0:05:23.280 just mentioned that the number of jobs. That is a 0:05:23.320 --> 0:05:28.040 survey of about four hundred thousand firms out of eleven 0:05:28.080 --> 0:05:32.000 point three million firms in the United States, about four 0:05:32.080 --> 0:05:36.240 hundred thousand of them have agreed voluntarily to send in 0:05:36.760 --> 0:05:41.800 a pretty complex survey response. Every month, they send that 0:05:41.839 --> 0:05:45.960 survey to electronic collection centers. There's one in Chicago, there's 0:05:45.960 --> 0:05:51.000 one in Atlanta. And that survey has to contain the 0:05:51.160 --> 0:05:54.680 twelfth day of the month, if it is a work day, 0:05:55.279 --> 0:05:58.880 or the closest workday to that twelfth of November, or whatever, 0:05:59.240 --> 0:06:01.039 and the reason for that is we want to get 0:06:01.120 --> 0:06:06.400 at least one pay period in the report. Okay, So 0:06:06.680 --> 0:06:11.800 they submit that. That data continues to be collected through 0:06:11.920 --> 0:06:15.440 almost the end of the month, not quite, but almost 0:06:15.480 --> 0:06:18.920 the end of the month. It goes from regional offices, 0:06:19.080 --> 0:06:23.960 these regional collection centers to the national headquarters in Washington, 0:06:24.440 --> 0:06:28.640 where about forty people sometimes less, out of the two 0:06:28.680 --> 0:06:32.320 thousand people that work there massage the data, which means 0:06:32.320 --> 0:06:35.640 that they take the data which is in the survey. 0:06:35.720 --> 0:06:38.599 It's a sample, it's just a fragment of the total population, 0:06:39.160 --> 0:06:43.520 and they multiply the responses by what are called weights, 0:06:43.800 --> 0:06:47.640 and that gives us the national number. So that number 0:06:47.760 --> 0:06:52.479 is all done pretty much by Tuesday night, preceding the Friday, 0:06:54.120 --> 0:06:58.960 and on Wednesday, the Commissioner gets to hear the data. 0:06:59.120 --> 0:07:03.160 The Commissioner, by the way, plays no role whatsoever in 0:07:03.440 --> 0:07:08.680 massaging the data or multiplying the survey results times the weights. 0:07:09.560 --> 0:07:13.200 Then I would brief the White House. The White House 0:07:13.200 --> 0:07:17.160 would then brief the Federal Reserve Ward chairman, the Secretary 0:07:17.200 --> 0:07:21.400 of the Treasury on Thursday. That Thursday night, usually they're 0:07:21.440 --> 0:07:25.880 sworn to secrecy, and then at eight o'clock in the morning. 0:07:25.920 --> 0:07:28.920 I would walk over to the Department of Labor, because 0:07:29.000 --> 0:07:32.360 blas's headquarters is about a twenty minute walk from the 0:07:32.360 --> 0:07:35.680 Department of Labor. I would brief the Secretary of Labor 0:07:36.320 --> 0:07:39.920 and for an hour from eight thirty to nine thirty, 0:07:40.240 --> 0:07:42.760 the Secretary of Labor and his staff had to remain quiet, 0:07:43.440 --> 0:07:46.160 and then at nine thirty they could answer questions of 0:07:46.200 --> 0:07:49.720 the press. So that basically is the process for the 0:07:49.840 --> 0:07:54.040 jobs report. There's another survey there. It interrupt me any time, 0:07:54.120 --> 0:07:57.240 but that's where we get the unemployment raise, called the 0:07:57.280 --> 0:08:02.040 Household Survey, and that's a survey of sixty thousand households 0:08:02.720 --> 0:08:06.280 selected out of a sampling frame. We can talk about 0:08:06.280 --> 0:08:10.160 that term to be representative of the entire United States 0:08:10.400 --> 0:08:13.000 by demographic segments. You know how many people are in 0:08:13.040 --> 0:08:17.840 the certain age group, male, female, the racial and ethnic compositions, 0:08:17.840 --> 0:08:22.520 and geographic locations. So it's a complex sample. That sample 0:08:22.720 --> 0:08:26.720 is done usually by Monday of the week prior to 0:08:27.880 --> 0:08:31.240 the release on Friday, and then again I get to 0:08:31.240 --> 0:08:33.439 see the results, so the commissioner gets to see the 0:08:33.480 --> 0:08:38.200 results on Wednesday. So that's kind of the calendar. Both 0:08:38.240 --> 0:08:42.400 the household Survey and the Employment survey cover the same 0:08:42.720 --> 0:08:45.959 two week period in the month. Now there may be 0:08:46.000 --> 0:08:50.520 a day or two switch there. Sometimes the household survey 0:08:50.600 --> 0:08:53.920 has a few more days in it than the establishment survey, 0:08:53.960 --> 0:08:56.880 but they all have that twelfth of the month because 0:08:56.880 --> 0:08:59.280 we're trying to get the middle of the month. By 0:08:59.320 --> 0:09:01.720 the way, that's really important because if you have a 0:09:01.800 --> 0:09:04.920 natural disaster that happens prior to the twelfth or after 0:09:04.960 --> 0:09:07.920 the surveys are closed, that does not affect the results, 0:09:08.000 --> 0:09:10.000 even though you would think since it happened in the 0:09:10.040 --> 0:09:12.600 month prior that you know, the hurricane might have affected 0:09:13.040 --> 0:09:14.240 implument results. 0:09:29.840 --> 0:09:33.840 So one thing you mentioned is companies, you know, voluntarily 0:09:34.080 --> 0:09:37.520 responding to these surveys. And one thing I always wondered 0:09:37.559 --> 0:09:42.880 about is how how I guess onerous or resource intensive 0:09:43.480 --> 0:09:46.040 is responding to these surveys. So do I have to 0:09:46.080 --> 0:09:50.079 have like a person who's dedicated to doing this every month? 0:09:50.280 --> 0:09:52.839 Does it consume resources? How long is this going to 0:09:52.880 --> 0:09:56.280 take me? And then also does the BLS ever try 0:09:56.320 --> 0:10:00.760 to fact check some of these self reported data points? 0:10:02.320 --> 0:10:06.400 Those are great questions. So the survey is not just 0:10:06.559 --> 0:10:09.760 one or two questions. It takes about thirty minutes to 0:10:09.800 --> 0:10:13.960 fill it out completely. We do get responses that are incomplete, 0:10:13.960 --> 0:10:18.160 and we accept those incomplete responses. The survey is usually 0:10:18.240 --> 0:10:22.880 filled out by someone who can access the company's data 0:10:23.480 --> 0:10:29.040 on wages, on total employment, on employment by supervisory or 0:10:29.240 --> 0:10:33.120 non supervisory job functions. In other words, it's a person 0:10:33.360 --> 0:10:37.640 usually in the operations office of a company. Now we 0:10:37.679 --> 0:10:39.880 ask companies that are very large, you know, like the 0:10:39.960 --> 0:10:42.960 big box retailers, all the way down to really small 0:10:43.040 --> 0:10:47.719 corporations or small firms, and it becomes more and more 0:10:47.760 --> 0:10:50.320 of a difficulty for the smaller firms, and we're very 0:10:50.360 --> 0:10:54.240 cognizant of that. So BLS has reduced the scope of 0:10:54.240 --> 0:10:57.679 the survey to get it as small as possible and 0:10:57.760 --> 0:11:02.320 as free of burdens and compliance as possible. It's also 0:11:02.360 --> 0:11:05.400 an electronics survey so that they can fill it out 0:11:05.440 --> 0:11:09.160 on their computer screens and hit a submit button, which 0:11:09.240 --> 0:11:13.080 greatly greatly improved our response rate, by the way, so 0:11:13.200 --> 0:11:15.640 that survey is not so hard now to fill out, 0:11:15.640 --> 0:11:18.600 though I think it still takes about thirty minutes, maybe 0:11:18.640 --> 0:11:20.360 more if you have to dig out the data if 0:11:20.360 --> 0:11:23.560 you're not in possession of it. The household survey is 0:11:23.640 --> 0:11:27.360 a totally different beast. It is a long survey. It 0:11:27.400 --> 0:11:31.040 consists of over one hundred and twenty questions. It really 0:11:31.600 --> 0:11:35.480 is hard to fill out if you kind of don't 0:11:35.480 --> 0:11:38.559 know the terminology. Let me just give you one. We 0:11:38.600 --> 0:11:41.960 asked the question on employment. Are you working or have 0:11:42.080 --> 0:11:44.319 you looked for work in the past four weeks? That's 0:11:44.400 --> 0:11:46.720 the key question. We've been asking that question for since 0:11:46.800 --> 0:11:51.000 nineteen I think nineteen forty three. If they answer I'm working, 0:11:51.280 --> 0:11:54.640 then they are considered employed. We'll have follow up questions 0:11:54.679 --> 0:11:57.839 about that, about their industry, about their tasks and so forth. 0:11:58.559 --> 0:12:01.880 If they say no, I'm not working, then we say, 0:12:01.880 --> 0:12:04.120 well did you look for work in the past four weeks? 0:12:04.520 --> 0:12:09.439 And they'll say, well, you know, maybe I thought about it. 0:12:09.640 --> 0:12:12.600 Well thinking about it. It's not enough you can look 0:12:12.640 --> 0:12:14.760 for one day in the past four weeks and be 0:12:15.440 --> 0:12:19.720 considered in the labor force but unemployed. So that sometimes 0:12:19.720 --> 0:12:23.120 we have to have a verbal follow up rather than 0:12:23.160 --> 0:12:26.200 they're just you know, just filling out a form. There 0:12:26.240 --> 0:12:29.160 has to be a conversation. That's why we don't do 0:12:29.240 --> 0:12:32.239 more than sixty thousand households because it is it's burdensome 0:12:32.880 --> 0:12:35.400 for the person who's the questioner. 0:12:35.640 --> 0:12:38.480 This leads right into I think what I was about 0:12:38.520 --> 0:12:40.679 to ask you, which is, you know we're having this 0:12:40.760 --> 0:12:44.640 conversation April twenty twenty five. Your term of office at 0:12:44.640 --> 0:12:47.400 the BLS ended March twenty twenty three, so you're actually 0:12:47.440 --> 0:12:50.360 under both the two recent presidents, under the first Trump 0:12:50.400 --> 0:12:54.120 administration then Biden. But let's say, okay, so let's say 0:12:54.120 --> 0:12:58.640 it's March twenty twenty three. How much costlier is that 0:12:58.800 --> 0:13:02.680 process of commune unicating with the households than it was, 0:13:02.800 --> 0:13:05.680 say in twenty thirteen or two thousand and three. 0:13:07.720 --> 0:13:10.720 It was more and more costly, certainly, and that's because 0:13:10.760 --> 0:13:14.960 of inflation and how inflation affected the wage bill for 0:13:15.040 --> 0:13:19.400 the survey. Okay, the survey is taken by the Census 0:13:19.480 --> 0:13:23.240 Bureau under a contract by BLS, So we pay the 0:13:23.240 --> 0:13:25.760 Census Bureau to take this to go out in the 0:13:25.760 --> 0:13:29.480 field and conduct the survey. Here's an idea of how 0:13:29.520 --> 0:13:32.480 much more expensive it was. In twenty nineteen, which is 0:13:32.480 --> 0:13:34.800 the first year of my term, I didn't have to 0:13:34.840 --> 0:13:39.760 find any additional funding for the household survey. It might 0:13:39.760 --> 0:13:44.040 have been just a small amount, but nothing significant. By 0:13:44.160 --> 0:13:46.640 twenty twenty three, we were having to find almost four 0:13:46.679 --> 0:13:50.280 million five million dollars. More so, the costs of the 0:13:50.320 --> 0:13:53.760 survey is that ye per month, that would be for 0:13:53.840 --> 0:13:58.319 the entire year. The cost of the survey went up significantly. 0:13:58.440 --> 0:14:00.840 Now your listeners will be thinking about the federal budget, 0:14:00.840 --> 0:14:03.239 which is in the trillions of dollars. You'd say, oh, 0:14:03.280 --> 0:14:07.000 that's not very much money. But the Bureau's budget had 0:14:07.400 --> 0:14:12.600 has not really changed for almost a decade, and our 0:14:12.640 --> 0:14:15.880 costs have gone up, you know, progressively, especially during the 0:14:15.920 --> 0:14:18.839 period of significant inflation which we had starting in twenty 0:14:18.880 --> 0:14:22.360 twenty one, so that you know, finding additional money is 0:14:22.440 --> 0:14:26.720 really hard. During the pandemic, I found millions of dollars 0:14:26.720 --> 0:14:31.680 of couch money, unused conference fees, unused travel fees, so 0:14:31.920 --> 0:14:35.960 I could apply that couch money if it were, you know, 0:14:36.040 --> 0:14:40.240 to p purposes like buoying up the CPS or building 0:14:40.280 --> 0:14:42.960 a data We built a brand new data center out 0:14:43.000 --> 0:14:48.760 in the suburb of Washington. But once we were back 0:14:48.800 --> 0:14:53.120 in business and fully using our entire budget, it was 0:14:53.240 --> 0:14:56.760 very difficult to find those dollars to fill the hole. 0:14:57.320 --> 0:14:59.880 And I think you could safely say I've been arguing 0:14:59.880 --> 0:15:03.320 this now for over two years. Our surveys, but especially 0:15:03.720 --> 0:15:06.200 the current population survey, the one that gives us the 0:15:06.280 --> 0:15:10.920 unemployment rate, the labor force data are dying, They're decaying. 0:15:11.280 --> 0:15:14.200 They're in very serious trouble and we will have if 0:15:14.600 --> 0:15:17.200 unless we modernize that survey, we will see a time 0:15:17.240 --> 0:15:20.480 when we will be like the British right, unable to 0:15:20.600 --> 0:15:24.720 publish portions of it that just don't have sufficient sample 0:15:25.400 --> 0:15:27.120 for statistical release. 0:15:27.760 --> 0:15:30.520 Okay, I'm going to ask the obvious question here, but 0:15:30.760 --> 0:15:34.320 why have costs of conducting these surveys gathering this data 0:15:34.360 --> 0:15:37.840 actually gone up? And I understand, you know, building data 0:15:37.880 --> 0:15:41.720 centers is probably an expensive process. But on the other hand, 0:15:41.960 --> 0:15:45.760 you know, going electronic instead of mailing out thousands and 0:15:45.840 --> 0:15:49.080 thousands of surveys in theory, I would imagine maybe that 0:15:49.240 --> 0:15:53.280 saves you some money. So what exactly is causing the 0:15:53.440 --> 0:15:54.920 increase in expenses here? 0:15:56.360 --> 0:15:59.440 The problem is not with the electronic surveys. They will 0:15:59.480 --> 0:16:03.960 eventually be very costly and they're there. We're suffering from 0:16:04.000 --> 0:16:06.440 a different problem, which is just public support for the 0:16:06.480 --> 0:16:11.320 electronic surveys. So the jobs report, that is that portion 0:16:11.400 --> 0:16:14.920 of it that's electronically captured by these collection centers in 0:16:15.080 --> 0:16:18.680 Chicago and Atlanta. I think those are going along all right, 0:16:18.920 --> 0:16:24.280 for the time being, Our real problem is in the surveys, 0:16:24.760 --> 0:16:28.760 the data of which is collected by people survey survey, 0:16:28.840 --> 0:16:32.880 field survey people. They go out and they talk to 0:16:32.920 --> 0:16:36.480 the households and they're The cost is the obvious one. 0:16:36.520 --> 0:16:42.200 It is the wage bill. These people are highly skilled interviewers. 0:16:42.320 --> 0:16:44.840 They are not high school graduates, and I have nothing 0:16:44.880 --> 0:16:48.440 against high school graduates, but they are educated, very trained, 0:16:48.920 --> 0:16:53.960 oftentimes economists, who will go out and speak to people 0:16:54.400 --> 0:16:58.840 and to follow up interviews, oh my gosh, and they'll 0:16:58.880 --> 0:17:01.880 spend hours and hours do it this. Well, okay, fine, 0:17:01.920 --> 0:17:05.520 that's great, but it costs a lot to keep those 0:17:05.560 --> 0:17:09.080 people employed. They have other opportunities, so we have to 0:17:09.160 --> 0:17:13.520 have competitive wages. The wage bill is driving that. Collection 0:17:13.680 --> 0:17:17.080 costs are a little bit higher, that is the processing costs, 0:17:17.520 --> 0:17:19.640 but the big part of that is the wage build 0:17:19.640 --> 0:17:21.240 and the wage bill is not going to go away. 0:17:21.400 --> 0:17:24.960 That's just going to continue to go up. So we 0:17:25.000 --> 0:17:28.240 need to do with the household survey what we have 0:17:28.359 --> 0:17:32.600 done with the establishment survey, the firm survey, and that 0:17:32.760 --> 0:17:37.439 is modernize it so it is more electronically collected. And 0:17:37.480 --> 0:17:40.960 then we also need to integrate data which can be 0:17:41.000 --> 0:17:44.800 obtained through the Internet on households. Households maybe go on 0:17:44.880 --> 0:17:48.040 to a platform and with a tablet or something and 0:17:48.160 --> 0:17:50.959 supply Maybe we go to a million households and they 0:17:51.000 --> 0:17:54.280 supply five or six questions and we combine that or 0:17:54.320 --> 0:18:00.360 blended with a person to person collected data. That's that's 0:18:00.359 --> 0:18:04.080 what we call modernization. I mean, I mean that sounds 0:18:04.080 --> 0:18:08.080 so innovative to your listeners, but it is very innovative 0:18:08.080 --> 0:18:11.480 for the statistical system. Unless we do that, the cost 0:18:11.560 --> 0:18:14.000 will continue to rise. The response rates, by the way, 0:18:14.080 --> 0:18:17.919 are continuing to decline. So that's public support and we 0:18:18.000 --> 0:18:21.560 will be we just won't have these surveys in the future. 0:18:21.600 --> 0:18:22.720 There is too expensive. 0:18:22.840 --> 0:18:25.159 I want to get to the modernization in a second, 0:18:25.240 --> 0:18:28.320 but just talk to us about I mean, I think 0:18:28.320 --> 0:18:31.280 you said these the surveys dying, which is pretty dramatic. 0:18:31.400 --> 0:18:36.320 Given the centrality of this between the response rates, the 0:18:36.480 --> 0:18:41.320 increased wage bill, et cetera. How severe is this crisis? 0:18:41.359 --> 0:18:43.600 How much time are we talking about because I imagine 0:18:43.600 --> 0:18:46.200 it gets harder and harder to find that couch change 0:18:46.240 --> 0:18:46.880 to keep it going. 0:18:48.119 --> 0:18:50.920 Yeah, Actually, we're out of time. We have to cut 0:18:50.960 --> 0:18:54.200 down on the sample and I'll give you I'll give 0:18:54.240 --> 0:19:00.320 you the evidence, my successor, the current Commissioner of Labor Statistics, 0:19:00.359 --> 0:19:03.480 at the end of last year that we would have 0:19:03.560 --> 0:19:07.720 to reduce the sample in the CPS, that's the household 0:19:07.800 --> 0:19:11.359 survey by five thousand households in order to just publish 0:19:11.800 --> 0:19:16.199 the rest of the sample. Okay, by cutting back on 0:19:16.280 --> 0:19:22.080 the sample, you cut back on publishing details in the 0:19:22.160 --> 0:19:26.600 current report. You may not have enough to publish on 0:19:26.680 --> 0:19:31.919 teenagers anymore. On certain demographic groups, you may not be 0:19:32.000 --> 0:19:36.360 able to publish details on certain age intervals, like everyone 0:19:36.480 --> 0:19:39.840 between the ages of fifty five and sixty or sixty 0:19:39.880 --> 0:19:44.120 five and seventy five. Every time you reduce the sample 0:19:44.560 --> 0:19:47.800 or your response rate falls and it kind of reduces 0:19:47.840 --> 0:19:52.360 it for you. We can talk about that next. What 0:19:52.440 --> 0:19:55.600 suffers are the details. Let's go to the CPI, which 0:19:55.640 --> 0:20:00.000 BLS also does. The CPI is a very big survey, 0:20:00.080 --> 0:20:03.240 important survey, but it is oftentimes the case that we 0:20:03.359 --> 0:20:07.280 don't have details for certain parts of the basket of goods, 0:20:08.080 --> 0:20:12.200 and so we will average across some months and published 0:20:12.240 --> 0:20:15.760 for another month based on averages of prior months. But 0:20:15.800 --> 0:20:18.960 then we run out of the integrity of that average 0:20:19.320 --> 0:20:23.440 and we can't publish the bread price or something like that. 0:20:23.440 --> 0:20:28.199 That happens more and more because we're getting less and 0:20:28.280 --> 0:20:33.600 less cooperation by retailers to allow our surveyors into their 0:20:33.640 --> 0:20:36.800 stores and to go around grabbing the potato chip bag 0:20:36.840 --> 0:20:39.200 and counting the number of potatoes in the back. So 0:20:40.520 --> 0:20:44.400 we've had to innovate on the CPI, on the producer 0:20:44.480 --> 0:20:48.920 price Index, on the import export price indexes. I hope 0:20:48.960 --> 0:20:51.679 you haven't noticed the innovations because you're not supposed to, 0:20:52.320 --> 0:20:57.159 but those innovations have saved, absolutely saved the import and 0:20:57.200 --> 0:21:00.679 export price. We had response rates down in the twenty centiles, 0:21:01.200 --> 0:21:05.399 completely unpublishable detail. During the time I was commissioner. We 0:21:05.440 --> 0:21:09.439 went to using data from the Commerce Department that is 0:21:09.480 --> 0:21:13.119 collected by the customs officials and it's working out just great, 0:21:13.560 --> 0:21:17.760 so we completely left the surveys there. The PPI has 0:21:17.840 --> 0:21:21.080 problems just like the British are having. Now not as bad, 0:21:21.600 --> 0:21:24.960 and we're using more and more data from private companies 0:21:25.720 --> 0:21:29.080 who will give us their data from the Internet, where 0:21:29.119 --> 0:21:31.760 we collect that data and we combine that with the 0:21:31.800 --> 0:21:36.120 survey data to keep the things alive on the CPI. 0:21:36.520 --> 0:21:40.200 It's very interesting. We get all of our retail gas 0:21:40.240 --> 0:21:44.680 prices now from a private company that aggregates gas prices 0:21:44.760 --> 0:21:49.000 for the retail gasoline industry, and it's a very good source. 0:21:49.160 --> 0:21:51.600 A lot of our housing prices now you mentioned housing, 0:21:52.000 --> 0:21:56.080 We're going to more and more to aggregators, private aggregators. 0:21:56.119 --> 0:21:59.400 So there are strategies for saving the survey by blending 0:21:59.440 --> 0:22:02.840 in private data with the survey data. That is not 0:22:03.000 --> 0:22:06.440 the case with the household survey of the labor force. 0:22:07.359 --> 0:22:11.680 Nobody really collects that data except BLS. And that's why 0:22:11.760 --> 0:22:16.280 these demographic surveys, Census has a bunch of them. Health 0:22:16.640 --> 0:22:20.680 the health department, the Department of Health has a lot 0:22:21.240 --> 0:22:22.639 very in very serious trouble. 0:22:23.680 --> 0:22:26.760 So I take the point about innovations. But one of 0:22:26.840 --> 0:22:29.560 the things that's been happening recently is we have been 0:22:29.600 --> 0:22:33.720 seeing bigger and bigger revisions in a lot of the data. 0:22:33.880 --> 0:22:37.520 Is that because of the response issues, So more late 0:22:37.760 --> 0:22:41.159 responses mean that you know you're going to get revisions 0:22:41.240 --> 0:22:45.040 later on as those responses are incorporated. And the reason 0:22:45.080 --> 0:22:48.960 I ask is I remember speaking to someone at the BLS, 0:22:49.040 --> 0:22:52.000 and here shout out to the people working at the BLS, 0:22:52.040 --> 0:22:55.040 because you can actually just call them and ask questions. 0:22:55.040 --> 0:22:59.520 They're really really responsive, amazingly, like the only government department 0:22:59.520 --> 0:23:01.439 that I really know where they'll just pick up the 0:23:01.440 --> 0:23:04.760 phone and talk to you about methodology. Anyway, I was 0:23:04.800 --> 0:23:08.880 asking about revisions to PPI for mayonnaise, of all things, 0:23:09.280 --> 0:23:11.879 and the data there had been revised from like five 0:23:11.960 --> 0:23:16.080 percent to ten percent, which is a pretty big change 0:23:16.320 --> 0:23:18.840 in the space of a couple months. And he was 0:23:18.880 --> 0:23:21.680 talking about how the PPI is subject to revision up 0:23:21.680 --> 0:23:25.400 to four months after it's published, and it gets updated 0:23:25.400 --> 0:23:28.520 as new replies come in. So I imagine if response 0:23:28.600 --> 0:23:31.879 rates are going lower, then maybe that's contributing to some 0:23:32.000 --> 0:23:33.320 of the revision issue. 0:23:33.560 --> 0:23:36.800 By and large, our issues have to do with response rate. 0:23:37.480 --> 0:23:40.480 You're absolutely right, and I'm before I go further, I'm 0:23:40.560 --> 0:23:44.120 very very happy you said that about BLS because BLS 0:23:44.160 --> 0:23:47.760 prize itself and being responsive and transparent by the way, 0:23:47.880 --> 0:23:50.399 so that's a good thing. So I think there are 0:23:50.400 --> 0:23:54.439 two sources of this problem. The first is really hidden 0:23:54.480 --> 0:23:56.240 and no one talks about it. So let me just 0:23:56.320 --> 0:24:01.439 mention when we refresh survey, and you have to do 0:24:01.480 --> 0:24:04.600 this because people are asked to give their survey responses 0:24:04.640 --> 0:24:06.760 only for a few months, right, and then they drop 0:24:06.800 --> 0:24:10.639 out and you have to find new people. The regional offices, 0:24:10.680 --> 0:24:12.600 there are six of them in the United States for 0:24:12.640 --> 0:24:15.639 the BLS, they have people who go out and do 0:24:15.760 --> 0:24:20.400 what's called initiate a new survey respondent. It's getting more 0:24:20.440 --> 0:24:24.160