Food Availability and Food Security in Rajasthan: An Application of Random Effect Model

72

Food Availability and Food Security in Rajasthan: An Application of Random Effect Model

Narain Sinha*

S.K.Kulshrestha**

Abstract

Growth of Agriculture sector in India in general and in the state of Rajasthan in particular

has been phenomenal in recent years in spite of the fact that about 60% of area in the State is

under desert. Nevertheless in the State most of the major crops are grown in all the districts

albeit with varying proportion. We have chosen for the analysis two important crops one

each from desert and non desert regions namely; pearl millet from among desert crops and

wheat among non-desert crops which together contribute more than 50% in cereal

consumption in the State. Using panel data approach this paper analyzes factors affecting

production of pearl millet and wheat vulnerability over time from 1981 to 2007 covering all

districts. Also in this paper we include effect of infrastructure and other factors which deem

affecting the financial position of the farmers. Gross Cropped Area and energized well are

significant in pearl millet production while rainfall, fertilizer consumption, cropping

intensity, gross irrigated area and energized well, play statistically significant role in

production of wheat. Inclusion of a time invariant factor, i.e., desert dummy shows the effects

of production in desert and non-desert areas. Desert affected the production of pearl millet

positively while that of wheat negatively.

Key words: Food Security, Farmers‟ suicides, Rural Credit, Random Effect Model, Hausman

test, Panel data.

JEL Classification: Q180, A140, C230, C510

Introduction

* Prpfessor of Economics(Retd.) e-mail [email protected]; ** Central University of Haryana, Mahendragarh (India)

mailto:[email protected]

2012 (pp 72-82) Food Availability and Food Security in Rajasthan: An Application of Random Effect Model Arthshodh

VOLUME IV (2) ISSN 2230-7877

Area wise, Rajasthan is the largest State in India in terms of geographical area with its unique

environment conditions, with about 60 % area under desert (known as Thar Desert).The pearl

millet occupies an important position in the state mainly because it is an arid zone crop. The

state is divided into two parts by the Aravali ranges –the western part called the Thar Desert,

spanning twelve districts out of thirty three districts while remaining are the non-desert

districts which lie on the eastern side. During the 11th Plan (i.e. 2007-08 to 2011-12) the

growth performance of agriculture in Rajasthan at 7.4 % was the highest among the Northern

States which include traditionally more developed agriculturally like Punjab (1.6%), Uttar

Pradesh (3.3%) and Haryana (3.3%) (GOI, 2013). We have chosen for the analysis two

important crops one each from desert and non-desert regions namely; wheat and pearl millet

which together contribute more than 70% in cereal consumption in the State former being a

non-desert crop and the later a desert crop. These crops are important from the point of view

of the Food Security bill currently debated in India which promises to provide coarse grain,

which includes inter alia pearl millet, wheat and rice, at less than two US cents per kg at

current exchange rate (August 2013) to the poor. In international market the same coarse

grains are priced at a staggering US$10 and above per kilogram. Crop diversification in

major pearl millet growing states of Haryana, Gujarat and Rajasthan has led to decline in the

production of pearl millet. However, at present Rajasthan occupies first position among the

pearl millet growing states. Food production and its availability is a necessary condition for

food security. India is more or less self-sufficient in cereals but deficient in pulses and

oilseeds. Agriculture being a state subject under the Indian constitution, its overall

performance largely depends on local conditions at the state level. Besides the availability of

inputs, institutional and infrastructure factors also affect the agricultural production.

Availability of rural credit has failed in India during the 1990s, pushing farmers towards

moneylenders. In the absence of any formal direction issued by the Central Government on

lowering the interest rates, crop loans in Rajasthan were being disbursed at the rate of 12 to

13 per cent as. Wide variation in the agriculture performance is observed among the Indian

states. Plan of the paper is as follows. After the review of literature, methodology is

presented in the following section.

Review of Literature



The exploratory studies have shown that many factors determine the agricultural production

and productivity. The technological changes in agriculture in Rajasthan are also associated

with specific institutional developments in banking, co-operative and agricultural extension

including the development of infrastructure and communication media (Adams and Bumb,

1979). Debate on food security in India involves three inter-related aspects viz. availability,

access, and absorption or nutrition and have been examined by Dev and Sharma (2010) who

observed that credit, infrastructure and water management are the major obstacles in food

availability. In the present study we consider these three aspects with special references two

food crops which are important in Rajasthan ,viz., pearl millet and wheat mainly grown

under arid and semi-arid agro-climatic conditions, respectively (Pathan et al., 2008). Millets

is considered because it possesses more food, feed and fodder values and is more

environment friendly and resilient to climate changes. Because of higher content of protein,

fiber, calcium and minerals as compared to wheat and rice, majority of millet crops are

termed as “Nutri-cereals” (Basavraj et al., 2010) .In the present study we employ panel data

analysis at district level. Battese and Coelli (1992) used panel data for the analysis of paddy

productivity in India for wheat production in Pakistan, while Battese and Broca (1997) used

data envelopment analysis. Agricultural crops production and productivity vary across

different agro climatic zones, some crops are more productive in desert areas while others are

in non-desert areas.

Methodology

Besides the traditional factors such as area, fertilizer consumption and irrigation which

determine the agriculture production, in the present paper three obstacles in food availability

as pointed out in Dev and Sharma (2010), are also considered ,viz., access to rural credit,

road length and energized wells in the specification of the model. Agriculture credit plays an

important role in improving agricultural production and productivity in mitigating the

distress of farmers. Several measures for improving agricultural credit flow to farmers have

been taken at the government level. Share of formal financial institutions (commercial banks,

RRBs and cooperatives) accounts for about 66 per cent of the total credit to cultivator

households. No accessibility of farmers to the formal credit system and having no means to

insure themselves against unexpected climatic shocks leaves them vulnerable to the informal



money lenders. Improved agriculture performance is essential for improved food security and

farm livelihoods only if other components in the value-chain such as infrastructure including

transport, storage, processing and marketing facilities for agricultural products are also

developed simultaneously (GOI, 2013), we have considered road length as a proxy for

transport facilities.

The study is based on a balanced panel data for the period 1981-2007, considering all the 27

districts† in Rajasthan. The dependent variable is production of individual crop pearl millet

and wheat separately. There are number of factors which determine the agricultural

production but this paper include factors like rainfall, fertilizer consumption, cropping

intensity, gross irrigated area, gross cropped area, credit of cooperatives societies, road

length and energized well as independent variables. We have included credit from

cooperatives because to investigate the agriculture credit plays an important role in

improving agricultural production, productivity and mitigating the distress of farmers in

India. Government has taken several measures for improving agricultural credit flow to

farmers. Over the years, a significant increase in the share of formal financial institutions

(commercial banks, RRBs and cooperatives) has taken place which accounted about 66 per

cent of the total credit to cultivator households by the early 1990s in the total credit availed

by cultivator households. Credit is undoubtedly the most important factor in the agriculture

development. The regional distribution of agricultural credit by commercial banks, both in

terms of quantum of credit and the number of accounts, has been skewed. There is a

significant concentration in the southern states followed by the northern and western states.

In contrast, the share of the eastern and the north-eastern states has been low. Further, nearly

three quarters of the farmer households still do not have access to the formal credit system

and this leaves them vulnerable to the informal money lender which is the main reason for

increasing incidents of farmers‟ suicide particularly in the Southern States. Road length has

been considered as an indicator of infrastructure because improved performance at farm level

will result in improved food security leading to greater food availability and improved farm

† During period of the study, some administrative changed were made in the State and some new

districts were culled out of existing districts. Necessary adjustments have been made to make the

data consistentt (See Nag et al., 2009).



livelihoods only if other components in the value-chain such as infrastructure including

transport, storage, processing and marketing facilities for agricultural products are also

developed simultaneously. Continuous innovation to improve productivity and

competitiveness of the agriculture sector are necessary to create jobs, generate income and

alleviate rural poverty (GOI, 2013).

Model- The basic framework for the purpose of econometric estimation is based on

= … … … (1)

There are K regressors in Xit and Yit is the dependent variable (Production) where i =

(1,….,27 Districts) and t = (1,….,27 Years) and β‟s are the coefficient for independent

variables. Individual state effects are considered in where contains intercept and a set of

individual group specific variables all of which are constant over time. If are unobservable

but assumed to be correlated with xit then we assume αi= and the model is known as Fixed

Effect Model given below.

= … … … (2)

If the unobserved individual heterogeneity can be assumed to be uncorrelated with included

variables then the model becomes Random Effect Model and is expressed as Random

Effect Model and can be rewritten as

= … … …. (3)

i = 1. . . N and t = 1. . . T

The individual-specific (district) effects αi are assumed to be realizations of iid random

variables with distribution [0,] and the error εit is iid [0,]. The nonrandom scalar intercept α is

added. Random effects can be normalized to have zero mean. The model can alternatively be

viewed as a special case of a random coefficient or varying coefficient model, where only the

intercept coefficient is random so in clearer terminology it is the random intercept model.

The model can be re-expressed as = µ + β +, where the error term uit has two components uit

= αi + εit.

For this reason the random effects model is also called the error components model

(Cameron and Trivedi 2005). The rationale behind random effects model is different from

that of the fixed effects model in that the variation across districts is assumed to be random

and uncorrelated with the independent variables included in the model. Random effects



model assumes that the units‟ error term is not correlated with the predictors which allows

for time-invariant variables to play a role as explanatory variables. Random effect model is

useful to solve the problem of heterogeneity across the districts and over the years.

Empirical Results

For the purpose of econometric estimation, we have considered Xkit represents observation

on kth independent variable in the ith district in the tth year . To be specific

X1 = Rainfall (Cm),

X2 = Fertilizer consumption (kg ha-1),

X3 = Cropping intensity (GCA/NCA*100),

X4 = Gross irrigated area („000 ha),

X5 = Gross cropped area („000 ha),

X6 = Credit of societies - loan advances (Lac Rs),

X7 = Road length (Km) and

X8 = Energized and tube wells (Number)

uit is the error term and E (uit) ~ N (0, σ2).

With the inclusion of dummy variable, which show the effect of desert area on production,

the model becomes,

… (4)

D= 1 if Desert, D= 0 if Non Desert

Considering a balanced panel data for the period 1981-2007, covering all the 27 districts in

Rajasthan, above model has been estimated. The dependent variable is production of

individual crop pearl millet and wheat separately. The coefficients of different factors

affecting the production of wheat and pearl millet as estimated using Random Effect Model

are presented in Table 1.

Table 1. Factors affecting the production of wheat and pearl millet in Rajasthan state.

Crop Wheat Pearl millet

Variables Coefficient p-value Coefficient p-value

Constant -122407 0.00480 -87991.8 0.15908



Rainfall 31.4457 0.01484 18.0928 0.22586

Fertilizer Consumption 587.243 0.00084 -386.68 0.07014

Cropping Intensity 1041.09 0.00296 407.846 0.34737

Gross Irrigated Area 721.194 <0.00001 -29.9509 0.64023

Gross Cropped Area 1.66118 0.89681 76.8728 <0.00001

Credits -0.102735 0.78285 -0.0227399 0.63243

Road -2.85182 0.54421 -2.38354 0.70916

Wells 1.03713 0.00338 3.31802 <0.00001

Dummy -59022.1 0.00529 116320 0.04740

Breusch Pagan test

Ho: Variance of the

unit-specific error = 0

Asymptotic test statistic:

= 506.797

p-value = 3.15546E-112


=3793.08

p-value =0.000

Hausman test

Ho : GLS estimates are

consistent


= 10.0322

p-value = 0.26277


= 11.2206

p-value = 0.189511

Estimated regression coefficients of different factors are affecting the production of wheat

and pearl millet in Rajasthan state can be examined in terms of their coefficients and their

significance is judged by corresponding p-value. Results of panel estimation reveal that

Rainfall is most crucial factor in the production of wheat while its effect on pearl millet is

non-significant, mainly because pearl millet is an arid zone crop and requires little rains in

the dry land regions. Fertilizer consumption (per hectare) has statistically positive effect on

wheat but in case of pearl millet it is statistically non-significant which might be due to more

acreage of pearl millet under dry land conditions. Cropping intensity plays statistically

significant role in production of wheat but it does not affect statistically the pearl millet

production which is an arid crop. Gross irrigated area shows statistically important factor in

production of wheat, but it does not influence the production on pearl millet. Gross cropped

area in Rajasthan has significant role on production of pearl millet because it is an arid crop

but it is non-significant in production of wheat as it depends more on irrigation facilities.



Accessibility to financial credit from the institutions is statistically significant neither in

wheat nor in pearl millet. There are two important implications drawn from this result. First

the credit societies are ineffective in rural areas which may imply that unlike in other Indian

states such as Maharashtra and Southern States, the farmers in Rajasthan are more prudential

financially and this could be one of the reasons why farmers‟ suicides are almost

nonexistence in spite of the fact that in Rajasthan about 61% area is under desert and risk

involved is huge. Road length as a proxy for infrastructure is not contributing significantly in

enhancing the production of wheat and pearl millet. The district effect is uncorrelated with

road infrastructure. Well (energized) and tube wells are effective statistically in increasing the

production of both wheat and pearl millet. Both wheat and pearl millet are affected by desert

dummy variable but its effect in the production of these crops is in opposite direction, i.e.,

wheat in negative while pearl millet in positive direction. The more arid is the region, the

smaller is the production of wheat, but higher is the production of pearl millet .In other

words, it implies that an increase in area under pearl millet in arid zone leads to higher

production of pearl millet.

The results of Hausman test fail to reject the null hypothesis which suggests that Random

Effect Model is good for estimators because they are consistent. The Breausch Pagan test

corresponds to the individual specific random effects and rejects the null hypothesis of zero

variance of unit specification errors assumption of iid errors.

Concluding Remarks

Rajasthan is the only state in India where more than 61% area has scanty rainfall but because

of road and irrigation infrastructure rural farmers are more comfortable financially. Fixed

effect model is found more suitable and district effects are not significant if wheat, which is

mainly no desert crop, and pearl millet, which is a desert crop, is considered. Farmer‟s

suicides are perhaps the reflection of the breakdown of institutional safety net in agriculture

sector. Farmer‟s suicides are in Rajasthan are almost nonexistence because farmers in

Rajasthan do not depend on rural institutional credit as indicated by the credit variable in the

model. Availability of food is not going to be a problem in Rajasthan. Similar study if

replicated for other Indian States may suggest availability of food which is a prerequisite



condition for the success of food security bill.

References

Adams, J. and Balu Bumb (1979).“Determinants of Agricultural Productivity in Rajasthan,

India: The Impact of Inputs, Technology, and Context on Land Productivity”, Economic

Development and Cultural Change, 27(4), 705-22.

Basavaraj G, Parthasarathy Rao P, Bhagavatula S and W. Ahmed (2010). “Availability and

Utilization of Pearl Millet in India”, Journal of SAT Agricultural Research, 8.erformance

Battese, G. and Broca, S. (1997). “Functional Forms of Stochastic Frontier Production

Functions and Models for Technical Inefficiency Effects: A Comparative Study for Wheat

Farmers in Pakistan”, Journal of Productivity Analysis, 8: 395-414.

Cameron, A. C. and P K. Trivedi, (2005). Micro econometrics Methods and Applications,

Cambridge University Press, Ed. (1): 756-757.

Dev, S. Mahendra and A.N.Sharma. (2010). Food Security in India: Performance,

Challenges and Policies, Oxfam India working papers series September.

GOI (2013). State of Indian Agriculture 2012-13, Ministry of Agriculture, Department of

Agriculture and Cooperation, Directorate of Economics and Statistics, Government of India,

New Delhi (India) .

Nag, A.K., S.K. Kulshrestha and Narain Sinha (2009).”Growth of Cropping Intensity in

Rajasthan: District wise Variation”, Annals of Experimental Agriculture & Allied Sciences,

4, (1&2).

Pathan, A.R.K., Gill, O.P. and A.K Nag. (2008). “Productivity Potential and Economics of

Various Crop Sequences under Diversification in Semi-arid Eastern Plain zone of Rajasthan”,

http://econpapers.repec.org/article/ucpecdecc/





Annals of Plant and Soil Research, 10 (1):19-22.

Food Availability and Food Security in Rajasthan: An Application of Random Effect Model

Documents