Blogs

Jim Manzi has a long blog post today about the Oregon Medicaid study that got so much attention when it was released a couple of weeks ago. Along the way, I think he mischaracterizes my conclusions, but I'm going to skip that for now. Maybe I'll get to it later. Instead, I want to make a very focused point about this paragraph of his:

When interpreting the physical health results of the Oregon Experiment, we either apply a cut-off of 95% significance to identify those effects which will treat as relevant for decision-making, or we do not. If we do apply this cut-off...then we should agree with the authors’ conclusion that the experiment “showed that Medicaid coverage generated no significant improvements in measured physical health outcomes in the first 2 years.” If, on the other hand, we wish to consider non-statistically-significant effects, then we ought to conclude that the net effects were unattractive, mostly because coverage induced smoking, which more than offset the risk-adjusted physical health benefits provided by the incremental utilization of health services.

I agree that we should either use the traditional 95 percent confidence or we shouldn't, and if we do we should use it for all of the results of the Oregon study. The arguments for and against a firm 95 percent cutoff can get a little tricky, but in this case I'm willing to accept the 95 percent cutoff, and I'm willing to use it consistently.

But here's what I very much disagree with. Many of the results of the Oregon study failed to meet the 95 percent standard, and I think it's wrong to describe this as showing that "Medicaid coverage generated no significant improvements in measured physical health outcomes in the first 2 years."

To be clear: it's fine for the authors of the study to describe it that way. They're writing for fellow professionals in an academic journal. But when you're writing for a lay audience, it's seriously misleading. Most lay readers will interpret "significant" in its ordinary English sense, not as a term of art used by statisticians, and therefore conclude that the study positively demonstrated that there were no results large enough to care about.

But that's not what the study showed. A better way of putting it is that the study "drew no conclusions about the impact of Medicaid on measured physical health outcomes in the first 2 years." That's it. No conclusions. If you're going to insist on adhering to the 95 percent standard—which is fine with me—then that's how you need to describe results that don't meet it.

Next up is a discussion of why the study showed no statistically significant results. For now, I'll just refer you back to this post. The short answer is: it was never in the cards. This study was almost foreordained not to find statistically significant results from the day it was conceived.

• Print
• Email
• Tweet

Peter Kirsanow thinks that L'Affaire IRS (501gate? Cincygate? Teagate?) should be thoroughly investigated. I'm on board with that. But Kirsanow wants to go further. Much, much further:

But the investigation shouldn’t be limited to the IRS. Until last week, the IRS was denying that conservatives were being targeted by the agency. Now we know those denials were completely false. What about the Department of Labor, or for that matter, any federal agency with authority to investigate, regulate, or fine individuals and businesses? With few exceptions, the permanent bureaucracy in Washington leans heavily left. If IRS employees could target conservatives, what prevents the same mindset from prevailing in other agencies?

Congress must use its time and resources judiciously. But it would be shortsighted not to take seriously the complaints that citizens — regardless of ideology — have made about other agencies as well. Hey, we conservatives might be paranoid. But it looks like this time someone was, indeed, out to get us.

Good idea. This could be an excellent WPA-style works program, and it's one that Republicans in Congress would be willing to fund generously. I recommend a citizen investigating force of at least 3 million drawn from all walks of life. There's no sense in thinking small here.

• Print
• Email
• Tweet

House Speaker John Boehner, according to Politico, is obsessed with Benghazi. And last week, after ABC News revealed the revised talking points crafted by the Obama administration following the September 11 attack that left four Americans dead, Boehner demanded that the administration release emails related to these talking points. "The truth shouldn't be hidden from the American people behind a White House firewall," Boehner declared. "Four Americans lost their lives in this terrorist attack. Congress will continue to investigate this issue, using all of the resources at our disposal." But thanks, in part, to the Republicans, the truth isn't being hidden. Boehner and his fellow Republicans had access to those emails—and used them for a public report they issued weeks ago that scooped the ABC News story.

In March, Boehner, according to a senior administration official, was invited to a White House-arranged briefing where the emails and other Benghazi-related material could be privately reviewed. Boehner did not attend; he sent staff, who attended with other House Republicans. Asked why Boehner did not participate in this session and why he did not at that time demand the release of the emails, Brendan Buck, his press secretary, says, "This is embarrassing pushback. Do you recall the report we put out in April? The committees were compiling information as part of their investigation and when the report was done, the committees requested the release of the emails." In an April 23 letter, five GOP House committee chairs did ask the White House to turn over to their committees the documents it had allowed the GOPers to review.

[Read more]

• Print
• Email
• Tweet

Robots! That's the topic of my latest piece in the current issue of the magazine. I've blogged on this subject a fair amount, but this is the first time I've tried to put everything together and explain what I really think robotics is likely to mean over the next few decades. Some of you are going to nod right along, some of you are going to think I'm crazy, and any economists in the audience are going to be rolling their eyes at my rather casual use of macroeconomic trend statistics to help make my point. But I'm pretty sure none of you will be bored.

So what is my point? First off, it's the obvious one that I think computer hardware and software are progressing fast enough that we're not very far away from true artificial intelligence. Along the way I break exciting new ground in describing Ray Kurzweil's "back half of the chessboard" analogy, which illustrates how continuous growth can look insignificant for a long time and then suddenly explode. After immense amounts of research, I decided on Lake Michigan as the key to my explanation of the chessboard analogy, but you'll have to click the link to see what this means. It even comes with a nifty little graphic that our art department created to illustrate how Lake Michigan is just like a digital computer.

Why spend so much time on all of this? Because whenever the subject of AI comes up, everyone wants examples but people like me can't come up with any. That's because AI doesn't exist yet. So we haul out Watson and driverless cars and so forth, and it all seems like pretty weak tea. But Lake Michigan explains why it's not. All these examples that seem pretty lame and not really very AI-like are exactly what you'd expect as mileposts along the road to AI. They aren't demonstrations of how far away we are, but exactly the opposite. They're demonstrations of how close we are. When this all finally happens, it's going to happen fast.

That's the first half of the piece. The second half is about what all this means. If AI is coming—and coming quickly—what does it mean for the economy? In the long run, it will be great, an era of both infinite leisure and material progress. But in the medium run I think the consequences will be fairly grim: more and more people will be put out of work, and no, there won't be new jobs that open up for them along the way. This will very decidedly not be a replay of the Industrial Revolution. What's worse, it will all happen so slowly that we're going to spend a long time denying what's unfolding before our eyes, and a whole lot of people are going to suffer because of it.

In fact, I think automation has been affecting our economy in nontraditional ways since about 2000 or so. Only by a tiny amount, though, which means it's impossible to demonstrate its impact conclusively. Still, you can amass evidence, and that's what I do. There may be other explanations for each of the trends I talk about, but when you put them all together I think the simplest collective explanation is that they point in the direction of automation already having a slight effect on employment and capital intensity. Slight but growing. Two or three decades from now, the Warren Buffetts of the world will own all the robots and a big chunk of the world will be permanently unemployed.

Do I prove this? Not by a mile. But in the end, I meant for this piece to be read as provocation more than anything else. So go ahead and be provoked, one way or the other. Click the link.

• Print
• Email
• Tweet