Consumption, habit formation, and savings: Evidence from a rural household panel survey

• Author: Ashok K. Mishra,S Nedumaran,Aditya R. Khanal
• Date: 01 February 2019
• Published date: 01 February 2019
• DOI: http://doi.org/10.1111/rode.12536

REGULAR ARTICLE

Consumption, habit formation, and savings:

Evidence from a rural household panel survey

Aditya R. Khanal

1

|

Ashok K. Mishra

2

|

S Nedumaran

3

1

College of Agriculture, Tennessee State

University, Nashville, Tennessee

2

W. P. Carey School of Business,

Arizona State University, Mesa, Arizona

3

International Crops Research Institute

for the Semi-Arid Tropics, Hyderabad,

Telangana, India

Correspondence

Aditya R. Khanal, Department of

Agricultural and Environmental Sciences,

College of Agriculture, Tennessee State

University, 3500 John A. Merit Blvd,

Nashville, TN 37209.

Email: akhanal1@tnstate.edu

Abstract

Protecting consumption from the effects of uninsured risk

is vital for rural farming households, who tend to be poor

and live close to subsistence level. Income uncertainty

and habit formation play important roles in the consump-

tion and savings. Variability in weather conditions has a

strong linkage with variability in agricultural income in

developing countries. This study analyzes consumption

and saving decisions of rural farm households in India.

Using household panel data for 4 years, we estimated

consumption equation accounting for habit formation

under income uncertainty. Our findings suggest an evi-

dence for habit formation among rural households. Addi-

tionally, we found that both annual and seasonal weather

risks significantly influence savings among rural house-

holds. Findings from this study also suggest a robust and

vibrant farm economy and that the nonfarm economy

could contribute to the economic well‐being of rural

farming households.

1

|

INTRODUCTION

The life‐cycle hypothesis implies that individuals plan their consumption and savings over a time

horizon (life) and smooth their consumption in the best possible way. Under life‐cycle or perma-

nent income hypothesis (Friedman, 1957; Modigliani & Brumberg, 1954), the consumption deci-

sion is an intertemporal allocation of resources available during a lifetime; the typical consumer

maximizes utility by choosing an optimal level of resources, in each period, subject to lifetime

budget constraints. A plethora of literature has examined this theory empirically (Flavin, 1981;

Hall, 1978; Hall & Mishkin, 1982; Kazarosian, 1997; Mishra, Uematsu, & Powell, 2012). One of

the challenges to the life‐cycle hypothesis is the prospect of risk and uncertainty associated with

income, which induces a demand for precautionary savings. Under liquidity constraints, however,

DOI: 10.1111/rode.12536

256

|

© 2018 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/rode Rev Dev Econ. 2019;23:256–274.

consumption growth should be sensitive only to increases in income because consumers can

smooth consumption by using savings if they expect future revenues to decrease.

In contrast, consumers with access to credit would follow a conventional life‐cycle model pre-

diction. However, empirical evidence reveals a deviation from this prediction. For example, Garcia,

Lusardi, and Ng (1997) and Shea (1995) found that, for credit‐unconstrained consumers, the level

of consumption is affected by the negative realization in income. A plausible explanation for such

an anomaly is the presence of asymmetric preferences. If preferences exhibit inertia, typically in

the case of habit formation, studies (Deaton, 1992; Meghir & Weber, 1996) show that households

adjust their consumption, but slowly. Only a few studies have examined consumption decisions in

the presence of habit formation and income uncertainty.

1

For example, Alessie and Lusardi (1997)

concluded that consumption depends not only on permanent income and income risk but also on

past consumption—a case of habit formation. However, Guariglia and Rossi (2002) pointed out

that a negative exponential utility function is not a good representation of preferences because it

does not rule out the possibility of negative consumption. Guariglia and Rossi's (2002) presentation

of the utility model is based on the generalization of Weil's (1993) model

2

and accounts for habit

formation. Using this generalized model and panel data from British households, Guariglia and

Rossi (2002) estimated changes in consumption. However, it should be noted that Guariglia and

Rossi (2002) examined mostly urban households in England, and the income risk of urban house-

holds may be less than that of rural households, especially rural families involved in farming as a

way of living. Recall that income from farming is risky and is significantly affected by uncertain-

ties in weather conditions (Mishra & Goodwin, 1997). Finally, weather risk is a primary source of

fluctuations in income for rural households in developing countries (Morduch, 1995).

Interestingly, a few studies have examined the consumption‐savings responses in the presence

of uncertainties in farm income. For example, Kochar (1999) examined consumption‐savings

behavior in the face of crop income shocks in agriculture using longitudinal data. The author found

that rural households may respond to crop income shocks by increasing their market (off‐farm)

hours of work. However, the author stresses that empirical results need to be confirmed with a lar-

ger sample size. Earlier, Paxson (1992) examined farmers’savings behavior as a response to rain-

fall shocks, with an assumption that variable rainfall resulted in variable household income in rural

Thailand. With rainfall shocks as a proxy for income variability, the author's findings suggested

that farmers have a higher propensity to save out of transitory income than farmers who do not

experience rainfall shocks. However, Paxson (1992) stressed that income variability from panel

data would have been a better indicator of income risk. We overcome this limitation in our study.

Finally, we cannot discount the importance of investigating the consumption‐savings behavior

of rural households in developing countries

3

because it has both micro‐and macro‐level implica-

tions. For example, if the behavior is based on saving and non-saving (spending) rates of farmers

in low‐income countries like India, in particular, weather variability also plays a signi ficant role in

the income and consumption behavior. Additionally, in rural areas with low or no irrigation facili-

ties (Hussain & Hanjra, 2004), with a lack of proper storage and processing infrastructures (Abass

et al., 2014), agricultural production depends highly on weather conditions. Further, Mondal et al.

(2014) and Lobell, Schlenker, & Costa‐Roberts (2011) argue that India is one of the most vulnera-

ble countries to future climate changes. Recall that variability in weather conditions has a strong

linkage to variability in agricultural income in rural areas. Protecting consumption from weather's

effects on agricultural profits is therefore vital for farming households living close to subsistence

level—especially in India. If farm families can save and non-save (spend) while adjusting con-

sumption by a one‐to‐one ratio, then policies concerning income variability may be less relevant.

Additionally, farm households behave differently under weather and income uncertainties (Paxson,

KHANAL ET AL.

|

257