Telling macro stories with micro | E-Axes

Not a member yet? Click here.
Forgot your Password?
Archives - Categories
On Inequality
On the Eurozone Debt Crisis
On Monetary Policy and Central Banking
On Global Economic Growth
On the Greek Debt Crisis
On the Banking and Financial Sectors
On Brexit
On China
On India
On Global Inflation
On Currencies
On the US Debt
On the "Economics" of the Arab Spring
Working Papers
Books suggested by members

Telling macro stories with micro

Author(s): Claudia Sahm

From Claudia Sahm's blog:

Don't let the equations, data, or jargon fool you, economists are avid storytellers. Our "stories" may not fit neatly in the seven universal plots but after awhile it's easy to spot some patterns. A good  paper in economics, according to David Romer, has three characteristics: a viewpoint, a lever, and a result.

Most blog or media coverage of an economics paper focuses on the result. Makes sense given the audience but buyer beware. Economists dissecting a paper spend more time on the lever, the how-did-they-get-the-result part. And coming up with new levers is a big chunk of research. The viewpoint--the underlying assumptions, the what's-central-to-the-story--tends to get short shrift. Of course, the viewpoint matters (often that's what defines a a story as economics), but it usually holds across many papers. Best to focus the new stuff. 

Except when the viewpoint comes under scrutiny, then the stories can really change. 

How much does micro matter for macro?

One long-standing viewpoint in economics is that changes in the macro-economy can largely be understood by studying changes in macro aggregates. Ironically, this viewpoint even survived macro's push to micro foundations with a "representative agent" stepping in as the missing link between aggregate data and micro theory. As a macro forecaster, I understand the value of the aggregates-only simplification. As an applied micro researcher, I am pretty sure it fails us from time to time. Thankfully, an ever-growing body of research and commentary is helping to identify times when differences at the micro level are relevant for macro outcomes. This is not new--issues of aggregation in macro go waaay back--but our levers, with rich, timely micro data and high-powered computation, are improving rapidly. 

I focus in this post on differences in household behavior, particularly related to consumer spending, since that's the area I know best. And I want to discuss results from an ambitious new paper: "Macroeconomics and Household Heterogeneity" by Krueger, Mitman, and Perri. tldr: I am skeptical of their results, above all, the empirics, but I really like what they are trying to do, to shift the macro viewpoint. More on this paper below, but also want to set it in the context of macro storytelling. 

How did we get here? Who invited this Representative Agent?

In a 2013 speech, "Aspects of Inequality in the Business Cycle," then-Fed Governor Sarah Bloom Raskin pushed back on the aggregates-only viewpoint but she also laid out the case for the status quo:

"... The typical macroeconomic analysis focuses on the general equilibrium behavior of "representative" households and firms, thereby abstracting from the consequences of inequality and other heterogeneity across households and instead focusing on the aggregate measures of spending determinants, including current income, wealth, interest rates, credit supply, and confidence or pessimism. In certain circumstances, this abstraction might be a reasonable simplification. For example, if the changes in the distribution of income or wealth, and the implications of those changes for the overall economy, are regular features of business cycles, then even an aggregate model without an explicit focus on distributional issues would capture those historical regularities."

Macro models are always going to abstract from some details of the economy. In the past nine years, have seen models that condition on aggregate measures of income, wealth, interest rates, sentiment, and credit conditions do a pretty good job explaining the changes in aggregate consumer spending. Of course, these models are not perfect (and out-of-sample forecasting and general equilibrium are whole other headaches), but adding micro heterogeneity to macro models is one in a long list of possible improvements. Adding a more realistic financial sector, exploring non-linearities, relaxing rational expectations, and extracting a better signal from noisy aggregate data are all in the queue too. Plus with "reform" pieces like Blanchard's "Do DSGE Models Have a Future?" ... I suspect the Representative Agent is not getting voted off macro island any time soon. 

What do people tell us? Dare we ask them?

I appreciate an easy-to-explain, well-fitting time-series model, as much as any forecaster. Much of my forecasting work is trying to explain the 'whys' of a model simulation and keeping an eye out for what those models might be missing. Even so, my greatest joy as an economist is working with micro survey data--where the Representative Agent gets lost in the crowd and my macro intuitions are regularly tested. 

Sifting through individual responses to find bits that are relevant for macro questions, like the impact of fiscal stimulus on overall consumer spending, is painstaking. And the growing multitude of micro data sets often leads to a wide range of estimates. Finally, on the margin, taking heterogeneity into account may not change macro results all that much ... which is almost a necessary condition for shifting viewpoints. In my latest paper with Shapiro and Slemrod on the 2011-2012 payroll tax cut, we identify households who used the stimulus to rebuild their balance sheets, to the point that they cut spending when the stimulus expires. The behavior we uncovered in the survey from asking households directly what they do with stimulus can explain a half percentage point of a two-percentage-point upside surprise in the saving rate in 2013. In this case, the micro adds some clarity relative the standard aggregate model, but improvement is neither large nor necessarily driven by the heterogeneity we focused on. Nevertheless, a third of households who don't fit neatly in standard economic models feel like a story worth telling to me.

What do we learn from this new paper? More work to do. 

Much of the work on how micro heterogeneity can affect macro outcomes (including mine above) fits in the genre of 'let the data speak.' Krueger, Mitman, and Perri in "Macroeconomics and Household Heterogeneity" take this a step further and develop a macro model that can actually generate some of those different voices. At 80+ pages, you might guess this is a complex story to tell, including relevant prior literature. (See also theirVoxEU summary). I will stick here to a some high-level reactions:

Things I like about the paper:

  • Persistent income. The very low levels of wealth that many households have is typically a struggle for economic models to generate. The authors find (see Table 7) that by adding highly persistent income shocks, as we see in the real world ... "Thus, the [model] economy contains a share of households with close to permanently low earnings, even in the absence of unemployment. These households, located predominantly in the lowest wealth quintile, have had no opportunity to accumulate significant wealth." Sound familiar? Differences in patience that they also add to their model are needed to get the high wealth closer to data (and even there they fall short). Unemployment insurance is also playing a role with the low levels of wealth but quantitatively the persistent income shocks do most of the work. 
  • Regime switching. They are trying to understand how the distribution of wealth may have affected the aggregate response of consumption to the Great Recession. I like the way that they use a regime-switching approach, which allows all the behavioral responses to vary by whether the economy is deep recession or not. Their model's insight that low wealth, impatient households would have a particularly strong precautionary saving motive in a recession (when the risk of losing a job is high) was interesting. In normal times, such bad outcomes are too distant a prospect for these impatient folks to get worked up about and save. With only two U.S. recessions that fit their definition of a severe recession in the post World War II era, obviously it would be tough to recover that much nuance from the aggregate time series.
  • Presentation. It might seem that the heavy lifting comes from solving these models. And I do respect the efforts here. One chapter of my dissertation was with dynamic programming models of portfolio choice, following solution techniques that Chris Carroll kindly posts on his website. Working with these models (much simpler than in this paper) did not end up being my comparative advantage, but I am still an eager consumer of them, especially ones that clearly lay out the economic intuition for their results. Viewpoints require intuition. 

Things I don't like about the paper:

    Wealth effects. It is not surprising but with all their focus on generating a realistic wealth distribution, the authors manage to super-size the role of wealth on consumption. They don't hide this fact and the consumption dynamics of the model get points for matching "qualitative" but not "quantitative" patterns in the Great Recession. In fact, their model predicts a drop in the aggregate saving rate (Table 11), not the rise that we actually saw. To be honest, I would have been more skeptical of the paper if they had managed to match those patterns given their model, but that just shows there are some missing mechanisms. The progress is real.
    • Empirics. This might seem like a fatal flaw (it's not) but I did not find the motivating empirical facts all that convincing. There is a big, ongoing debate about whether the debt cycle and subsequent pullback in consumer spending in the Great Recession wasbroad-based across households or whether it was concentrated among certain groups, such as the highly levered. My reading of the empirical evidence, including my work with Bricker and Krimmel on auto spending in the SCF and expectations in the Michigan survey, comes down on the side of broad based. Even so, I concede this is a (maddeningly) open question. What bugged me in this paper is that the consumption growth response to the Great Recession, as they measured in the Panel Study of Income Dynamics (in Table 4) is quite similar across wealth groups. It's an anomalously small drop in income for the lowest wealth group that leads to the disproportionate change in their consumption rate out of income. That seems a bit too delicate an empirical result for their model in which the correlation between wealth and consumer spending is very strong. And stronger than the data.
    • Patience. Seems reasonable that households have varying degrees of patience, that is, that they value current versus future consumption differently. But patience is not directly measured here so the authors are basically a relabeling of residuals. This is common storytelling device in macro ... keep that in mind the next time someone tells you about total factor productivity (at least we used to call it the Solow residual). But we shouldn't overstate how much we really know by just giving it a name. Empirical work on patience could be brought to bear on this analysis but I raise this more caveat than a complaint.

    On balance, I liked this paper a lot and would highly recommend those interested to give it a careful read. It could even be an important step in shifting the way we tell macro stories and the degree to which we understand about macro dynamics.

    Are we there yet? No. 

    Up to this point, I have focused on how economists tell their stories ... our viewpoints, levers, and results. Economics is not supposed to be about economists, but sometimes our stories can feel that way, especially to non economists. And to be fair, the viewpoints that economists bring to their work do have an impact on the results, if nothing else by what we choose to study.

    I listened this week to a recent Fed Up event hosted by the Kansas City Fed. One thing that struck me in the questions and answers was how the viewpoints varied across the participants. I am not weighing in on the substance and you will hear plenty of agreement at the event too, but it was a good reminder that the economist viewpoint I am most familiar with is one of many. 

    I thought the question from Rod Adams of Minnesota Neighborhoods Organizing for Change (starting at 28:45 in the video) illustrated this well:

    "... recently you have said that we are near full employment, that it is now time to intentionally slow down the economy by raising the interest rates. Now I don't understand how you can think that. The unemployment rate for African Americans in Minnesota is still near 9 percent. Underemployment is double that amount. Many people want full-time work but just can't get it. And there are others who have given up looking for work out of total despair, because the jobs aren't there. If the labor market were truly healthy, people in my community would all be able to find full-time jobs at decent wages. Now the economy is recovered for much of white America but for black and Latino workers it is not ..."

    Starting with his viewpoint for what is a healthy economy, might lead to different lever, a different result on macro outcomes. Recall persistent, individual-level income shocks played a powerful role in the model discussed above, but the process was fairly simple. Their model wouldn't capture the fact, as Kocherlakota highlighted recently, that "the black unemployment rate has been about 1.9 times the overall rate for more than 40 years" and that is a proportional, not a constant gap, so it widens in recessions. It is unclear whether or how such persistent, micro differences would change our views on macro fluctuations and how to best address them. Still this feels like a opportune time for economists to revisit their assumptions and explore some new viewpoints.

    © 2011–2017 e-axes. All rights reserved. | Credits | Contact Us | Privacy Statement | Sun 21 Jan, 2018 14:05:04 PM
    e-axes is proudly powered by Norder - Creative Solutions