On Perverse Incentives and Replication in Science

   No comments
Stephen Hsu has a nice blog post on this topic. He writes about this common pattern:

1. Study reports results which reinforce the dominant, politically correct, narrative.
2. Study is widely cited in other academic work, lionized in the popular press, and used to advance real world agendas.
3. Study fails to replicate, but no one (except a few careful and independent thinkers) notices.

#1 is spot-on for economics. Woe be to she who bucks the dominant narrative. In economics, something else happens. Following the study, there are 20 piggy-back papers which test for the same results on other data. The original authors typically get to referee these papers, so if you're a young researcher looking for a publication, look no further. You've just guaranteed yourself the rarest of gifts -- a friendly referee who will likely go to bat for you. Just make sure your results are similar to theirs. If not, you might want to shelve your project, or else try 100 other specifications until you get something that "works". One trick I learned: You can bury a robustness check which overturns the main results deep in the paper, and your referee who is emotionally invested in the benchmark result for sure won't read that far.

Hsu then writes:


"one should be highly skeptical of results in many areas of social science and even biomedical science (see link below). Serious researchers (i.e., those who actually aspire to participate in Science) in fields with low replication rates should (as a demonstration of collective intelligence!) do everything possible to improve the situation. Replication should be considered an important research activity, and should be taken seriously"

That's exactly right. Most researchers in Economics go their entire careers without criticizing anyone else in their field, except as an anonymous referee, where they tend to let out their pent-up aggression. Journals shy away from publishing comment papers, as I found out first-hand. In fact, much if not a majority of the papers published in top economics journals are probably wrong, and yet the field soldiers on like a drunken sailor. Often, many people "in the know" realize that many big papers have fatal flaws, but have every incentive not to point this out and create enemies, or to waste their time writing up something which journals don't really want to publish (the editor doesn't want to piss a colleague off either). As a result, many of these false results end up getting taught to generations of students. Indeed, I was taught a number of these flawed papers as both an undergraduate and a grad student.

What can be done?

Well, it would be nice to make replication sexy. I'm currently working on a major replication/robustness project of the AER. In the first stage, we are checking whether results are replicable, using the same data sets and empirical specifications. In the second stage, we plan to think up a collection of robustness checks and out-of-sample tests of papers, and then create an online betting market about which papers will be robust. We plan to let the original authors bet on their own work.

Another long-term project is to make a journal ranking system which gives journals points for publishing comment papers. Adjustments could also be made for other journal policies, such as the extent to which a particular journal leeches off the academic community with high library subscription fees, submission fees, and long response times.

The AEA should also come out with a new journal split between writing review articles (which tend to be highly cited), and comment papers (which tend not to be). In that case, they could do both well and good.

As an individual, you can help the situation by writing a comment paper (maybe light up somebody who isn't in your main field, like I did). You can also help by citing comment papers, and by rewarding comment papers when you edit and serve as a referee. As an editor, do you really care more about your journal's citations than truth? You could also engage in playful teasing of your colleagues who haven't written any comment papers as people who aren't doing their part to make economics a science. (You could also note that it's also a form of soft corruption, but I digress...)


How to read and write csv using jackson library

     No comments
The sample code below demonstrates how we can read a csv file into an array of objects. Also it writes an array of objects into csv.

To use don't forget to include in your pom.xml 

<dependency>
 <groupId>com.fasterxml.jackson.dataformat</groupId>
 <artifactId>jackson-dataformat-csv</artifactId>
 <version>2.7.0</version>
</dependency>

import java.io.File;
import java.io.FileReader;
import java.io.Reader;
import java.util.ArrayList;
import java.util.List;

import com.fasterxml.jackson.core.JsonGenerator;
import com.fasterxml.jackson.databind.MappingIterator;
import com.fasterxml.jackson.databind.ObjectReader;
import com.fasterxml.jackson.databind.ObjectWriter;
import com.fasterxml.jackson.dataformat.csv.CsvMapper;
import com.fasterxml.jackson.dataformat.csv.CsvSchema;

/**
 * @author Edward P. Legaspi
 **/
public class CsvTest {

 private final String FILE_NAME = "offerTemplateCategory.csv";

 public static void main(String args[]) {
  try {
   CsvTest app = new CsvTest();
   app.testCsvRead();
   app.testCsvWrite();
  } catch (Exception e) {
   e.printStackTrace();
  }
 }

 private void testCsvRead() throws Exception {
  System.out.println("read csv");

  // load file from resource
  ClassLoader classLoader = getClass().getClassLoader();
  File file = new File(classLoader.getResource(FILE_NAME).getFile());

  // configure the schema we want to read
  CsvSchema schema = CsvSchema.builder().addColumn("parentCategoryCode").addColumn("code").addColumn("name").addColumn("description").build();
  CsvMapper mapper = new CsvMapper();

  // configure the reader on what bean to read and how we want to write
  // that bean
  ObjectReader oReader = mapper.readerFor(OfferTemplateCategory.class).with(schema);

  // read from file
  try (Reader reader = new FileReader(file)) {
   MappingIterator mi = oReader.readValues(reader);
   while (mi.hasNext()) {
    System.out.println(mi.next());
   }
  }
 }

 private void testCsvWrite() throws Exception {
  // initialize our list
  List list = new ArrayList<>();
  list.add(populateOfferCat(1));
  list.add(populateOfferCat(2));
  list.add(populateOfferCat(3));

  // initialize and configure the mapper
  CsvMapper mapper = new CsvMapper();
  // we ignore unknown fields or fields not specified in schema, otherwise
  // writing will fail
  mapper.configure(JsonGenerator.Feature.IGNORE_UNKNOWN, true);

  // initialize the schema
  CsvSchema schema = CsvSchema.builder().addColumn("parentCategoryCode").addColumn("code").addColumn("name").addColumn("description").build();

  // map the bean with our schema for the writer
  ObjectWriter writer = mapper.writerFor(OfferTemplateCategory.class).with(schema);

  File tempFile = new File("c://temp//output.csv");
  // we write the list of objects
  writer.writeValues(tempFile).writeAll(list);
 }

 /**
  * Initialize an OfferTemplateCategory using index as suffix.
  * 
  * @param index
  * @return
  */
 private OfferTemplateCategory populateOfferCat(int index) {
  OfferTemplateCategory o1 = new OfferTemplateCategory();
  o1.setParentCategoryCode("PARENT_" + index);
  o1.setCode("CAT_" + index);
  o1.setName("CAT_NAME_" + index);
  o1.setDescription("CAT_DESCRIPTION_" + index);

  return o1;
 }

}

Just create your OfferTemplateCategory class with the fields in schema (parentCategoryCode, code, name, description). Maybe add another field unknown. Try to remove mapper.configure(JsonGenerator.Feature.IGNORE_UNKNOWN, true) and you'll see the error I've mentioned.

WSJ Article on Trump's Fed Vacancies

   No comments
Given the importance of monetary policy, I've been waiting for signs about who Trump will appoint to the Fed. The media gave Obama's Fed picks very little attention. In Geithner's bio, which was largely about Obama administration economic policy, he didn't even mention Obama's Fed choices. He didn't consider them important. In my view, however, the Fed is crucial in getting the US economy back to normal growth. Will Trump pick hard or soft money types? It's anybody's guess. The WSJ has some details, although they leave out the views on monetary policy of those on the short list.

In any case, here are some key grafs for those that can't get behind the WSJ's paywall:

President Donald Trump will be able to recast the Federal Reserve by filling three or more vacancies on its seven-member board of governors, and is leaning toward candidates with banking and financial world experience rather than academic economists.
       After his campaign criticism of the central bank’s low-interest-rate policies, many observers speculated he would seek more “hawkish” candidates who would favor higher borrowing costs. But his choices may be driven less by these issues and more by their practical experience, judging from his early picks for other top economic policy posts in the administration—drawn from investment banking, private equity and business—and the pool of early contenders for the Fed jobs.
...

      David Nason, an executive at General Electric Co.’s financing arm and former Treasury Department official during the 2008 financial crisis, emerged as a leading candidate ...  Other potential candidates the administration has considered include former BB&T Corp. Chief Executive John Allison, and Rep. French Hill (R., Ark.), a former banker just re-elected to his second House term, these people said. 
...
        Mr. Trump’s views on monetary policy are unclear—he both praised and criticized the Fed’s low-rate policies during the campaign. He also promised to roughly double the pace of annual economic growth to 4%, which would be harder if the central bank boosts interest rates more aggressively to restrain inflation.
        “There’s no reason to think Trump would go for one of these really predictable, hawkish academics who’s schooled in balancing inflation,” said Sarah Binder, a political-science professor at George Washington University who has studied the Fed’s relationship with Congress. “He seems to want people, as best we can tell, with a background in finance or a background in business that brings some sort of real world—from Trump’s perspective—appreciation for the good stuff that low rates can do,” she said.
         The seven-member Fed board has two vacancies and will have a third in April when governor Daniel Tarullo steps down in April.

Yes, Monetary Policy Works in a Liquidity Trap, and Fiscal Stimulus Stimulates

   No comments
Here's a nice short article in VoxEU by Benoit Mojon showing the decoupling of US and Euro interest rates after the Taper Tantrum, and the increase in US interest rates after the US election of Donald Trump raised likelihood of fiscal stimulus. 

Here are the key graphs: 


Interestingly, the expected Trump fiscal stimulus increased interest rates in the US, and in Europe, although to a lesser extent. But, longer term, European rates rose just as much. Interesting stuff.




Ranking of Economists, by # of Twitter Followers

   No comments

What's wrong with the US economy? A theory, in one blog post

   No comments
Contrary to popular belief among my fellow liberal economists, the US economy is far from great. The graph below compares Real US GDP relative to its long-run trend. It shows that US RGDP in 2016 was some 25% below its long-run trend. While one might be able to quibble a bit with the trend, it's clear that US economic growth has slowed considerably since 2000, with the bulk of this coming since 2008.


How about US manufacturing? A friend who recently published a paper on manufacturing in a top 20 journal complained how hard it was to publish, despite being one of his better papers. For some reason, the topic seems to bring out the crazy in a lot of economists. Why? I'm not sure, but we economists like to think we know something about manufacturing, and now in the Age of Trump, many economists have a completely understandable desire to try to show that everything Trump says is wrong. Like shooting hair-spray at a flame, this has the effect of making the usual Dunning-Kruger effect even worse than normal when the topic turns to the decline of US manufacturing. The general feeling among many economists is that the decline in manufacturing employment is all about robots, productivity, and sectoral shift toward services. People who think trade are part of the story must be ignorant of economics. I was once taken in by this view as well. However, as I laid out in my own research (Susan Houseman does here and here), the data is not supportive of this assertion as an explanation for the relative decline in manufacturing output and exports, or the decline in the level of manufacturing employment. Productivity growth in manufacturing since the mid-1990s has been, if anything, lackluster. Part of the confusion is that there was a long-term trend decline in manufacturing as a share of the population (even while the level was flat) which is due to productivity growth and sectoral shift. Yet, after the Asian Financial Crisis in 1997, the US dollar appreciated substantially, while Asian countries began hoarding dollar reserves. Add in the rise of China and China's WTO accession, and the Bush tax cuts and 9/11 spending, and you have the perfect confluence of factors to cause a "Surprisingly Swift" collapse in the level of manufacturing employment.

The graph below, of real manufacturing output, does not support the thesis that the problem with US manufacturing employment is simply fast productivity growth. Why would fast productivity growth and the introduction of robots cause a large trade deficit and a decline in real output relative to trend? Inquiring minds would like to know.

























However, while the collapse in manufacturing output and employment may have been a chief cause of the initial slowdown in overall GDP growth, GDP was only about 5% below trend in the 2000s. What has made things worse has been the reaction of policy to the shock of the rise of China and the dollar appreciation. First, low interest rates spurred by savings out of Asia and the carnage to manufacturing in the US led to a housing bubble. Then, poor bank regulation of the shadow banking sector meant that the collapse of the housing bubble meant a financial crisis. And then the liquidity trap, and difficulty that central banks apparently have managing the economy at the ZLB meant that this temporary shock had a long-term impact. An impact that ultimately results in the Tea Party congress of 2010 to 2020, and the election of Trump. (Here's a slightly longer version of this old bedtime story.)

In the early days of the zero lower bound, it was believed by many that monetary policy couldn't be effective. I also believed that rates couldn't go negative. Others simply believed that monetary policy was much less effective. However, since then we've learned that ST rates can go negative. Perhaps as low as -1.5%. We've also learned that monetary policy announcements still affect stock, bond, and exchange rate markets in ways consistent with monetary policy being effective. An example of this was when the Fed hiked the discount rate in April 2010, causing the dollar to appreciate. And this was when the economy still deeply depressed -- precisely the opposite of what your Econ 1 textbook would recommend in this situation. I once asked Jeremy Stein, who later served on the FOMC, what in the world the Fed was thinking. He tried to tell me that this never happened.

What we've also witnessed during the past few years is that the Fed has repeatedly decided not to alter its monetary policy stance even as it marks down its own forecasts for economic growth and inflation. This can be seen in the graph below. After the first round of QE in 2008, the Fed did almost nothing again until the end of 2010 -- a gap of two years -- even though the economic recovery was much slower than it anticipated. Then, even though growth was slower in 2011 and 2012, they waited another 2 years -- the end of 2012 -- before acting again. What the Fed has come to decide over time is not that they need to do more stimulative monetary policy, but that the economy can't grow like it used to. And given that the Fed can always make the economy grow slower if it wants, this situation has developed not necessarily to America's advantage. Time to dust off Bernanke's "self-induced paralysis" thesis.






A reply to Lawrence White

   No comments
People are generally not accustomed to the idea of a central bank altering the maturity structure of outstanding government debt in a manner that might confer a financial benefit to the government (and by extension, to the citizens it represents). The purpose of my previous post was to make people aware of this possibility, using the recent experience of the Federal Reserve's quantitative easing (QE) program as an example.

The gist of my story went something like this. Suppose that the Fed makes a one-time purchase of a 10-year treasury bond yielding a risk-free 2.5% annual nominal coupon. Suppose that the purchase is financed by "printing" reserves. (Note, this asset purchase and method of finance describes the basic nature of QE). Suppose further that the interest paid on reserves remains below 2.5% for the duration of the bond. Then the Fed makes a profit off the rate of return differential in each of the 10 years it holds the bond on its books. If I understand my critics correctly, none would dispute the financial benefit associated with this Fed intervention as far as the public purse is concerned; at least, as long as it is somehow known ex ante that the short rate will remain below the long rate in the manner assumed.

One objection to this story asks why the Fed is needed to convert a 10-year bond into a lower-cost short-term instrument when the Treasury itself could have just issued a short-term debt instrument to begin with. That's a good question. I'm not sure what factors determine the Treasury's choice over the maturity structure of its debt. The key question to ask is whether the structure chosen is optimal from a social standpoint. The benefit associated with the Fed intervention I described in my post assumes that it is not. What justifies this assumption? No one really challenged me on this. Another objection relates to the optics of the enterprise from a political perspective. Even if the intervention makes sense from an economic perspective, paying interest on reserves held by big banks with the effect of reducing Fed remittances to the Treasury just looks bad (even though it's equivalent to the Treasury paying those big banks for holding treasury debt).

The main objection to my post seems to rest on the claim that I have ignored "duration risk." Let me quote directly from Lawrence White, who was kind enough to take the time to critque my post:
When the Fed borrows short from the banks to lend long to the Treasury, it does not do so costlessly. Duration transformation carries a risk of capital loss, also known as duration risk. Suppose the yield curve shifts up, both short and long interest rates rising together. The Fed will experience a decline in present value of its assets that will swamp the smaller decline in the present value of its liabilities. Such an event is not unknown: in 1979-81 it rendered insolvent about two-thirds of US thrift institutions, who were financing 30-year fixed-rate mortgages with 1- and 2-year deposits. In cash-flow terms, such an event would mean that the Fed would quickly have to start paying higher interest rates to borrow, while its asset portfolio continues to pay low yields and roll over much more slowly. The Fed’s annual net transfer to the Treasury might even go negative. Smaller or negative transfers from the Fed to the Treasury would mean a sudden jump in the present value of the public’s ordinary tax liabilities. Net interest income from playing the yield curve is not a free lunch.
To assess the merit of this critique, let's consider the following thought experiment. First, imagine that there is no central bank (End the Fed wins). Next, imagine that the Treasury wants to borrow money. One option is to issue the 2.5% 10-year bond described above. But another option would be to issue very short-term treasury debt, rolling it over for 10 years. If the Treasury behaved in this manner, it would in effect be replicating the Fed intervention I described above. There would be obvious cost savings to the government and taxpayer, at least, assuming that short rates remained below 2.5%. So what could possibly go wrong?

Well, the risk is that the short-term rate might rise and stay on average over 2.5% for the next 10 years. In that event, which is certainly a risk to consider, the Treasury will have made a loss relative to locking in its debt at the long-term rate of 2.5%. But from the perspective of the taxpayer, it matters not one iota whether this prospective loss is incurred because of the Treasury action or an equivalent action on the part of a Fed QE program.

So, if the point is that financing short-term at low rates entails risk relative to financing long-term at high rates, then I agree. The question, however, is not whether QE exposes the taxpayer to duration risk. Of course it does. The same would be true if the Treasury was to shorten the average duration of its debt on its own. Whether the Fed or Treasury performs this operation is immaterial (from an economic perspective, not necessarily from a political perspective). The fundamental question then is whether government debt is structured optimally to begin with. To the extent that the fiscal authority relies too heavily on long-term debt, then some QE on the part of the Fed could have the public finance benefit I described in my post.

Does the Treasury rely too heavily on long-term debt? I honestly do not know. Those who warn against the Fed shortening duration must implicitly assume that Treasury debt-structure is optimal. Maybe they're right. But as the following figure shows, the spread between long and short rates is usually (though not always) positive.


If this pattern could be relied on to hold in the future, then the question to ask of a QE program is not whether it entails duration risk (it does). Rather, the question is whether the endeavor constitutes a positive net-present-value project.  The data above suggests that it may very well be. This is not the same as claiming there's a "free lunch." The Fed has saved the government over $80 billion a year over the past seven years. That's about $600 billion in cost savings--money that the government would have had to secure through other means to service its debt. It was fluke, the critics will say. An ex ante gamble that pays off ex post is not a justification. Sure. But billions in remittances, year after year, one fluke after another?

Let me briefly address a few other points made by Larry. In the quoted passage above, he warns that the Fed is exposed to suffering a capital loss on its assets should market interest rates rise, that this has happened to private banks in the past with dire consequences, that the result may be lower (possibly negative) remittances to the Treasury.

I think these concerns are overstated. Warning against the prospect of having to reduce remittances is not an economic argument against exploiting a profit opportunity (although, one could make the case on political grounds).


Yes, if the market price of bonds fell, the market value of Fed assets would decline. Should this be a concern? U.S. treasury bonds are nominally risk-free and the Fed normally holds its assets to maturity. Larry's reference to the plight of U.S. thrift institutions is not, in my view, relevant to the issue at hand. Because Fed liabilities are money (essentially equity shares accepted by all as a payment instrument), the Fed cannot go insolvent in the same way a private company can (although politically, the optics would be bad). In any case, the proper way to view the issue is from the perspective of a consolidated Fed/Treasury balance sheet. The issue concerns optimal maturity structure and the impact on the government's consolidated balance sheet as market interest rates change.

It is also Larry's view that the Fed has been carrying "significant" default risk through its holdings of mortgage-backed securities. While I think there are good reasons to let the MBS portion of the Fed's balance sheet run off, I think Larry is once again overstating concerns (even if I grant his point about potential market distortions). The MBS purchased by the Fed are not legacy assets--they constitute senior tranches of securities issued well after the real estate market bottomed out. In short, they're about as safe as securities backed by private assets get. That's not to say that the Fed should be holding such assets. My own view is that the Treasury should have issued bonds against such securities. Doing so would have gone some way to alleviating the worldwide shortage of safe assets.

Larry concludes with a plea for "normalization." Let's go back to a world where bank reserves are scarce. But why? In what world does it make sense to render liquidity scarce? Milton Friedman argued that optimal debt policy entails eliminating all liquidity premia. Almost every economic model I am aware of supports this notion. An appeal to follow historical norms is not an economic argument.
 
Let me sum up. I think there are two types of arguments one can bring to bear on the question of central bank balance sheet policy as a fiscal debt-management device. The first is economic and the second is political.

The political argument is based on the idea that an independent and accountable central bank is a good idea. To protect central bank independence, it would be unwise for a central bank to operate with the expressed purpose of managing the debt--a realm guarded jealously by the fiscal authority. I think this is a good and practical argument against becoming too enthused with large central bank balance sheets. But this is not the argument that Larry is making.

Larry's argument is an economic one. An economic argument against the Fed helping to manage the debt, suggested to me by my colleague Steve Williamson, is that the Fed does not really have the tools to do the job properly. In particular, the Fed cannot issue debt that can circulate widely. The only liability it can issue--reserves--is clearly inferior to short-term treasury debt. In particular, reserves can only be used by banks with reserve accounts, while treasuries can circulate outside the banking system. This is a good point. It basically says "since the Treasury can do the job better in principle, better the Fed not do it at all." But this not the argument Larry is making either.

Larry's argument implicitly assumes (possibly correctly) that the Treasury has structured its debt optimally. If it has, then there is no "free lunch" for the Fed. If this is the case, then I would have to agree. (And it's heartwarming to see Larry thinking so highly of a government agency's ability to run its operations!)

As I reflect back on the original argument I put forth, I think I would now put it in the following way. First, I would not advocate that the Fed explicitly pursue balance sheet policy as a way to help government financing. As I made clear in my original post, the recommendations I was making were predicated on the assumption of the Fed achieving its dual mandate. If in the course of achieving the dual mandate the Fed is saving the Treasury billions of dollars, then what's the rush to reverse operations? The fear of exposing the taxpayer to duration risk may be a reason--if you feel the Treasury already has its debt structured optimally. But the fact of consistently positive and sometimes very large remittances from Fed to Treasury for years (and even decades) might lead one to question that assumption.