The Case for Paranoia
Not a BLS truther, but an anti-anti-truther: We are getting very exercised denouncing those who doubt the BLS report of a startling September increase in employment, aren’t we? I haven’t heard so much self-righteous post-truth stance-taking since the MSM rose as one to declare that Obama hasn’t undermined welfare reform’s work requirements (which he has).
Here’s Chuck Todd:
This is really making me crazy … The Federal Reserve gets questioned now for politics these days. The Supreme Court and John Roberts get – we have got, we have corroded – what we’re doing, we are corroding trust in our federal government in a way. And, one-time responsible people are doing it. And the idea that Donald Trump and Jack Welch, rich people with crazy conspiracy theories, can get traction on this is a bad trend … We have mainstreamed, ‘when did you stop beating your wife.’
You know, people who don’t know much about Washington may think that everyone around here is hyper-politicized, but if you actually go into the bowels of the federal government, there are a lot people who don’t care that much about politics.
They’re numbers geeks. They do their jobs. They go home. They’re not that political. And I guarantee you the people in the BLS are totally committed to the numbers. If somebody tried to introduce politics in their work, there would be mass resignations and there would be a lot of calls to reporters at various institutions saying this is happening. So I guarantee you, I feel very strongly it’s not happening.
We’ve hit that moment in the election when people begin to lose their minds. … Let’s take a deep breath. Jobs reports are about the economy, not about the election.
Why has Jack Welch doubled down on the false, inflammatory, and slanderous tweet that he sent out five minutes after the jobs report came out on Friday? …, . So when he tweets like the grumpy Republican partisan he is, he will immediately see a pretty angry stream of at-replies. Those replies will come from Democrats, of course; but they will also come from people who think that it’s a good idea to have some evidence before accusing the president of a felony that could result in a jail term and/or impeachment; and generally from technocrats on both sides of the aisle who have great respect for the excellent job that the Bureau of Labor Statistics does every month in an enormous and highly complex economy. ….
Does he have any evidence that the Chicago guys might be manipulating data? No. Does he think it’s even possible for the Chicago guys to be manipulating the data? Evidently, he does. What makes him say that? He won’t say. But is it a legitimate question to raise? In Welch’s eyes, absolutely, yes. If you’re Jack Welch, it seems, any time there’s US data which makes the government look good, the question can and probably should be raised: might the data be wrong? Or, might the government be manipulating it?
The paranoid style in American Politics is nothing new: it was famously diagnosed by Richard Hofstadter in 1964. …
And so, with one unretracted tweet, Welch has effectively rendered himself irrelevant in the so-called thought-leadership world he has dominated for so long.
I am not a BLS “truther.” If Gary Burtless of Brookings–who presumably knows the people at BLS–says they “are much more loyal to their professional duties than they are to the person who happens to be holding office at the time,” that carries a lot of weight with me. The explosion of jobs in the bureau’s household survey also jibes with–and would help explain–the mysterious rise in consumer confidence in the last few weeks. (It can’t all just be from Bill Clinton telling us we “will feel it.”) Even if the numbers are off, they are probably off for normal, non-political reasons. The two surveys often diverge, it’s been pointed out. The household sample is relatively small compared with the establishment survey. There are always outliers.
So why do I find myself sympathizing with the BLS-skeptics? Because while I don’t buy the skepticism, I don’t share in the righteous MSM claim that the skepticism is crazy. Maybe if you are a BLS stat-geek like Burtless you just know the numbers aren’t cooked. But most people aren’t BLS stat geeks, including those who assure us they “know … about Washington” like David Brooks and Chuck Todd. And Jack Welch. For the vast universe of non-BLS-geeks, there seem to be plenty of reasons for at least worrying about manipulation of the household survey, even if you still think it’s unlikely. Here’s my list;
1. It’s a big surprise increase. “[E]mployment growth of 873,000 in September, which pushed the unemployment rate down to 7.8 percent from 8.1 percent in spite of a surge of new workers into the labor force.”
2. Many people have suggested that the household survey reflected a surge of self-employment. One of them, economist Lawrence Mishel–before being “angry” and outraged and sad, etc.–notes that the actual, unadjusted growth in self-employment in September was 93,000. How did that translate into, or “drive” a seasonally adjusted increase of 873,000? Obviously the “seasonal adjustment” was a big part of the positive result. When I’ve seen that adjustment described, it seems to have a lot of big moving parts–massive numbers being added or subtracted. BLS even has to estimate the rate of illegal immigration–something the acknowledged experts at Pew have been unable to do without large, embarrassing errors. Burtless concedes “There are a lot of steps between when the raw data are collected and we end up with estimates of the unemployment rate, where either mistakes or purposeful manipulation could affect things.” Well …
3. Don’t tell me, reassuringly, that these BLS people aren’t political appointees but career civil servants. Civil servants have strong job protections, which means they have some extra leverage when it comes to resisting inappropriate political orders–but that doesn’t mean they don’t have political views and agendas of their own. I dealt with a lot of career civil servants when I was reporting in Washington. Most were competent people, and honest people–a few of them the most honest I’ve ever met. But if there was one that wasn’t political–in terms of pushing for what they regarded as good policies–I don’t remember him or her. Which is to say they are normal human beings. And I would guess at least 80% of them were Democrats–certainly in the Department of Labor. If you were a Republican, would you go work in the Department of Labor at some mid-level analyst’s job?
4. Human beings will lie if the stakes are high enough, even normally honest professionals. I know a poverty expert in DC with an impeccable record–but when the fate of welfare reform was in the balance, this person told reporters lies (a new experience for that person, to his or her credit). Nothing in the welfare reform debate was as politicized and conspiratorial as it must have seemed to people “who do not know much about Washington.” It was all much more politicized and conspiratorial, at least if you really knew what was going on.
Most of the world’s bureaucracies, after all, churn out self-serving BS even when the stakes are low. Most American bureaucracies, public and private, churn out self-serving BS. If we’ve created some small pockets of statisticians we can actually trust to tell it straight–at BLS, and maybe the Fed–that’s a rare and hard-won achievement.
I wouldn’t trust a welfare number that came out of Obama’s HHS, for example, any more than I’d believe a Pentagon body count from Vietnam– except maybe the basic welfare caseload data. It would be shocking if that were fudged. But Obama’s election didn’t turn on the monthly welfare caseload number. It arguably did turn on the monthly unemployment number. The stakes could hardly be higher. Labor department bureaucrats could rationally conclude that their agency’s basic mission and budget–and the jobs of their friends at OSHA, not to mention their colleagues who research and set the Davis-Bacon rates–depend on avoiding the election of an anti-union Republican government-cutter. You have to set that against Burtless’ assurances. Crazy, I know. Next I’ll be telling you that the internationally trusted LIBOR index is rigged. **
5. Welch raises a different possibility: That the numbers weren’t skewed by the DC bureaucrats who massaged them, but at the bottom level where they originated.
The unemployment data reported each month are gathered over a one-week period by census workers, by phone in 70% of the cases, and the rest through home visits. In sum, they try to contact 60,000 households, asking a list of questions and recording the responses.
Some questions allow for unambiguous answers, but others less so.
So let me get this straight: Thousands of leftish census workers, acting without central direction, made hundreds of thousands of subjective judgments that erred on the side of painting a positive September employment picture in order to boost Team O? That’s … actually not all that implausible, is it? You’d have to be an exceptionally clueless Census worker not to realize what was at stake in the election’s home stretch, and not to talk about it over lunch with your colleagues. And how many census workers are Dems? 80%? 85%? These are people who travel around New York urging residents to be counted so they can get their “fair share of funding.” 90%? Are they bilingual workers? How are the GOPs doing among Latinos? 95%?
OK, I’m a hopeless paranoid. But one of the worst things about the respectable smackdown of the BLS skeptics is how inept it is at assuaging paranoia. Hostage negotiators will tell you that “calm down” is the worst thing to tell an angry man. And “shut up, how dare you question the professionalism of civil servants” is probably the least effective way to convince a paranoid citizen.
A text book example is the National Journal piece touted by NJ editor Ron Fournier (who was also a particularly self-righteous welfare anti-truther). “Jobs data not rigged. Period,” tweeted Fournier, who as proof linked to an article breathlessly describing Statistical Policy Directive Number 3, which sets out rules for
keeping a large volume of data processed by a large number of employees top-secret ahead of a pre-scheduled release time.
“Doors to offices where embargoed data is reviewed are locked,” for example. The BLS Commissioner himself isn’t told the household data until “Monday of the week it was made public”! All very impressive, except the charge isn’t that the BLS data are not secret, that they are leaked to enable insider trading. The charge is that, while being rigorously kept secret, they are fudged. Does the National Journal think BLS Truthers don’t understand this distinction, that they’ll be snowed by all the locked-briefcase secrecy? Not a good way to win them over–especially since some of the measures taken to ensure secrecy–“compartmentalizing” the data sector-by-sector, for example, so only “a handful of employees” who “gather in closed-door meetings to compile the final reports” see the full picture–might actually make it easier for the few who do see the full picture to manipulate it, and harder for those who are compartmentalized to understand the manipulation.***
Again, I’m not saying this happened! I’m saying National Journal’s attempt to convince skeptics that it didn’t happen was wildly ineffective, because it treated them like fools. What might convince skeptics? A completely transparent step-by-step description of the entire complex process by which the data are compiled and “adjusted,” with names of the various officials involved and descriptions of the checks (including public dissemination of data) along the way.**** That might bore Jack Welch to death, but it would be worth it. Nothing less has much chance of working.
Trust me. When it comes to paranoia, I’m a life-tenured professional.
Backfill: Yes, I’m late with this! I was doing my taxes. Shafer’s near-real-time report was annoyingly thorough.
Update: So now this week’s jobless claim numbers seem much better than they really are because the Labor Department omitted a key caveat from the press release? No way that could have been influenced by politics! Elaborate safeguards are in place.
**–Note that there are powerful, rational reasons to prettify jobs numbers quite apart from politics–because if people feel good about the economy that is good for the economy: Consumers will spend more and businesses will invest more. If you were an American patriot without party affiliation or inclination, you might goose the data just to goose the recovery.
***–NJ also notes that Greece and Argentina did fudge their numbers, which seems more like an illustration of point 4, above, than an argument that it couldn’t happen elsewhere (e.g. here).
****– This WaPo piece doesn’t come close to doing the job. Like the National Journal report, it tries to snow readers with descriptions of secrecy measures as opposed to anti-manipulation measures.