Climate Change, Rainfall Variability,

and Adaptation through Irrigation:

Evidence from Indian Agriculture ∗
†
Ram Mukul Fishman
November 28, 2011

∗ I thank Wolfram Schlenker, Upmanu Lall, Douglass Almond, Bernard Salanie, Jeffrey

Sachs, Ruth Defries, Solomon Hsiang, Chandra Kiran Krishnamurti, Naresh Divedani, Shama
Parveen, Tobias Siegfried, Jesse Anttila-Hughes, Gordon McCord, Kapil Narula, and the
participants of the Sustainable Development Seminar in Columbia University. I thank the
Columbia Water Center for Technical and Financial support .
† Harvard Kennedy School. email: ram [email protected]

1
 Abstract

Alongside warmer temperatures, climate change is expected to also

result in increasingly irregular and extreme precipitation patterns, and

both of these trends can have potentially severe consequences for food

production. Rain fed agriculture in developing countries is expected to be

particularly vulnerable to these changes, leading many to argue that the

expansion of irrigation is one of the best feasible and proven adaptation

strategies for attenuating climate change’s impacts. I combine detailed

data on daily weather, irrigation and crop yields across India over the last

four decades to evaluate the adaptive resilience achieved by the expansion

of irrigation in the country during this time period. I begin by providing

the first estimate of the impact of increased intra-seasonal rainfall variabil-

ity on yields, and show that its magnitude and significance rival that of

total seasonal rainfall. I then provide evidence that irrigated yields are less

vulnerable to the impacts of rainfall irregularities, and estimate that the

expansion of irrigation to cover all the rice cultivating farmland in India

could eliminate 90% of the projected precipitation-driven climate change

impact on rice yields. I also provide evidence suggesting that farmers ex-

pand irrigation in response to adverse rainfall conditions, in accordance

with a simple model reflecting the flat rate pricing of irrigation water in

India. In contrast to its attenuation of precipitation shocks, I find that

irrigation mitigates the impact of increased seasonal heat exposure only

partially (in the case of wheat) if at all (in the case of rice). Because

temperature increases turn out to drive most of the projected impact on

rice yields, these results together suggest the expansion of irrigation, as

actually practiced by farmers in India, has at best limited potential to

adapt agriculture to the changing climate.

2
1 Introduction

1.1 Summary of Findings

Projecting the consequences of climate change for economic productivity is es-

sential for a cost benefit analysis of possible mitigation strategies. However,

while the understanding of these impacts on current modes of production is

progressing, there is still a paucity of evidence on the feasibility and costs of

various adaptation strategies that can shift production to less vulnerable modes,

making any such cost benefit analysis incomplete (Tol [2009]).

Agriculture is a case in point. Perhaps more than any other sector of the

economy, agriculture has always been and has remained vulnerable to chang-

ing weather conditions, because of the strong effect temperatures and moisture

availability have on the key processes involved in crop growth (Rosenzweig and

Hillel [2008]). As a result, some projections of the agricultural impacts of in-

creasing temperatures and altered precipitation patterns associated with climate

change are alarmingly large (depending on the scenario and the region) (Parry

[2007]). Given the sensitivity of food security and world food prices to supply

shocks, this is a major concern. Although various potential strategies have been

proposed to protect crop yields in the changing climate (Parry [2007],Howden

et al. [2007],Mendelsohn and Dinar [1999]), there is little evidence to suggest how

effective they will be when deployed on a large scale in actual field conditions.

Such an analysis is especially important in developing countries. Not only

are the greatest declines in crop yields projected to occur in these countries

(Parry [2007]), but because a large fraction of the population is still dependent

on agriculture, often as small holder subsistence farmers, the welfare conse-

quences can be even more severe. For this same reason, market imperfections

and behavioral factors mean that evaluating the potential and large scale ef-

fectiveness of agricultural adaptation technologies, purely on the basis of their

3
performance in controlled plot experiments, is insufficient.

In this paper, I approach this problem using detailed weather and agricul-

tural data spanning three decades from all across India. India produces food

for about a sixth of the world’s population but it’s agricultural and economic

fortunes have always been dependent on the vagaries of the monsoon rainfall

(Krishna Kumar et al. [2004]) and the production of major crops in the country

is still highly variable.1

Many studies have demonstrated the impact of total annual rainfall on crop

production in India (Krishna Kumar et al. [2004], Auffhammer et al. [2006],

Guiteras [2008]) and globally (Lobell et al. [2011]). However, it is widely believed

that crop yields are also highly sensitive to irregular precipitation events like dry

spells and intense downpours. Moreover, there is a general expectation in the

climate science literature that global warming will intensify the hydrological

cycle, and result in more uneven intra-seasonal distribution of precipitation

(Trenberth et al. [2005], Hennessy et al. [1997]). The IPCC’s fourth assessment
2
report concludes that

“...in a warmer world, precipitation tends to be concentrated into more in-

tense events, with longer periods of little precipitation in between. Therefore,

intense and heavy downpours would be interspersed with longer relatively dry

periods...

The results in this paper provide what are, to the best of my knowledge,

first estimates of the actual impact of intra-seasonal variability on crop yields.

They indicate that changes in the intra-seasonal distribution of daily rainfall

can have as large and statistically significant impact on yields as does total

rainfall. For example, I estimate that each additional rainless day reduces rice

yields in the rainy season by 0.4% (keeping total rainfall fixed). In comparison,
1 for example, according to the data set I use, in a median rice growing Indian district, the

growth rate of rice production varies over time by some 25% around a mean growth of 1.6%
2 ‘‘http://www.ipcc.ch/publications_and_data/ar4/wg1/en/faq-10-1.html’’

4
the impact of the associated decrease in total rainfall (about 10mm per day on

average) would reduce yields by only about 0.15%, and the associated average

increase in temperature degree days (about 1 degree-day lower in a rainy day, on

average) would reduce yields by 0.09%. In fact, I calculate that the incorporation

of the frequency of rainless days into climate change projections in India (for

example Krishna Kumar et al. [2003]) overturns the positive projected impact

of increased precipitation in the country, making the net effect of the changes

in precipitation patterns negative.

In terms of the impact of rising temperatures, my results are consistent

with the findings of Guiteras [2008]. Even though the impact of typical annual

fluctuations is smaller or comparable to that of fluctuations in precipitation

(for example, I find that a one standard deviation increase in the number of

rainless days decreases the yields of most rainy season crops by some 4%-6%,

whereas a one standard deviation increase in seasonal temperatures decreases

yields of only some crops by about 2%-3%), the projected relative magnitude

of temperature increase is much higher, so that most of the negative impacts of

3
climate change are expected to be related to rising temperatures.

For millennia, irrigation has been one of the most commonly practiced ways

to buffer crop yields from weather anomalies in many parts of the world. Irri-

gation can supply an alternative source of water to substitute deficient rainfall

(in terms of both quantity and timing) and to supply the additional evapo-

transpirative water demand of crops that are exposed to increased temper-

atures. An expansion of irrigation coverage is therefore an often mentioned

means of adaptation to climate variability and change (Howden et al. [2007],

Parry [2007]). For example, crop simulation suggest that the expansion of ir-

rigation can potentially reduce the agricultural losses resulting climate change,
3 But for this same reason, as pointed out by Guiteras [2008], these estimates involve
projecting the results outside of the sample used to derive them and should be interpreted
with caution.

5
especially in developing countries (Rosenzweig and Parry [1994]). Estimating

how effectively irrigation, as actually practiced in farmers’ fields, has been able

to buffer crop yields from weather fluctuations is therefore an important step

towards better assessing its actual potential as an adaptation strategy.

In India especially, the high variability and uneven temporal distribution of

rainfall within and across years has made irrigation essential for agricultural pro-

ductivity and food security, and its expansion has been one of the central pillars

of India’s agricultural development policy for many decades (Shah [2008])(fig-

ure 1). India is now the world’s largest consumer of groundwater, with over ten

million wells, and the world’s third largest dam builder, with over 4,000 large

dams.

However, since irrigation is often accompanied by the use of different crop

varieties, farming practices, and potentially other forms of risk preserving be-

havior, because it relies on water resources that are themselves dependent on

rainfall, and because of pervasive inefficiencies in water management and dis-

tribution, its performance in the field and in controlled experiment may differ,

and scholars actually debate its effectiveness in stabilizing agricultural growth

in India (Moench [1992], Hanumantha Rao et al. [1988]) 4 .

The analysis in this paper resolves some of this debate. The results suggest

that, first, irrigated yields tend to be higher: irrigated rice yields are 20% higher

and irrigated wheat yields are 34% then their rainfed counterparts. Second, they

suggest that irrigation is highly effective in protecting rainy season crops from

rainfall irregularities. For example, if irrigation would expand to fully cover

4 For example, Moench (Moench [1992]) echoes the common notion that:“In addition to
serving as a regular source of supply, ground water plays a critical buffering role during periods
of drought when surface flows are limited and unreliable.”In contrast, Rao (Hanumantha Rao
et al. [1988]), in one of the few existing works on the issue, uses state level crop statistics,
and argues that“The instability in agricultural production has increased on account of rise
in the sensitivity of output to variations in rainfall [due to the] high complementarity of new
seed-fertilizer technology with wateralthough area under irrigation has increasedthis irrigation
is itself dependent on rainfallthe uncertainty of irrigation has been increasing in the recent
period”

6
 2.5 2.5

Annual Rainfall Anomalies (in Stan. Dev)

Yield, Cultivated Area (relative to 1970)

2
2 1.5
Fraction of Area Irrigated

1
1.5 0.5
0
1 -0.5
-1
0.5 -1.5
-2
0 -2.5
1969 1974 1979 1984 1989 1994 1999 2004
Year
Annual Rainfall Cultivated Area

Yield Fraction of Area Irrigated

Figure 1: Time series of all india total food grain yield (dark green) and area
(light green), relative to 1970 levels, the fraction of area irrigated (blue line),
and the standardized deviation of annual rainfall (blue bars, right axis). The
correlation between yields, and to a lesser degree area, with total rainfall may
have declined somewhat in recent years, and while cultivated area has stopped
increasing, irrigation is continuing to expand.

7
all of the area cropped with rice, the impact on rice yields of the projected

increase in the numbers of rainless days would be reduced by 90%. They also

suggest irrigation can reduce the impacts of decreases in total rainfall on rainy

season crops, but the results are less robust (In any case, since total rainfall

is projected to increase in India, the associated impacts on yields are actually

positive, but modest, to begin with). Furthermore, in contrast to the ‘dumb

farmer’ assumption that no adaptation will take place (Patt et al. [2009], Tol

et al. [1998]), I find evidence suggesting that farmers do expand irrigated areas

if it can help respond to rainfall deficiencies, in accordance with a simple model

that reflects the flat rate pricing of irrigation water in India.

However, my estimates also suggest that while irrigation is highly effective

in protecting crop yields from the negative consequences of altered precipitation

patterns, it seems to have a limited capacity (for dry season wheat), if any (for

rainy season rice), to counteract the impact of increased temperatures. Since

I also show that the main impact on yields in due to rising temperatures, the

scope for irrigation as an effective adaptation strategy seems limited, even if we

ignore the impact of climate change on water resources themselves (vrsmarty

et al. [2000]).

1.2 Approach

My empirical strategy utilizes random year to year fluctuations in weather con-

ditions to statistically identify their impact on the yields of most major crops,

following the panel approach used to estimtate climate change impacts in the

U.S. (Deschenes and Greenstone [2007], Schlenker and Roberts [2009]) and more

recently in India (Guiteras [2008]). However, in addition to including measures

of intra-seasonal variability in the weather variables, I also extend the method

by studying temporal and spatial heterogeneity in the response of crop yields

8
and its correlation with irrigation coverage.

Projecting the impacts of the expansion of irrigation to currently rainfed

areas can be misleading if irrigation coverage is correlated with other, unob-

served factors that are the ones really driving the reduction in the vulnerability

of crops to weather fluctuations. While the data available to me does not allow

me to completely overrule this possibility, several lines of evidence reduce its

likelihood. First, I provide evidence to show that the effect of irrigation on crop

vulnerability to rainfall variability and dry season heat exposure is independent

of arbitrary unobserved factors that are determined by state dependent time

trends (which precludes regional factors, agricultural policies and other institu-

tional factors) or by time independent district attributes (such as soil quality).

In other words, the increase in irrigation within districts over time correlates

with reductions in crop sensitivity, controlling for unobservable constant district

attributes.

Second, the different impacts I find irrigation has on the vulnerability of dif-

ferent crops in different seasons to different types of weather shocks are in broad

qualitative agreement with phenological predictions. It is natural to expect ir-

rigation to be highly effective in substituting for deficiencies in soil moisture

resulting from rainfall failure. However, the impacts of high temperatures on

crop growth occur through a host of channels and are not limited to increased

water demands. For example, accumulated heat exposure affects the maturation

period and therefore the final yield of the crop, irrespective of water availability.

Irrigation water cannot therefore be considered a perfect substitute for high tem-

peratures. Observations of yields in controlled conditions in research stations

in Asia show that irrigated rice yields are significantly affected by temperatures

(Peng et al. [2004], Welch et al. [2010]), whereas Guoju et al. [2005] shows that

irrigating spring wheat in arid regions of China can offset some of the impacts

9
of increased temperatures, in broad agreement with my findings.

Third, if irrigation is more prevalent where crops are, to begin with, more

vulnerable to weather and hence where the returns to irrigation are higher, then

my estimates for its impact are conservative. For example, I offer evidence, both

dynamic and cross sectional, that irrigation is more prevalent where and when

rainfall tends to be more deficient, and therefore, where crops are likely to be

more sensitive.

Methodologically, my approach attempts to bridge some of the gap between

two prominent approaches to climate change agricultural impact projection: the

panel approach and the cross sectional (Ricardian) approach, used to estimate

the impacts of climate change on U.S. agriculture by Mendelsohn et al. [1994]

(and in India by Krishna Kumar et al. [2004], Dinar [1998], Kumar and Parikh

[1998], Sanghi and Mendelsohn [2008], Kumar and Parikh [2001]).

In the cross sectional approach, average land value or farm profits are re-

gressed on long-term climate variables like mean temperature and precipitation

to project the impacts of long-term changes in those variables. A potential

weakness of this approach is that is may fail to disentangle climate variables

from unobserved confounding factors that impact farm profits, a hurdle that is

partially overcome in the panel approach by controlling for unobserved, time in-

dependent spatial fixed effects and basing the estimates on year to year weather

fluctuations within locations. On the other hand, the Ricardian approach may

be more likely to take into account the long-term adaptation possibilities avail-

able for farmers, whereas the panel approach relies on short term changes, so it

probably over estimates the real long term impacts.

This paper uses the panel approach, but explicitly incorporates data on

irrigation coverage, a prominent adaptation method, in the analysis in order

to test whether it explains the geographical heterogeneity in crops’ response

10
to weather. Previous papers addressed the issue of irrigation within the cross

sectional approach: Schlenker et al. [2005] shows that the response in irrigated

and non-irrigated areas in the U.S. is different and runs the analysis of land

values separately in the two sub samples; Kurukulasuriya and Mendelsohn [2007]

also finds differences in the response of yields in a cross sectional analysis in

Africa (the authors also address the potential bias due to endogenous irrigation

use, but find the results are not significantly altered). A notable dynamic result

is provided by (Duflo and Pande [2007]), who show that downstream from a

dam, the sensitivity of agricultural production to total rainfall is reduced after

the dam’s construction.

The remainder of the paper is organized as follows. Section 2 describes the

data and some broad features of agriculture and irrigation in India. Section 3

analyzes the overall impact of weather fluctuations on crop yields. Section 4 in-

corporates irrigation in the analysis, and estimate the degree to which irrigation

is effective in reducing the impact of weather fluctuations on yields, as well as

the degree to which irrigated areas are expanded in response to weather fluctu-

ations. Section 5 presents a simple model of irrigation that bears out the main

empirical findings of section 4. Section 6 uses these estimates to project the

consequences of climate change for rice and wheat production in a few stylized

climate and irrigation scenarios. Section 7 concludes.

2 Data

2.1 Agricultural Data

The Indian agricultural calendar is organized around the monsoon, during which

most of India’s rainfall occurs (figure 2). The two main growing seasons are the

rainy season (Kharif, June to September) and the drier season that follows

11
Figure 2: A sketch of the Indian agricultural Calendar. Cropping decisions
(dashed lines) and crop growth (solid lines) for the Rainy and Dry season are
shown along a time line superimposed with sample daily precipitation (above)
and its accumulation (below), taken from actual rainfall in one of the districts
in the year 2000.

it (Rabi, roughly October to February) 5 . Crops in both seasons rely on the

monsoon rainfall to satisfy their water requirements, either directly, in the rainy

season, or by drawing on accumulated rainwater, be it as soil moisture or in

artificial storage, in the dry season.

Food production is dominated by rice cultivation in the rainy season and

on rice and wheat cultivation in the dry season. These two crops are relatively

water intensive in comparison with the other crops in their respective seasons

(especially rice which is mostly cultivated in flooded plots). They were also the

main focus of the green revolution, which consisted of the use of high yielding

varieties and increased application of irrigation water and fertilizer. As a result,

they have the largest coverage by irrigation compared to other crops (figure 5).

While my analysis is focused on rice and wheat, which also have the largest

number of observations, I also look at other major crops from the coarse cereal

and pulse groups, traditional crops that are common to rainfed agriculture in the

semi arid tropics, but have lagged far behind rice and wheat in terms of research,

development and intensification. These include maize, barley, groundnuts, two

coarse cereals, Sorghum (Jowar) and Pearl Millet (Bajri) and two pulses, Pigeon
5 In some areas, a third summer crop is also grown, but I have no data on it, and it entirely

irrigated

12
 Rainy Season (Kharif) Dry Season (Rabi)
Arhar Rice
Gram

Maize
Rice Maize
Jowar Jowar Wheat

Bajra
Coarse Cereals
Pulses

Figure 3: The distribution of area cultivated with Foodgrains in India, by crop

(average of 2003-2008). Crops analyzed in this paper are labeled, and organized
into rice, wheat, coarse cereals and pulses. Source: Agricultural Statistics at a
Glance, Directorate of Economics and Statistics, Dept. of Agriculture, Govt. of
India.

Pea (also known as tur, arhar or red gram) and Chickpea(Gram), as well as the

water intensive cotton and sugarcane. Together, these crops cover the great

majority of cultivated area in India (figure 3).

Data on the yield, production and area (production divided by yield) of these

crops is obtained from the Indian Harvest data set, produced by the Center for

the Monitoring of the Indian Economy (CMIE). Observations are reported on

the levels of districts (Indian sub-state administrative units, of which there are

about 580 in the sample) in 18 states for the years 1970-2004. The data also

provides figures on irrigated areas.

2.2 Weather Data

I use gridded daily precipitation and temperature figures produced by the In-

dian Meteorological Department through a re-analysis (with physical climate

models) of observational data (Srivastava et al. [2009], Rajeevan et al. [2005]).

The gridded data is converted to district level data, in order to match it with

13
agricultural data, by averaging (area weighted in the case of precipitation) over

grid points falling within a given district.

To capture the impact of temperatures, I use growing degree days, a measure

of heat exposure used to predict crop yield in phenology (and shown by Schlenker

et al. [2006] to provide a better fit of observed yields), and defined by

X
DDS = D(Tavg,d ) (1)
d

where Tavg,d is the average daily temperature in day d, the summation is

over the days of the growing season and D(T ) reflects the ability of crops to

absorb heat in the temperature range from 8 ◦ C to 32 ◦ C degrees, i.e.


if T ≤ 8◦ C




 0,


D(T ) = T − 8, if 8◦ C < T ≤ 32◦ C (2)




if T ≥ 32◦ C

24,


Even though nonlinear effects have been detected by using discrete tempera-

ture bins in the U.S. context (Schlenker and Roberts [2009]), Guiteras [2008], in

his analysis of Indian agriculture and temperature, does not find strong differ-

ence in climate predictions when using the two methods. I therefore only include

seasonal degree days in each of the two seasons in my regressions, depending on

the crop.

From the daily precipitation figures, I construct ten summary statistics.

The first five include total monsoon rainfall and monthly rainfall for each of the

monsoon months, namely June, July August and September. In addition, I use

measures of the dispersion of rainfall within the monsoon season that are used in

the climate science literature to capture increased rainfall variability in climate

change simulations. These include the frequency of rainy days (with precipita-

14
tion over 0.1 mm, May [2004]), the duration of the longest dry spell (Tebaldi

et al. [2006]), and the shape parameter of the gamma distribution which is fitted

to the distribution of daily rainfall within the rainy days of the season. Gamma

distributions are commonly used to describe the distribution of observed and

model-generated daily rainfall (Gregory and Mitchell [1995],Stephenson et al.

[1999]), and the shape parameter measures the skewness of the distribution:

higher values of the shape parameter indicate a more even distribution of rain-

fall between the season’s rainy days (dry days are not incorporated into the

gamma fit). 6 .

These weather parameters are all correlated to some degree. On average, an

increase of one wet day tends to increase total rainfall by 10mm, reduce degree

days by about 1 degree per day, and increase the shape coefficient by 0.003

(panel regression with fixed districts effects).

3 Weather Fluctuations and Crop Yields

3.1 Empirical Strategy

My empirical strategy follows the panel approach (Deschenes and Greenstone

[2007],Schlenker and Roberts [2009],Guiteras [2008]) to estimate the propor-

tional impact on annual changes in weather patterns on crop yields. The basic

regression I run is

log Ysdt = v · Wdt + fs (t) + λt + pd + sdt (3)

Here, Ydst is the yield of a given crop in district d, state s and year t; Wdt

is a vector of weather variables including total Monsoon rainfall, seasonal de-

6 Since I control for total seasonal rainfall, I do not incorporate rainfall intensity (total

rainfall divided by the number of rainy days) or the scale parameter of the fitted gamma
distribution (which is related to the rainfall intensity), both of which are also often used in
the climate science literature

15
gree days (DDS), calculated in either the dry or rainy season, depending on

the crop in question, and variables describing the intra-seasonal distribution of

daily rainfall, described in section 2.2: the number of rainy days, the duration of

the longest dry spell, and the shape parameter of the fitted gamma distribution.

The estimated coefficients vector v describes the vulnerabilities of crop yields to

these weather variables. The regressions also include controls for unobservable,

time invariant district attributes pd that may affect yields, such as soil quality

or other geographical attributes; Quadratic time trends fs (t) that reflect tech-

nological progress and productivity gains, which I allow to differ from state to

state because of the large variance in agricultural performance across India; and

aggregate unobservable year effects λt , to separate annual country-wide factors,

such as climatic fluctuations (e.g. ENSO) from changes in the weather patterns

that I include in the regressions. Because of potential spatial and serial corre-

lation in both weather outcomes and yields, I allow the unobserved errors sdt

to be correlated across years and districts in the same state.

I will mostly report regression results estimated with the use of standardized

weather variables 7 , which will facilitate an easier interpretation and comparison

of the estimates across crops. I also run the regressions using absolute values of

these variable, and the pattern of the results is unchanged.

3.2 Results

3.2.1 Regression Estimates

Regression results are reported in table 1 for rainy season crops and in table 2

for dry season crops.

7 For a variable Xdt that varies with district d and year t, the standardized form I use is
Xdt − µd
(4)
σd
where µd is the district specific mean (taken over time) and σd is the district specific standard
deviation.

16
 Five of the eight rainy season crops are affected by total rainfall in a sta-

tistically significant manner (p < 0.1), and all of them in a negative way, with

a loss of about 3%-5% per standard deviation reduction in total rainfall. Most

crops are estimated to suffer from increased heat (degree days), but of these,

only that on rice yields is statistically significant (at 3% per standard deviation

increase). Maize, Sorghum and Sugarcane are the exceptions, with Sorghum

showing a statistically significant positive response of 2%.

In the rainy season, among the intra-seasonal distributional measures, the

frequency of rainy days emerges as the dominant one, in terms of both the size of

its impact, its statistical significance and its consistency across crops, not only

in comparison to other distributional measures, but even in comparison to the

traditional measure of total rainfall. All crops show a statistically significant

negative response to a reduction in the number of rainy days, even when total

rainfall is controlled for, with impact per standard deviation in the range of

4%-8%. When the number of rainy days is included, there is no statistically

significant response to other measures of intra-seasonal distribution.

This is also the case in the dry season. However, in the dry season, most

crops are affected by rainfall totals more strongly and significantly than by

the frequency of rainy days. In addition, dry season degree days has a more

consistent and negative impact on most crops, and very large for some (chickpea

and groundnut). The impact on wheat, while modestly negative, does not have

high statistical significance (p = 0.15) but wheat is a highly irrigated crop

(about 80% of the area cropped with wheat is irrigated) and we will see later

that controlling for irrigation will reveal a statistically significant effect.

The difference in the relative impact of the rainfall distribution and its sea-

sonal total on rainy and dry season crops is natural. Rainy season crops experi-

ence the flow of irrigation directly, and are therefore sensitive to its distribution

17
within the season. Dry season crops, in contrast, rely on the accumulation of

rainy season precipitation in either natural (as soil moisture or groundwater)

or in or artificial (reservoirs, tanks) storage. The efficiency with which rainfall

is captured in storage is likely to be higher when the rainfall is more evenly

distributed, but this effect is probably secondary to the direct effect of total

rainfall amount.

3.2.2 Non Parametric Fits

Figure (4) presents non parametric plots (kernel regressions) for the impact of

total rainfall, degree days and the frequency of rainy days on the yields of rainy

season rice and dry season wheat. The plot shows clearly the lack of impact

of rainy day frequency in the dry season, and the familiar concave response to

total rainfall in both seasons.

3.2.3 Robustness Checks

Tables 3 and 4 display the results of sensitivity checks for rainy season rice and

dry season wheat, respectively. Column 2 repeats the basic regression. Column 3

removes year effects. In column 4 I control for the previous year’s yield. Because

this can introduce a bias in panel estimation when errors are serially correlated,

I also use the Arellano-Bond estimator (Arellano and Bond [1991]) in column 5.

In Column 6 I replace district fixed effects with a smaller set of geographical fixed

effects (districts have split up over time, and in the basic regressions I include a

separate fixed effect for a district before and after it splits). In Column 7 I run

the regression on a sub sample of that of column 6 in which each location has

at least 25 years of observation, in order to address the concern that missing

observations introduce a bias in the estimation. The estimates of all weather

coefficients are stable across these alternative models. The only exception is the

impact of degree days on wheat yields, which is lower when year fixed effects

18
Figure 4: Local polynomial fits (kernel regressions), with 95% confidence in-
tervals (errors not spatially correlated), of rainy season anomalies of rice (top)
and wheat (bottom) yields (residuals from regressions on year effects, state spe-
cific quadratic time trends, other weather variables and district fixed effects) on
rainy day frequency (left), total rainfall (center) and seasonal degree days (right)
anomalies. Histograms display the distributions of the observed anomalies.

19
are omitted, but is more significant in some of the specifications. In addition,

I also re-estimate the basic regressions while omitting each of the states from

the sample (results not shown). The coefficient estimates remain stable and

significant.

4 Irrigation Usage and its Impact on Crop Yields

In this section, I examine evidence for the impact of irrigation on crop yields

and their sensitivity to weather conditions, including seasonal degree days, total

rainfall and the frequency of rainy days. Having detected no statistically sig-

nificant response to the other measures of intra-seasonal variability, I will drop

them from the analysis for simplicity.

4.1 Empirical Strategy

Unfortunately, the CMIE dataset does not report irrigated and unirrigated

yields separately, nor is there any data available on the amount of water applied

per unit area. The CMIE data does report the gross (annual total) area irri-

gated per crop in a given district in a given year, where irrigation is defined as

the application of water to the field at least once during the growing season. I

will therefore follow an indirect approach and examine weather higher irrigation

coverage implies a lowered sensitivity of aggregate yield to weather variables.

To measure the extent of irrigation coverage, I define the fraction of area

irrigated as

GIAsdt
F IAsdt = (5)
GAsdt

where GIA is gross irrigated area (annual totals of both seasons) and GA is

20
 Rice Maize Sorghum Bajra Arhar Groundnut Cotton Sugarcane Wheat Gram Barley
Rice Maize Sorghum Bajra Arhar Groundnut Cotton Sugarcane Wheat Gram Barley
(Median) Fraction of Area Irrigated 100%

75%

50%

25%

0%

Figure 5: Ranges (25th to 75th percentiles) of the fraction of area irrigated for
the rainy season (blue) and dry season (brown) crops analyzed.

8
gross cropped area (annual totals of both seasons).

I then test the hypothesis in two different ways.

4.1.1 Comparisons Across Districts

My first test will be to regress the basic yield model separately in each district,

i.e. regress

log Ysdt = vd · Wdt + τd t + πd + sdt (6)

separately within the time series of each district in the data, and obtain district

specific estimates vd (a vector that includes coefficients related to total rainfall,

the number of rainy days, and seasonal degree days), and then regress them,

across districts, on the mean irrigation ratio in that district F IAd (averaged

across time):

vd = v0 + δF IAd + ξ d (7)
8 For those crops that are exclusively cultivated in only one of the seasons, this measure

provides the season specific, and not just annual, ration of area that is irrigated. For those
crops that are grown in both seasons, including rice, sorghum, maize and groundnuts, the
fraction of gross area irrigated provides a measure of irrigation coverage in specific season
only under the assumption that the fraction of cropped area that is irrigated is equal in both
seasons. Because this is a strong assumption, I will run the regressions below on a subsample
of districts in which these crops are only grown in one of the seasons (the rainy season). Figure
5 displays the ranges of irrigation coverage for the crops studied here.

21
 A negative value of a particular component of δ will suggest that yields in

more irrigated districts are less sensitive to the effects of that particular weather

variable.

The main weakness of such cross-district comparisons is the possible endo-

geneity of irrigation, i.e. the possibility that irrigation coverage is correlated

with other district attributes and it is those attributes that are reducing the

sensitivity of crops there. An improvement can be accomplished by dividing the

sample into two time periods, before and after 1985, an approximate watershed

period for irrigation coverage in many parts of India, and then to estimate the

regression (6) separately, in each district, in each of the two time periods. This

estimation produces district and period specific coefficients ωd,i where i = 1, 2

and period 1 (2) refers to the years before (after) 1985. Similarly, one can also

average the fraction of irrigated area within each district and each period to

obtain F IAd,i . One can then run the regression:

vd,i = vd,0 + δF IAd,i + ξ d,i (8)

which control for unobservable, period independent district attributes vd,0 . The

estimates of the components of δ will then only reflect the impact of changes

in irrigation coverage within districts, over time, on the sensitivities of yields to

weather variables, and allow us to rule out the possibility that any other time

independent district property is driving the result.

4.1.2 Interacted Model

My second, and related, test of the hypothesis is to estimate the two ‘levels’ of

the model, i.e. (6) and (7) simultaneously. In other words, to combine the basic

yield model

log Ysdt = vdt · Wdt + fs (t) + λt + πd + sdt (9)

22
in which the coefficients on weather variables are now allowed to vary across dis-

trict and time, with a reduced form of the hypothesis that irrigation expansion

reduces the sensitivity of yields to rainfall patterns, i.e.:

vdt = v0 + δF IAdt + ξ dt (10)

Combining the two models yields a model with interaction terms:

log Ysdt = v0 · Wdt + fs (t) + λt + πd + φF IAsdt + δ · Wdt + sdt + ξ dt · Wdt (11)

where I have also included the un-interacted fraction of irrigated area on the

right hand side. The coefficient φ estimates the (proportional) increase in irri-

gated yields in response to an increase in the fraction of irrigated areas (i.e. the

proportional increase in irrigated yields in comparison to un-irrigated yields).

The three coefficients in the vector δ estimate the change in the sensitivity of

crop yields to the three weather variables in response to an increase in the frac-

tion of irrigated areas. Note that for OLS estimates of this model to be unbiased

we must assume that the error term, consisting of all terms in the third line of

(11) is uncorrelated with any of the other terms in the model.

The interacted model estimates jointly the impact of variation of irrigated

area across districts and across time on the levels and sensitivities of crop yields

to weather fluctuations. In this framework, removing the impact of unobserv-

able district specific time independent attributes on the sensitivity to weather

fluctuations is less attractive because it will involve the addition on several hun-

dreds estimating variables to the regression (consisting of the interactions of

each district dummy variable with the three weather variables). However, I

also attempt to isolate the impact of regional and policy related attributes on

the sensitivity of crop to weather fluctuations by separating the effect of state

23
dependent time independent attributes, as well as those that depend on state

specific quadratic time trends. To do so, I replace the ‘second level’ of the

model, equation (10), with

vsdt = v0 + δF IAdt + ν s + ξ dt (12)

or

vsdt = v0 + deltaF IAdt + ν s + µs t + ξ dt (13)

which will add terms like ν s · Wdt and µs · Wdt t to the interacted model.

4.2 Results

4.2.1 Comparisons Across Districts

Columns 1-4 of tables 5, 6 and 7 present regression estimates for the cross district

comparison (7) of the sensitivities to the three weather variables for rainy season

rice. The parallel results for dry season wheat are presented in tables 8, 9 and

10.

In each table, the first column present estimation of the model across the

entire sample of districts (outliers were removed), the second column restricts

the estimation to those districts in which the district-specific regression yielded

a coefficient which is sensitive at the 10% level, and the third column includes

unobservable state effects. The fourth column presents estimates of the model

(8) which includes unobservable district attributes and captures the impact of

changes in the irrigation coverage over time, within districts (amounting to a

“difference in differences” approach).

In the case of rice, the results indicate that irrigated districts tend to be less

sensitive to fluctuations in both total rainfall and the frequency of rainy days,

but do not indicate any such impact on the sensitivity to seasonal degree days

24
Figure 6: Quadratic fits of the district-wise estimated weather coefficients (total
rainfall in blue, frequency of rainy days in green, and seasonal degree days in
red) against the mean fraction of irrigated area in the district.

(note that for the rainfall variables, this is indicated by the negative slope, but

since the coefficients on seasonal degree days tend to be negative, an indication

of the buffering impact of irrigation would be provided by a positive slope). This

is also evident in figure 6 which displays quadratic fits of the scatter. However,

only the impact on the sensitivity to the number of rainy days is also significant

when estimated across time (column 4 in the table).

In the case of wheat, the only robust finding is with regard to the impact

of heat: more irrigated districts tend to be less vulnerable to the impact of dry

season degree days, both across districts and over time.

4.2.2 Interacted Model

The interacted model (equation 11) estimates the impact of higher irrigation

coverage across districts and over time simultaneously. An indication of the

results for rainy season rice is provided by figure 7 which is similar to figure 4,

except that observations in which irrigation coverage are above and below the

median value (0.5 in the case of rainy season rice) are plotted separately.

Estimates of the interacted regression are presented in table 11 for rainy

season crops and in table 12 for dry season crops. The coefficients of interest

25
Figure 7: Local polynomial fits (kernel regressions), with 95% confidence inter-
vals (errors are not adjusted for spatial correlation), of rainy season anomalies
of rice yields (residuals from regressions on year effects, state specific quadratic
time trends, other weather variables and state effects) on rainy day frequency
(left), total rainfall (center) and seasonal degree days (right) anomalies. His-
tograms display the distributions of the observed anomalies. The red plot is
estimated over observations for which less than half of the cropped area is irri-
gated and the blue plot is estimated where more than half the area is irrigated.

are the interaction coefficients (last three rows). In the rainy season, the in-

teracted coefficients for total rainfall and the number of rainy days tend to be

mostly negative, indicating a buffering effect for irrigation, but only five of the

estimates (including both for rice) are statistically significant at the 10% level,

possibly because the relatively small number of observations of irrigated area for

most crops other than rice. Taken at face value, the esimtates suggest that as

irrigation coverage moves from none (F IA = 0) to full (F IA = 1), the response

of rice yields to a one standard deviation drop in the number of rainy days falls

from 9% to 1% reduction in yield. In contrast, the interacted coefficients with

rainy season degree days do not show any particular pattern. Note also the

coefficients for the impact of increased irrigation on yields (fourth row), which

are only statistically significant for rice (irrigated rice yields are estimated to be

20% higher than non irrigated one in an average weather year) and for millet

(about 50% higher).

26
 In the dry season, the available irrigation data only allows me to estimate the

model for the three crops that are never grown in the rainy season. While the

signs of all interacted coefficients points in the direction of a buffering impact,

its relative size and statistical significance vary across the three crops. Wheat

is shown to be highly sensitive to dry season degree days, with an impact that

moves from 10% to 2% loss of yield per standard deviation of degree days,

as irrigation coverage moves from none to full. Irrigated Wheat and Barley

yields are estimated to be 34% higher than their non-irrigated counterparts in

an average year.

4.2.3 Robustness checks for Rice and Wheat

Tables 13 (for rice) and 14 (for wheat) report estimates of additional regressions

to test the robustness of the results for the impact of irrigation.

Column 1 repeats the estimation of the basic yield model (equation (3))

and column 2 reports the same estimation when the sample is restricted to

observations for which data on irrigated areas exist. Column 3 repeats the

estimation of the simple interacted model, for reference (equation (11)). In

Column 4, the ‘second level’ of the model, equation (10) is extended to include

state specific, time independent unobservable factors that might explain the

sensitivity to weather fluctuations, as in equation (12). This helps to separate

irrigation coverage from most agricultural, energy and water related polices that

are mostly determined by state governments, as well as geographical effects on

the scale of states. Column 5 goes a step further, and allows for unobservable

factors that are determined by state dependent time trends - as in equation

(13).

In Columns 6 and 7 I address the concern that correlations between irrigated

area and weather fluctuations could be driving the results (I will actually find

evidence for this in the next section). In Column 6 I first regress F IA on the

27
three weather variables, and then plug the residual from the regression into (10)

instead of the original F IA. In column 7 I replace F IA by its district mean

(taken over time). Both measures should remove any correlations between the

irrigation coverage and the same year’s weather anomalies.

The remaining robustness checks are similar to those performed for rice and

wheat yields. In column 8 I remove year effects. In column 9 I control for the

previous year’s yield. Because this can introduce a bias in panel estimation when

errors are serially correlated, I also use the Arellano-Bond (Arellano and Bond

[1991]) estimator in column 10. In Column 11 I replace district fixed effects

with a smaller set of geographical fixed effects (districts have split up over time,

and in the basic regressions I include a separate fixed effect for a district before

and after it splits). In Column 12 I run the regression on a sub sample of that

of column 11 in which each location has at least 25 years of observation, in

order to address the concern that missing observations introduce a bias in the

estimation.

In the case of rice, the main results, i.e. the statistically significant negative

coefficients on the interaction of irrigated area with both total rainfall and the

number of rainy days, are robust to the above robustness tests, with the ex-

ception that for total rainfall impacts, significance is lost and coefficient shrink

to almost zero in columns 4 and 5. This means that the buffering impact of

irrigation on total rainfall cannot be separated from state determined attributes.

In the case of wheat, a similar pattern holds for the buffering impact of

wheat on dry season heat exposure (degree days). The coefficient retains its

magnitude in all the models, even though its statistical significance is reduced

when state specific unobservable are allowed to interact with dry season heat

exposure.

28
4.3 Usage of Irrigation

If irrigation helps buffer crop yields from weather shocks, one might expect the

use of irrigation to expand in response to such shocks. However, both the supply

and the demand for irrigation water may depend on rainfall, so the prediction

is not clear cut.

Irrigated areas are likely to be serially correlated because the expansion of

irrigation often involves investments with lasting effects (e.g. drilling a well).

In regressing irrigated areas on weather fluctuations, I therefore also control for

lags of the irrigated area in the same district. In table 15 I report estimates of

the model

log IAsdt = a log IAsd,t−1 + v · Wdt + fs (t) + λt + πd + sdt (14)

where IAsdt is irrigated area of the given crop in state s, district d and year t.

It is well known that including lagged variables in a panel regression with

fixed effects can introduce a bias. I therefore repeat the estimation by using

the the Arellano-Bond estimator (Arellano and Bond [1991]). The results are

reported in table 16.

In both tables, estimates for rainy season crops are presented in columns

1-8 and for dry season crops, in columns 9-11. 9 . The pattern of the evidence

seems to suggest that the irrigated areas of rainy season crops tend to shrink in

response to ‘positive’ anomalies in the number of rainy days or in total rainfall.

In contrast, it seems that the irrigation of dry season crops expands in response

to an abundant monsoon, and to a lesser extent, to increase heat exposure.

9 Sugarcane and cotton are actually long duration crops that extend into the dry season

29
5 Interpretation

This section presents a simple model of irrigation usage and its effects. The

main assumption in the model is that irrigation water is priced at a flat rate.

This assumption, which is a distinctive feature of India’s irrigation economy,

turns out to explain the main empirical findings of the previous section.

5.1 Model

In India, access to and the use of irrigation are determined by a host of com-

plex institutional, physical and infra-structural factors, and irrigation water is

certainly not allocated through an efficiently functioning market. Rather, Shah

[2008] use the term ‘Anarchy’. There is no real regulation of access or property

rights for common groundwater resources, and pervasive inefficiencies plague

surface irrigation.

The two prominent forms of irrigation in the country are surface irrigation

and groundwater irrigation (which has steadily grown since the 1970s to displace

surface irrigation as the largest share of irrigated area). In India, the dominant

cost of access to both of these sources is flat, and independent of the amount of

water used. In the case of surface (canal) irrigation, users are normally charged a

certain flat price per unit of land, per season, and there is no volumetric charge.

Groundwater is not directly priced or regulated. The cost of access is the cost of

drilling a well, and the only running costs are the costs of energy for pumping,

which can come from either electricity or diesel. Almost everywhere in India

where groundwater use has boomed (the Northwest and much of the interior of

peninsular India), it has done so by using highly subsidized, publicly provided

electricity that is priced at a flat (if any) tariff (Shah [2008]), whereas diesel

users (mostly located in eastern India) pay much higher and sticky marginal

costs, (Shah et al. [2009]) and therefore irrigate substantially less area. Not

30
all irrigation water is priced at a flat tariff: water markets are prevalent in

many parts of India (Saleth [1998]). Still, access to these markets is often

distorted by various social and other factors (Anderson [2011]) and water prices

are often not a balance of supply and demand (Dubash [2002]), partly because of

the difficulty of transporting water over substantial distances and the resulting

localized monopolies. Assuming a flat rate costing structure for irrigation, per

unit land, is therefore a likely reasonable approximation and this is the modeling

approach I take here.

Consider a weather variable W that influences the supply of moisture to a

crop (total rainfall, rainy day frequency or the negative of temperature related

evapo-transpirative demand), so that the yield (per unit area), in the absence

of irrigation, can be described as a rising, concave function of W :

Yf,N I = pf y(W ) (15)

where the subscript N I stands for non-irrigated, and p is a basic farm pro-

ductivity factor (e.g. soil, capital or technology), which I assume will vary across

farmers: I will order farmers by a continuos parameter 0 < f < 1 that orders

them according to their basic productivity factor pf .

I will assume that the cost of all inputs, other than irrigation, is CIN per

unit area, so that the net profit per unit of (unirrigated) area is

Pf,N I = pf y(W ) − CIN (16)

Irrigation decisions are made on the basis of a fixed cost of entry, CIR , per

season, per unit area. Entry provides an amount of water w per unit area which

I assume to be a perfect substitute to the weather related water supply/demand.

I will also assume that the cost of other inputs is unchanged. In other words,

31
the irrigated yield is

Yf,IR = pf y(W + w) (17)

and the net profit per unit of irrigated area is

Pf,IR = pf y(W + w) − CIN − CIR (18)

I will assume that the irrigation entry cost CIR is independent of the current

weather conditions. As discussed above, this is probably a reasonable first

approximation in India, where most costing is flat and insensitive to actual water

usage. For this reason, it is reasonable to assume irrigating farmers will utilize

the entire amount of available water. However, the amount of water available

per unit area to those farmers who choose to irrigate can depend on weather

conditions. For example, in the case of groundwater irrigation, the amount

of water available to farmers with access to wells and a certain energy supply

depends on on the water table, which in turn depends on the rainy season’s

rainfall (this is shown by Fishman et al. [Forthcoming]) using rainfall and water

table data from two regions of India). Similarly, the amount of water that flows

in irrigation canals depends on reservoir levels, which in turn determined by

rainfall capture.

In this simple mode, the distinction of whether the supply of water per unit

land in irrigated plots, w, depends on current weather or not, will turn out to

be key to predictions about irrigation use and its impact on yields.

To see this, assume first that the amount of water per unit area w available

to an irrigating farmer is mostly determined by factors that are un-correlated

with in-season weather. Notice first that because the yield function is rising and

32
concave, the irrigated yield is higher

Yf,IR = py(W + w) > py(W ) = Yf,N I (19)

and less sensitive to current weather:

dYf,IR dYf,N I
= py 0 (W + w) < py 0 (W ) = (20)
dW dW

Second, farmer f will choose to irrigate her plot if

pf (y(W + w) − y(W )) − CIR > 0 (21)

so the threshold productivity level for irrigation (with more productive farmers

irrigating and less productive ones not irrigating) is

CIR
pf ∗ = (22)
f (y(W + w) − y(W ))

and is clearly decreasing in W , because the difference between irrigated and

non-irrigated yields is decreasing in W (20), so that better weather results in

less irrigated area.

Now assume instead that the amount of water applied in irrigated fields w

increases in W . In this case, (19) still holds, i.e. it is still true that the irrigated

yield is higher, but it is not necessarily true that it is less sensitive to the weather

W , because the direction of the inequality in (20) depends on weather

y 0 (R + w(W ))(1 + w0 (W )) T y 0 (W ) (23)

which depends on the magnitude of w0 (W ), i.e. on how strongly the available

water supply depends on W .

33
 The sign of this inequality also determines how irrigated area depends on

weather fluctuations, because as before, it determines whether the difference in

yields, and hence the returns to irrigation increase or decrease in W .

In summary, the simple model presented here predicts that when irrigated

water supply per unit area is uncorrelated with the same weather conditions that

influence yields, irrigated yields will be higher and less sensitive to weather, and

the irrigated area will decrease in response to ‘good’ weather. However, when

this water supply is correlated with weather conditions, irrigated yields will still

be higher, but not necessarily less sensitive to weather conditions, and the use of

irrigation may not necessarily decrease in ‘good’ weather. If the water supply is

strongly responsive to weather conditions, yields may actually be more sensitive

and irrigated area may actually increase in response to ‘good’ weather.

5.2 Relation to the Empirical Findings

The empirical findings of this paper can be summarized as follows. Rainy season

crops (mainly rice) are vulnerable to heat exposure, total rainfall and its distri-

bution within the season. Dry season crops (mainly wheat) are sensitive to both

heat exposure and the (accumulated) total rainfall in the rainy season. Irrigated

rainy season yields are effectively protected from the impacts of rainless days,

and to a lesser degree (the evidence is weaker) from the impacts of lower total

rainfall (this is natural: supplemental irrigation through modest storage facil-

ities can be more effective in overcoming uneven rainfall distributions than in

producing the additional water needed to substitute for absolute deficiencies in

the total amounts of rainfall). In the dry season, in contrast, irrigation does not

seem to be able to buffer crop yields from deficiencies in total rainfall accumu-

lation from the rainy season. Finally, irrigation does not seem to protect rainy

season crops from increased heat exposure, but has some success in protecting

34
dry season crops from it.

One of the key differences between the rainy season and the dry season is

in the different relative importance of same year’s monsoon rainfall to the sup-

ply of irrigation water. In the rainy season, water availability for irrigation,

especially early in the season, is largely determined by sources that are uncor-

related with the same year’s monsoon, but probably more by precipitation in

the previous year or in upstream locations. In the dry season, however, the

availability of irrigation water in natural and artificial storage is largely deter-

mined by the capture of rainfall in the immediately previous rainy season (this

is shown by Fishman et al. [Forthcoming] for groundwater aquifers in central

and northwestern India).

The model presented above would therefore predict that rainy season crops,

if irrigated, will be less vulnerable to rainfall deficiencies, and their irrigation will

expand when rainfall is deficient. It also predicts that irrigation may not protect

dry season crops from deficiencies in total rainfall, and that the irrigated area of

these crops will respond positively to more abundant rainfall. These predictions

are in accordance with the main empirical findings, as summarized above.

6 Simple Climate Change Simulations

In this section, I apply the empirical results to estimate some stylized implica-

tions of climate change to rice and wheat yields in India. For illustration, I use

the median projections of the IPCC’s scenario A1B for the period 2080-2099,

which include a precipitation increase over South Asia of about 11%, and an

increase in mean daily temperature of about 2.7◦ C during the rainy season and

about 3.5◦ C during the dry season. While changes in the frequency of rainless

days are not reported by the IPCC, several studies cited therein report increases

in this frequency, including, May [2004] and Krishna Kumar et al. [2003], who

35
report a decrease of up to 15 days in the number of rainy day across much of

India in the 2050s. I therefore add an absolute uniform decline of 10% in the fre-

quency of rainy days during the monsoon to the A1B projection. I assume total

precipitation rises by an average 1mm per day, roughly 10% of the 1980-2000

mean.

I restrict the estimation to the impacts on yields, assuming cropped areas

remain identical to their 1980-2000 levels. The statistical model for rainy season

rice yields is estimated over this sample with the use of the absolute (vs. stan-

dardized) weather variables, including squares of the total precipitation amount

to reflect the concavity revealed in the non-parametric estimation.

Figure 8 shows the simulation results of this climate scenario for the all

India average yield of rainy season rice, calculated over the subset of districts

in India for which sufficient data on rice yield and area was available during the

period 1980-2000. The figure presents the impacts associated with the changes

in each weather variable, using the point estimate for the response as well as

one standard deviation below and above it (first row).

The calculations show that when the intra-seasonal distribution of rainfall is

accounted for, the positive impacts of the projected increase in total rainfall are

diminished (also because of the concave shape of the response to total rainfall)

and are easily outweighed by the decrease in the number of rainy days. Even

putting temperature increase inside, the estimates show that the net impact

of the altered precipitation patters projected in India are going to be negative

(-7%), in contrast to the standard predictions.

The impact of the increased temperature dominates the precipitation impact,

at -36%, consistently with Guiteras [2008], but, as he emphasizes, the prediction

needs to be interpreted with caution because the associated rise in temperature

is much larger than the temperature fluctuations with which the crop response

36
 All India Rice Production, Rainy Season

-10% +2.7 C +10 mm/day

Wet Days Degree Days Total Rainfall
Model Mean + s.d. - s.d. Mean + s.d. - s.d. Mean + s.d. - s.d.

Full Sample -7% -5% -9% -36% -23% -46% 1% 2% -1%

In Irr. Sample 1.0 -7%

Current Irrigation 0.8 -5%
Fully Irrigated 0.1 -1%
No Irrigation 1.8 -12%

All India Wheat Production, Dry Season

-10% +3.5 C +10 mm/day

Wet Days Degree Days Total Rainfall
Model Mean + s.d. - s.d. Mean + s.d. - s.d.

Full Sample -11% -5% -16% 2% 2% 1%

In Irr. Sample 1.0 -17%
Current Irrigation 0.9 -15%
Fully Irrigated 0.8 -14%
No Irrigation 1.3 -22%

Figure 8: Simplified simulation of climate change impact on all India Rice (top)
and Wheat (bottom) production. Full sample refers to a simulation based on
the estimation of the response of yields to weather fluctuations across the entire
CMIE data set. In the second row (In Irr. Sample) the model is estimated on
the sub sample of the data set for which irrigation observations exist. Rows
3-5 apply estimates of the interacted model to the simulation to calculate the
weather impacts of those variables whose impact is robustly shown to respond to
irrigation coverage, i.e. rainy days frequency in the rainy season and degree days
in the dry season. The simulation is run in three scenarios (rows 3-5): using
the current levels of irrigation, assuming irrigation is expanded to cover the
entire cropped area, and assuming irrigation completely ceases. The figures in
the green shade are the ratios of the projected impact in each of the irrigation
scenarios in comparison to an estimate made without taking irrigation into
account, but over the same sub-sample (the simulation in the second row). As
explained in the text, these ratios are less sensitive to the choice of the actual
climate change scenario.
37
model was estimated.

In the dry season as well, the negative impact of increased temperatures (-

11%) easily outweighs the gains from increased rainfall. The impact of changes

in the frequency of rainy days is statistically insignificant and was ignored.

In row 2, I repeat, for comparison, the basic simulation on the sub-sample

for which irrigation data exists. In rows 3-5 of each of the tables I report the

results of simulations in which irrigation is taken into account in determining

the district specific responses to the changing climate conditions. I focus on

the single weather variable for which I found robust evidence for the impact

of irrigation: the frequency of rainy days in the rainy season and the seasonal

degree days in the dry season.

Row 3 reports simulation results for the current irrigation scenario. To the

left of the revised estimates (in the shaded area), I calculate the ratio of the

revised estimate in comparison to a simulation which does not take irrigation

into account. The simulation of the impact in the current irrigation scenario

produces projections that are lower, due to the fact that the baseline production

of rice and wheat is higher in those districts that are more irrigated. In the case

of rainy season rice, taking irrigation into account reduces the projected impact

by 20%. In the case of wheat, it reduces it by 10%.

Row 4 simulates the impact under a scenario of full irrigation coverage. The

impact of the change in the frequency of rainy days on rainy season rice is

reduced by 90%. The gains for wheat are much more modest: the impact of

increased dry season temperatures is reduced by only 20%, largely because most

wheat is already irrigated at present.

On the other extreme, row 5 reports the changes in the impact if irrigation

was to vanish, for example in an extreme scenario of groundwater depletion

and large declines in the flow of major rivers. The impact on rice yields from

38
changes in the rainy day frequency is then increased by 80%. The impact on

wheat is increased by 30%, largely because irrigation is estimated in this model

to possess only a partial ability to protect crop from increased heat exposure.

Note that to first order in the magnitude of climate change, the impact

on total average yield, in percentage terms, is proportional to the size of the

impact (assuming it is uniform), with a proportionality factor that reflects a

weighted average of districts’ specific vulnerabilities, base yields, and cropped

areas. For this reason, the relative impact on India’s average crop yields across

various irrigation scenarios, as reported in the shaded areas of figure 8, is not

very sensitive to the details of the specific climate scenario used.

7 Conclusion

The expansion of irrigation in India has been a hallmark of its agricultural de-

velopment. The analysis of this paper confirms that this expansion has been

quite effective in increasing yields and in protecting food production in the coun-

try from deficiencies in the amount or the timing of rainfall. Putting aside a

complete cost-benefit analysis, the contribution of irrigation to economic devel-

opment has likely been substantial.

Since the expected size of the temperature shift associated with climate

change is substantially higher than the expected shift in rainfall totals or the

frequency of rainy days (in relation to their inter-annual variability) the results

in this paper indicate that it will dominate climate change’s mean impact on

crop yields. Since irrigation is found to be less effective in protecting crops

from temperature shifts, its overall effectiveness as an adaptation strategy seems

limited. This does not mean, however, that irrigation will discontinue to be an

effective strategy for addressing inter-annual variability in precipitation patters.

In fact, there are some predictions that this variability itself may increase, and

39
a complete welfare evaluation of the benefits of irrigation expansion will have

to take this into account.

The success of India’s irrigation as a buffering strategy has come at high

costs. Irrigation, as it is practiced in India at present is unsustainable. Sur-

face irrigation has proven highly inefficient and the construction of large scale

dams of questionable social merit (Duflo and Pande [2007], on Dams [2000]).

Groundwater irrigation, which has emerged as the principal and most success-

ful irrigation source, consumes enormous amount of energy and is unsustainably

depleting the country’s aquifers (Fishman et al. [Forthcoming]). The results of

this paper should not be interpreted as suggesting that water resources have

little value for India’s future agricultural performance. On the contrary. They

confirm the valuable services these water resources can provide, services that

may be compromised at a time in which they may turn out to be even more

important.

References

S. Anderson. Caste as an impediment to trade. American Economic Journal:

Applied Economics, 3(1):239–263, 2011.

M. Arellano and S. Bond. Some tests of specification for panel data: Monte

carlo evidence and an application to employment equations. The Review of

Economic Studies, 58(2):277, 1991.

M. Auffhammer, V. Ramanathan, and J.R. Vincent. Integrated model shows

that atmospheric brown clouds and greenhouse gases have reduced rice har-

vests in india. Proceedings of the National Academy of Sciences, 103(52):

19668, 2006.

O. Deschenes and M. Greenstone. The economic impacts of climate change:

40
 evidence from agricultural output and random fluctuations in weather. The

American Economic Review, 97(1):354–385, 2007.

A. Dinar. Measuring the impact of climate change on Indian agriculture, vol-

ume 23. World Bank Publications, 1998.

N.K. Dubash. Tubewell capitalism: groundwater development and agrarian

change in Gujarat. Oxford University Press, 2002.

E. Duflo and R. Pande. Dams*. The Quarterly Journal of Economics, 122(2):

601–646, 2007.

R. M. Fishman, T. Siegfried, P. Raj, V. Modi, and U. Lall. Over-extraction from

shallow bedrock versus deep alluvial aquifers: Reliability versus sustainability

considerations for india’s groundwater irrigation. Water Resources Research,

Forthcoming.

JM Gregory and JFB Mitchell. Simulation of daily variability of surface tem-

perature and precipitation over europe in the current and 2× co2 climates

using the ukmo climate model. Quarterly Journal of the Royal Meteorological

Society, 121(526):1451–1476, 1995.

R. Guiteras. The impact of climate change on indian agriculture. University of

Maryland Mimeo, 2008.

X. Guoju, L. Weixiang, X. Qiang, S. Zhaojun, and W. Jing. Effects of tempera-

ture increase and elevated co2 concentration, with supplemental irrigation, on

the yield of rain-fed spring wheat in a semiarid region of china. Agricultural

water management, 74(3):243–255, 2005.

CH Hanumantha Rao, S.K. Ray, and K. Subbarao. Unstable agriculture and

droughts: Implications for policy, volume 47. Vikas Pub. House (New Delhi),

1988.

41
KJ Hennessy, JM Gregory, and JFB Mitchell. Changes in daily precipitation

under enhanced greenhouse conditions. Climate Dynamics, 13(9):667–680,

1997.

S.M. Howden, J.F. Soussana, F.N. Tubiello, N. Chhetri, M. Dunlop, and

H. Meinke. Adapting agriculture to climate change. Proceedings of the Na-

tional Academy of Sciences, 104(50):19691, 2007.

K. Krishna Kumar, NR Deshpande, PK Mishra, K. Kamala, and K.R. Kumar.

Future scenarios of extreme rainfall and temperature over india. In Proceed-

ings of the Workshop on Scenarios and Future Emissions, Indian Institute of

Management (IIM), Ahmedabad, July, volume 22, pages 56–68, 2003.

K. Krishna Kumar, K. Rupa Kumar, RG Ashrit, NR Deshpande, and

JW Hansen. Climate impacts on indian agriculture. International Journal of

Climatology, 24(11):1375–1393, 2004.

K. Kumar and J. Parikh. Climate change impacts on indian agriculture: the

ricardian approach. 1998.

KS Kumar and J. Parikh. Indian agriculture and climate sensitivity. Global

environmental change, 11(2):147–154, 2001.

P. Kurukulasuriya and R. Mendelsohn. Endogenous irrigation: The impact of

climate change on farmers in africa1. World, 2007.

D.B. Lobell, W. Schlenker, and J. Costa-Roberts. Climate trends and global

crop production since 1980. Science, 2011.

W. May. Simulation of the variability and extremes of daily rainfall during the

indian summer monsoon for present and future times in a global time-slice

experiment. Climate dynamics, 22(2):183–204, 2004.

42
R. Mendelsohn and A. Dinar. Climate change, agriculture, and developing

countries: does adaptation matter? The World Bank Research Observer, 14

(2):277, 1999.

R. Mendelsohn, W.D. Nordhaus, and D. Shaw. The impact of global warming

on agriculture: a ricardian analysis. The American Economic Review, pages

753–771, 1994.

M.H. Moench. Chasing the watertable: Equity and sustainability in groundwa-

ter management. Economic and Political Weekly, pages 171–177, 1992.

World Commission on Dams. Dams and development: a new framework for

decision-making: the report of the World Commission on Dams. Earth-

scan/James & James, 2000.

M.L. Parry. Climate Change 2007: impacts, adaptation and vulnerability: con-

tribution of Working Group II to the fourth assessment report of the Inter-

governmental Panel on Climate Change. Cambridge Univ Pr, 2007.

A.G. Patt, D. Schröter, A.C. de la Vega-Leinert, and R.J.T. Klein. Vulnerabil-

ity research and assessment to support adaptation and mitigation: common

themes from the diversity of approaches. Assessing vulnerability to global en-

vironmental change: making research useful for adaptation, decision making

and policy, page 1, 2009.

S. Peng, J. Huang, J.E. Sheehy, R.C. Laza, R.M. Visperas, X. Zhong, G.S.

Centeno, G.S. Khush, and K.G. Cassman. Rice yields decline with higher

night temperature from global warming. Proceedings of the National Academy

of Sciences of the United States of America, 101(27):9971, 2004.

M. Rajeevan, J. Bhate, JD Kale, and B. Lal. Development of a high resolution

43
 daily gridded rainfall data for the indian region. Met. Monograph Climatology,

22:2005, 2005.

C. Rosenzweig and D. Hillel. Climate variability and the global harvest: impacts

of El Niño and other oscillations on agroecosystems. Oxford University Press,

USA, 2008.

C. Rosenzweig and M.L. Parry. Potential impact of climate change on world

food supply. Nature, 367(6459):133–138, 1994.

R. Saleth. Water markets in india: Economic and institutional aspects. Markets

for Water, pages 187–205, 1998.

A. Sanghi and R. Mendelsohn. The impacts of global warming on farmers in

brazil and india. Global Environmental Change, 18(4):655–665, 2008.

W. Schlenker and M.J. Roberts. Nonlinear temperature effects indicate severe

damages to us crop yields under climate change. Proceedings of the National

Academy of Sciences, 106(37):15594, 2009.

W. Schlenker, W.M. Hanemann, and A.C. Fisher. Will us agriculture really ben-

efit from global warming? accounting for irrigation in the hedonic approach.

The American Economic Review, 95(1):395–406, 2005.

W. Schlenker, W.M. Hanemann, and A.C. Fisher. The impact of global warming

on us agriculture: an econometric analysis of optimal growing conditions.

Review of Economics and Statistics, 88(1):113–125, 2006.

T. Shah. Taming the anarchy: groundwater governance in South Asia. Earth-

scan, 2008.

T. Shah, M. Ul Hassan, M.Z. Khattak, P.S. Banerjee, OP Singh, and S.U.

Rehman. Is irrigation water free? a reality check in the indo-gangetic basin.

World Development, 37(2):422–434, 2009.

44
AK Srivastava, M. Rajeevan, and SR Kshirsagar. Development of a high reso-

lution daily gridded temperature data set (1969–2005) for the indian region.

Atmospheric Science Letters, 10(4):249–254, 2009.

DB Stephenson, K.R. Kumar, FJ Doblas-Reyes, JF Royer, F. Chauvin, and

S. Pezzulli. Extreme daily rainfall events and their impact on ensemble fore-

casts of the indian monsoon. Monthly weather review, 127(9):1954–1966, 1999.

C. Tebaldi, K. Hayhoe, J.M. Arblaster, and G.A. Meehl. Going to the extremes.

Climatic Change, 79(3):185–211, 2006.

R.S.J. Tol. The economic effects of climate change. The Journal of Economic

Perspectives, 23(2):29–51, 2009.

R.S.J. Tol, S. Fankhauser, and J.B. Smith. The scope for adaptation to climate

change: what can we learn from the impact literature? Global Environmental

Change, 8(2):109–123, 1998.

K.E. Trenberth, J. Fasullo, and L. Smith. Trends and variability in column-

integrated atmospheric water vapor. Climate Dynamics, 24(7):741–758, 2005.

C.J. vrsmarty, P. Green, J. Salisbury, and R.B. Lammers. Global water re-

sources: vulnerability from climate change and population growth. science,

289(5477):284, 2000.

J.R. Welch, J.R. Vincent, M. Auffhammer, P.F. Moya, A. Dobermann, and

D. Dawe. Rice yields in tropical/subtropical asia exhibit large but opposing

sensitivities to minimum and maximum temperatures. Proceedings of the

National Academy of Sciences, 107(33):14562, 2010.

45
 Table 1: Log Yields, Rainy Season, Standardized Weather Variables
(1) (2) (3) (4) (5) (6) (7) (8)
Rice Maize Sorghum Millet Pigeon Pea Groundnut Cotton Sugarcane
Rainy Days 0.04∗∗∗ 0.06∗ 0.04∗∗ 0.08∗ 0.04∗∗ 0.05∗∗ 0.04+ 0.04∗∗
(0.00) (0.01) (0.00) (0.03) (0.00) (0.00) (0.07) (0.00)

Shape Parameter 0.01 0.00 -0.01 -0.01 -0.00 0.02 0.01 -0.01
(0.13) (0.56) (0.10) (0.18) (0.79) (0.10) (0.50) (0.51)

Longest Dry Spell -0.00 0.01+ -0.00 -0.01 0.00 -0.01 0.02 0.01
(0.83) (0.08) (0.93) (0.59) (0.60) (0.22) (0.22) (0.35)

Rainfall, Monsoon 0.04∗ -0.01 -0.00 0.03∗ 0.03∗ 0.05∗∗ 0.05+ 0.01
(0.01) (0.67) (0.89) (0.01) (0.03) (0.01) (0.05) (0.54)

46
Degree Days, Rainy Season -0.03+ 0.03 0.02+ -0.02 -0.02 -0.02 -0.11 0.01
(0.06) (0.28) (0.05) (0.19) (0.20) (0.28) (0.19) (0.86)
N 8861 3869 6470 2940 10129 5163 1978 2980
adj. R2 0.767 0.621 0.623 0.708 0.605 0.468 0.639 0.839
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001
 Table 2: Log Yields, Dry Season, Standardized Weather Variables
(1) (2) (3) (4) (5) (6) (7)
Wheat Rice Maize Barley Sorghum Chickpea Groundnut
Rainy Days -0.00 0.02∗ -0.00 0.00 0.02∗ 0.03∗∗ -0.00
(0.81) (0.02) (0.80) (0.53) (0.02) (0.01) (0.67)

Shape Parameter 0.00 -0.01 0.01 0.01 0.01 0.01 0.00

(0.12) (0.41) (0.27) (0.27) (0.43) (0.43) (0.63)

Longest Dry Spell -0.00 0.01+ -0.01 0.01 0.01 0.01 -0.01
(0.89) (0.09) (0.35) (0.12) (0.21) (0.21) (0.15)

Rainfall, Monsoon 0.03∗∗ 0.01 0.00 0.03∗ 0.03∗ 0.05∗∗∗ 0.04∗∗∗

(0.00) (0.17) (0.68) (0.02) (0.03) (0.00) (0.00)

47
Degree Days, Dry Season -0.03 0.00 -0.05+ -0.03∗ -0.15 -0.09∗∗ -0.10+
(0.15) (0.94) (0.07) (0.03) (0.15) (0.01) (0.05)
N 11234 3493 1459 6683 2593 9798 2069
adj. R2 0.817 0.737 0.862 0.730 0.507 0.572 0.421
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001
 Table 3: Log rice Yield, Rainy Season, Standardized Weather Variables, Additional Regressions
(1) (2) (3) (4) (5) (6)
Rainy Days 0.04∗∗∗ 0.06∗∗∗ 0.04∗∗∗ 0.04∗∗∗ 0.04∗∗ 0.04∗∗
(0.00) (0.00) (0.00) (0.00) (0.00) (0.01)

Shape Parameter 0.01 0.01∗ 0.01 0.01 0.01 0.00

(0.13) (0.03) (0.10) (0.17) (0.19) (0.50)

Longest Dry Spell -0.00 -0.01 -0.01 -0.01 -0.00 -0.01

(0.83) (0.39) (0.19) (0.26) (0.69) (0.23)

Rainfall, Total 0.04∗ 0.05∗∗ 0.03∗∗∗ 0.04∗∗∗ 0.03∗ 0.02

(0.01) (0.00) (0.00) (0.00) (0.01) (0.20)

48
DDS -0.03+ -0.03∗ -0.03∗∗∗ -0.03∗∗∗ -0.03∗ -0.03∗
(0.06) (0.02) (0.00) (0.00) (0.04) (0.04)
N 8861 8861 5975 5975 8812 4991
adj. R2 0.767 0.754 0.766 0.737
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001
 Table 4: Log wheat Yield, Dry Season, Standardized Weather Variables, Additional Regressions
(1) (2) (3) (4) (5) (6)
Rainy Days -0.00 0.01 0.00 0.01∗ -0.00 0.00
(0.81) (0.21) (0.59) (0.04) (0.91) (0.71)

Shape Parameter 0.00 0.01∗ 0.00 0.00 0.00 0.00

(0.12) (0.01) (0.37) (0.84) (0.11) (0.19)

Longest Dry Spell -0.00 0.00 -0.00 0.00 -0.00 -0.00

(0.89) (0.51) (0.88) (0.54) (0.83) (0.81)

Rainfall, Total 0.03∗∗ 0.03∗∗ 0.03∗∗∗ 0.03∗∗∗ 0.03∗∗ 0.03∗∗

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

49
DDS -0.03 -0.01 -0.03∗∗∗ -0.03∗∗ -0.02 -0.05∗∗
(0.15) (0.27) (0.00) (0.00) (0.18) (0.01)
N 11234 11234 9579 9579 11133 9076
adj. R2 0.817 0.811 0.814 0.829
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001
Table 5: Cross Sectional Regression, Rainy Season rice, Total Rainfall
(1) (2) (3) (4)
∗∗∗ +
FIA -0.09 -0.04 -0.09 0.03
(0.00) (0.21) (0.05) (0.56)

Constant 0.08∗∗∗ 0.16∗∗∗ 0.10∗∗

(0.00) (0.00) (0.00)
N 265 79 374 604

Table 6: Cross Sectional Regression, Rainy Season rice, Rainy Days

(1) (2) (3) (4)
FIA -0.05∗∗ -0.08∗ -0.11+ -0.16∗
(0.00) (0.03) (0.10) (0.01)

Constant 0.08∗∗∗ 0.19∗∗∗ 0.10∗

(0.00) (0.00) (0.01)
N 265 79 374 604

Table 7: Cross Sectional Regression, Rainy Season rice, Degree Days

(1) (2) (3) (4)
FIA -0.06∗∗∗ 0.02 0.01 -0.07
(0.00) (0.43) (0.59) (0.22)

Constant 0.00 -0.12∗∗∗ -0.04∗

(0.74) (0.00) (0.01)
N 265 57 374 604
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001

50
Table 8: Cross Sectional Regression, Rainy Season wheat, Total Rainfall
(1) (2) (3) (4)
FIA -0.05∗∗∗ -0.01 -0.01 0.04
(0.00) (0.60) (0.72) (0.33)

Constant 0.06∗∗∗ 0.11∗∗∗ 0.03

(0.00) (0.00) (0.19)
N 365 81 365 638

Table 9: Cross Sectional Regression, Rainy Season wheat, Rainy Days

(1) (2) (3) (4)
FIA -0.00 0.06 -0.04 -0.04
(0.66) (0.38) (0.63) (0.26)

Constant 0.01 0.00 0.04

(0.12) (0.98) (0.41)
N 365 36 365 638

Table 10: Cross Sectional Regression, Rainy Season wheat, Degree Days
(1) (2) (3) (4)
FIA 0.03∗∗ 0.17∗∗∗ 0.07 0.06+
(0.00) (0.00) (0.38) (0.09)

Constant -0.02∗ -0.11∗∗ -0.06

(0.01) (0.00) (0.26)
N 365 60 365 638
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001

51
 Table 11: Log Yields, Rainy Season, Interacted Model with Standardized Weather Variables
(1) (2) (3) (4) (5) (6) (7) (8)
Rice Maize Sorghum Millet Pigeon Pea Groundnut Cotton Sugarcane
Rainy Days 0.09∗∗ 0.06+ 0.06+ 0.17∗∗ 0.03 0.11∗∗∗ -0.00 0.06
(0.01) (0.07) (0.08) (0.01) (0.16) (0.00) (0.94) (0.15)

Rainfall, Total 0.07∗ -0.02 -0.06 0.08∗ 0.03 0.02 0.02 -0.03
(0.01) (0.14) (0.13) (0.05) (0.16) (0.42) (0.38) (0.56)

DDS -0.04 0.06∗ -0.03 0.00 -0.09∗ -0.07 0.03 0.10

(0.20) (0.04) (0.42) (0.94) (0.02) (0.13) (0.80) (0.49)

FIA 0.20∗∗ -0.01 0.14 0.49+ -0.15 0.06 0.85 -0.09

52
(0.00) (0.81) (0.39) (0.05) (0.55) (0.61) (0.26) (0.52)

FIA X Rainy Days -0.08+ -0.01 -0.07 -0.23∗ -0.06 -0.07 0.07 -0.03
(0.08) (0.82) (0.11) (0.03) (0.23) (0.19) (0.68) (0.66)

FIA X Total -0.07∗ 0.06 0.01 -0.05 -0.13∗ -0.02 -0.26+ 0.04
(0.03) (0.18) (0.91) (0.46) (0.03) (0.78) (0.09) (0.51)

FIA X DDS -0.00 -0.05 0.01 0.07 -0.15∗ 0.03 -0.69 -0.13
(0.99) (0.20) (0.86) (0.15) (0.01) (0.59) (0.23) (0.33)
N 3333 1080 1078 638 1456 915 561 536
adj. R2 0.786 0.610 0.684 0.809 0.640 0.542 0.664 0.792
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001
 Table 12: Log Yields, Dry Season, Interacted Model with Standardized Weather Variables
(1) (2) (3)
Wheat Barley Chickpea
Rainy Days 0.01 0.02∗ 0.02
(0.35) (0.04) (0.10)

Rainfall, Total 0.05∗ 0.03+ 0.07∗∗

(0.03) (0.07) (0.00)

DDS -0.10∗∗ -0.06∗ -0.05

(0.01) (0.02) (0.33)

FIA 0.34∗∗∗ 0.34∗∗ 0.07

53
(0.00) (0.00) (0.16)

FIA X Rainy Days -0.01 -0.03∗∗ -0.00

(0.44) (0.00) (0.94)

FIA X Total -0.02 -0.00 -0.04+

(0.47) (0.82) (0.09)

FIA X DDS 0.08+ 0.05 0.06

(0.05) (0.32) (0.18)
N 6077 4021 4698
adj. R2 0.863 0.749 0.574
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001
 Table 13: Log rice Yield, Rainy Season, Interacted Model with Standardized Weather Variables, Additional Regressions
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Rainy Days 0.05∗∗ 0.06∗∗ 0.09∗∗ 0.15∗∗ 0.09 0.04∗∗ 0.08∗∗ 0.12∗∗ 0.09∗ 0.07∗∗∗ 0.09∗∗ 0.12∗
(0.00) (0.00) (0.01) (0.00) (0.98) (0.00) (0.00) (0.01) (0.01) (0.00) (0.01) (0.05)

Rainfall, Total 0.03+ 0.03 0.07∗ 0.08∗∗ -2.36 0.03+ 0.08∗∗ 0.06∗ 0.05∗ 0.08∗∗∗ 0.07∗∗ 0.04
(0.06) (0.13) (0.01) (0.00) (0.34) (0.09) (0.00) (0.04) (0.02) (0.00) (0.01) (0.15)

DDS -0.03+ -0.04+ -0.04 -0.02 -5.16 -0.03 -0.01 -0.04+ -0.03 -0.02 -0.04 -0.05
(0.09) (0.09) (0.20) (0.42) (0.15) (0.10) (0.50) (0.08) (0.30) (0.22) (0.20) (0.33)

FIA 0.20∗∗ 0.24∗∗ 0.24∗∗ 0.21∗∗ 0.22∗∗ 0.14∗ 0.29∗∗ 0.17∗∗ 0.18+
(0.00) (0.00) (0.00) (0.00) (0.00) (0.03) (0.00) (0.00) (0.07)

54
FIA X Rainy Days -0.08+ -0.13∗ -0.13+ -0.08+ -0.06∗ -0.08 -0.08 -0.07∗∗ -0.07+ -0.13
(0.08) (0.04) (0.05) (0.08) (0.03) (0.11) (0.12) (0.00) (0.07) (0.11)

FIA X Total -0.07∗ 0.01 0.01 -0.07∗ -0.11∗∗ -0.05 -0.06+ -0.09∗∗∗ -0.07∗ -0.06
(0.03) (0.72) (0.86) (0.02) (0.00) (0.17) (0.05) (0.00) (0.01) (0.12)

FIA X DDS -0.00 0.00 -0.01 0.00 -0.02 -0.00 -0.00 0.00 -0.00 0.01
(0.99) (0.90) (0.70) (0.94) (0.42) (0.85) (0.88) (0.86) (0.98) (0.78)
N 6217 3333 3333 3333 3333 3333 5621 3333 2647 1902 3253 1714
adj. R2 0.770 0.780 0.786 0.808 0.808 0.787 0.778 0.770 0.795 0.791 0.795
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001
 Table 14: Log wheat Yield, Dry Season, Interacted Model with Standardized Weather Variables, Additional Regressions
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Rainy Days 0.00 0.00 0.01 0.01 -4.74∗ -0.00 0.00 0.01 0.02 0.01 0.01 0.00
(0.96) (0.88) (0.35) (0.74) (0.02) (0.94) (0.83) (0.28) (0.28) (0.16) (0.30) (0.95)

Rainfall, Total 0.03∗∗ 0.03∗∗ 0.05∗ 0.07∗∗ 2.65 0.03∗∗ 0.05∗∗ 0.05∗ 0.05+ 0.04∗∗∗ 0.05∗ 0.06∗∗
(0.00) (0.00) (0.03) (0.01) (0.11) (0.00) (0.00) (0.02) (0.05) (0.00) (0.03) (0.00)

DDS -0.03 -0.05∗ -0.10∗∗ -0.18∗∗ 3.68 -0.02 -0.06∗∗ -0.12∗∗ -0.10∗∗ -0.09∗∗∗ -0.09∗ -0.13∗∗∗
(0.16) (0.02) (0.01) (0.00) (0.69) (0.36) (0.00) (0.00) (0.00) (0.00) (0.03) (0.00)

FIA 0.34∗∗∗ 0.35∗∗∗ 0.35∗∗∗ 0.34∗∗∗ 0.34∗∗∗ 0.29∗∗∗ 0.25∗∗∗ 0.33∗∗∗ 0.33∗∗∗
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

55
FIA X Rainy Days -0.01 -0.01 -0.03 -0.02 -0.00 -0.01 -0.01 -0.00 -0.01 0.00
(0.44) (0.77) (0.51) (0.39) (0.80) (0.41) (0.51) (0.73) (0.42) (0.82)

FIA X Total -0.02 -0.01 0.00 -0.02 -0.03∗ -0.02 -0.02 -0.02 -0.02 -0.03
(0.47) (0.74) (0.88) (0.48) (0.03) (0.35) (0.42) (0.17) (0.45) (0.17)

FIA X DDS 0.08+ 0.06 0.07 0.08+ 0.05∗∗ 0.09+ 0.09∗ 0.07∗∗ 0.08+ 0.13∗∗
(0.05) (0.34) (0.26) (0.08) (0.00) (0.06) (0.01) (0.00) (0.08) (0.00)
N 11234 6077 6077 6077 6077 6077 10172 6077 5672 4789 6060 5449
adj. R2 0.817 0.858 0.863 0.867 0.868 0.863 0.828 0.859 0.870 0.863 0.861
p-values in parentheses
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001
 Table 15: Irrigated Areas
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Rice Maize Sorghum Millet Pigeon Pea Groundnut Cotton Sugarcane Wheat Chickpea Barley
Rainfall, Total -0.01 -0.07 -0.12∗∗ -0.15∗∗ -0.07∗ -0.04 0.01 -0.01∗ 0.06∗∗ 0.03 -0.02+
(0.51) (0.10) (0.00) (0.00) (0.01) (0.33) (0.56) (0.04) (0.00) (0.15) (0.07)

Wet Days -0.02 -0.11+ -0.05 -0.14 -0.04 -0.18∗ -0.01 0.02+ -0.01 -0.02 -0.03∗
(0.19) (0.05) (0.64) (0.14) (0.28) (0.05) (0.84) (0.07) (0.29) (0.40) (0.05)

DDS -0.01 0.07 0.04 -0.07 0.08 -0.07 -0.22∗∗ -0.07∗∗ -0.05+ -0.08 -0.08
(0.79) (0.18) (0.57) (0.52) (0.35) (0.59) (0.00) (0.00) (0.07) (0.25) (0.32)
N 3704 3214 1119 2129 972 1200 2827 4690 4956 3689 3396
adj. R2 0.955 0.892 0.818 0.834 0.842 0.849 0.946 0.976 0.960 0.890 0.945
p-values in parentheses

56
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001
 Table 16: Irrigated Areas, Arellano Bond Estimation
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Rice Maize Sorghum Millet Pigeon Pea Groundnut Cotton Sugarcane Wheat Chickpea Barley
Rainfall, Total 0.00 -0.06∗∗ -0.08∗ -0.11∗∗ -0.08∗ -0.03 0.02 -0.02∗∗ 0.05∗∗∗ 0.03∗ 0.01+
(0.73) (0.00) (0.05) (0.00) (0.05) (0.45) (0.18) (0.00) (0.00) (0.01) (0.10)

Wet Days -0.02 -0.10∗∗∗ 0.03 -0.16∗∗∗ 0.01 -0.14∗∗ -0.02 0.00 0.00 0.00 -0.01
(0.21) (0.00) (0.54) (0.00) (0.84) (0.01) (0.28) (0.55) (0.60) (0.93) (0.37)

DDS 0.02 0.12∗∗ 0.30∗ -0.07 0.10 0.06 -0.13∗∗ -0.07∗∗ -0.02 -0.00 -0.03
(0.41) (0.00) (0.02) (0.40) (0.42) (0.62) (0.00) (0.00) (0.20) (0.93) (0.22)
N 2938 2744 807 1768 664 877 2251 4006 4750 3420 3286
adj. R2
p-values in parentheses

57
+ p < 0.1, ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.0001