Wednesday, September 12, 2018

Human Resources And "Employer Learning"

I recently read Bryan Caplan's book The Case Against Education. It contains several pages about "firing aversion" (a term Caplan used on his blog but not in the book) and "employer learning." This was all very interesting to me because of a recent series of events at work (which don't directly affect me but of which I have intimate knowledge).

Caplan's challenge in the book is to explain why educational credentials, high school diplomas and college degrees, have such a strong impact on earnings. People don't appear to learn very much from high school or college, and certainly they forget much of what they learned. When tested, they fail to recall most of the material they learned and fail to apply knowledge to new problems. (If you doubt this, seriously, go read the book! There is a thorough literature review backing this surprising claim.) And yet employers will pay a premium for a degree, over and above what they'd pay for an otherwise identical candidate without the degree. Caplan's story is that a degree signals pre-existing ability. Four years of intensive study demonstrate a combination of intelligence, conscientiousness, and agreeableness that can't otherwise be faked.

Okay, but why can't employers just use a "hire, observe, and dump" strategy? Hire someone who shows some promise of becoming a productive employee, and just fire them if they don't work out. After a few months of working for you, the employee will have demonstrated their ability (or lack thereof). The "everyone gets a degree" signalling equilibrium should break down if employers start using this strategy. But it doesn't happen. So this is another mystery that Caplan needs to explain. He does so beautifully.

He describes the literature on "employer learning", which tries to answer the question "How long does it take for employers to get a good idea of their employees' abilities?"

When researchers explicitly gauge the speed of employer learning, the process seems to take years or decades, not months—especially for the sizable majority that doesn’t finish college.  
Labor regulations and lawsuits aside, firms are not run by robots. When humans work side by side, they develop fraternal feelings for one another. As long as their business is not in jeopardy, many employers retain moderately subpar employees indefinitely. And even if the boss is bereft of empathy, most of their employees won’t be. Disgruntled workers are less productive workers.
 ...Give people a chance, observe how they do, fire them if they don’t measure up: a “Hire, Look, Flush” personnel policy sounds both profitable and fair. Yet group identity and pity get in the way. After a firm hires you, you’re part of the team. If you don’t measure up, firing you isn’t like returning a blender to Walmart. Your teammates either have to live with your poor performance, or feel sorry to see you go. ...The more firms fear to fire, the more educational signaling matters. Once employers get hirer’s remorse, they’re stuck in an awkward position. Relying on credentials is a good way to avoid getting stuck in the first place.
In the real world, it can be very hard to fire someone even if the firm has a good reason.

Here is the process as it was described to me by an HR person. If the employee isn't  performing, they generally get placed on a "productivity improvement plan." The manager discusses the employee's poor performance with them and lays out some objective goals to meet. If the employee meets those goals, the slate is swept clean. The employee is thought to have "graduated", the problem resolved. This process is emotionally exhausting for the manager and humiliating for the employee, and I suspect the real purpose of these plans is to make the under-performer feel so uncomfortable that they seek employment elsewhere. It is HR's way of avoiding a wrongful termination lawsuit, which can happen even though the employer is an at will employer and can fire with or without cause. (It's also probably HR's check against unreasonable/incompetent/arbitrary managers, which is a real problem legitimately needing a solution. But I've seen cases where this process gets in the way when there is an unambiguous case for termination.)

The recent example (the one involving people I know) involved someone who does statistical work. He's not an assembly line worker building widgets. His manager can't simply write an improvement plan that says, "Build 20 widgets a day, or your job will be terminated." Measuring performance for some kinds of jobs is subjective. It would be hard to develop a clear metric of success. And even if someone did write down clear goals for an improvement plan, that's the kind of thing that a desperate employee can game. (Caplan says, "With a job at stake, even a slacker will work like a dog." As in, I can temporarily be more productive and conscientious if I really try, but I will revert to my resting state when the pressure is off.)

The HR person's description of a "performance improvement plan" deeply bothered me, particularly the part about the slate being swept clean if the problem employee meets the plan. Suppose a bad employee gets placed on such a plan. The employee is no good, and manager and employee both know it. But they work their butt off to avoid getting canned. The manager sees the effort, feel sorry for the worker, and reconsiders whether their performance is really so bad they need to be fired. The manager rolls her eyes and checks the "Meets Goals" box.  So they "meet the plan." But the relationship between the manager and employee will never quite be the same. They both share a mutual understanding that the manager does not respect the worker. The worker transfers to another manager when another position opens up within the same company. Suppose they're still no good in their new position. They go on an "improvement plan" under their new manager, and the pattern repeats itself. What should be seen as a history of poor performance and a wake of disappointed managers is instead seen as a perpetually "clean slate." The historical record gets erased every time the under-performer "meets plan."

This can be made worse when a weak-willed manager decides to avoid having these tough conversations with their problem employees. It is emotionally exhausting to have those conversations, and some people don't have the stomach for it. In many companies, a manager has to do an annual performance review. This is an official document of the employee's performance for the year, usually filed with Human Resources. Come the end of the year, when a review is due, the manager might realize they haven't communicated any of their concerns with the problem employee. It would be "unfair" to give the person a bad review without any prior warning. A warning might have given the employee a fair chance to do something about the problem. So the manager might just say "Screw it!" and give the employee an average/adequate evaluation. If this continues year after year, perhaps under a series of different managers, HR will not have any official record of which employees are under-performers.

Given all this, Caplan's discussion of "employer learning" makes sense. It does not surprise me at all that an employee can under-perform for years or decades without getting fired. Productivity problems can be hard to measure, difficult to document (especially if an employer has a policy of purging the relevant history!), and emotionally difficult to confront.

A manager recently described to me an issue with a problem employee. I wouldn't have been able to understand the problem without my specific technical background. Trying to explain this problem to an HR person could be extremely difficult. You end up having to say, "Take my word for it: this person is not qualified for his job." A potential solution to this problem was to make the under-performer acquire a certificate or credential from some third party. (Most of the people I work with have various credentials from various accrediting organizations. Career tip: get one!) That would be an unambiguous test. If the employee can't pass the test required for the credential, we have a neutral third-party confirming our suspicion that they're no good. In this light, credential inflation makes a lot of sense! "It's hard to describe...he's just not qualified" can be a tough sell. "He doesn't have the industry-approved certificate to do this work" is a pretty unambiguous test.

Sunday, August 5, 2018

One Cheer for Buy Local?

I’ll start by saying the “buy local” movement is deeply misguided in many ways. The belief that locally grown food is more environmentally friendly is just plain wrong for fairly obvious reasons.

I don’t want to be confused with these people who fail to do basic cost-accounting. The “buy local” crowd simply does not understand basic concepts of comparative advantage and the gains from trade. Thus only one cheer.

Where to start? When I was in high school my friends and I played some role playing games from the White Wolf universe. (Full of vampires, werewolves, and other creatures, mostly in the modern world.) White Wolf had a “support your local comic store” philosophy. As in, if you see this role-playing guidebook at a big book store in an out-of-town trip, go back home and buy it or order it from your local comic store. (Presumably this idea also applied to internet sales.)  I think this is actually a good idea.

Why? Can’t a big book store or an internet vendor carry a bigger selection and more efficiently supply us with role-playing books? Yes. Maybe. But the local comic book store is supplying something else. Maybe the store owner has deep knowledge of the products he sells. Maybe he can answer questions, even hard-to-google questions, about Dungeons and Dragons or comic books. You get access to this knowledge “for free”. Meanwhile the comic store has fixed costs: inventory, keeping the lights on, paying staff, etc. It would be a little bit rude to pump the store owner for his specialized knowledge, which he dispenses for free, then turn around and price-shop from someone who will undercut him. (The internet vendor has lower inventory costs, because they can make-to-order or keep a few copies on hand for a much larger population than what the local store serves. At some point, the local comic store can't compete with these economies of scale.)

In this sense, maybe there is a case to be made for supporting local boutique-style shops over internet vendors or big stores in the nearest big city. The higher price you pay is like paying “club dues”, and you get access to specialized knowledge for paying these dues. You might also get a sense of community. I would often run into people I knew and have nerdy conversations at my local comic shop. If you value this kind of interaction, it’s probably worth supporting your local shop.

Maybe there’s an even better model, where nerds and artisans get paid directly for their expertise, so we do away with this “marked-up retail” model of reimbursing them. Pay directly for the expertise, then shop for the best price anywhere we can find it. Or maybe there are enough online forums and discussion groups that this information is freely available or easily search-able. (Careful with this. Ever run into a question that’s hard to Google? I sure have. I alluded to this problem above.) But it’s worth thinking about whether there’s some wisdom to this “buy local” stuff.

Some Thoughts on Teaching At Work vs. College

Cross-posting from my Facebook page.

I’ve been trying to learn some data science concepts and teach them to my team at work. A few thoughts and observations.
  1.  It’s different from teaching physics to engineering students (my previous experience with teaching). My team at work needs to actually learn this stuff because it affects their ability to do their jobs. The engineering students were eager to get their A (or B or perhaps C) and then forget everything. Best case scenario, they might use 10% of what they learned in their physics class. The class was mostly useless and they knew it. Motivation is important.
  2.  There is a powerful and tempting illusion that your listeners understand you as well as you understand yourself. Resist this illusion. Do you remember how long it took for these concepts to sink in when *you* learned them? Nobody soaks it up in a single lecture. Be realistic about your expectations. (I was ecstatic when someone told me he understood ~30% of my discussion.) It usually takes time and repeated exposures and some effort studying the material outside of class.
  3.  I have a pretty good memory (sometimes), and I have to remind myself that not everyone has this gift. (I remember explaining a chemistry lab to the valedictorian of my class in high school. Something about how the experiment we were doing was a straightforward application of a previous unit, and he’d said something like “Nobody remembers all that crap.”) When I have baffling conversations with colleagues, I remember them. I sometimes understand them *years* later, when it finally clicks, when I finally acquire the necessary context to understand. “Oh, *that’s* what that guy was talking about!” I think most people just forget about stuff they don’t understand and it’s gone for good? Maybe other people have this experience? (If this is a more common experience, that would make me more hopeful.)
  4.   If you’re going to teach something, know your shit. Practice. Review the material multiple times. Know it inside and out. Rehearse explaining it to a skeptical or confused audience. In the reading group I’m leading at work, we’re reading a chapter out of a data science textbook. I’ve read it through three times now, and read some passages 10 or 20 times to get the point. The same principle applied when teaching physics. I knew some grad students who didn’t prepare adequately or at all, and their students probably resented them for it.
  5.  Related to 2) and 4), there is something called the “illusion of explanatory depth”. You feel like you understand things more deeply than you do. You might read something and be nodding your head, but it might not be sinking in as well as you think. Stop yourself, summarize what you just read in plain English, pretend you are explaining it to an audience, etc. Better learn you’re confused now than while standing in front of an audience, whose respect you’re trying to hold.
  6.  Watch your audience. Look at their faces and scan for signs of confusion. You may just need to slow down and reiterate a point. Or maybe you need to ask if anyone has questions. This is a difficult skill, and I’m not really sure I have it. It’s much easier to accommodate a group of five or ten motivated listeners than a room full of 30 students, though.
  7.   Not everyone will “get it.” That’s okay. Be willing to write off some percent of your audience. If a few of them absorb a substantial fraction of the material, that’s a win!
  8.  There is some combination of “here are practical things you need to know to use these methods” and “here are the academic/theoretical underpinnings of what we’re doing.” You can still do quite a lot if you only have the former but lack the latter. But ultimately you should be able to describe what you’re doing and answer tough questions from people who are skeptical of your analysis. You need the theoretical underpinnings to do this. Imagine someone designing a building entirely by using engineering software vs. someone who understands the physics of force diagrams and could validate some of what the software is doing by hand or from first principles. My guess is that the one with a better understanding of the physics is less likely to make a huge conceptual mistake. Just so with modern statistics.
  9. I recently read Bryan Caplan’s The Case Against Education. There are long and depressing sections about how little people actually retain from their educations (high school *and* college). Also, “transfer of learning”, the application of concepts from one field to a related problem in a different field, is virtually impossible to find in the real world. (As in, studies of this topic fail to find it, despite heroic efforts to search for it.) Not to be too cynical here, but try to temper your enthusiasm with a little skepticism.
  10.  Teaching a physical skill is somewhat easier in the following sense. You can witness someone trying and failing, critique their form, and have them try again. It’s harder to observe understanding, because that takes place inside someone’s head. (Getting people to speak up when they’re confused is difficult, as any teacher will tell you.) Also, people doing physical activities are usually more motivated and are explicitly “opting in.” I’m thinking mostly of the various martial arts I’ve done. Particularly jiujitsu, where you often have to use a technique against an unwilling opponent who is trying to stop you.

Wednesday, July 25, 2018

Economists Have Done a Great Job Making the Case Against Tariffs

Famously, economists have very different opinions from the lay public about economic policy. They are largely pro-free trade. Until fairly recently there was broad agreement among economists that the minimum wage causes unemployment. (Which probably means it’s a bad idea, but not necessarily.) They are broadly pro-free-market.

So what’s going on with the public? We have this whole tribe, this entire profession, who spends their time thinking deeply about these topics, collecting data, building formal models, and doing other kinds of research. And they come to agreement on at least a few topics. So why is this consensus so widely at variance with what the general public believes?

I sometimes hear professional economists comment on this belief gap. They say things like, “Economists have done a terrible job of making the case for free trade.” (Or whatever the topic.) I’ll insist otherwise. Economists have done a fantastic job of making the case for free trade. A single microeconomics course cured me of a lot of sloppy thinking on economic topics. It’s not like they just jot down a bunch of equations and throw in some stodgy commentary for text-filler, just to fill up a 400+ page textbook. The formulas are just there to keep your thinking disciplined. The text is there to explain the logic of free trade in plain English. I’ve seen many very elegant examples of this argument (this one about a car-crop in Iowa is probably my favorite). Some of the econ textbooks I've read are downright quotable, converting a logical, mathematical argument into beautiful prose. If someone writes a short, beautiful, easily digestible proof, it’s not their fault that the lesson fails to register on the narrow bandwidth of the public’s short attention span.

Russ Roberts has done some important work trying to communicate economics to the public. His podcast Econtalk is excellent. It covers a wide range of topics in a conversational style, but (almost) always manages to give the economist’s perspective. He’s written some books that try to viscerally communicate the lessons of economics, using fiction (The Invisible Heart, The Choice, The Price of Everything). It’s important work, and this is probably an effective way to reach people. But I want to push back against the notion that formal logic is too stodgy, or that we need to be “touchy-feely” to reach everyone.

Dry, textbook economics cured me of my leftism. I’ve seen the critiques of basic economics (mostly I’m speaking of micro/price theory), by left-wing economists and by non-economists. I think these critiques are mostly lame and totally unconvincing. Even when they land a solid critique (“Humans aren’t always and everywhere rational actors…”) these critiques fail to revive the interventionists policies supported by those critics (“…therefore, what, the government should be able to intervene in all our life decisions?”). What I’m trying to say is: textbook economics is very convincing. I’ve been there. I’ve had this play out inside my own brain. I know what it feels like to be convinced by pure logic, to discard a political orientation that was once a part of my identity because it’s not tenable. When I hear a different kind of argument, the kind that goes “Let’s discard our sense of reason because of this touchy-feely consideration over here…” I always think it’s a little bit creepy. Like someone is trying to override my sense of proportion or scale by fixating me on something emotionally compelling. Emotionally compelling examples might be necessary to "bring the lesson home" so to speak, but the underlying logic has to be there, too. Otherwise the next guy who comes along with a different compelling example will sway me back, and so on in a pointless random walk.

____________________
There's something to be said for convincing people with bad-but-convincing arguments, but it feels totally sleazy. It always leaves me with the dueling reactions: "I didn't want it this way" and "Meh, I'll take it." Also, it tends to back-fire. The person who is convinced by a bad argument might not stay convinced for long. Also, people observing the discussion will come away with the impression that your side is disingenuous, and as Exhibit A they can point to this shining example you've just given them. 

Sunday, July 22, 2018

Does Scott Alexander Read My Blog?

I had written these two posts a few weeks ago. Then Scott Alexander wrote a post last week with a nearly identical theme. Scott's post is longer than mine and covers a lot more ground, but is built on the same theme. It's kind of eerie. It's not like this was something from the news cycle that we both decided to comment on.

I'm sure it's a coincidence, actually, because I have this feeling all the time. I get it with other bloggers I follow (Bryan Caplan, Scott Sumner, David Henderson). Maybe it's an over-active pattern-matching routine in my head. Or maybe I'm actually taking a cue from them. I'm sure Scott Alexander hears from the same kinds of people I do on the "personal responsibility/laziness" question. So actually, maybe it's no coincidence he has something to say in response to these people.

Tuesday, July 17, 2018

Drug Poisonings and Chronic Health Conditions


In an earlier post, I wrote that about 20-25% of drug poisoning deaths involved some kind of chronic health condition. See what I wrote. I think this calls into question whether these were truly drug overdoses. If a death certificate for a supposed opioid poisoning has “sleep apnea” and “obesity” listed on it, that calls into question whether, in a “but for” sense, the death was caused by the drugs. It's more accurate to say that the death was a combination of a drug overdose (drug interaction, in most cases) and a pre-existing health condition.

I’ve done some more thorough analysis, and it looks like I was on the right track with the 20-25% figure. I can’t give any details just now, but I’m glad I stumbled on this path. Will share details and acknowledgements when I can.

I also know quite a lot more about how the death certificate is filled out, how the causes of death get coded in the ICD-10 codes, how an “underlying cause” is selected from the various causes listed on the certificate, how the different parts of the certificate work, etc. When I wrote that post (linked above) a couple of years ago, I had no idea about any of this. I guess I just assumed it was a bunch of contributing causes listed in no particular order. The story is a little more complicated than that. I may have to do a “death certificate explainer” post when the paper I’m working on comes out. I don’t think the stuff I know now invalidates anything I wrote before, but the details are interesting.

Shoulder Flexibility


I’ve been taking private gymnastics lessons, one or two a months since last November. I’d always wanted to do those cool gymnastics tricks (front handsprings, back handsprings, back flips, aerial cartwheels, etc.) After eight lessons, I have a solid front handspring, and I’m close to a back flip and an aerial. My back handspring is not at all ready for prime time, but I can do them on a trampoline. I’m fairly pleased with my progress. In my most recent lesson (last week), we figured out I could do a roundoff after about five minutes of instruction. That was satisfying.

I said this before in a previous post : It’s really important to get a coach with stuff like this. You can sometimes make impressive gains on your own, but there’s no substitute for having an expert telling you what you’re doing wrong. Try this for anything you want to get better at. Are you struggling with a computer language or some other skill at work? Find someone who knows better than you and ask them. Even offer to pay a tutor. Come up with a set of questions or skills you want to learn, then set up a meeting with a co-worker or paid tutor who is willing to teach you. You might find you’re not even asking the right questions.

One thing I never would have thought about is my shoulder flexibility. I had very stiff shoulders, which my gymnastics coach spotted right away. I’ve been stretching them daily since November, and they’re a lot more flexible now. I have very good leg flexibility (almost down to the splits), but never stretched my shoulders. This deficiency was masked from my view, because in my mind “I’m pretty flexible.” I never would have thought to work on this. My front handspring (which I could already do to a reasonable standard before I started stretching) has gotten much better. The push off the ground isn’t at the proper angle unless you can reach your arms far enough above/behind your head.

The moral of the story: get a coach.