Estimating Excess Sensitivity and Habit Persistence in Consumption Using Greenbook Forecasts
| DOI | http://doi.org/10.1111/obes.12333 |
| Author | Vipul Bhatt,Hardik Marfatia,N. Kundan Kishor |
| Date | 01 April 2020 |
| Published date | 01 April 2020 |
257
©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.12333
Estimating Excess Sensitivity and Habit Persistence in
Consumption Using Greenbook Forecasts
Vipul Bhatt,*N. Kundan Kishor,†Hardik Marfatia‡
*Department of Economics, James Madison University, 421, Bluestone Dr., MSC
0204, Harrisonburg, VA 22807, USA (e-mail: bhattvx@jmu.edu.)
†Department of Economics, University of Wisconsin-Milwaukee, Bolton Hall, Milwaukee, WI
53201, USA (e-mail: kishor@uwm.edu.)
‡Department of Economics, Northeastern Illinois University, 5500 N St Louis Ave, BBH
344G Chicago, IL 60625, USA (e-mail: h-marfatia@neiu.edu.)
Abstract
In this paper, werevisit the issue of excess sensitivity of consumption to income and address
the weak instrument problem that is well documented in this literature. Using quarterly
data for the US economy, we first highlight the weak instrument problem by showing
that the use of conventional instruments tends to overestimate the share of rule-of-thumb
consumers. To address this weak instrument problem, we propose a newinstr ument for en-
dogenous disposable income growth in the consumption function, namely, the Greenbook
forecast of real disposable income growth. We show that this instrument encompasses the
information contained in the conventional set of instruments, and is a superior predictor of
income growth. We find that using our proposed instrument ameliorates the weak instru-
ment problem and provides a much smaller estimate for the rule-of-thumb consumers. We
also extend our empirical framework to allow for habit persistence and provide an estimate
for this important parameter of the consumption function. Finally, we use a time-varying
specification of consumption function that allows for endogenous regressors, and docu-
ment a decline in the share of rule-of-thumb consumers and a rise in the habit-persistence
parameter in the US over our sample period.
I. Introduction
The standard intertemporal consumption model of the household predicts that the changes
in marginal utility from consumption cannot be predicted based on past information
(Hall, 1978). There is a large literature on empirically testing this property of the stan-
dard model using aggregate as well as household level consumption data.1Campbell and
Mankiw (1989) provided a framework that extends the standard model of consumption by
JEL Classification numbers: E21, C22, C26, C53.
1See Flavin (1981); Zeldes (1989); Jaeger (1992); Shea (1995); Parker (1999); Jappelli and Pistaferri (2000)
among others.
258 Bulletin
incorporating consumers who use a fraction of their current income for consumption as a
rule-of-thumb. Their framework allows for testing the excess sensitivity of consumption
to current income changes by estimating the economic and statistical significance of the
estimated fraction of rule-of-thumb consumers under the assumption of constant real in-
terest rate.2An important methodological constraint imposed in the implementation of this
framework is the availability of instruments for the potentially endogenous income growth
variable in the consumption function. Many studies have implemented the framework of
Campbell and Mankiw (1989) and have used lagged valuesof income growth, consumption
growth, interest rates, inflation, stock returns, etc as instruments.3
A valid instrument must be both exogenous and relevant.Although the literature in this
area has paid substantial attention to the issue of exogeneity, the issue of weak instruments
remains a key problem (Weber, 2000; Kiley, 2010). Given very noisy feature of quarterly
disposable income growth, it is often difficult to predict this variable,and hence most of the
instruments used in the existing literature suffer from the weak instrument problem (Weber,
2000; Kiley, 2010). An important contribution of our paper is to providean instr ument that
is both exogenous as well as strongly correlated with income growth.
In this paper, we propose to use the Greenbook (GB, henceforth) forecast of real dis-
posable income growth as an instrument for income growth. GB forecast of disposable
income growth is a natural candidate as it is both exogenous, and more importantly, it is
directly relevant in terms of predicting contemporaneous disposable income growth. First,
the exogeneity is ensured by the fact that these forecasts are performed in the past.4Be-
cause these forecasts are made in the past, any unanticipated shock to contemporaneous
consumption growth will be uncorrelated with the GB forecast. More importantly, these
forecasts have strong predictive power for future income growth, as they are based on so-
phisticated models and thus are able to aggregate the information about future movements
in real disposable income growth. Many studies in the forecasting literature have docu-
mented the usefulness of these forecasts in terms of providing valuable information about
the future movements in important macroeconomic variables (for example, see Romer
and Romer, 2000; Sims, 2002; Faust and Wright, 2009, among others). We find that the
one-period ahead GB forecast of real disposable income growth is a strong predictor of
the disposable income growth. Further, we find that the GB forecast encompasses the in-
formation contained in the conventional set of instruments used in the existing literature,
and in this sense provides a superior forecast of income growth.5
2The benchmark specification used in our analysis is derived from a consumer Euler equation based on a constant
real interest rate assumption that equals the subjective discount rate. Under an alternative assumption about the real
interest rate the form of the consumption Euler equation and the resulting specification would be different. Section
IV provides further details on this specification issue.
3Weber (2000) providesan excellent survey of empirical studies that provide an estimate of the fraction of rule-
of-thumb consumers.
4The data on GB forecasts are available at: https://www.philadelphiafed.org/research-and-data/real-time-center/
greenbook-data/pdf-data-set.
5One can argue that the superiority of the Greenbook forecasts in predicted future income could be due to access
to information that is not readily available to public. Such information asymmsectionetry between household and
the staff at the Federal Reserve Board can have implication for the exogeneity argument we make here. However,
there is evidence that the superiority of the GB forecasts is not due to this asymmetry in the information set between
the households and the staff at the Federal Reserve Board. In section ‘The Superiority of the Greenbook Forecastof
Disposable Income Growth’we provide a more detailed exposition of this issue.
©2019 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial