Testing Ricardian Equivalence with the Narrative Record on Tax Changes
| Date | 01 April 2020 |
| DOI | http://doi.org/10.1111/obes.12339 |
| Published date | 01 April 2020 |
| Author | Alfred A. Haug |
387
©2019 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 82, 2 (2020) 0305–9049
doi: 10.1111/obes.12339
Testing Ricardian Equivalence with the Narrative
Record on Tax Changes*
Alfred A. Haug
Department of Economics, University of Otago, Dunedin 9054, New Zealand
(e-mail: alfred.haug@otago.ac.nz)
Abstract
The Ricardian equivalence hypothesis is tested empirically with a subcategory of the nar-
rative measures of US tax shocks developedby Romer and Romer (in, American Economic
Review 2010;100:763). The present valueof tax increases motivated solely by concerns for
improving the fiscal health of the government is used. These tax news represent a switch
from debt to tax financing that should have no effects on real output and consumption. For
the post-1982:IV period, fiscal anticipation plays an important role as many of the tax in-
creases are implemented with substantial delays. Anticipated tax hikes increase economic
activity in the delay period. Ricardian equivalence is rejected.
I. Introduction
An important benchmark for fiscal policy is the Ricardian equivalence hypothesis. It states
that it does not matter for the economy whether government expenditures are financed by
current taxation or instead by issuing government bonds (Barro, 1974). Ricardian house-
holds are only concerned about the present value of their intertemporal tax liabilities, which
are in turn determined by the present value of the stream of government expenditures and
currently outstanding government debt. The timing of taxes does not matter. Government
bonds are simply seen as postponed taxes that will have to be paid for at some future date.
Hence, a switch from tax to bond financing, or vice vera, has no effect on output or con-
sumption. Barro (1974) argued that Ricardian consumers react, as long as the stream of
government spending is kept fixed, to a tax cut financed by an increase in the government
deficit by increasing savings by an equal amount. The tax cut leads to a dollar-for-dollar
increase in bond holdings. Therefore, neither output, or output growth, nor interest rates
change. The Ricardian equivalence hypothesis is based on several restrictive assumptions,
such as lump-sum taxes, perfect capital markets where forward-looking households do
not face liquidity constraints and can borrow at the same interest rate as the government,
altruistic operative bequests that link generations, and no uncertainty about the future tax
incidence.
JEL Classification numbers: C51, E62, H62.
*The author thanks, without implicating, anonymousreferees, David H. Romer and participants at the third IAAE
Annual Conference, WAMS and NZESG for very helpful comments on an earlier version.
388 Bulletin
A large number of theoretical and empirical papers have studied Ricardian equivalence,
particularly during the period of high federal US government budget deficit to GDP ra-
tios in the 1980s. This literature has been surveyed by Seater (1993) and more recently
by Ricciuti (2003). On a theoretical level, relaxing the very strict assumptions necessary
for Ricardian equivalence can lead to government bonds having either positive or nega-
tive net wealth effects for households (Barro, 1974, 1989). Under Ricardian equivalence,
government bonds are not being considered net wealth. For example, Judd (1987) demon-
strated this in a theoretical model with distortionary taxes, finite lives, and adjustment
costs. Positive and negative wealth effects can cancel each other or be altogether negli-
gible. Therefore, the fact that tax systems are generally not based on lump-sum taxation
does not invalidate Ricardian equivalence, as, for example, transportation costs do not
invalidate the assumption of perfect competition in many applications. The issue cannot
be settled on theoretical grounds.1However, the empirical evidence is not conclusive. The
recent global financial crisis and the ensuing sovereign debt crises in Europe have sparked
renewed research interest in fiscal policy issues. For example, Hayo and Neumeier (2017)
study Ricardian equivalence with German population survey data, carried out in 2013,
asking about changes in consumption and savings behaviour in response to large increases
in public debt in the period from 2008 to 2012.
In this paper, we use the present value of narrative US tax changes that wereintroduced
in order to reduce inherited government budget deficits and not to finance new government
spending. Therefore, such legislated tax news constitute a switch from bond financing of
deficits to tax financing. If Ricardian equivalence is a good approximation, this switch
should have no statistically significant effects on real GDP and real private consumption.
In addition, another novel feature of our paper is that we study the role of exogenous
deficit-driven tax changes that are implemented with a delay of several quarters, i.e. the
role of fiscal foresight or anticipation for deficit-driven taxes.
Romer and Romer (2010) pioneered the method to construct narrative US tax shocks
for the post-WWII period from a range of government documents.2However, they did not
analyse separately the present-value time-series for ‘deficit-driven’ tax revenue changes
motivated by inherited budget deficits, introduced for reasons unrelated to current macroe-
conomic fluctuations or government spending. We use this subcategory of Romer and
Romer’s (2010) exogenous (as opposed to endogenous) deficit-driventax changes in order
to empirically test Ricardian equivalence. The goal is to shed new light on the controver-
sial issue of whether the economy displays Ricardian equivalence features, which is of
relevance in particular to orienting theoretical models and for understanding some of the
effects of fiscal consolidation.
Section II describes aspects of the data and theory relevant for the study and section III
presents results for the full post-WWII sample and for a sub-sample post-1982:IV.Account-
1See Evans (1991). Furthermore, Evans, Honkapohja and Mitra (2012) recently showedthat rational expectations
are not necessary for Ricardian equivalence to hold and a certain adaptive learning rule instead can produce equiv-
alence. However, learning may happen only gradually and not uniformly across individuals, as demonstrated in a
laboratory experiment by Meissner and Rostam-Afschar (2017).
2These tax schocks have been used by severalothers, such as Favero and Giavazzi (2012) and Mertens and Ravn
(2012). Furthermore, the narrative approach was also employedby Ramey (2011) to study instead the fiscal multiplier
effects of large US governmentmilitary spending shocks. See Ramey (2019) for a recent sur veyon fiscal multipliers.
©2019 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial