Tuesday, April 17, 2018

Test Post 2 (Optional)

Something isn’t right with my software. Need to make 2 test posts.

Tax Complexity Interview

For the day income taxes are due, The Washington Post ran an interview with two tax scholars about why the U.S. personal income tax system is so complex. Check it out: you may be surprised at their two main answers.

Read the whole thing, entitled “Why the U.S. tax system is so complicated — but Americans are proud to pay taxes anyway” in the April 12th edition of John Sides Monkey Cage column.

Infographic of Logical Fallacies

Last year, a few students commented towards the end of the semester that leaning about logical fallacies had helped them a lot as economists. I agree: these things are all over public discussion of macroeconomic issues.

Logical fallacies are everywhere, and it’s useful to list out the possibilities and consider ones you haven’t thought of before.

You need to click through to get the whole thing (with interactivity), but here’s a screen capture:

Capture of Logical Fallacy Infographic 2

In the original, these are more readable, and sortable.

Barriers to Entry and Income Inequality

There’s new research showing across the cross-section of countries, that higher barriers to starting new businesses is associated with higher levels of income inequality.

A problem with this sort of research is that it’s not clear if barriers cause inequality, or if inequality causes barriers. Causality is hard to establish in non-experimental settings.

However, economists are better than most social sciences at purging their results of the feedback between two possible causes, and this research does a good job of that.

The data on barriers comes from the World Bank, which publishes a cross-sectional data set on the number of consecutive steps an entrepreneur must complete prior to opening for business. These range from as few as one out into the twenties, and might include things like educational and training requirements, land or equipment ownership, licensing, testing, or health and environmental conditions.

The results show that between otherwise identical and typical countries, that the one with one more step involved in starting a business has a higher income share for the rich (roughly 31% vs. 29%), and a larger Gini coefficient (a limited and basic but very common measure of overall income inequality).

Income inequality is, of course, a macroeconomic concern. However, the microeconomic mechanism at work here is well known: more barriers means less entry in response to positive profit signals, and less competitors leads to bigger markups and profits.

The real world is no doubt a mix of these, but these results are consistent with these polar stories: 1) well-meaning governments erect barriers to protect consumers and those make the rich richer (presumably because they have the resources to overcome the barriers), or 2) the rich influence government to erect barriers to keep competitors out and this helps them get richer.

The research by Chambers, McLaughlin and Stanley appeared in Public Choice, and is entitled “Barriers to Prosperity: The Harmful Impact of Entry Regulations on Income Inequality”. The article is not available for download, but it can be viewed online for free.


Part of the problem of understanding the interactions of macroeconomics and policy is … the policy choices that get linked together under the banner of parties are often nonsensical.

Consider the U.S.: Democrats tend to be more interested in erecting barriers to business formation, but are also more worried about income inequality, even though this research indicates the two go together. Republicans grandstand with an opposing pair that’s just as conflicted.

Macroeconomists are too dull to come with crazy stuff like this on their own ;-D

Friday, April 13, 2018

Zuckerberg, Facebook, and Congress

Magicians will tell you that magic works because they get you to focus on the wrong thing.

And there’s a branch of the public choice subfield of economics named after Peltzman’s research on regulation.

The big news this week is Mark Zuckerberg, the founder and CEO of Facebook, testifying before Congress for two days about how information users divulged on Facebook might be used by others.

Alex Tabarrok has a piece on Marginal Revolution this week pointing out that these hearings are not what they seem to be. Think about magic. You’re probably focused on the wrong thing, and in this case the Peltzman model can tell you what you should focus on. It’s good enough to just quote in full:

If you want understand the Facebook hearings it’s useful to think not about privacy or  technology but about what politicians want. In the Peltzman model of regulation, politicians use regulation to tradeoff profits (wanted by firms) and lower prices (wanted by constituents) to maximize what politicians want, reelection. The key is that there are diminishing returns to politicians in both profits and lower prices. Consider a competitive industry. A competitive industry doesn’t do much for politicians so they might want to regulate the industry to raise prices and increase firm profits. The now-profitable firms will reward the hand that feeds them with campaign funds and by diverting some of the industry’s profits to subsidize a politician’s most important constituents. Consumers will be upset by the higher price but if the price isn’t raised too much above competitive levels the net gain to the politician will be positive.

Now consider an unregulated monopoly. A profit-maximized monopolist doesn’t do much for politicians. Politicians will regulate the monopolist to lower prices and to encourage the monopolist to divert some of its profits to subsidize a politician’s most important constituents. Monopolists will be upset by the lower price but if the price isn’t lowered too much below monopoly levels the net gain to the politician will be positive. (Moreover, a monopolist won’t object too much to reducing prices a little since they can do that without a big loss–the top of the profit hill is flat).

With that as background, the Facebook hearings are easily understood. Facebook is a very profitable monopoly that doesn’t benefit politicians very much. Although consumers aren’t upset by high prices (since Facebook is free), they can be made to be upset about loss of privacy or other such scandal. That’s enough to threaten regulation. The regulatory outcome will be that Facebook diverts some of its profits to campaign funds and to subsidize important political constituents.

Who will be subsidized? Be sure to watch the key players as there is plenty to go around and the money has only begun to flow but aside from campaign funds look for rules, especially in the political sphere, that will raise the costs of advertising to challengers relative to incumbents. Incumbents love incumbency advantage. Also watch out for a deal where the government limits profit regulation in return for greater government access to Facebook data including by the NSA, ICE, local and even foreign police. Keep in mind that politicians don’t really want privacy–remember that in 2016 Congress also held hearings on privacy and technology. Only those hearings were about how technology companies kept their user data too private.

Understanding macroeconomic policies can be problematic because elected officials are often not doing what they seem to be doing. In this case, that’s worrying about your privacy.

On the other hand, unelected officials are often more straightforward to understand. In this case, Facebook’s problem is not really with Congress (which wants a deal) but with Margrethe Vestager who wants more sweeping changes.

Or in the case of something like U.S. trade policy, worry less about Trump, and more about his advisors.

Thursday, April 12, 2018

Oversupply of Housing and the Financial Crisis

It’s become a factoid about the financial crisis of 2007-9 that construction companies were building too many homes that eventually were abandoned when owners could no longer make mortgage payments on them.

Maybe not.

Kevin Erdmann, an author and visiting scholar at the Mercatus Center† has written a fascinating paper about this (available here in full, and required). It is written at a level accessible to undergraduates.

What I like most about this is the careful stock to stock, and flow to flow, comparisons; and also median to median comparisons. From these he builds an argument that the structures themselves could not have been the problem.

  • First, his comparison of the (stock of) houses to the (stock of) population shows a steady or declining ratio from 30 years ago. No problem there.
  • Second, he shows that starts of new homes were in line with population growth. No problem there either.
  • His Figure 3 is more problematic: this sort of chart is difficult to read without distorting one’s view of the data. What seems clear is that there was a shift in the 2000’s out of manufactured homes and into traditional ones. I’m not sure what to make of that. Erdmann sees a decline in multi-family housing there, but I don’t.
  • His Figure 4 shows housing expenditures being steady as a share of income (or declining a bit) over the last 35 years. This result is the opposite of popular perception: people claim housing has gotten more expensive, and it hasn’t in proportional terms. No problem there either.
  • But, he does show in Figure 5 that while median rent and median income (a great comparison) are in line across the country. But what we do have is half-a-dozen cities where both incomes and rents are super high: people are getting paid more to afford housing there. Figure 6 shows that all of those cities do poorly on affordability of housing. No problem there either.
  • Those 6 cities are all well-known for tight housing markets with limited new construction. Figure 7 shows how emmigration from those 6 cities matches up well with immigration into nearby cities without limits on construction. In short, it was like spreading contagion: not enough construction in those 6 cities drove up prices there, so people moved nearby and drove up prices there too. And then people stopped moving as much in 2006. We don’t know why that happened.
  • The last two figures focus on Phoenix, and show that while there was huge immigration into Phoenix, it was matched by construction starts. Then both fell, and vacancies for rentals went up.

Here’s the conclusion: there was no oversupply of housing. There was a big drop in regional migration, but it was matched by reductions in new construction. I am sketchier on Erdmann’s other conclusion: that that there was an undersupply of buyers prior to the financial crisis, and this was transmitted into owners who were unable to sell putting their homes up for rent.

† The Mercatus Center is a libertarian think tank associated with George Mason University. I found Erdmann’s work to be largely free of libertarian positions until the last paragraph which gives an Austrian twist that I’m not sure is merited. The rest of it is solid.

FWIW: lots of us were watching the housing market in the 2000’s. I do remember hearing, more than once, that the age of the U.S. housing stock was older than it ever had been, and that this was motivation for new construction. That fact which was so common 15 years ago has seemingly been forgotten for the last ten.

Wednesday, April 11, 2018

Forecasting with Autoregressive Behavior

In class we estimate autoregressive models of ln(real GDP). Case 3 is an AR(1), Case 4 is an I(1), and Case 5 is an ARI(1,1).

The I indicates the special case where the coefficient on a lagged dependent variable (sometimes with a little algebra and the Engle-Granger Representation Theorem) is restricted to equal one. You don’t get a stochastic trend without that restriction. So Case 4 is a restriction on Case 3, and Case 5 could be thought of as an AR(2) with one of those restricted.

TS asked after class the other day if these sort of models are used for serious forecasting. My response was that, after removing a lot of details, all serious economic and financial forecasting models have AR(p) and ARI(p,1) processes at their core.

Using any of these is a craft, and one of the things you learn is that while they can do pretty good forecasts in the short-run, their long-run forecasts are sort of uninteresting. They’re not necessarily wrong, but they may not say much.

The reason for this is those lags. They contain all the information that makes the forecasts work. This works for forecasting one period ahead (t+1) because you have the data from time t to use in your model. But what if you want to forecast t+2? Where do you get the data from t+1 for the lags? One thing you can do is use your forecast of t+1 that you made at t as an input to make your forecast for t+2. This works OK. But the forecast made at t of t+1 is missing whatever shocks do happen at t+1 to make that period interesting, so the forecasts for t+2 tend to be a little plain and less volatile. The further you go into the future, the worse that problem gets.

The end result of this is that forecasts from AR(p) and ARI(p,1) models tend to converge to the average after a few periods.

Now combine that with the idea of the stochastic trend that you get from the I(1) part of those models. This is saying that there is no central tendency for the trend to return to. It’s always there, but it comes off the most recent data point you’re at. You can be above or below it in the short-run, but the best thing we can say is that in the long-run you’ll settle down to that particular stochastic trend. If there’s a shock next period, your stochastic trend will shift, but the new forecast is that you’ll still settle down to that new one.

So, check out these 10-year Treasury Bond rate forecasts:


These are collected from Blue Chip, which surveys the most popular economic forecasters, who are (no doubt) using an ARI(p,1) model for these.†

What each colored curve is showing is the forecast from a particular point in time. They all wobble a bit in the short-run before converging to a flat line in the long-run. This is just an example. When you see this behavior in future publications, you know where it’s coming from.

You can check out the source post entitled “Losing Interest” at the Lawrence Economics Blog, but the whole thing is not required.

† Technically, the rates themselves are probably not an ARI(p,1). But rates are formed as a ratio of coupon payments and total amount borrowed, both of which do follow that sort of process.