Links – 05/21/2017

  • The Nation published a piece about the history of Modern Monetary Theory. I’m very skeptical about MMT, but I’ve been meaning to read some of the literature about it for some time now. If you know a good introduction, please share the reference in the comments.
  • The Atlantic published a very interesting profile of Richard Spencer by a former classmate. Unlike most other articles about Spencer, the author takes his ideas seriously, but doesn’t exaggerate the threat he poses.
  • The Census Bureau published its estimates of voter turnout in the last presidential election and it turns out that uneducated white people were about as likely to vote as in 2012. I’m not really surprised by that, it just confirms that Trump only won because a lot of people who had voted for Obama in 2012 voted for him, which was already clear immediately after the election. The data also confirms that the wave of Hispanic voters that is going to destroy the Republican party, which pundits have been announcing before every presidential election for a few decades, also didn’t happen. I can already tell you that it also won’t happen in 2020, but it won’t prevent people from announcing it again.
  • A recent study argues that automation will not warp the labor market as radically as you might think. I have to say that I’m totally unconvinced, but I haven’t read the study yet and the article contains a lot of links to other pieces that seem interesting, so I wanted to flag it.
  • Politico has a good piece on recent psychological research showing that liberals are just as intolerant as conservatives. This is one of many cases where social scientists discover what anyone who is not living in a bubble already knew. In fact, if you read the literature on that issue, you will see that, for many years, it was relying on poor methodological decisions that basically guaranteed that social psychologists would find the results they wanted.

11 thoughts

  1. Automation is going to kill over the road trucking within 10 years. Along with that will go tons of jobs associated with roadside services. Retail has five years to live, tops. Crappy restaurants that basically feed people fried onions while they shop are dead too. That’s all just the opening round that is literally around the corner. Those people are smoking crack.

    1. Yeah, I’m open to be convinced otherwise, but I doubt it’s ever going to happen. My view is that, by the middle of the century, most jobs will be automatized except the ones that require a very high level of skills. Since I don’t think it will be possible to raise the level of skills of most people to the point where they are employable, I believe it will bring about a social revolution compared to which even the industrial revolution will look like a picnic. So we better start to think about how we can deal with this right now. Eventually, even jobs that require a very high level of skills will be automatized once artificial intelligence becomes sophisticated enough to replace humans, which I have no doubt will happen. In fact, I suspect it will happen during my lifetime, but I have no idea what’s going to happen after that.

      1. Yeah, people are putting their heads in the sand if they do not consider the relentless advance of A.I. a substantive threat.

        A.I. is quickly becoming less and less of a ‘tool’ like the tech advances earlier generations ushered in. Moreover, the distributive nature of the cloud, together with deep learning has made it easy and (for the first time ever) feasible to advance the technology to new levels without any substantive iteration. You just feed the learning algorithms vast amounts of data. When you couple that with predictable advances in hardware, the development of newer and better A.I. develops at a pace far grewater than expected for reaching near-parity with most human intelligence. In addition, even before near-parity, it can be placed within a firm and learn how to automate all the white-collar jobs within an ever increasing number of divisions it can automate.

        I read the report, and they fail to mention that in the mid-70’s productivity growth (for production and nonsupervisory workers) broke away from wage growth (for that same class of workers), and has exponentially outpaced it ever since. So, somehow job churn is stagnating, wages for blue-collar and non-senior level white-collar jobs are stagnating, blue-collar and non-senior level white-collar productivity is at an all-time high, but declining per capita productivity shows A.I. is not a symptom or potential future cause of job erosion. That is a remarkable level of blindness.

        If you lineup those graphs – both the ones they present and the ones they ignore – you get the sense that most people will be screwed unless they are an owner or investor. That is, if we don’t start making substantive plans now, for how we will deal with this new era of tech advancement.

  2. The point you make about Obama voters voting for trump isn’t supported by the article you link — it attributes Trump’s victory more to minority groups (predominantly democratic voters) not voting in 2016 when they did vote in 2012. I don’t think the data in the WaPo article gives us much of a bearing on people voting Obama 2012 and Trump 2016.

    A better source for the question you’re interested in is the two “owl of Minerva” blog posts found here:
    http://justthesocialfacts.blogspot.com/2017/05/the-owl-of-minerva.html
    http://justthesocialfacts.blogspot.com/2017/05/the-owl-of-minerva-part-2.html

    As always, thanks for both your blog and your recommendations 🙂 I read “The Triumph of Politics” based on something you said a month or so ago and I thoroughly enjoyed it.

    1. It does support that claim insofar as it shows that uneducated white voters turned out to vote at basically the same in 2016 as in 2012. Since Trump did significantly better with among these voters than Romney in 2012, and this was essential to his victory, it means that many of them who had voted for Obama in 2012 voted for him in 2016. Of course, it’s possible that they are not the same people who voted for Obama in 2012 (i. e. that many uneducated white voters who voted for Obama in 2012 didn’t vote at all in 2016, whereas other people in the same group who hadn’t voted in 2012 turned out in 2016 and voted for Trump), but it’s overwhelmingly unlikely given that voting behavior is highly correlated across time at the individual level (i. e. if you voted in 2016, it’s very likely that you had also voted in 2012). I’m pretty sure the Census data could confirm that, since they must have asked people how/whether they had voted in 2012, but I don’t have time to check. The drop in black turnout was also a factor, but it’s not sufficient to explain Trump’s victory in many states, not by a long shot. For instance, in Ohio, there was a 10-point swing in favor of the Republicans between 2012 and 2016. There is really no doubt that Trump wouldn’t have won if many uneducated voters who had voted for Obama in 2012, particular in the Midwest, had not voted for Trump in 2016. This is why Democrats are really fooling themselves if they think they can win in 2020 without swinging these people back. The drop in black turnout contributed to it, but it doesn’t even come close to explaining all of it, and I doubt it would have been sufficient for Trump to win if uneducated white voters had not swung in such huge numbers. (Again, it would be easy to check that, but I don’t have time.) This is what I meant, but you’re right that the author of that article doesn’t explain that, it was just a personal observation from the data he presented. Thanks for the blog posts, I will try to read them later. And I’m glad you enjoyed The Triumph of Politics, it’s a very interesting book, which destroys a lot of myths that are widespread among both Republicans and Democrats! By the way, if you’re interested in understanding the results of the presidential election, I strongly recommend Sean Trende and David Byler’s analysis, which I linked to in a previous batch of links.

      1. Trump won Michigan with a margin of 10,000, Wisconsin with a margin of 30,000, and Pennsylvania with a margin of 70,000. So yeah, he wouldn’t have won without Obama 2012 voters. There’s a good chance he also wouldn’t have won if minority turnout had been closer to 2012 levels. Whether or not those Obama 2012 voters make a bigger difference than a lower turnout for democrats is one big question Democrats are concerned about, and it’s uncertainty means that “Democrats are really fooling themselves if they think they can win in 2020 without swinging these people back” is way too cavalier. To know the answer I’d want a justifiable estimate of the number of votes ‘gained’ from each source.

        An NYT article from right after the election (I can email it if it’s paywalled; linked below) claims “For every one voter nationwide who reported having voted for Obama in 2012 and Trump in 2016, at least five people voted for Trump after not having voted four years ago.”

        It’s not that easy to get at the questions you don’t have time for because (a) the Census survey doesn’t in fact ask about 2012 votes and (b) even if they did, the results would be heavily biased because voters mis-remember voting for Obama instead of Romney. I’ll do a little digging and follow up 🙂 I just signed up to get data from American National Election Studies, but I don’t have complete faith in my ability to get good evidence from data. will post again after I’ve tried though.

        NYT article: https://www.nytimes.com/2016/11/20/opinion/sunday/the-democrats-real-turnout-problem.html?_r=0

        1. This NYT article doesn’t provide the details of the poll used to draw the conclusion you quote, i. e. that for every one voter nationwide who reported having voted for Obama in 2012 and Trump in 2016, at least five people voted for Trump after not having voted four years ago. But I don’t even need to see the details to know that it’s false, because there is no way it could be true or, if you prefer, it’s extremely unlikely to be true.

          In order to assess how plausible that conclusion is, what matters is not the margin by which Trump won the swing states, but rather by how much the vote swung in these states. Now, even though Trump’s margin of victory was often slim, the swing was huge. For instance, even in Michigan, where Trump only won by a margin of 0.23%, there was almost a 10-point swing in favor of the Republicans compared to 2012. Almost none of that swing can be explained by the difference in black turnout, which apparently isn’t even statistically significant.

          For a swing of that magnitude to have taken place, it just has to be the case that a huge number of people who had voted for Obama in 2012 voted for Trump in 2016. Indeed, were this not the case, it would mean that a lot of people voted in 2016 but not in 2012, because there would be no other way to explain such a huge swing. However, as I already noted, this is extremely unlikely, because past voting behavior is a very strong predictor of future behavior, i. e. the people who vote from one election to the next are generally the same. (This is not specific to the US, but is observed almost everywhere.) Of course, they are not exactly the same and, in a close election, turning out people who didn’t vote the last time can make the difference, and perhaps it did last November in some states, but this can only explain a relatively small part of the kind of swing we have seen in most key states.

          The data about differential turnout at the county-level depending on Trump’s vote share given in the NYT article you quoted doesn’t even come close to explaining the kind of swing we saw in state-wide results. In fact, my argument would only be strengthened by disaggregating and looking at county-level results instead of state-wide results, because in a lot of counties, including many that traditionally vote Democrat, the swing between 2012 and 2016 was even more ridiculous.

          Now, even if what I just said is true, it could still be that the Democrats could win in 2020 without swinging back the uneducated white voters who, despite voting for Obama in 2012, voted for Trump in 2016. But I don’t believe for a second they could. Given how small Trump’s margin of victory was in Wisconsin and Michigan, you’re probably right that, in those states, the drop in black turnout could have been enough to make Clinton lose. Indeed, in the case of Michigan, I have no doubt that it was enough. (I’m more skeptical about Wisconsin, where black turnout apparently didn’t really change, and more importantly the margin was larger.) But as long as he won Pennsylvania, Trump didn’t need Wisconsin and Michigan.

          However, not only do I not think that black turnout alone can explain Clinton’s defeat in Pennsylvania, but even if it could, it wouldn’t mean that the Democrats could easily do something about it. First, in the case of Pennsylvania, I did a back-of-the-envelope calculation and found that, even if black turnout had decreased by 5 points, which it almost certainly didn’t (the difference between 2012 and 2016 apparently isn’t even statistically significant), Clinton would still have come up short by a few thousands votes. (If in addition her margin among black voters had been as large as Obama’s, on the other hand, it would probably have been enough.)

          But even if Clinton would have won, had black turnout remained the same as in 2012, it doesn’t mean that Democrats can hope to win that way in 2020. Indeed, I don’t think they can, because I don’t think they can make black people turn out at the same rate in 2020 as they did when Obama was running. Even if they nominate a black candidate, I don’t think black turnout will go back to what it was in 2008 and 2012, because now there has been a black president already and there just won’t be the same connection between that hypothetical candidate and the black population.

          I guess I could be wrong about that, but counting on increasing turnout among a specific group strikes me as a very bad strategy, because it’s always very difficult. This is not just true in the US, it’s true everywhere. It’s almost always much easier to win an election by flipping voters who voted for the other side than to do it by turning out people who didn’t vote last time. I’m not saying you can never get any votes that way, but it’s a lot more difficult, because as I have noted above the people who vote from one election to the next are always roughly the same. This means that you have a much larger reservoir of potential voters among voters who voted for the other side last time but, because they are not ideologically very committed to either side, could be convinced to vote for you this time. And this tells me that, if the Democrats don’t flip a lot of the uneducated white people who voted for Obama in 2012 but for Trump in 2016, they are going to have another bad surprise in 2020.

          EDIT: I guess the claim I really want to make is not so much that the Democrats couldn’t win in 2020 without swinging back enough of the uneducated white people who voted for Obama in 2012 and for Trump in 2016, but the weaker claim that it’s a much better strategy to try to win by flipping enough of these voters rather than by increasing turnout among minorities. I think the stronger claim is probably true, but if you don’t think so I’m happy to concede this point to you, since I think the weaker claim is what is really important.

          1. Still futzing with R, but I appreciate the comment. I was initially just contesting the stronger point, yes, and I’m not sure about the weaker point (meaning I have no clue, not that I’m skeptical). Really I’m no longer contesting the stronger point though. I did a little reading on the side and since I last looked into this lots of people have concluded essentially your point — that even if better motivating the base might(??) have won the 2016 election, Trump-Obama voters are a much larger effect and their votes are more essential to the future of the democratic party (see what I read at the bottom, for the curious). Thanks for taking the time 🙂

            http://www.mcclatchydc.com/news/politics-government/article147475484.html
            http://nymag.com/daily/intelligencer/2017/05/it-was-obama-trump-voters-in-the-midwest-with-econ-anxiety.html

          2. Thanks for the links, I didn’t know about what they are talking about . It’s a good news for the Democrats if their strategists now understand that Clinton didn’t lose because of turnout, but they are going to have to convince the rest of the party and it won’t be easy given how much currency that myth has among Democrats. This is really hard to understand because it was already clear on the day after the election that it was false. Again, the problem is that Democrats keep focusing on the margins, when they should pay more attention to the magnitude of the swing in key states. I think one reason they do so is because focusing on the margins and Clinton’s popular vote victory is a way of reassuring themselves that Trump’s election was just a fluke, but this is really stupid, because it’s a recipe for another disaster in 2020. Still, if their strategists have understood that, they are still less confused than the GOP establishment after Romney’s defeat. The infamous “autopsy” of the defeat by the GOP was quite possibly the most incompetent political analysis in the history of politics…

Comments are closed.