Teaching macroeconomics as though Lehmans didn’t happen
September 15th marked the tenth anniversary of the fall of Lehman Brothers, destabilizing Western economies at levels not seen since the 1930s. It also marked the second week of fall classes, with many economics graduate students cranking through equations that define the discipline’s conventional macroeconomic models. With such names as New Classical, Real Business Cycle and New Keynesian, these models can all be traced to the rational expectations revolution of the 1970s, which sought to explain stagflation when the conventional Keynesian framework could not. The rational expectations approach attempted to provide more precise behavioral microfoundations than the Keynesian model by positing that economic actors can form expectations of future economic values, say inflation, such that on average, their predictions of future values tend to be correct. This assumes the actors share the same understanding of the structure of the economy and past economic data. This research program would come to dominate macroeconomic scholarship and strongly influence policy makers, culminating in the creation of the dynamic stochastic general equilibrium (DSGE) model, a popular forecasting and policy analysis tool used in central banks and finance departments.
This approach to macroeconomic modeling came under scrutiny following the 2008 crisis, with Nobel laureate Paul Krugman asserting that most of the macroeconomics over the past 30 years was “spectacularly useless at best, and positively harmful at worst”. While this did spark some soul-searching within the discipline, the debate has been inconclusive. Several policy-making bodies are taking seriously the limitations of 1970s macroeconomics. In its recent Medium-term Research Plan, the Bank of Canada recognises that the crisis has challenged its reliance on New Keynesian DSGE models, encouraging the exploration of alternative modeling paradigms, such as agent-based and stock-flow consistent models.
On Canadian campuses, however, where the next generation of macroeconomists are being trained, there is no clear signal that similar changes are being made in the curriculum of grad-level macroeconomics. A recent panel discussion among academic economists featured the admission that the 2008 crisis was the most embarrassing empirical failure of the profession since the Great Inflation of the 1970. Yet, in the same breath, that professor said he wouldn’t change a thing in his teaching. Indeed, a glance at the macroeconomics syllabuses of several top Canadian grad schools find little evidence of a shift away from teaching the rational expectations-grounded macro models that have come under criticism.
Professors tend to teach what they are taught. With the sunk cost of prepping for PhD macroeconomic comprehensive exams, they have little incentive to develop a new course involving subject matter in which they are not trained. Further reinforcing the status quo is the tendency to teach what you research. Working in a climate of publish or perish, macroeconomic profs have good reason to not deviate from the dominant research agenda, which remains wedded to 1970s macro. In the absence of strong leadership for change or a mandate from either the dean or the premier to sit down with one another and re-design the curriculum, teaching macro in the post-crisis era will continue to be business as usual.
Yet this is not in the public interest. Given the acute financial stress experienced ten years ago, we have a stake in knowing that the policy makers of tomorrow are well prepared to confront episodes of economic downturn and instability. Learning to use a larger modeling toolbox is part of such preparation.
So, what are Canada’s economics students to do in the meantime as they are grind through the math describing a DSGE model? As befitting any college course where critical thinking is one of the learning outcomes, here are some questions students may ask about the models they are taught:
1. Who is in the model? The basic models tend to have a single agent representing all consumers who are assumed to be sufficiently alike as autonomous rational optimizers sharing common knowledge. Can the model accommodate multiple actors who may differ by age, preference, belief, resources and class?
2. Is there room for “black swans”? The 2008 crisis was precipitated by the collapse of the U.S. subprime mortgage market, an event deemed of low risk but of high impact. How does the model address this and other examples of fundamental uncertainty?
3. What kind of markets are modelled? Models with perfect competition behave very differently from more realistic models with imperfect competition, information asymmetries, price rigidities and institutional constraints.
4. Is there a financial sector? Perhaps the strongest criticism of the 1970s macro models was the reduction of complex financial plumbing to a single interest rate variable. Can these models feature lenders and borrowers? Are there banks? How does money fit in?
5. Does the model have to move to equilibrium? Following an economic shock, standard models tend to instantaneously jump to a new equilibrium path. However, observations of macroeconomic variables as they unfold over time suggest that such adjustment may be a much slower, sequential process. Understanding this path of adjustment may be of greater importance than the equilibrium destination.
6. How are these models empirically tested? A model’s usefulness should be judged by how it explains actual economic history.
With these and other critical questions about the core macro teaching models, tomorrow’s dismal scientists should be better prepared to confront challenging economics times.