Selasa, 20 Juni 2017

Reis on the state of macro

Ricardo Reis has an excellent essay on the state of macroeconomics. "Is something really wrong with macroeconomics?"


In substantive debates about actual economic policies, it is frustrating to have good economic thinking on macro topics being dismissed with a four-letter insult: it is a DSGE. It is worrying to see the practice of rigorously stating logic in precise mathematical terms described as a flaw instead of a virtue. It is perplexing to read arguments being boxed into macroeconomic theory (bad) as opposed to microeconomic empirical work (good), as if there was such a strong distinction. It is dangerous to see public grant awards become strictly tied to some methodological directions to deal with the crisis in macroeconomics.
There have been lots of essays lately bemoaning the state of macroeconomics. Most of these essays are written by people not actively involved in research, or by older members of the profession who seem tired when faced with the difficulty of understanding what the young whippersnappers are up to, or by economic journalists who don't really understand the models they are criticizing. I am old enough to feel this temptation and have to fight it.

Many bemoan the simplifications of economic models, not recognizing that good economic models are quantiative parables. Models are best when they isolate a specific mechanism in a transparent way.

Critics usually conclude that we need to add the author's favorite ingredients -- psychology, sociology, autonomous agent models, heterogeneity, learning behavior, irrational expectations, and on and on -- stir the big pot, and somehow great insights will surely come. This is the standard third-year PhD student approach to writing a thesis, and explains why it takes five years to get a PhD.


I'm a bit guilty too -- with "Discount Rates" and "Macro Finance" full of my ideas on how to stir the pot and maybe get somewhere that I can't quite work out yet. Ricardo captures the feeling well:
 In turn, it would have been easy to share my thoughts on how macroeconomic research should change, which is, unsurprisingly, in the direction of my own research.
But at least I didn't argue that everyone else is wrong or bad or stuck, just these are my hunches on good things to try!

Others bemoan "too much math" in economics, a feeling that seldom comes from people who understand the math. The fact is, we have too little math in economics. There are so many phenomena we'd like to capture, so many frictions and real-world complications we would like to add and understand, but just don't have the tools to do it. Especially in finance, policy discussions go on and on about channels that we have very little clue to model.

Good economics is about answers, not questions; it's about finding the few simple ingredients that work. Economic models are so sensitive to ingredients that if you just pour and stir, you get garbage. Economics remains quite different from physics in that way. The underlying ingredients of (say) a climate or aircraft design model are very well understood, so you can make complex models that work. The underlying ingredients of economic models are not so well understood -- how much more will people work if their wages rise, how do they interpret statements by government officials, how do companies change their prices, and so on -- that small changes in the little ingredients make big differences in the economy-wide outcomes. Hence, good models are clear quantiative parables, not stone-soup melanges of popular ingredients.

Ricardo starts by evaluating current macroeconomics, empirically, by what active reasearchers are actually doing
... accurate measures of the state of macroeconomics are what the journals have recently published, or what the recent hires of top departments are working on.
After summarizing the research of 8 recent star new macroeconomics PhDs,
...this is all exciting work, connected to relevant applied questions, and that takes data and models seriously. In contrast, in the caricatures of the state of macroeconomics, there are only models with representative agents, perfect foresight, no role or care for inequality, and a cavalier disregard for financial markets, mortgage contracts, housing, or banks. Supposedly, macroeconomic research ignores identification and does not take advantage of plentiful microeconomic data to test its models, which anyway are too divorced from reality to be useful for any real world question. 
Compare this caricature with the research that I just described: the contrast is striking. Not a single one of these bright young minds that are the future of macroeconomics writes the papers that the critics claim are what all of macroeconomic research is like today. Instead, what they actually do is to mix theory and evidence, time-series aggregate data and micro data, methodological innovations and applied policy questions, with no clear patterns of ideology driven by geography.
After summarizing some other critiques, more data
Table 3 reports the articles published in the latest issue of the top journal in macroeconomics, the Journal of Monetary Economics...These include: theoretical papers on sovereign debt crises and capital controls, applied papers on the interrelation between financial indicators and macroeconomic aggregates, papers looking at extreme events like catastrophes and liquidity traps, and even purely empirical papers on measuring uncertainty in micro data and on forecasting time series in the macro data. There is originality and plurality, and a significant distance from the critics’ portrayal of research.
...Yet, according to De Grauwe (2009) “The science of macroeconomics is in deep trouble.” while Skidelsky (2009) thinks that there has already been a “...discrediting of mainstream macroeconomics”. These opinions express feelings more than facts, so it is hard to debate them.
Critics who bemoan rational expectations, DSGE, representative agent modeling, are largely complaining about what the young turks of the 1980s were doing, upsetting the Cambridge ISLM consensus of the 1970s. If you go to macroeconomics seminars today it's almost painful that every single paper has heterogeneous agents, some irrational expectations, and financial frictions galore. I actually long for the simplicity and transparency of the single agent rational expectations benchmark.

How about the charge that macroeconomic policy advice has been a failure?

Ricardo correctly points out that we are not nearly as influential as we think we are. (And policy makers are just as deaf to microeconomists, for example on free trade!)
macroeconomists are very far from running the world....Most macroeconomists support countercyclical fiscal policy, where public deficits rise in recessions, both in order to smooth tax rates over time and to provide some stimulus to aggregate demand. Looking at fiscal policy across the OECD countries over the last 30 years, it is hard to see too much of this advice being taken. Rather, policy is best described as deficits almost all the time, which does not match normative macroeconomics....Critics that blame the underperformance of the economy on economists vastly overstate the influence that economists actually have on economic policy.
I would double this observation for finance and the financial crisis. There is a trope in the media that efficient markets finance caused the crisis. This shows appalling ignorance. If you listen to efficient market finance, it says to invest passively in the market index, not risky tranches of mortgage backed securities pools, or the hedge funds that buy them.  And no academic told our regulators to set up the preoposterous system of mortgage subsidy and too big to fail bailouts.
One area where macroeconomists have perhaps more of an influence is in monetary policy. ...Looking at the major changes in the monetary policy landscape of the last few decades —central bank independence, inflation targeting, financial stability—they all followed long academic literatures. ...In the small sub-field of monetary economics, one can at least partially assess its successes and failures in the real world by judging how central banks have done over the past few decades....Every central bank that I know of in the developed world is in charge of keeping inflation low and stable.
Everyone loves to complain about the Fed. But their mandate is price level stability, now interpreted as less than 2% inflation, maximum employment, at least as much as money can affect it, and low interest rates. Check, check, check. I'm not sure how they did it, but it's hard to get too excited.

Mostly, by learning the lessons of history and not screwing up:
following the collapse of Lehman or the Greek default, news reports were dominated by non-economists claiming that capitalism was about to end and all that we knew was no longer valid, while economists used their analytical tools to make sense of events and suggest policies. In the United States in 2007-08, the Federal Reserve, led by the certified academic macroeconomist Ben Bernanke, acted swiftly and decisively.... While the recession was deep, it was nowhere as devastating as a depression. The economic profession had spent decades studying the Great Depression, and documenting the policy mistakes that contributed to its severity; these mistakes were all avoided in 2008-10.
Furthermore, macroeconomics is not special in its strengths and weaknessses
A separate criticism of macroeconomic policy advice accuses it of being politically biased....Yet, labor economics also has a history of heated debates and strong ideological priors, as well as continuous re-examination of truths previously held as obvious, such as the effects of the minimum wage on employment or of immigration on wages.  ..Macroeconomics does not stand out from labor and public economics in the features that the critics point out as the source of its crisis...macroeconomics is not all that special relative to the other fields. Economists across all fields were in part surprised by the crisis, but also eager to study it and analyze it.
Ricardo goes on to describe some frustration with how macroeconomics is taught and here I think he falls a bit into the temptation that befell those he criticizes. This is really, I think, a call for synthesis, for us all to spend some time seeing what the robust and teachable lessons are of the new models. Distillation is research. It took a long time for economists to figure out what Keynes' book really meant. 

Ricardo calls for 
...one could teach a macroeconomics class where the baseline model has (i) finite lives with overlapping generations, (ii) preferences over non-durables and housing, (iii) naive hyperbolic discounting, (iv) sticky information in forming expectations, (v) incomplete markets for individual income risk with maximally tight borrowing constraints, (vi) monopolistic competition and firm entry with fixed costs, (viii) nominal rigidities, (viii) simple banks with a net worth constraint (ix) distortionary taxes and government spending, and (x) a desire for liquidity for exchanges in decentralized markets.
Yes, but, there is some wisdom in the old joke of the drunk who looks for his car keys under the light, not near the car where he dropped them. The simple stochastic growth model has clear intuition and lessons. Each of these frictions adds a departure from that simple model. Just what basic intuitions emerge from the soup of all these ingredients is still something that we, as researchers, have not accomplished. And that is a sign of vibrancy too. Having accomplished a lot that still needs distillation is a sign of a vibrant field.

A big strain of macro and finance criticism berates us for not forecasting the great recession and financial crisis. Ricardo ends with a good reiteration of why prowess at unconditional forecasting is not a measure of economic science. The theory is efficient markets, not clairvoyant markets. That's like berating climate science because weather forecasters can't tell you if it will rain two weeks from now. It's worse, as most theories, especially in finance, predict with great clarity that crises and consequent recessions will not be predictable. It's like saying probability theory is wrong because you can't use it to outsmart the slot machines at Las Vegas. Ricardo has a lovely analogy, that medicine is quite useful even though your doctor can't predict the date of your death with great accuracy. 

In sum, a nice closing paragraph
Current macroeconomic research is not mindless DSGE modeling filled with ridiculous assumptions and oblivious of data. Rather, young macroeconomists are doing vibrant, varied, and exciting work, getting jobs, and being published. Macroeconomics informs economic policy only moderately and not more nor all that differently than other fields in economics. Monetary policy has benefitted significantly from this advice in keeping inflation under control and preventing a new Great Depression. Macroeconomic forecasts perform poorly in absolute terms and given the size of the challenge probably always will. But relative to the level of aggregation, the time horizon, and the amount of funding, they are not so obviously worst than those in other fields. What is most wrong with macroeconomics today is perhaps that there is too little discussion of which models to teach and too little investment in graduate-level textbooks.