Testing the Empirical Relevance of the ‘Saving for a Rainy Day’ Hypothesis in US Metro Areas
| DOI | http://doi.org/10.1111/obes.12310 |
| Published date | 01 December 2019 |
| Date | 01 December 2019 |
| Author | Ragnar Nymoen,André KallåK Anundsen |
1318
©2019 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 81, 6 (2019) 0305–9049
doi: 10.1111/obes.12310
Testing the Empirical Relevance of the ‘Saving for a
Rainy Day’ Hypothesis in US Metro Areas*
Andr´
e Kall ˚
aK Anundsen† and Ragnar Nymoen‡
†Housing Lab, Oslo Metropolitan University, Oslo, Norway
(e-mail: andre-kallak.anundsen@oslomet.no)
‡Department of Economics, University of Oslo, Oslo, Norway
(e-mail: ragnar.nymoen@econ.uio.no)
Abstract
The joint implication of the consumption Euler equation and cointegration between income
and consumption is that savings predict future income declines, the ‘saving for a rainyday’
hypothesis. The empirical relevance of this hypothesis plays a key role in discussions of
fiscal policy multipliers, and it holds under the null that the permanent income hypothesis
is true. We find little support for this hypothesis using time series data for the 100 largest
US Metropolitan Statistical Areas for the period 1980q1–2015q4. Our approach is to test
for cointegration and weak exogeneitybetween income and consumption, and by exploring
the direction of Granger causality between the two time series. We find that income more
often predicts consumption and saving than the converse.We also give evidence that house
price changes played a role in US income and consumption dynamics, before, during and
after the Great Recession.
I. Introduction
Consumer expenditure is by far the largest component of spending in the US econ-
omy, and in most other countries as well. The study of saving and consumption dy-
namics is therefore of great importance for both economic policy analysis and economic
forecasting.
JEL Classification numbers: C22, C32, C51, C52, E21, E62.
*Weare thankful to the editor, Anindya Banerjee, and an anonymous reviewerfor comments that have contributed
to improve the paper considerably. Large parts of this paper were written while Anundsen was working at Norges
Bank. The views expressed are those of the authors and do not necessarily reflect those of Norges Bank. Earlier
versions of the paper was presented at the conference ‘Econometric Modelling in a Rapidly Changing World’ in
Oxford, September 2014, at the 2014 European Winter Meeting of the Econometric Society in Madrid, December
2014, at the 37th Annual Meeting of the Norwegian Economic Association in Bergen, January 2015, at the 23rd
Symposium of the Society for Nonlinear Dynamics and Econometrics in Oslo, March 2015, and at seminars and
workshops in the Central Bank of Hungary, Norges Bank and Statistics Norway. Weare g rateful to the participants
at these events for their comments. We thank Farooq Akram, Bruce Hansen, Veronica Harrington, Tord Krogh and
Asbjørn Rødseth for comments and discussions. All results may be reproduced using an Ox-code that will be made
available on the authors’websites.
Testing the‘saving for a rainy day’ hypothesis 1319
It is well known that the rational expectations permanent income hypothesis (PIH
hereafter) due to Hall (1978) is consistent with unit-root non-stationarity of income and
stationarity of saving, see e.g. Muellbauer and Lattimore (1995, Ch. 3.2). When combined
with the famous theoretical result of Hall (1978), stating that consumption follows a first-
order Markov-process, we obtain the implication that Granger causation runs from lagged
saving to current income and not from saving to consumption. In this paper, we explore
the empirical relevance of these theoretical conjectures by testing the direction of Granger
causality between consumption and income using quarterly time series data for the 100
largest US Metropolitan Statistical Areas (MSAs) over the period 1980q1–2015q4. The
alternative hypothesis is that consumption is not well approximated by the Euler equation.
Instead consumption adjusts (partly) to predictable income changes, and savings is thereby
affected dynamically.
A common ground is represented by the idea that the savings rate may be a stationary
variable, even though there are stochastic trends in the time series of both income and
consumption. This common ground allows the analysis to be held within a vector autore-
gressive (VAR) framework. To account for stochastic trends in income and consumption,
we apply econometric methods that are robust to non-stationarity. Under the hypothesis
that the statistical relationship between consumption and income describes the PIH, a fall
in saving anticipates a future increase in income and a rise in saving anticipates future in-
come declines (see Campbell (1987)).1This also explains why the result has been dubbed
the ‘saving for a rainy day’hypothesis, cf. Attanasio (1999).
In his seminal paper, Campbell (1987) referred to (Granger) causation running from
the savings rate to income growth – and not the other (Keynesian) way around – as the
weak implication of the permanent income hypothesis. Empirically, using aggregate US
data for the period 1953–84, Campbell found that the implication of the PIH for the
direction of Granger causality preserved even if other implications of the PIH fared less
well empirically.2In particular, the Euler equation found little support on aggregate data,
see Flavin (1981), Campbell and Mankiw (1989) among others.
The conclusion that the PIH is at best partly correct, and that it needs to be supplemented
by several factors to account for the many features of consumption dynamics that we are
trying to understand, is well known (see e.g. Romer, 2006; Carroll, 2009; Jappelli and
Pistaferri, 2010; Attanasio and Weber, 2010). Nevertheless, the PIH continues to be one
of the core elements of modern macroeconomics (see e.g. Ljungqvist and Sargent, 2004,
p. 3), it is the centre-piece of macroeconomic DSGE models, and it is essential to the
iterative forward solution of these models, cf. Muellbauer (2016).
One reason why macroeconomists hold on to the principles of PIH, despite the empir-
ical weakness of the consumption Euler equation, may be because it is thought that PIH
nevertheless gets the system properties right: Income and consumption is cointegrated and
income is the most important equilibrating variable, and savings can therefore hold some
predictive power about future income. The objective of this paper is to contribute to the
system assessment of the PIH.
1Campbell showedthis for an infinitely lived consumer with quadratic utility function, equal and constant subjective
discount rates and no credit constraints.
2Campbell (1987, p. 1267).
©2019 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial