Optimal Fiscal and Monetary Policy with Commitment

Mikhail Golosov and Aleh Tsyvinski 1

May 31, 2006

Forthcoming in New Palgrave Dictionary of Economics and Law

Abstract
“Optimal fiscal and monetary policy with commitment” is a policy of choosing taxes and
transfers or monetary instruments to maximize social welfare. “Commitment” refers to
ability of a policymaker to make binding policy choices.

Optimal fiscal policy under commitment

Ramsey approach to the optimal taxation

“Ramsey approach to optimal taxation” is a solution to the problem of choosing optimal

taxes and transfers given that only distortionary tax instruments are available.

A starting point of a Ramsey problem is postulating tax instruments. Usually, it is

assumed that only linear taxes are allowed. Importantly, lump sum taxation is prohibited.
Another assumption crucial to this approach is that all activities of agents are observable.

Given the set of taxes, a social planner (government) maximizes its objective function
given that agents (firms and consumers) are in a competitive equilibrium. Usually, it is
assumed that government’s objective is to finance an exogenously given level of
expenditures. It is important to note that, if lump sum taxes were allowed, then the first
welfare theorem would hold, and the unconstrained optimum would be achieved.

There are two common approaches to solving Ramsey problems. The first is the primal
approach, which characterizes a set of allocations that can be implemented as a
competitive equilibrium with taxes. By implementation we mean: for a set of taxes find a
set of (consumption and labor) allocations and equilibrium prices such that these
allocations are a competitive equilibrium given taxes. Conversely, a set of (consumption
and labor) allocations is implementable if it is possible to find taxes and equilibrium
prices such that these allocations are a competitive equilibrium given these prices and
taxes. Implementation often makes it possible to simplify a Ramsey problem by
reformulating a problem of finding optimal taxes as the problem of finding
implementable allocations. This reformulation of the problem is referred to as a primal
approach to Ramsey taxation.
1
Golosov: MIT and NBER; Tsyvinski: Harvard University and NBER
Main lessons of Ramsey taxation: uniform commodity taxation, zero capital tax in
the long run, and tax smoothing.

One of the central results of the literature on Ramsey taxation is uniform commodity
taxation (Atkinson and Stiglitz 1972). Consider a model with a finite set of consumption
goods that can be allocated between government and private consumption. All of these
goods are produced with labor. Assume that each consumption good can be taxed at a
linear rate. Then, under certain separability and homotheticity assumptions, commodity
taxation is uniform, i.e. the optimal taxes are equated across consumption goods.

Ramsey taxation provides a compelling argument against taxing capital income in the
long run in a model of infinitely lived households. The Chamley-Judd result (Chamley
1986, Judd 1985) states that in a steady state there should be no wedge between the
intertemporal rate of substitution and the marginal rate of transformation, or,
alternatively, that the optimal tax on capital is zero. The intuition for the result is that
even a small intertemporal distortion implies increasing taxation of goods in future
periods in contrast to the prescription of the uniform commodity taxation. Therefore,
distorting the intertemporal margin is very costly for the planner. Jones, Manuelli, and
Rossi (1997) extend the applicability of the Chamley-Judd result by showing that the
return to human capital should not be taxed in the long run. Chari, Christiano, and Kehoe
(1994) provide the state-of-the art numerical treatment of optimal Ramsey taxation over
the business cycle and conclude that the ex ante capital tax rate is approximately zero.

There has been a long debate on the optimal composition of taxation and borrowing to
finance government expenditures. Barro (1979) considered a partial equilibrium economy
and argued that it is optimal to smooth distortions from taxation over time, a policy
referred as tax smoothing. The implication of this analysis is that optimal taxes should
follow a random walk. Lucas and Stokey (1983) considered an optimal policy in a
general equilibrium economy without capital, and showed that if government has access
to state contingent bonds optimal taxes inherit the stochastic process of the shocks to
government purchases. Chari, Christiano and Kehoe (1994) extended this analysis to an
economy with capital and showed the Lucas and Stokey results remain valid in that set up
with or without state contingent debt, as long as the government can use taxes on capital
to effectively vary the ex-post after tax rate of return on bonds. Finally, Aiyagari et al.
(2002) analysis showed that if ex-post taxation of returns is impossible, the optimal taxes
follow a process similar to a random walk. They also showed the conditions under which
the tax smoothing hypothesis is valid.

Mirrlees Approach to optimal taxation

The Mirrlees approach to optimal taxation is built on a different foundation than Ramsey
taxation. Rather than stating an ad hoc restricted set of tax instruments as in Ramsey
taxation, Mirrlees (1971) assumed that an informational friction endogenously restricted
the set of taxes that implement the optimal allocation. This setup allows arbitrary
nonlinear taxes, including lump-sum taxes.

The informational friction posed in those models is unobservability of agent’s skills: only
labor income of agents can be observed. Therefore, from a given level of labor income it
cannot be determined whether a high skill agent provides a low amount of labor or effort,
or whether a low skill agent works a prescribed amount. The objective of the social
planner (government) is to maximize ex-ante, before the realization of the shocks, utility
of an agent. This objective can be interpreted as either insurance against adverse shocks
or as ex-post redistribution across agents of various skills. An informational friction
imposes incentive compatibility constraints on the planner’s problem: allocations of
consumption and effective labor must be selected such that an agent chooses not to
misrepresent its type.

In summary, the objective of the Mirrlees approach is to find the optimal incentive-
insurance tradeoff: how to provide the best insurance against adverse events (low
realizations of skills) while providing incentives for the agents to reveal their types
(provide high amount of labor).

Main lessons of Mirrlees approach in a static framework.

Theoretical results providing general characterization of the optimal taxes in the static
Mirrlees environment are limited. The central result is that the consumption-leisure
margin of an agent with the highest skill is undistorted, implying that the marginal
income tax at the top of the distribution should be optimally set equal to zero. Saez
(2001) is a state-of-the art treatment of the static Mirrlees model in which he derives a
link between the optimal tax formulas and elasticities of income. Mirrlees (1971) was
also able to establish broad conditions that would ensure that the optimal marginal tax
rate on labor income was between 0 and 100 percent.

Main lessons of dynamic Mirrlees literature: distorted intertemporal

margin
Recent literature starting with Golosov, Kocherlakota, and Tsyvinski (2003) and Werning
(2001) extends the static Mirrlees (1971) framework to dynamic settings. Golosov,
Kocherlakota, and Tsyvinski (2003) consider an environment with general dynamic
stochastically evolving skills. An example of a large unobservable skill shock is disability
that is often difficult to observe (classical example is back pain or mental illness).
Golosov, Kocherlakota, and Tsyvinski (2003) show for arbitrary evolution of skills that,
as long as the probability of agent’s skill changing is positive, any optimal allocation
includes a positive intertemporal wedge: a marginal rate of substitution across periods is
lower than marginal rate of transformation. The reason for this is that this wedge
improves the intertemporal provison of incentives by implicitly discouraging savings.
This result holds even away from the steady state and sharply contrasts with the
Chamley-Judd result that stems from the exogenous restriction on tax instruments.
Werning (2001) and Golosov, Kocherlakota, and Tsyvinski (2003) show that, in a case of
constant types, a version of uniform commodity taxation holds and the intertemporal
margin is not distorted.

Implementation of dynamic Mirrlees models is more complicated than implementation of

either static Mirrlees models, which are implemented with an income tax, or than
implementation of Ramsey models of linear taxation. By implementation we mean
finding tax instruments such that the optimal allocation is a competitive equilibrium with
taxes. One possible implementation is a direct mechanism that mandates menus
consumption and labor allocations for each date. However, such mechanism can include
taxes and transfers never used in practice. Three types of implementations were proposed.
In Albanesi and Sleet (2004), wealth summarizes agents' past histories of shocks that are
assumed to be i.i.d. and allows us to define a recursive tax system that depends only on
current wealth and effective labor. Golosov and Tsyvinski (2006) implement an optimal
disability insurance system with asset-tested transfers that are paid to agents with wealth
below a certain limit. Kocherlakota (2005) allows for a general process for skill shocks
and derives an implementation with linear taxes on wealth and arbitrarily nonlinear taxes
on the history of effective labor.

Optimal monetary policy

The theory of the optimal monetary policy is closely related to the theory of optimal
taxation. Phelps (1973) argued that the inflation tax is similar to any other tax, and
therefore should be used to finance government expenditures. Although intuitively
appealing, this argument is misleading. Chari, Christiano and Kehoe (1996) extended the
Ramsey approach to analyze optimal fiscal and monetary policy jointly in several
monetary models, and found that typically it is optimal to set the nominal interest rate to
be equal to zero. Such policy, called a Friedman rule, after Milton Friedman, who was
one of the first proponents of zero nominal interest rates (Friedman (1969)). To
understand intuition for the optimality of Friedman rule, it is useful to think about a
distinctive feature of money from other goods and assets. In most models, money play a
special role of providing liquidity services to households that cannot be obtained by using
other assets such as bonds. Inefficiency arises if the rates of return on bonds and money
are different, since households, by holding money balances lose the interest rate. When a
nominal interest rate is equal to zero, which in deterministic economy implies that
inflation is negative, with nominal prices declining with the rate of households time
preferences, the real rates of return on money and bonds are equalized, and this
inefficiency is eliminated.

The optimality of the Friedman rule stands in a direct contrast with Phelps arguments for
use of the inflationary tax together with other distortionary taxes such as taxes on
consumption or labor income. The reason for this is that money, unlike consumption or
leisure, is not valued by households directly, but only indirectly, as long as it facilitates
transactions and provides liquidity. Therefore, it is more appropriate to think of money as
an intermediate good in acquiring final goods consumed by households. Diamond and
Mirrlees (1971) established very general results about undesirability of distortion of the
intermediate goods sector, which in monetary models implies that inflationary tax should
not be used despite the distortions caused by taxes on the final goods and services.

The intuition developed above is valid under assumption that nominal prices are fully
flexible, and firms adjust them immediately in response to changes in market conditions.
However, even a casual observation suggests that many prices remain unchanged over
long periods of time, and Bils and Klenow (2004) documented inflexibility of prices for a
wide variety of goods. Inflexible or sticky prices lead to additional inefficiencies in the
economy that could be mitigated by monetary policy. For example, an economy wide
shock, such as an aggregate productivity shock or a change in government spending may
call for a readjustment of real prices. If adjustment of nominal prices is sluggish, the
central bank can increase welfare by adjusting nominal interest rates and affecting real
prices.

It is important to recognize that a government is also able to affect real (after tax) prices
using fiscal instruments instead. In fact, Correia, Nicolini and Teles (2002) show that, if
fiscal policy, is sufficiently flexible and can respond to aggregate shocks quickly, then
the Friedman rule continues to be optimal even with sticky prices, with fiscal instruments
being preferred to monetary ones. In current practice, however, it appears that it takes a
long time to enact changes in tax rates, while monetary policy can be adjusted quickly.
Schmitt-Grohe and Uribe (2004) showed that as long as tax levels are fixed or
government is not able to levy some of the taxes on goods or firm’s profits, then the
optimal interest rate is positive and variable.

Most of the applied literature on the monetary policy is based on the joint assumption of
sticky prices and inflexible fiscal policy. Woodford (2003) provides a comprehensive
study of the optimal policy in such settings. This analysis examines how central bank
response should depend on the type of the shock affecting the economy, degree of
additional imperfections in the economy, as well as choose the policies that would rule
out indeterminacy of equilibria. Two common policy recommendations for central banks
share many of the features of the optimal policy responses in this analysis. One of such
recommendations – a Taylor rule (see Taylor (1993)) – calls for the interest rates to be
increased in response to an increase in the output gap (difference between actual and a
target level of GDP) or inflation. Another recommendation, inflation forecast targeting,
requires that the central bank commits to adjust interest rate to ensure that projected
future path of inflation or other target variables does not deviate from the pre-specified
targets.

In addition to the analysis described above, several new, conceptually different

approaches to the analysis of monetary policy have emerged in the recent years. For
example, da Costa and Werning (2005) reexamine optimal monetary policy with flexible
prices in Mirrleesian settings and confirm the optimality of the Friedman rule there.
Seminal work by Kiyotaki and Wright (1989) gave raise to a large search-theoretic
literature seeking to understand the fundamental reasons that money differs from other
goods and assets in the economy. Lagos and Wright (2005) provide a framework for the
analysis of optimal monetary policy in such settings.

Bibliography

Albanesi, S. and Sleet, C., 2004, Dynamic Optimal Taxation with Private Information,
Discussion Paper 140, Federal Reserve Bank of Minneapolis.

Atkinson, Anthony B. and Joseph E. Stiglitz, 1972, The structure of indirect taxation and
economic efficiency, Journal of Public Economics 1, 97-119.

Bils, M. and P. Klenow. 2004. Some evidence on the importance of sticky prices. Journal
of Political Economy 112, 947-985.

Chamley, Christophe. 1986. Optimal taxation of capital income in general equilibrium

with infinite lives. Econometrica 54 (May): 607–22.

Chari, V. V.; Christiano, Lawrence J.; and Kehoe, Patrick J. 1994. Optimal fiscal
policy in a business cycle model. Journal of Political Economy 102 (August): 617–52.

Chari, V.V., L. Christiano and P. Kehoe. 1996. Optimality of the Friedman rule in
economies with distorting taxes. Journal of Monetary Economics 37, 203-223.

Correia, I., J-P. Nicolini and P. Teles. 2002. Optimal fiscal and monetary policy:
Equivalence results. Manuscript, Federal Reserve Bank of Chicago.

da Costa, C. and I. Werning. 2005. On the Optimality of the Friedman Rule with
Heterogeneous Agents and Non-Linear Income Taxation. Manuscript, MIT.

Diamond, P.A. and J.A. Mirrlees. 1971. Optimal taxation and public production I:
Production efficiency. American Economic Review 61, 8-27.

Friedman, M. 1969. The optimum quantity of money. In The optimum quantity of money
and other essays, 1-50. Chicago: Aldine.

Golosov, M. and Tsyvinski, A., 2006, Designing Optimal Disability Insurance: A Case
for Asset Testing, Journal of Political Economy 114(2).

Golosov, M., Kocherlakota, N., and Tsyvinski, A., 2003, Optimal Indirect and Capital
Taxation, Review of Economic Studies, 70 (3), 569-587.

Judd, Kenneth L. 1985. Redistributive taxation in a simple perfect foresight model.

Journal of Public Economics 28 (October): 59–83.

Jones, L., R. E. Manuelli and Peter E. Rossi,"On the Optimal Taxation of Capital
Income," (1997) Journal of Economic Theory, Vol. 73, No.1, 93-117.
Kiyotaki, N. and R. Wright. 1989. On money as a medium of exchange. Journal of
Political Economy 97, 927-954.

Kocherlakota, N. R., 2005, Zero Expected Wealth Taxes: A Mirrlees Approach to

Dynamic Optimal Taxation, Econometrica 73 (2005), 1587-1622.

Lagos, R. and R. Wright. 2005. A unified framework for monetary theory and policy
analysis. Journal of Political Economy 113, 463-484.

Phelps, E. 1973. Inflation in the theory of public finance. Swedish Journal of Economics,
75 67-82.

Schmitt-Grohe, S. and M. Uribe. 2004. Optimal fiscal and monetary policy under sticky
prices. Journal of Economic Theory 114, 183-209.

Taylor, J. 1993. Discretion versus policy rules in practice. Carnegie-Rochester

Conference Series in Public Policy 39, 195-214.

Werning, I., 2001, Optimal Unemployment Insurance with Hidden Savings, University of
Chicago, mimeo.

Woodford, M. 2003. Interest and prices. Princeton University Press.