Greg Kaplan

Prof Economics @ University of Chicago
Lead Editor @ JPE Macro
Chair @ e61 Institute
Book: "Macro: The Economic Models that Shape Our World"
(coming Nov 2026)

[International and Monetary Economics Network @int_mon_econ]

This looks like a must-read!

"Macro: The Economic Models That Shape Our World" by Greg Kaplan.

https://t.co/zoptrR1SZq

"In clear and engaging prose, Chicago economist Greg Kaplan demystifies how our everyday behavior connects to the biggest questions in fiscal and monetary policy. Tracing the evolution of economic thinking from Keynes to today, he guides readers through the models used to explain our economy, uncovers their flaws, and reveals what cutting-edge research tells us about managing the economy in good times and bad. These earlier models are valuable, Kaplan argues, but those that underlie much of conventional wisdom are fundamentally out of touch with the realities of how people spend and save...The journey culminates in the HANK (Heterogeneous Agent New Keynesian) model, the current state of the art that is more consistent with reality. HANK accounts for our diverse spending and saving habits and will transform the way we think about economic policy. Written for anyone who wants to truly understand how our economy works, Macro is the long-overdue book that bridges the gap between pioneering research and real-world behavior."

https://t.co/zoptrR1SZq

[Greg Kaplan @GregWKaplan]

@EichMartin @HannoLustig Expected inflation ????

[Greg Kaplan @GregWKaplan]

@ludwigstraub @a_auclert @ChristianKWolf @JesusFerna7026 @luigi_bocola Sadly will be on a plane on my way to Boston, but if the WiFi allows will be keenly watching the fireworks!

[NBER @nberpubs]

Building a model that combines heterogeneous households with incomplete markets and state-dependent pricing with strategic complementarities by firms to analyze the effects of large macroeconomic shocks and policy interventions, from @GregWKaplan and Ken Miyahara https://t.co/DZKwW3R4co

[Greg Kaplan @GregWKaplan]

Nothing revolutionary in terms of computation. Here is an overview for those who are interested.

We express the steady-state of the model as a mean-field game, described by a system of HJBs and KFEs, which we compute using standard finite difference methods as in https://t.co/OKLoWMJjLV

We stack the steady-state equilibrium conditions alongside the steady-state calibration targets as a system of non-linear equations and solve these using standard non-linear solvers. Thats for the steady-state.

To find impulse responses to a single sequence of shocks, we cast the full dynamic system as a system of non-linear equations and again use a standard trust-region solve to solve for it.

Allowing for a series of unexpected shocks involves solving for the first shock, then using the implied aggregate state as the initial condition for the second shock, solving for that equilibrium and so on. So just a loop over the single MIT shock.

To estimate the shock realizations, i.e. the non-linear inversion, we stack the sequence of residuals between model and data alongside the equilibrium conditions so that at each step its the same type of system of equations just bigger.

Two "secret weapons" (as Tom Sargent would say) make this all feasible:

(1) Automatic differentiation: we use AD to compute all derivatives with respect to prices and shocks and to compute the Jacobian. This is what makes it fast and easy to solve the large systems of non-linear equations. Without automatic differentiation, all the non-linear systems would be much more difficult to solve.

(2) Homotopy: we start with small deviations from steady-state by shrinking the GDP and inflation deviations we target back towards 0. For small deviations it's easy to find a solution because the solutions are close to steady-state and Jacobians are very informative. We then use those equilibria and estimated shocks as initial guesses for a larger deviation, and so on, up to the desired size.

For the Calvo model the whole thing (20 MIT shocks + shock estimation) solves in Matlab on my MacBook Air in about 4-5 minutes. With state-dependent pricing it's more like 20-30 minutes because there are additional equilibrium objects to solve for. We have not tried to optimize for speed yet - Im sure with some help from AI this could be made much faster.

[Greg Kaplan @GregWKaplan]

New paper: "How Does Monetary and Fiscal Policy Affect the Economy in the Face of Large Shocks?" (with @KenMiyahara ) Paper: https://t.co/76PmxQm9He

Two model features are important for understanding the effects of policy in the COVID era: heterogeneous households + state-dependent pricing.

How would the US economy have evolved under different fiscal and monetary responses? 🧵

[Greg Kaplan @GregWKaplan]

Paper: https://t.co/1pmAqWAE1b

[Greg Kaplan @GregWKaplan]

Could we have done something better?
A *funded* stimulus could have delivered similar short-run support for output and inflation, with far less permanent price increase and larger, more broadly shared welfare gains.

[Greg Kaplan @GregWKaplan]

On 14-15 Dec, @E61Institute and @UNSW will host the latest edition of the Macro4Micro workshop in Sydney, Australia. Details here: https://t.co/nQOaPOi9mG

Confirmed speakers include @glviolante, @itskhoki, @JonSteinsson and Emi Nakamura.

Would love to see you in Sydney - please consider submitting a paper and spread the word!

[Greg Kaplan @GregWKaplan]

Wow, thats cool! I don't know enough about Dulac functions, Floquet Multipliers and Liouville's theorem is not sufficient to verify. But yes, I should have asked AI for other ways to prove it. A year ago when I was working on this I asked Claude and ChatGPT to help me find a Lyapunov function to no avail... Thanks!

[Piotr Żoch @pzoch5]

It seems that @GregWKaplan wrote a much needed general public friendly book on modern macro.

https://t.co/mWqn8s4blU