Friday, July 1, 2016

The four macros and the way forward

Noah Smith identifies four different versions of macroeconomics and feels that the formal academic macroeconomics has failed everyone,
The first is what I call “coffee-house macro,” and it’s what you hear in a lot of casual discussions. It often revolves around the ideas of dead sages - Friedrich Hayek, Hyman Minsky and John Maynard Keynes. It doesn’t involve formal models, but it does usually contain a hefty dose of political ideology. The second is finance macro. This consists of private-sector economists and consultants who try to read the tea leaves on interest rates, unemployment, inflation and other indicators in order to predict the future of asset prices (usually bond prices). It mostly uses simple math, though advanced forecasting models are sometimes employed. It always includes a hefty dose of personal guesswork.
The third is academic macro. This traditionally involves professors making toy models of the economy -- since the early ’80s, these have almost exclusively been DSGE models. Though academics soberly insist that the models describe the deep structure of the economy, based on the behavior of individual consumers and businesses... they contain so many unrealistic assumptions that they probably have little chance of capturing reality. Their forecasting performance is abysmal. Some of their core elements are clearly broken. Any rigorous statistical tests tend to reject these models instantly, because they always include a hefty dose of fantasy. The fourth type I call Fed macro. The Federal Reserve uses an eclectic approach, involving both data and models. Sometimes the models are of the DSGE type, sometimes not. Fed macro involves taking data from many different sources, instead of the few familiar numbers like unemployment and inflation, and analyzing the information in a bunch of different ways. And it inevitably contains a hefty dose of judgment, because the Fed is responsible for making policy.
And on the way forward,
the new macroeconomics will focus on empirics and falsification -- in other words, looking at reality instead of making highly imaginative assumptions about it... macro will be fertilized by other disciplines, such as psychology and sociology, and will incorporate elements of behavioral economics... I think the new macroeconomics won’t just be new kinds of models and a more empirical focus; it will redefine what “macroeconomics” even means. As originally conceived, macro is about explaining national-level data series like employment, output and prices. Eventually, economists realized that to explain those things, they would need to understand the smaller pieces of the economy, such as consumer behavior or competition between companies. At first, they just imagined or postulated how these elements worked -- that’s the core of DSGE. Economists now realize that consumers and businesses behave in ways that are much more complicated and difficult to understand. So there has been increased interest in what’s called “macro-focused micro” -- studies of businesses, competition, markets and individual behavior that have relevance for macro even though they weren’t traditionally included in the field. Examples of this would include studies of business dynamism, price adjustment, financial bubbles and differences between workers.
This presentation by Justin Wolfers captures the problems with academic macro. But I am not sure that academic macro-economists are likely to discard their models in a hurry. For a start, it is difficult to shed layers on layers of orthodoxy accumulated over a long career so easily. More importantly, macro-focused-micro, involving disciplines like agent-based modelling, and rigorous empiricism and cross-disciplinary approaches do not lend itself to being neatly and consistently researched, comprehended and disseminated as a unified narrative. 

In this context, I am reminded of a course taught by Dani Rodrik a few years back. After each class, I would hear disappointed course mates complain that there were no clear and actionable takeaways. That's precisely the point.

I feel a clearer way to look at modern macro would be to use these newer lenses to acknowledge the complex nature of the underlying problem, become aware of all the instruments and tool-kits for its examination, select the most appropriate model(s) as the specific situation demands, and finally apply your informed judgement to choose the right course of action. This is very different from algorithmic application of a model to determine policies.

1 comment:

Karthik Dinne said...

Regarding "actionable points" - I observed that it's quite ubiquitous among many across the spectrum. Two reasons in my understanding apart from those mentioned by you in one of your earlier posts titled "why do we gloss over state capacity issues"

1) People, especially those in charge of implementation fail to differentiate between loose talk "ye hona chahiye, wo hona chahiye" (which they get to hear a lot) and genuine theories showing goal posts but not solutions. Implementors seem to think that the academic theories which can be used in reasoning are also gyan like the former category.

Of course, you know it better!

Dani Rodrik also says in his interview with Tyler Cowen that he meets two types of PMs or economic ministers - a) who want 'actionable solutions' b) who ask probing questions to know insights from research so that they can use the insights in their decision making. I am yet to meet people of 2nd kind, though!

2) "Need for Closure" (NFC). There's good literature on this in psychology. Every problem triggers a loop in us and troubles unless solved. The fear of uncertainty forces people to arrive at a situation where they feel that they have an answer.

This is the reason many people end up terming superficial reasons as the binding constraints or root-cause. They end up deriving strong conlusions from initial impressions.

For instance, if I visit a school and find that students don't have text books, my whole campaign for reform will revolve around textbooks, if I am not exploring the issue in depth (which is the case with most people). Having to say "something as the reason" is relieving for people.

It takes careful RCTs to help us break such NFC biases and help us see behind the initial impressions.