View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Panel Discussion
Structural Economic Modeling:
Is It Useful in the Policy Process?

James Bullard
President and CEO

Federal Reserve Bank of St. Louis

International Research Forum on Monetary Policy
Washington D.C.
26 March 2010
Any opinions expressed here are my own and do not necessarily reflect those of the Federal Open Market Committee participants.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

Y ES

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

IT IS USEFUL

Is calculus useful for engineering?
Is knowledge of anatomy useful for surgery?
Conclude:
Structural modeling is an indispensible tool for macroeconomists.
Having a tool is necessary but not sufficient for good policy design.
Pretending to not have a structural model in mind is a
falsehood—all approaches to macroeconomics have a theory
behind them, implicitly or explicitly.
Explicit theories have the virtue of laying the assumptions bare for
examination and criticism.

But ...
... the models we need are big and complicated.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HIS

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

TALK

We all know what needs to be done:
Organize a more intensive national research effort in macroeconomics.

What we have today:
Back rooms (academia) and front rooms (policymakers) of
macroeconomics.

The back room research has been brilliant ...
... but has often been resisted in the front room.
As a result, we are nowhere near where we need to be in terms of
having a useful, comprehensive macroeconomic model that we
can use to get the economy to perform at its peak level.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

B ACK ROOMS

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

AND FRONT ROOMS

Sargent’s amusing description of academia—back rooms–versus
policymaking—front rooms.
My caricature of the back room versus the front room:
Back room: States and parameters are known by all, and policy
choice is menacingly complicated.
Front room: States and parameters are known by none, but policy
choice itself is disarmingly simple.
Back room: Stationarity assumptions are commonplace, but data
are often untrusted.
Front room: Stationarity considered unlikely, but thirst for data is
unquenchable.

Implication: The focus in the policy world is on determining the
state of the system by looking at lots of data.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

A COMMENT ON FORECASTING
Really, we track the economy.
We try to understand the quarters immediately past, the current
quarter, and the coming quarter.
Beyond that, many forecasts are a near random walk, with only
slow reversion to mean.
It is possible that the forecastable component of economic
activity may never be much greater than it is today.
Problem: A credible forecast is itself influencing the actions of
forward-looking actors in the economy.
Bottom line: Better forecasting is welcome but is not the ultimate
objective.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

B ETTER

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

POLICY IS THE PRIMARY GOAL

Better policy is the main objective.
Improved policy could deliver better outcomes—possibly
dramatically better—even in a world in which the forecastable
component of real activity is small.
Actual policy is like best practice medicine.
Actual policy is slow to adapt to suggestions from the research
frontier: Stick with policy adjustments which seem to have
worked well in the past.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

T HE BACK ROOM

I MPROVED M ACROECONOMICS

C ONCLUSIONS

RESEARCH

The academic research in macroeconomics over the past 30 years
has been brilliant.
The advances in the level of understanding of the intricacies of
the economy have been astonishing.
The main lesson:
The global economy is no simple system governed by a few
equations.
The fact that the economy is populated by purposeful,
forward-looking actors has huge ramifications for policy.
Optimal policies may not look like the policies currently in place.
To put a satisfactory, comprehensive macroeconomic model
together is a Herculean task.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

T HE STICKING POINT

The policy community—the front room—continually resisted the
developments in macroeconomics for decades.
Instead, there was general condemnation from this community
of the new approaches as unrealistic and impractical.
There was more truth in this early on, and much less truth in it
today.
The current crisis cannot be blamed on Lucas and Prescott
having too much influence in the policy world!

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

I MPROVED MACROECONOMICS

We need a more intensive national research effort in
macroeconomics.
Our current effort is not sophisticated enough to handle the
challenges that lie ahead.
We are beyond the point where a few professors or a small group
of researchers can make a major advance.
We need a determined effort to put together a satisfactory,
comprehensive model.
We should not allow ourselves to be off the research frontier.
Daunting? Yes, but we build aircraft carriers and space stations.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

C OMPREHENSIVE ENOUGH

Keep very negative outcomes as a possibility inside the model,
so that policy choice can be evaluated with the very negative
possibilities in mind.
Leading example: Financial panic.

Be able to understand trade-offs of policy choice on a more
global level.
With small models, policymakers have to guess what the effects
might be on aspects of the economy outside the analysis.
Leading example: Low interest rates as a prelude to a bubble.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

F INANCIAL INTERMEDIATION
We know from the literature that frictions in the intermediation
process can profoundly affect the general equilibrium.
Why assume these frictions away?
We also have a long line of literature following Diamond and
Dybvig (1983) concerning bank runs and related phenomena.
This should be integrated into our macro models, so that we can
understand how our policies are affecting the probability of a
run.
Also: global versus local policy analysis—
The run is a departure from the neighborhood of the targeted
equilibrium.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

M ONEY AND

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

LIQUIDITY

There is a large theoretical literature that attempts to be more
rigorous about the nature of money demand.
This literature should move closer to the data from financial
markets.
Especially interesting are Gorton’s (2008) ideas about repos as
privately-issued money.
We cannot make progress on this issue with money-in-the-utility
function assumptions.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

I NTERNATIONAL

Globalization is upon us.
Our understanding of the international linkages between
monetary and fiscal policies is tenuous at best.
Still, a critical challenge going forward is to understand the
impact of rapid development in much of the world on the global
equilibrium.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

H ETEROGENEOUS

I MPROVED M ACROECONOMICS

C ONCLUSIONS

HOUSEHOLDS

We know that policies have an uneven impact.
The last decade has witnessed a burgeoning research agenda on
heterogeneous households.
With this technology we can better analyze the general
equilibrium effects of policies across households.
It is hard work, computationally intensive.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

M ULTIPLE EQUILIBRIA AS
PHENOMENA

I MPROVED M ACROECONOMICS

C ONCLUSIONS

“ BUBBLE - LIKE ”

Two decades, two bubbles.
It is time for the policy community to embrace the concept of
multiple equilibria from the macroeconomic research world.
That would give us policymakers at least one way to coherently
address these issues.
The typical policy response in the literature: Adopt a policy that
kills off the undesirable equilibria.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

L EARNING

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

AND RATIONALITY

Much of the criticism of macroeconomics is associated with the
idea that there is “too much rationality.”
But the rational households in the models can readily be
replaced with learning households, and the analysis can proceed
from there.
This raises new issues, such as learnability of an equilibrium.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

G ROWTH AND

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

HUMAN CAPITAL

The pace of long run growth is the most important aspect of
economic performance.
We should be analyzing stabilization policies in conjunction with
growth policies.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

O THER

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

AREAS

These are just examples of the types of features a satisfactory
model would have.
We would like to be able to understand the trade-offs between
many policy choices.
Above all, we would like to be able to understand where the real
dangers are:
Big ticket welfare losses are associated with leaving the
neighborhood of a targeted equilibrium.
This has to be a possibility inside the models.

I NTRODUCTION

P OLICY C HOICE V ERSUS D ATA

T HE BACK ROOM RESEARCH

I MPROVED M ACROECONOMICS

C ONCLUSIONS

C ONCLUSIONS

The nation needs a more aggressive and better-funded research
effort in macroeconomics.
The current level of effort is unlikely to meet the many
challenges the U.S. faces going forward.
The level of complexity is beyond what individual researchers
can handle.
Small, shortcut models are going to be wildly wrong on many
dimensions.

The task is daunting, but the nation will spend a great deal on
other large projects arguably of lesser consequence.

Federal Reserve Bank of St. Louis
stlouisfed.org

Federal Reserve Economic Data (FRED)
research.stlouisfed.org/fred2/

James Bullard
research.stlouisfed.org/econ/bullard/