Friday, June 19, 2015

All Mathy and No Place to Go

If you're having trouble understanding Paul Romer's elaborations on mathiness, this post from last week probably won't help much either. As Romer admits, he's fighting mathiness with wordiness, and for me the words aren't always making sense.

Here are some of the words which seem to be at the heart of his argument, which is a response to David Andolfatto's post:
The provably false statements that economists like Andolfatto make (and he is certainly not alone) may be more than mere signals. They might be an irreversible commitment to stay at an institution where his club is already in control because they prevents someone from ever being employable at a competitive institution where logic is still valued.
I read that several times, and I think I'm closer to understanding it than after my first shot at it, but I'm still not quite clear on what Romer is trying to say. Let's work through this:

1. "The provably false statements that economists like Andolfatto make..." Andolfatto made some statements. They are "provably false" statements, which would be a bad thing if it were true, especially if David knows they are false. Then David would be a liar of course. Further, there other economists "like Andolfatto" who are doing the same thing. He's apparently a member of an identifiable group of economists who are doing something systematic, perhaps knowingly making false statements. Potentially we have a group of lying scoundrels on our hands.

2. "...may be more than mere signals." So making false statements is a signal of something, according to Romer. If you have been reading Paul's missives, you'll know that a "provably false statement" is a signal of a general state of "mathiness," an affliction - indeed, potentially a conspiracy to deceive - whereby the perpetrators subdue unwitting readers with a torrent of impenetrable math designed to cover up poor economics.

3. "They might be an irreversible commitment to stay at an institution where his club is already in control..." The institution is the place where Andolfatto and his ilk work, apparently. There may be a club controlling the place where he works, and possibly David has made an "irreversible commitment" to stay there - on the scale of a pact with the devil. David went down to the crossroads, and came back playing some serious bottleneck guitar, but at what cost?

4. "...because they prevents someone from ever being employable at a competitive institution where logic is still valued." This is the cost. Make "false statements" and you're stuck working where you're working now, because people who work in places where "logic is still valued" won't hire you.

Well, as far as I can tell, Paul Romer's drive for an eradication of "mathiness," and his support of "truth," and "norms of science," has little to do with valuing logic, and everything to do with Paul's own professional disputes, which are really of little interest to most of us. It's nice that people have an opportunity to discuss the economics of growth and innovation, which is fundamental in economics - though we perhaps have not progressed as far as we might have hoped in understanding it. But Romer's conspiracy theories are an a par with grassy knoll theories. Start probing, and there's nothing there. In particular, there are no falsehoods in Andolfatto's blog post, as far as I can tell. I disagree with David on a daily basis, but he's one of the more honest and un-clubby economists I have known in my professional career. How many other economists attach their names to their referee reports?

So, my advice for Paul Romer is to lighten up. I think he would rather be remembered for his fine early-career research on economic growth than for late-career grumpiness and nastiness.

Monday, June 8, 2015

Economics and Deception

I was reading Noah Smith's "Economic Argments as Stalking Horses," which at the minimum got me interested in finding out what a stalking horse is. Animals understand which other animals will cause them harm, they key into the movements of those other animals, and they'll get out of the way quickly if they understand that something potentially harmful is in the vicinity. Birds, for example, don't feel threatened by horses, but they are smart enough to figure out that humans are bad news. So, if you hide behind a horse, you can get close enough to a bird to shoot it. Indeed, you can train a horse to be the thing you hide behind to hunt birds, and then "stalking horse" becomes its job description. Interesting.

This may not be the best analogy for the idea Noah wants to get across, but it at captures the notion that deception is involved, so we'll go with it. Noah is interested in this quote from Russ Roberts, on Twitter (which I avoid like the plague):
Just a curious coincidence that economists who like stimulus want bigger government and those who oppose it prefer smaller.
One can find counterexamples, of course - John Taylor and Greg Mankiw wouldn't object to being called Keynesians - but I think we could characterize this as a strong empirical regularity. And in principle the existence of this regularity may be completely innocent, with no deception involved. Maybe the world consists of two sets of people. There is a set of non-interventionists convinced by the data that governments are somewhat inept, and should be given a fairly narrow set of tasks to do. The government may be viewed by these people as being just as inept at stabilization policy as at running a health care system, for example. Or non-interventionists may think that the economy works in a self correcting way in that, even if the government is not inept, its ability to intervene is extremely limited. The second set of people - the interventionists - thinks that governments are smart. Indeed, interventionists may think the average Jane/Joe is quite stupid - so stupid that she or he needs the government to help her or him make economic decisions. From the point of view of interventionists, the government should have a lot on its plate, including stabilization policy. The private sector is messed up in long-run and short-run fashions, and needs fixing.

In principle, the non-interventionists and the interventionists could both be honest, and using - to the best of their respective abilities - the best available economic theory and evidence. But, you may wonder, what is the best available economic theory and evidence on the matter at hand? What does economic science say the government should do? There has certainly been plenty of work on optimal taxation, both in static and dynamic contexts, and with and without private information. Also, mechanism design in part grew out of problems in which we want to elicit preferences. For example, we have something - national defense for example - which is clearly a public good. How much of it do we want? Well, if you ask people how much they want and make it clear that you will tax them more the more national defense they say they want, they'll say they don't want much. The result will be under-provision of national defense. If you apply mechanism design to that problem, you can figure out how to get an efficient quantity of public goods provision, while getting everyone to truthfully report how much they want. Perhaps surprisingly (but not surprising if you pay attention), Mike Woodford's work on New Keynesian models was motivated in part by work in public finance. In the foreword to Woodford's "Interest and Prices," he describes the genesis of his ideas:
My advisor, Bob Solow, always insisted on the unity of microeconomics and macroeconomics, and wore both hats with equal flair. He challenged me, while I was still writing my dissertation, to try to integrate sticky prices into the kind of intertemporal general-equilibrium models that were then becoming the dominant paradigm for macroeconomic analysis. Another of my teachers, Peter Diamond, insisted upon the importance of public economics that would allow an integrated treatment of inefficient allocation of resources across sectors and macroeconomic stabilization.

So, in modern economics, the theory of government could be thought of as an integrated whole. If we're thinking about "stabilization" and the size of government, those are just parts of the same problem. You really can't separate "short run" from "long run." How much of that thinking actually finds its way into discussions among people who actually have some influence on fiscal policy decisions? Very little. But monetary policy is another story. I think it's fair to say that central bankers are pretty serious about using available economic theory and evidence.

So perhaps we should focus more narrowly on monetary policy. Think of this as part of government, and that we could have non-interventionist and interventionist views about it. As I stated above, the non-interventionists and the interventionists could in fact be completely up-front about what they are doing - no deception involved. But ... we know how people are. Sometimes people shade the truth; they leave out facts that might be important to the case at hand; they misrepresent the views of people with a different take on things; they may even lie outright. How are you going to figure that out? Well, we are all students of human behavior, and we've encountered deceptive people since we started interacting with other two-year-olds. But let's confine attention here to economists - who of course may sometimes behave like two-year-olds, just like anyone else.

Why might an economist be motivated to be deceptive?

1. Follow the money: What is this person paid to do? If they are telling you something that might keep the gravy train rolling, or elicit another gravy train, maybe you should be suspicious.
2. Follow the money (more subtle version): If you are thinking that academic economists are immune, forget it. A successful paradigm pays off big time. Maybe you can make your paradigm successful by exaggerating the virtues of your own paradigm, or trashing your competitors.
3. Too old, too lazy, or insufficiently talented, to change: The new paradigm may be the better one. But maybe you invested in the old paradigm. So the new paradigm will depreciate your human capital severely. It's lower cost for you to oppose the new paradigm than to invest.

So, (1), (2), and (3) are impediments to science, and we would like to root these things out if we could. It's a policing problem. But, good economic science is a public good, so how do we get people to cooperate and do the policing? Maybe we have to depend on the fact that people are hard-wired to punish bad behavior. We had a long discussion about this in the comments section of this post. In any case, before we even get started with policing, we have to know how to identify the bad behavior. Understanding motivation obviously helps, but there are also some signals we should pay attention to - here, think in terms of a Spence signalling model:

i) A high ratio of destruction to construction: Constructive arguments are costly, and the more information the deceiver provides, the easier it is to catch him/her. So the artful deceiver uses up a lot of words in trashing his/her perceived opponents, and few words in detailed explanation of his or her own ideas.
ii) Repetition: If you want to tell lies, you clearly can't admit what you're up to. No matter how bald-faced the lie, the deceiver will keep repeating it, hoping that repetition will somehow create the ring of truth.
iii) Behaves like a politician: Politicians will try to impugn the character of their opponents. If you can get people to believe your opponent is a bad person, maybe they'll believe your opponent also does bad science.
iv) Uses up a lot of words discussing general bad behavior (of others of course): It takes one to know one.

Now, to help you out, let's do examples. Alan Meltzer wrote a piece for the Wall Street Journal in May 2014 called "How the Fed Fuels Future Inflation." Keep in mind here that a year has passed, but the world doesn't look so different now from a year ago, in regard to what Meltzer is discussing. Basically, the Fed's balance sheet is large, and inflation is low. But, Meltzer says:
Never in history has a country that financed big budget deficits with large amounts of central-bank money avoided inflation. Yet the U.S. has been printing money—and in a reckless fashion—for years.
The quote pretty much encapsulates the basic ideas in the piece. Let's do some fact-checking:

1. Was the U.S. budget deficit "big" in 2014, when Meltzer wrote his piece? At 2.8% of GDP, this was larger than at some times. But in the three previous cyclical troughs (looking only at the federal government deficit), the numbers were 9.8% (2009), 3.4% (2004), and 4.5% (1992). Further, in 2014 we were in a slow recovery phase, with GDP much below its long-term 3% growth trend, which would tend to make tax revenue low. So, I think it's a stretch to call the federal government's budget deficit "big." We know that out-of-control fiscal policy can be associated with big inflations, but it's hard to make the case that the federal government is out of control in this instance.

2. Is the creation of "central bank money" always associated with inflation? Well, apparently not, and we have an important example sitting under our nose of course. It's well known that, as Meltzer says, base money has increased by several-fold since the beginning of the last recession in the United States. Here's the what's happened to the inflation rate - raw PCE deflator and core PCE:
Both of these indicators have been below the Fed's 2% inflation goal for more than 3 years. Even though some of the recent low inflation reflects low oil prices, there is no sign that this is turning around. And the U.S. is not an anomaly. In Japan, the central bank's balance sheet increased by 120% from January 2008 to May 2014, and by 155% from January 2008 to May 2015. In Japan, the year-over-year inflation rate was 1.4% in May 2014 (excluding tax effects and food), and -1.7% in April 2015. In Switzerland, which has the world's largest central bank balance sheet (relative to GDP) in the world, as far as I know, the monetary base increased by 690% from January 2008 to May 2014, and by 870% from January 2008 to April 2015. But, the year-over-year CPI inflation rate in Switzerland was 0.2% in May 2014, and -0.9% in March 2015. So, a more accurate statement would be: "As yet, large-scale asset purchases by central banks, given zero or near-zero nominal interest rates, has failed to produce higher inflation."

Is Meltzer deceiving us? First, we have to ask whether he's motivated to do so. I don't think this is a "follow the money" motivation - either the obvious or subtle version. But Meltzer is definitely wedded to a paradigm, which is the quantity theory of money. He might be able to modify that paradigm (though I have my doubts) to reconcile it with the observations he's confronted with, but clearly he's not trying too hard. My guess is that the motivation is #3. Meltzer is 87, so the fact that he is alive and writing for the WSJ is a feat in itself. If I'm putting sentences together at that age I'll be amazed. Nevertheless, the quantity theory is what he knows, and it appears he's going to stand by it. Learning is done in this case, I think.

Is Meltzer giving off signals of deception? Definitely. He spends more words impugning policymakers than making the case for his worldview. As well, if you Google "Krugman Meltzer," Krugman will show you many cases where Meltzer makes the same charges, so there is repetition. However, though his opponents are named, he doesn't make them out to be bad characters. There's some politics in the piece though. In particular, when discussing the deficit, Meltzer mentions only the executive branch. We all know that the legislative branch plays an important role in fiscal policy.

Another good example is Paul Krugman, which is a lot more fun. Here, we'll look at two blog posts and his most recent column in the NYT. The first blog piece is from two days ago, and is a direct response to Noah's blog post. Krugman is trying to convince us that he's innocent. He argues that he's only looking at the evidence and drawing the obvious conclusions:
First of all, the case for viewing most recessions — and the Great Recession in particular — as failures of aggregate demand is overwhelming.
The idea is that thinking of most recessions in this way will be helpful. "Overwhelming" means this is pretty much universal. We'll all benefit from this insight. A "failure of aggregate demand" means something within a particular paradigm. Basically, "viewing" recessions as failures of said aggregate demand requires that you work within that paradigm, which is AD/AS, IS/LM. That's obvious if you've read Krugman. He thinks that paradigm is the cat's meow. So, within the paradigm, "aggregate demand failure" means the aggregate demand curve shifted left. You'll note first that, even for a well-schooled AD/AS, IS/LM fanatic, this is very non-specific. What happened to shift the AD curve? Was it an autonomous drop in consumption, an autonomous drop in investment, a shift in the money demand function, or what? Shouldn't we care?

Since we're into overwhelming territory, we would have to make the case that at least someone would find the information that the AD curve had shifted to the left useful. Suppose I am Ben Bernanke, and it's October 2008. Paul Krugman walks in the door and tells me the AD curve has shifted left. Does this help me to decide how to get banks to take my discount window loans, which I'm concerned they are not taking up for fear that they might be stigmatized as a result? Does this tell me which large financial institutions I should support, and how, and which ones I should just let go? Does this tell me why there are now large differences among particular market interest rates that did not exist several months previously? You get the idea.

On the other hand, suppose I am an empirical macroeconomist, and I'm trying to understand the Great Recession. I have coffee with Paul Krugman, and he tells me not to worry. This looks pretty much like other recessions - just an insufficiency of demand. As an empirical macroeconomist, I may not know what Paul is talking about, as there now exist practicing macroeconomists who have never seen AD/AS, IS/LM. But, for example, Christiano, Eichenbaum, and Trabandt know AD/AS, IS/LM (at least I'm pretty sure the first two do). They're well-known empirical macroeconomists, and they're not coming out and saying the Great Recession is about aggregate demand deficiency. They have a model, and the model has stochastic shocks which they give names to, and none of those shocks appear to have the name "aggregate demand." They have taken the time to write a whole paper in which they try to figure out how the shocks account for the Great Recession, and what the propagation mechanism is. Conclusion:
We argue that the vast bulk of movements in aggregate real economic activity during the Great Recession were due to financial frictions interacting with the zero lower bound.
Sorry, but that's not in AD/AS, IS/LM.

Now, consider another person. This one is older than the average undergraduate - old enough to actually care about the Great Recession, and not think of this as ancient history. This person is, say, 30, and worked on Wall Street during the financial crisis. She saw a lot of stuff. Volatile financial market activity, unemployed people, large financial institutions in trouble, etc. Now she's motivated to go back to school and take some economics so she can understand all that stuff. She takes an intro-to-macro course, learns AS/AD, and is told that the Great Recession is just like all the other recessions - AD shifts left. Given her experience in the world does this get her excited? Does this information somehow put all her practical experience into perspective? I don't think so.

So, Krugman's post is looking suspicious. He's making strong claims without much to back it up. You can say it's all in his other blog posts, but I've read those too, and I don't buy it. There's not much dissing in this particular post, but Krugman sneaks something in at the end:
The point is that while it’s definitely OK to scrutinize economists’ motives — to ask whether they’re responding to logic and evidence, or just talking their political book — assertions that it’s all politics deserve the same scrutiny. Is my behavior consistent with claims that my views are purely a reflection of my political preference? And if it isn’t — which I don’t think it is — what’s driving such claims? Might it be … politics, deployed on behalf of economic doctrines that have lost the substantive debate?
I forgot to include this above. It's a great strategy to muddy the waters by impugning the motives of those who are trying to police you.

The second exhibit is Krugman's NYT column from today. This is concerned with
...people who keep saying the same thing no matter how much evidence accumulates that it’s completely wrong.
Here, Krugman is in full destructive mode. Who is he fighting?

a) People (Meltzer, goldbugs, etc.) who are worried about too much inflation.
b) Bob Lucas who, he claims, "accused Christina Romer, the administration’s chief economist, of intellectual fraud."
c) People who don't like Keynesian economics.
d) Obamacare naysayers.

We'll leave him alone on (a) and (d), as he's got a case there, though he's been beating it to death. Bob Lucas certainly doesn't deserve this treatment, though. On (b), if you investigate, you will find the supposedly offending discussion in the transcript of a conference from 2009. So this is ancient news by now. Further, if you read carefully, you'll find that Lucas's remarks come in a panel discussion, and that Krugman is quoting him out of context to make him look bad. I say Lucas is not guilty.

On Keynesian economics, we've got another bold claim, that Keynesianism is
...an approach that, among other things, correctly predicted quiescent inflation...
Here, I had to look up "quiescent" to make sure I know what it means. He's saying that Keynesian economics correctly predicted the low inflation that we've had. Of course, basic AS/AD, IS/LM doesn't "predict" inflation. Inflation is not in there, as it's a static model. Presumably what Krugman means is that the Keynesian Phillips curve does a good job of predicting inflation. We should check that out. This is a scatter plot of quarterly year-over-year PCE inflation and the unemployment rate since the end of the last recession. The line connects the observations from 2009Q3 on the far right to 2015Q1 on the far left:
So, the Phillips curve would predict that, when unemployment is falling, in a world in which inflation expectations aren't moving around much (which we could argue is the case here), inflation should be rising. Do you see that in the chart? Neither do I. Indeed, some Phillips curve adherents have been confidently predicting for a long time that the falling "output gap" will lead to inflation moving back to 2%. That hasn't happened, and those people should feel embarrassed in the same way that Alan Meltzer should feel embarrassed. Thus, maybe we shouldn't give Krugman a pass on (a). He's just as wrong as Meltzer, but with more company, which is a deeper problem.

Krugman goes on in the same vein in this blog post, which is an attempt to defend himself against the charge of hypocrisy. Looks to me like he's digging himself in deeper.

So, is Krugman being deceptive? First, motivation:

1. Follow the money: The Krugman enterprise is a very successful one. He is well-paid by the New York Times, he commands a large speaking fee, and he also has a gig with CUNY. Krugman serves a large constituency that is not really interested in science, or in learning the intricacies of economics. Other people who read Krugman have a smattering of economics, and I'm sure they find it flattering, and assuring, to be told by a guy with a Nobel prize that what they know is all they need to know about what is going on in the world. Behaving in the way he does just feeds his success in his chosen occupation.
2. Follow the money (subtle version): Doesn't apply. Krugman is past caring about success in the academic journals.
3. Too old, too lazy, or insufficiently talented, to change: Plenty of talent here, of course, and I'm going to stick up for my cohort of over-60s, and say he's not too old either. There's an element of too-lazy though. At the peak of his academic career, Krugman became accustomed to being at the top of his field. He had a prestigious position at a top school, a well-respected academic record, and influence in the international trade community. I think he's not entirely content earning a good living as a writer/blogger/speaker and political pundit. He seems to want to be influential as a macroeconomist. But he can't do that in the conventional way, as it's too costly. So what else can he do but try to cut the macro profession down to size? I think that's part of the destructive impulse here.

Second, the signals:

i) A high ratio of destruction to construction: If you've been reading Krugman, you've seen a lot of this. Typically he can't write a blog post or NYT column without slamming someone - real or imagined. The last NYT column is probably the worst it gets - it's quite unremitting.
ii) Repetition: Again, using the last NYT column as an example, there's little in here he hasn't said before. For example, you can do a search on the supposed bad behavior of Bob Lucas vis-a-vis Christina Romer, and you'll find that repeated again, and again, and again, over the space of years.
iii) Behaves like a politician: In a discussion of how people stick to ideas contradicted by evidence, he for some reason slips in a claim that Bob Lucas is a bad guy. What for?
iv) It takes one to know one: Here's a selection of Krugman's helpful advice from the stuff I've been referencing:
...the place to start fighting is within yourself...making the same wrong prediction year after year, never acknowledging past errors or considering the possibility that you have the wrong model of how the economy works — well, that’s derp... the peddlers of politically inspired derp are quick to accuse others of the same sin...The first line of defense, I’d argue, is to always be suspicious of people telling you what you want to hear...Fighting the derp can be hard, not least because it can upset friends who want to be reassured in their beliefs. But you should do it anyway: it’s your civic duty.

I don't want to leave you on a negative note. After all, my goal is to try to help you to escape unnecessary negativity. Generally, I think that we haven't been thinking about the inflationary process in the right way. But if we give our models a chance, and pay attention to what the data is telling us, the answers are staring us in the face. In most monetary models I know about, and that includes New Keynesian models, low nominal interest rates in fact cause inflation to be low. That's a monetary theory of inflation. The central bank controls nominal interest rates. If nominal interest rates are high, inflation is high. If nominal interest rates are low, inflation is low. And the data is consistent with that theory. Japan has had low nominal interest rates for 20 years, and that has produced persistently low inflation. Many central banks in the world, including those in the U.K., the U.S., the Euro area, Denmark, Switzerland, Sweden, and Japan, have targeted nominal interest rates at persistently low levels. And inflation in those countries is persistently low.

The problem, as I argue in this paper (with David Andolfatto) is a policy trap. Taylor-rule policymakers, operating under the mistaken impression that low nominal interest rates will ultimately produce higher inflation, are stuck with perpetually low nominal interest rates and low inflation. There are elements of that idea that are not new, but we do some novel things in the paper, I think. Basically, though, I think this is an idea we should run with. I'm happy to run away from it once it doesn't work though. Wouldn't want to be derpy.

Monday, June 1, 2015