Tuesday, October 11, 2011

The 2011 Economics Nobel - Christopher Sims

(Note : This was earlier written as part of the previous post, so it begins abruptly where that one leaves off. It got so long that I split the post into two. I am beginning to think that even very knowledgeable economists may have fundamentally mischaracterized Sims when they see him as someone primarily useful for empirical techniques)

I found Chris Sims much more exciting - digging into his work was a whole new level of education. This dude should really have been blogging! First of all there is the fact that his VAR, SVAR models and techniques require as few structural assumptions about the economy as needed. This I now believe is an important objection to all mainstream/blogosphere macro that tackles policy issues - why the need to presume unidirectional causality arguments? (If logical causality was to become the dominant paradigm in macro, shouldnt we all be Austrians by now?)

Indeed, Scott Sumner's critique of VAR as possibly misidentifying nominal shocks and reversing causality looks exactly wrong at first cut - that's precisely the kind of pitfall that VAR models seem to be trying to avoid. Second, and more importantly, Sims is almost wholly free of policy and worldview biases in the issues that he tackles. This is absolutely remarkable - almost unique - in a macroeconomist who writes most frequently on monetary issues. Lastly, Sims is about so much more than the empirical techniques! Reading even 2 or 3 of his papers is proof enough.

Here are a few gems I found in Sims's research:

1) Limits to inflation targeting - "Inflation targeting may do more harm than good if there is a substantial chance that the central bank cannot in fact control inflation. .........These considerations suggest that in those countries where inflation control has in the past been most difficult, inflation targeting may be least useful......."

3) IS/LM critique (this one actually has several great sections worth reproducing in full)-

"Since the equations he wrote down did not include an explicit role for
expectations, the ISLM codification of Keynesian orthodoxy could plausibly ignore
them, and did so. This not only distorted Keynes’s thinking, it ironically weakened
Keynesian orthodoxy in the face of the rational expectations critique"

"A coherent Keynesian approach, accounting for endogenous expectations, implies very
strong effects of monetaryand fiscal policy and leads to greater attention to the
role of the government budget constraint in making the effects of monetary policy
conditional on prevailing fiscal responses, and vice versa"

"If money illusion, lack of foresight, and decision-making inertia are important
in macroeconomic dynamics, it seems more likely that they are important in labor
markets than in the majority (value-weighted) of investment and savings decisions"

"It is useful to recognize where we are introducing non-neutrality into a model and to bear in mind the limitations of non-neutrality assumptions. But it was one of Keynes’s central
insights that in this respect a little ad hockery is not too high a price to pay for
maintaining a model’s grip on reality"

"That an interest rate increase engineered by the central bank could exacerbate rather than reduce inflation, is commonly recognized in current policy discussions ... " (Is this mechanism at work in India these days?)

7) The biggest risks that the European Monetary Union faces : this one is beyond outstanding. It has so many great ideas that I can't even select properly. It describes most mechanisms at work in the Eurozone today, comprehensively. Oh, and it's written in 1999. Which makes it single best macro prediction I have encountered yet.

Phew, that's not a few, that's a lot. And there are many more. Chris Sims has basically tackled almost every major macro/monetary policy and theory conundrum head on in his career, and has outlined mechanisms that almost completely describe the macroeconomic funk (he does abstract out of finance and default) that developed economies find themselves in these days. He's done this about 10-15 years ago. He handles fiscal & monetary policy symmetrically, recognises the areas of overlap between the two, handles disequilibrium as well as rational expectations with equal ease, creates models that have monetary as well as fiscal non-neutralities, looks at open economy dynamics and explicitly recognises and models various deficit financing as well as budgetary constraint choices. Most importantly, he twists theories to suit facts.

And you're telling me that he is a VAR technician?

I feel almost obliged to link to an early 90's critique by Larry Summers of empirical macroeconomics, of exactly the kind of epitomised by Sargent & Sims. Summers is more appreciative of Sims's VAR models than of Sargent's 'deep parameter estimations'. I had enjoyed reading the piece earlier, agreeing with the basic philosophy though I had no real idea about the details. But I now think that Summers critique is basically summarized as 'empirical macroeconomics has not been useful yet' and I am less convinced. 'Yet' can be key. And as Noah Smith says here, the lack of conclusions from VAR models may be a feature not a bug. Could it be that the theoretical as well as 'pragmatic empirical' macro that Summers prefers pretends to tell us more than it really does?

I read one criticism somewhere that VARs are models with linear dynamics and as such may be unhelpful in describing the fundamental non-linearity of the macroeconomy. This sounds more reasonable to me. But then what do I know.

Either way, I found digging into Sims's work a mind-opening, nearly mind-bending endeavour. I invite everyone interested to do the same. You can begin with this interview here.

(Update : Have corrected some formatting/typo issues in this post as well as the previous one. )

The 2011 Economics Nobel

Thomas Sargent & Christopher Sims have received the 2011 Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel. Tyler & Alex posted some excellent summaries (1, 2, 3, 4) over at Marginal Revolution. Others (Krugman, Sumner, Noah Smith) posted their own sets of congratulations, inferences & comments.

I'd heard of both but was until yesterday extremely unaware of the work of either, barring a faint memory of having read something (by Robert Hall, perhaps) about Sargent's place on the freshwater - saltwater spectrum and knowing the full expansions of VAR and SVAR. So yesterday was a day of great education. It is noted that this was a prize for 'empirical macroeconomics' or even 'macroeconomic empiricism'. Everyone also made amply clear that as true blue empiricists, both Sargent and Sims hold views and have published papers that sit uneasy with almost all ideological and policy camps. I found that the theory and policy implications, as well as the historical context, were often more revealing than the sophistication of the techniques, which I'm ill equipped to comment on.

Tyler linked to two of Sargent's most influential papers (1, 2). While the broad takeaway from both seems to be that monetary policy and fiscal policy must usually act together for either to be really effective in a business cycle, it was interesting that Scott Sumner linked approvingly to a paper titled "Some Unpleasant Monetarist Arithmetic" which claims that monetary authorities may be powerless in managing the price level and open market operations may end up achieving the reverse of what they intend to. On reading the paper, you understand why - Sargent's paper is about the importance of expectations in general and uses ratex (though he doesn't assert or necessitate that ratex is the right way to model expectations). Sargent also uses definitions of 'tight money' and 'easy money' that belong to a different era and that Scott has completely disavowed. But I still find it a little bit of a cheat that Scott doesn't mention the government debt mechanism for this counterintuitive result (more on this later), nor does he so much as nod to the later introduction that this result may hold even in the short run and thus monetary policy effectiveness is fundamentally contingent on accommodation by fiscal policy.

Tyler also linked to this interview with Sargent that I did not enjoy much - he defended the 'state of modern macro' by essentially saying that macro researchers are more nuanced than they are being given credit for. This is true but it is a truism. Almost every academic worth talking about is more nuanced than popularly characterized - this does not tackle the basic substance of the critique. I also found it a bit disappointing that Sargent didn't subject his hypothesis on the Eurozone (pages 10,11 of 14) to his own strict empirical standards, otherwise he would have easily noticed that the default risks are better correlated with the size and volatility of the capital account surplus than the fiscal deficits. One interesting bit was when he talks about the Kareken-Wallace model of banks and bank runs as an alternative to the more popular Diamond-Dybvig. He compares the basic features, assumptions and gaps in both models and I recommend reading that section (pages 5,6 of 14) in detail. Noah Smith's take on Sargent is charming and well worth reading - he calls Sargent a 'badass' whose research sometimes proved too hot to handle for theorists who would have wanted him to side with them. (The last link is a rather damning indictment of Lucas & Prescott if meant in all seriousness and not as an exaggeration).

Update : There was a part on Chris Sims written earlier for this post. But as I kept discovering more while writing it, I realised it deserved a separate post.

Thursday, September 29, 2011

The 'Old Keynesian' deceit, the crisis & Europe

Paul Krugman, on Ireland. "And I have, of course, written repeatedly — both informally and in actual papers with diagrams and Greek letters — that in a deleveraging, liquidity-trap economy, wage reductions would reduce, not increase, employment "

Paul Krugman, on Ireland. "Look, standard Keynesian models, open-economy version, tell a very clear story about what happens when a country pegs its exchange rate at a level that leaves its industry uncompetitive. The country doesn’t stay depressed forever: high unemployment leads to actual or at least relative deflation, which gradually improves cost-competitiveness, which leads to rising net exports and gradual expansion. In the long run, full employment is restored; it’s just that in the long run we’re all, well, you get the picture"

So Professor, which is it? Does deflation reduce employment or increase employment? I know, I know. Reduce in the short run, increase in the long run. The challenge, is to tell us when the short run transforms into the long run in a deflationary economy without a change in the macroeconomic policy regime. When does the inflection point occur? What causes it?

The biggest empirical challenge to the Keynesian (ala Hicks, Tobin, Krugman) theories of the business cycle is not stagflation, as was earlier presumed. It is the paucity of the deflationary wage-price death spirals of the kind that you may expect or predict from these theories. Hyperinflations and sovereign defaults are much more common.

And if I understand my IS-LM (and I think I do), 'deleveraging' is not part of it. IS-LM is a plausible theory of the price of debt (of a very ideal, 'no credit risk' kind) , but it does not even begin to touch the level of debt, esp. risky debt. Among the many intellectual conceits of the Krugman/ De Long band in the past two years has been to casually add 'deleveraging' to the more standard terms of 'liquidity trap' and 'deflation'. I have nothing against adding to your intellectual repertoire. Indeed, it is the only way to approach anything even remotely complex. But to speak of deleveraging as if it has been part of their intellectual framework all along is just cheating.

For the past couple of years, the old macroeconomic camps and opponents have been mostly heat and very little light, with everyone either dropping their heads or dropping their pants. Initially, the finance and banking experts (Raghuram Rajan, Gary Gorton) were the ones to turn to. But now that the focus is back on the broader macroeconomy, I find it most edifying to turn to Ricardo Caballero & Ken Rogoff.

Among the bloggers, Scott Sumner remains the most consistently interesting, policy recommendation-wise. And the recently created David Glasner blog is just a goldmine. (Incidentally, has anybody else noticed that the RBI seems to be living a Sumnerian dream, by design or by default? India's annual NGDP growth for the past 4 odd years has consistently been stable in the whereabouts of 16-17%. 6% inflation, 10% growth ; 10% inflation 7% growth, etc.)

But the most radically simple and brilliant policy recommendation for 'balance-sheet-weakness-inducing-deficient-demand' type recessions has to be the modified 'helicopter drop' proposed by Steve Waldman here. He talks about the Fed, but there is nothing in that recommendation that can't be replicated elsewhere in the world. Or at least in India, should we need to.

The presumption, of course, is that there isn't a looming sovereign debt crisis. And that there is monetary independence. And enough fiscal consolidation to ensure that there is treasury's blessing for the central bank to undertake a quasi-fiscal action. Europe, therefore, is as a nice gentleman told me recently, quite 'shafted'.

Tuesday, August 16, 2011

Of Op-Eds

Recently, Ravikiran blogged after a long time. Reading his post took me to the days about 4-5 years back when I had just started blogging, and the 'Indian blogosphere' (or at least the part of it that I engaged with) was a virile, juvenile and occasionally illuminating virtual space that had not yet been neutered by Facebook, Twitter, work and maturity. Out of nostalgia, I browsed some of his older posts. Among the most recent ones was a link to an editorial by Shruti Rajagopalan. It caught my eye due to the recent controversy over the sponsorship of the London Olympics by Dow Chemicals, the company acquired Union Carbide, which was responsible for modern India's worst industrial disaster.

It's a piece worth discussing for two reasons. One, it is very well written, concisely argued, compact yet forceful. Two, and more importantly, it is an excellent demonstration of the sheer futility of the op-ed as a non-fiction form of great merit.

Shruti argues that by preventing American tort lawyers from doing their customary ambulance-chasing out of an intuitive, paternalistic yet misguided and ill-informed sense of protecting the Bhopas Gas Tragedy victims from exploitation, the Indian government actually ended up causing them (the victims) more harm. The civil lawsuit was settled for the presumably low sum of $470 Mn; the criminal charges took more than two decades to process and ended in ridiculously lenient sentences.

Ravikiran talks about strict liability and judicial reform in his post linking to that piece. I don't have any good conceptual framework to analyse strict liability, so I won't go there. The need for judicial reform in India is a no-brainer, so I won't go there either. But the specific claim that Shruti discusses at length in her op-ed, the value of the civil settlement, is worth discussing. For this discussion, please bear in mind that Shruti's argument is explicitly utilitarian in a rather precise sense - she is making the claim that the Bhopal Gas tragedy victims were stiffed of just compensation, monetarily.

Shruti mentions that American tort lawyers typically charge one-third as commission fees from any legal settlement. Thus, a $470 Million pay-out would have needed to be $705 Million or above for the America-tort-lawyer way to be more successful than the Indian-government way. (Losses due to the leaks in the government distribution mechanism can be assumed to be the same in either case, and so are not material to this analysis).

So what would have made the payout $705 Million or more? Shruti claims that bureaucrats underestimated the fatality and injury rate, partially because of incompetence, partially because their incentives were not aligned with correctly estimating the right numbers. She does not charge (or even insinuate) outright corruption, but one may well imagine that to be a third source of error and manipulation. Hence, Shruti concludes that the Bhopal Gas victims were stiffed.

But wait, wasn't that conclusion contingent on the American tort lawyers being able to extract $705 Million or more? Do we have any evidence on whether that would have been achieved? We have some clue - she mentions that estimates of the death toll range from 4000 to 15000 (though she doesn't mention the sources of those estimates - if the bureaucratic estimate was 3000, the range should have been 3000 to 15000, at the very least). If it is 15,000 and the government was off by 5x, $470 million is rather obviously a prima-facie rip-off. If it is 4000, then maybe not, because the pay-out also includes injuries etc.

So what is the convincing evidence in the op-ed that the American tort lawyers would have been able to extract $705 million? None, actually. And that is the fundamental problem with a policy op-ed. Even the best ones are arguing from assertion, at some level or the other. Some are more nuanced than others, take longer time to figure out, but they all suffer from this same fatal flaw. Shruti believes (more or less correctly) that the bureaucrats' incentives are not properly aligned, and this leads her to believe and argue the result of their actions was necessarily sub-market-optimal. She doesn't feel the need to investigate the evidence required to go from the premise to the conclusion, or at the very least, doesn't feel the need to present it.

Data-based argumentation in policy is difficult. There is only one history. There are way too many variables to consider. There isn't enough data. You can't set up Monte Carlo simulations. You're not even sure if those would be useful at all. And in any case, if you were pursuing any of the above for writing something, that would be a PhD paper, not an op-ed. Shruti could have tried to collect data or evidence on the average pay-outs in cases argued by private tort lawyers in the US vs those argued by government counsel. Or in some comparable legal systems. But would she have found comparable or useful data? Would she have been able to draw any meaningful patterns? Most importantly, was it worth her time to do any of this for a thousand word essay?

And that's the thing about op-eds. They leave your posterior belief distribution almost exactly where your prior was. If you agreed more or less with Shruti's point of view, you would have found yourself nodding along, relishing the forceful prose, being appalled by the sheer injustice of government action and inaction. If you didn't and thought hard enough, the sheer lack of facts being used to argue her case (as opposed to simply the facts of the case) would become apparent. And you would dismiss the article as yet another speciously argued anti-government piece.

And if you were sitting somewhere on the fence, like I was, you would swing from this view to the other. Many times. And then wonder about the futility of the op-ed.

Saturday, June 11, 2011

Of Light Bulbs

Tyler Cowen gets one wrong. He approvingly refers to this article by Virginia Postrel. Ms Postrel's point, most generally, is that taxes or other commercial incentives are better solutions than regulatory bans. To the first degree, I agree. However, it is interesting to see how frequently she is wrong about the specifics that she is using/referring to while making the general point.

Ms Postrel is referring to the impending ban on incandescent light bulbs (called GLS in the lighting industry). She quotes Instapundit's dissatisfaction with the non-reduction in his electricity bills even after switching over to CFLs (compact fluorescent lamps). Okay. So? Controlling analysis for other variables, anyone? She believes GLS is a low margin commodity while CFLs are 'high margin specialty wares'. Are you freakin' kidding me? What are the odds that Ms Postrel has never read even a rudimentary analyst report on the lighting industry?

She refers to the American unwillingness to shift out of incandescent bulbs into CFLs. Okay. For the longest time, Bengal did not vote the CPM out. People are inertial. In this inertia, they make stupid decisions. Like not going to that expensive gym they have just taken membership of. Or not paying their taxes/credit card bills on time though these days, those payments are often just a click away.

She talks about the moral hazard problem makes this rule inefficient - about how people who know that CFLs consume less electricity are no longer incentivized to be careful about electricity usage and may end up leaving the bulb on for longer hours. Right. CFLs are about 5-6 times as energy effective as incandescent bulbs. The average American leaves the light on in his house for 6-8 hours typically (standard assumption in lighting calculations). If he shifts to CFL, Ms Postrel posits that he may be likely to leave the light on for more than 36-40 hours every day.

But the funniest part is reserved for the last. "The bulb ban makes sense only one of two ways: either as an expression of cultural sanctimony, with a little technophilia thrown in for added glamour, or as a roundabout way to transfer wealth from the general public to the few businesses with the know-how to produce the light bulbs consumers don’t really want to buy."

(emphasis mine)

Yes, apparently CFL manufacturing technology is one of the best kept secrets in the industrial world.

I invite Ms Postrel to read just one industry report on lamp margins by category. Or to visit one of the many CFL assembly operations around the world (mainly in China) to figure out exactly how high tech they are. Or, to just do some simple arithmetic before making claims on inefficiency.

Ultimately, she has exactly one point in her favour. That of the freedom of consumer choice. Usually, it's strong enough, but in this particular case, it doesn't seem so compelling to me. Else, you will soon see big protests or an active underground market in incandescent bulbs. (After all, marijuana has been illegal since forever, hasn't it?) I don't see that happening anytime soon.

As an aside, the US is a funny country. CFL is 3x the price of GLS and the adoption rate is just 25%. In India, CFL is 10x the price of GLS and the adoption rate is already 20%. And remember the fact that India is a fundamentally cash constrained economy, and the levels of electricity theft render the energy saving proposition useless in a few regions. For most Indians, the abiding value of GLS is not in its ability to light up fancy restaurants, but rather its promise of the being the only lamp you can buy with exactly one ten-rupee note in your pocket, the lamp that you grudgingly buy because paying up a 100 Rs. upfront is just not a viable option now, whatever its benefits in the future.

So tell me, have you also experienced Gell-Mann amnesia recently?

(Full disclosure : I work in the lighting industry, and my job is such that broadly, I'm happier when CFL sales increase. If you ask me personally, the tube-light remains lighting wise the best solution before the recent advent of LEDs, for many reasons. But then the upfront cost can be nearly Rs. 500 if you're also buying the ballast, and it's not a bulb-retrofit. )

Friday, June 03, 2011

Bad books - bleg

One kid from the insti on an internship wanted some help on his project. He is supposed to recommend how to improve demand forecasting in one of our businesses. He has a recommendation, which is simple, intuitive but assumptive. He tests it on one data point. It works. He tests on another. It doesn't. He wishes away the evidence. I don't let him move on. He finally says something to the effect of - 'it's a theory, I studied it at the insti.' I don't remember studying anything of the sort, so I ask him where he read this theory. He whips out a book. International authors, fancy cover. 10 pages devoted to A/F ratio (the theory). I skim through it. It's 10 pages of tables with lots of data points, calculations of the mean and standard deviation of those data points, celebrating the normal curve and devotedly explaining why 99.8% may not be very different from 100% but 90% is (I'm not kidding).

Then, somewhere in between, there's one little sentence. 'Since the forecasting bias (A/F) of last season was x, we can assume that maybe it will be the same this season'. That's it, that's all. The entire model, the entire meat and juice, apologetically assumed way, to devote 10 pages to arithmetic, banal arithmetic at that.
I fancy a job as an editor sometimes.

Sunday, March 06, 2011

To keep it alive

Say you're researching/writing an article about the 'demographic dividend'. Say you're hypothesizing that it is the single biggest factor in India's recent economic growth. Say your evidence is the differences in inter state growth rates and inter-state working age population, between those states are doing well and those that aren't doing so well. Say your evidence shows you that the working age population in UP/MP/Bihar actually *declined* between 1991 and 2001.

What would your reaction be?

Option A. "Oh dude, I forgot. Migration. Let's dig deeper. and re-hypothesize."

Option B. "Uh-oh. Nevermind, who's gonna get that anyway. Let's publish."

Option C. "Damn! That's like the last piece in this tremendous body of evidence. Brilliant, let's publish!".

If your response is B or C, congratulations! You are now fit to write for Mint Lounge's economics coloumn.

Forgive the vitriol, it's been very long.