Presented in ACIAR-IFPRI two days Regional Dialogue on Machine Reforms’ for Sustainable Intensification of Agriculture in South Asia on July 21-22, 2017 in New Delhi, India

1. Adoption and Impact of Conservation Agriculture (CA)
Technology: Evidence From Haryana And Punjab
P. K. Joshi, Md. Tajuddin Khan and Avinash kishore
IFPRI, New Delhi
India

2. Objectives
• To map the adoption and impacts of CA technology
• Identify drivers of CA adoption and diffusion
• Evaluate impacts of ZT adoption on Yield and reducing
crop loss from Weather Shock

3. Data and Location of study
Haryana: 50 villages, 10 districts
and 717 farmers
Punjab: 61 villages, 12 district and
716 farmers
Creating panel data for Haryana
• 400 Households interviewed
earlier in 2003-04 (Ernstein et.al)
• We have only able to trace 286
farmers (As most of them have
left agriculture and some we did
not found)

4. The state of technology adoption in Haryana
and Punjab
Variable name Punjab Haryana t-test
Laser land Leveler 39.89 45.89 -2.29**
Zero tillage 8.65 36.54 -12.63***
Rotavator 49.37 72.39 -8.93***
Combine harvester 91.49 83.96 4.35***
Straw reaper 81.17 81.31 -0.07
Note: ***, ** and * denotes significance at 1, 5 and 10 percent level.
Tractor

5. Haryana: Adoption of technologies
0
100
200
300
400
500
600
700 1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
NO.OFFARMERS
ZT LLL Rotavator Combine Straw_Reaper

6. Punjab: Adoption of technologies
0
100
200
300
400
500
600
700 1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
NO.OFFARMERS
Combine Straw reaper Rotavator ZT LLL

7. Operated landholding size and the likelihood to adopt Zero tillage and
Laser land leveller

8. Farmer’s stated/realized reasons for technology
adoption
Zero-tillage
(N=353)
Laser land leveller
(N=680)
Rotavator
(N=911)
Increase yield 55.81 49.12 22.06
Save labour 58.36 28.68 48.08
Save water 27.20 86.47 11.20
Improve soil health 12.46 20.88 48.63
Save time 68.56 47.94 65.09
Unscientific perception
Non-adoption reasons: Based on farmers perception
Lack of machinery availability/knowledge, High cost, Benefits not clear, Poor yield

9. Source of information for technology adoption
Haryana Punjab
ZT LLL RV ZT LLL RV
Agriculture Department 3.73 1.61 3.14 22.58 11.73 15.95
Another farmer 54.58 69.44 53.05 8.06 4.89 54.05
Family member 22.37 12.06 23.48 33.87 61.56 9.73
Input dealer 13.90 13.94 17.93 8.06 5.21 12.43
Newspaper/TV 1.69 0.80 1.85 17.74 12.70 5.95
University scientist 1.02 1.61 0.37 3.23 2.28 0.81
Visit research station 2.03 0.00 0.18 1.61 0.33 1.08
Others 0.68 0.54 0.00 4.84 1.30 0.00

10. Correlation matrix of technology adoption
Haryana Laser land leveler Zero tillage Rotavator Combine harvester Straw reaper
Laser land leveler 1
Zero tillage 0.1557*** 1
Rotavator -0.0134 -0.2374*** 1
Combine harvester 0.1355*** 0.0791** 0.0956** 1
Straw reaper 0.1327*** 0.0666* 0.104*** 0.6387*** 1
Punjab
Laser land leveler 1
Zero tillage 0.0939** 1
Rotavator 0.1811*** 0.0336 1
Combine harvester 0.0238 0.0405 -0.0688* 1
Straw reaper -0.0011 0.0720* -0.031 0.4797*** 1

11. Transition Matrix: Adoption history of ZT
2003 (Yes) 2003 (No) Overall (717)
2015 (Yes) 67 (23.5%)
(Always Adopters)
52 (18.2%)
(Later Adopters)
262 (36.5%)
(Area under ZT: 35%)
2015 (No) 31 (10.8%)
(Dis-adopters)
136 (47.5%)
(Never adopters)
455 (63.5%)
Overall (400) 138 (34.5%)
(Area under ZT: 26%)
262 (65.5%)
❖ In 2015, more than 36 percent of all farmers in our sample practice ZT wheat in approximately 35 percent
of their total area under wheat up from 26% in the EMS survey in 2003-04 (EMS 2007).
❖ Very low dis-adoption and only a few new adopters. Growth on intensive margin,

12. Correlates of technology adoption: Multivariate probit model
• Adoption of specific technologies is not independent of other technological
choices on the same farm
• We adopts the multivariate probit (MVP) model that accounts for error term
correlation.
• MVP allow unobserved and unmeasured factors (error terms) to be correlated
(Belderbos et al. 2004; Lin et al. 2005).
• Our MVP model consists of three binary choice equations namely use of ZT,
LLL and RV. We have not considered combine harvester and straw reaper, as
these are widely accepted in both states.

13. Multivariate probit results
❖ Land size, education and
access to institutional
resources are important for
tech. adoption (ZT and LLL)
❖ Smaller pump owners adopt
LLL
❖ Farmers from backward/lower
caste & reg. farmer group are
more likely to adopt ZT
❖ Credit has negative association
with ZT and RV
❖ Adoption of these
technologies is higher in the
bordering/neighboring
districts.
LLL ZT RV
Age -0.0021 -0.0048 0.0028
land owned 0.0212*** 0.0196*** 0.0075*
water depth (feet) -0.0003 0.0022** 0.0016*
Pump size(HP) -0.0016*** -0.0002 -0.0008
Neighbor district 0.0074 0.6042** 0.0477
Primary 0.5337*** 0.1337 0.1128
Middle 0.4212*** 0.1260 0.1187
Higher secondary 0.4089*** 0.1285 0.0780
Graduate 0.4827*** -0.0170 0.0350
Small Farmers 0.1380 0.0871 -0.0333
Medium 0.2632** 0.1293 -0.1001
Large 0.1792 0.0465 0.0345
Upper caste 0.2264** -0.3022** 0.1690
Credit access 0.0609 -0.2572** -0.3440****
Replace seed every year -0.1217 0.3314*** -0.1012
MSP 0.2655** 0.5752*** -0.0328
Biofertilizer 0.0617 0.0077 0.2354***
Member if reg. farmer 0.5022*** 0.4514*** -0.0878
Distance input 0.0187* 0.0141 0.0144*
Distance Extension 0.0060 0.0020 -0.0171**
Distance Agri. equip -0.0160* -0.0132 -0.0197***
Constant -1.3539*** -1.8128*** 0.0591

14. Results
• Adoption of CA technologies are higher in Haryana.
• Highly skilled and educated farmers and farmers belongs from forward caste are more
likely to adopting LLL
• Extended family network is the main source of information for all CA technologies
• Role of government institutions in promoting the technologies was more visible in
Punjab.

15. Zero Tillage: “A Climate Smart Solution to Downside
Risk of Agriculture”?
• We also looked at the impact of ZT adoption on wheat yields in 2015
when unseasonal rainfalls occurred right before harvest
• Collected 4 years of recall data on wheat yields in ZT and CT plots
of farmers and
• Used this panel data to estimate the impact of unseasonal rains on
wheat yields in CT and ZT plots using Difference-in-Differences
method

16. 14.0
14.5
15.0
15.5
16.0
16.5
17.0
17.5
18.0
18.5
2011 2012 2013 2014 2015
YIELD(QUINTALS)
ZT CT
Unseasonal rain
and hailstorm
Recall data shows parallel trends in wheat yields of ZT
and CT plots before 2015

17. Distribution of rainfall (weekly total) over the wheat growing season
Source: J.P. Aryal et al. / Agriculture, Ecosystems and Environment 233 (2016) 325–335
Seasons’ average rainfall in a
week

18. District Wise Average Yield Loss
Difference between
Average Wheat Yield
in 2015 and 2014 (in percent)
Ambala 3.70
Fatehabad 6.63
Jind 9.98
Kaithal 1.98
Karnal 4.75
Kurukshetra 9.57
Panipat 6.34
Sirsa 3.22
Sonipat 12.55
Yamunanagar 6.22
Total 6.79

19. In bad year, yield loss in ZT plot is ~30% lower
Yield in ZT plot is higher during both bad and normal year
Wheat yield (quintal/acre) Pre-period = Mean of
2011 to 2014
Pre-Period =
2014 Only
Post-period Dummy (2015) -1.8127*** -1.4869***
(0.1268) (0.1198)
ZT Plots Dummy 0.9036*** 0.7078***
(0.2166) (0.2488)
ZT Dummy X Post 0.4374* 0.4188*
(0.2438) (0.2111)
Constant 17.2406*** 16.9843***
(0.0910) (0.0962)
No of Observations 1586 1532
Adj-R2 0.2994 0.2409
1. Less water stagnation
2. Less logging due to stronger root systems
Not just the yield
loss, even the quality
also affected
(hailstorm)

20. Concluding remarks
• Conservation agriculture methods like zero till (ZT) make crops more resilient
to weather shocks
• Yield in ZT plot is higher during both bad and normal year
• In bad year, yield loss was lower by 1.05–1.10 quintals per hectare in ZT plots
compared to 3.73-4.53 quintals per hectare in the CT plots.
• Adoption of ZT helped in reducing crop loss in wheat by 24–28 percent,
valued at 1,523–1,595 Indian rupees (Rs.) per hectare (approximately US$22.50
per hectare).
• Our findings suggest that ZT adopters should get a discount on the premium
for crop insurance.