1 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 Research on World Agricultural Economy https://ojs.nassg.org/index.php/rwae Copyright © 2022 by the author(s). Published by NanYang Academy of Sciences Pte. Ltd. This is an open access article under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) License. (https://creativecommons.org/licenses/by-nc/4.0/). DOI: http://dx.doi.org/10.36956/rwae.v3i3.559 Received: 17 June 2022; Received in revised form: 11 July 2022; Accepted: 19 July 2022; Published: 5 August 2022 Citation: Kumar, K.N.R., 2022. Technical Efficiency of Rice Farmers in Telangana, India: Data Envelopment Analysis (DEA). Research on World Agricultural Economy. 3(3), 559. http://dx.doi.org/10.36956/rwae.v3i3.559 *Corresponding Author: K. Nirmal Ravi Kumar, Department of Agricultural Economics, Agricultural College, Bapatla, Acharya NG Ranga Agricultural University(ANGRAU), Andhra Pradesh, India; Email: drknrk@gmail.com RESEARCH ARTICLE Technical Efficiency of Rice Farmers in Telangana, India: Data Envelopment Analysis (DEA) K. Nirmal Ravi Kumar* Department of Agricultural Economics, Agricultural College, Bapatla, Acharya NG Ranga Agricultural University (ANGRAU), Andhra Pradesh, India Abstract: It is known that inability of the farmers to exploit the available production technologies results in lower efficiencies of production. So, the measurement of technical efficiency in agricultural crops in developing countries like India gained renewed attention since the late 1980s from an increasing number of researchers. Accordingly, the present study has employed Data Envelopment Analysis (DEA) and Malmquist Total Factor Productivity Index to ascertain the Technical Efficiency of rice productivity (2021-2022) and its changes over the study period (2019-2020 to 2021- 2022) respectively in Telangana, India. This study was based on secondary data pertaining to rice productivity (output variable), fertilizer doses (NPK), seed rate, water applied and organic manure (input variables). The findings of Data Envelopment Analysis revealed that the overall mean technical efficiency score across all the Decision-Making Units was 0.860 ranged between 0.592 to 1.000. So, the Decision-Making Units, on an average, could reduce their input usage by 14 per cent and still could produce the same level of rice output. Further, fertilizers (60.54 kg/ha); seed (5.63 kg/ha); water (234.48 mm) and organic manure (3.76 t/ha) use can be reduced without affecting the current level of rice productivity. Malmquist Total Factor Productivity indices (2019-2020 to 2021-2022) revealed that the mean scores of technical efficiency change, pure technical efficiency change and scale efficiency change are more than one (1.153, 1.042 and 1.009 respectively), unlike technological change (0.983). All the Decision-Making Units showed impressive progress with reference to technical efficiency change (1.112) and it is the sole contributor for Total Factor Productivity change in rice cultivation. The DEA results suggest that farmers should be informed about the use of inputs as per the scientific recommendations to boost the technical efficiency of rice productivity in Telangana. It also calls for policy initiatives for distribution of quality inputs to the farmers to boost technical efficiency in rice production. Keywords: Constant returns to scale; Malmquist total factor productivity index; Decision Making Units; Telangana 1. Introduction FAO during the International Year of Rice of 2004 stated that “Rice contributes to many aspects of soci- ety and therefore can be considered a crystal or prism through which the complexities of sustainable agriculture and food systems can be viewed. The issues related to http://dx.doi.org/10.36956/rwae.v3i3.559 mailto:drknrk@gmail.com https://orcid.org/0000-0002-0041-572X 2 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 rice production should not be viewed in isolation but in the framework of agricultural production systems through ecological and integrated systems” [1]. This statement highlights rice not only as one of the most important food crops world-wide but also an intricate part of socio-cultural influencer of many people’s lives. Rice is grown in about 120 countries and China leads other countries in the world with a production of 214 million tonnes followed by India with 116 million tonnes and these two countries together contribute over 50 percent of the world’s output in 2019. Nine out of the top ten and 13 out of the top twenty rice- producing countries are in Southeast-Asia [2]. Rice contributed more than 40 percent of the total food grains production in India in 2019 and accounted for 21 percent of global rice production. West Bengal, Uttar Pradesh, Punjab, Andhra Pradesh, Odisha and Telangana are the leading rice producing States in India [3]. Boosting the yields of rice is very much critical for the well-being of millions of rice producers and consumers in India, as around 22 percent of the population still lie Below Poverty Line (BPL) in 2018 [4]. Further, the demand for rice is pro- jected at 137.3 million tonnes by 2050 [5]. To accomplish these goals, the rice yields must be increased by around 42 percent i.e., from the present level of 2393 kg/ha (in 2011-2012) to 3400 kg/ha. Telangana State is emerging as the ‘Rice Bowl of India’ because, in a short span of five years, the area un- der rice cultivation has doubled from 0.91 million hec- tares in 2014-2015 to 1.93 million hectares in the 2018- 2019. Recently, with the completion of Kaleshwaram Lift Irrigation Scheme, the extent of rice cultivation in Telangana has increased in just one year from 1.93 million hectares in 2018-2019 to 2.88 million hectares in 2019- 2020 and accordingly, production shot up from 6.6 mil- lion tonnes to 10.5 million tonnes during this reference period 2022 [6]. So, the adequate water resources and other inputs like seed, fertilizers subsidy, free power etc., being provided by the State Government enabled the farmers to take up rice cultivation. However, the statistical data available in the offices of Joint Director of Agriculture in Telangana has revealed drastic variations in rice produc- tivity and resources usage. These variations in resources usage contributed to low productivity of rice (compared to potential) and this may arise owing to lower Technical Efficiency (TE). This is an indicator of presence of techni- cal inefficiency in rice productivity across the districts in Telangana. Considering the socio-economic importance of rice farming in this state, there seems to be a research need for investigating the extent of such inefficiencies. It, therefore, calls for a scientific inquiry on TE of rice pro- duction in Telangana, which would be of much relevance for farmers, researchers, policymakers and other stake- holders to take appropriate measures for enhancing TE in rice productivity, efficient management practices and con- sequent, sustainable agricultural planning. In this context, this study formulated the following three research ques- tions viz., what is the TE of rice productivity across all the districts in Telangana? What is the trend in TEs of rice productivity over a period of time? What input quantities are required to produce at the technically efficient point on the production frontier? [4] So, this study gives an impor- tant direction to farmers for employing right combination of productive resources in the rice production programme. Further, the lack of empirical studies in Telangana on this pertinent issue has prompted the researcher to conduct sci- entific enquiry across the 32 rice producing districts with the following specific objectives: ● To estimate TEs in rice productivity across the dis- tricts or Decision-Making Units (DMUs) in Telan- gana ● To find out the potentials for reduction in the levels of critical inputs across the DMUs. ● To analyze the trends in TE and sources of TFP of rice over the study period. 2. Review of Literature There have been a sizeable number of studies on ef- ficiency measure in the field of agriculture through apply- ing DEA approach because of its non-parametric nature. A review of literature on application of DEA in measuring efficiency in crop productivity is presented here under. Tolga et al. (2009) [7] measured TE and determinants of TE of rice farms in Marmara region, Turkey. Their study revealed that mean TE score of sample rice farms was 0.92 and ranged between 0.75 to 1.00 implying that they can reduce the inputs usage by eight per cent without affecting the level of output. Fabio (2015) [8] studied both technical and scale ef- ficiency in the Italian citrus farming through employing both DEA and Stochastic Frontier Analysis (SFA). The findings revealed that though the estimated TE from SFA is on par with the DEA, the scale efficiency realized from SFA is found higher compared to DEA. Both the models revealed that TE and scale efficiency were positively in- fluenced by farm size, unlike number of plots of land and location of farm in a less-favoured area. Sivasankari et al. (2017) [9] employed DEA to analyze the TE of rice farms in Cauvery delta zone of Tamil Nadu. The findings revealed that TE index ranged from 0.41 to 1.00 under both Constant Returns to Scale (CRS) and 0.48 to 1.00 under Variable Returns to Scale (VRS) speci- fications with mean TEs of 0.76 and 0.81 respectively. 3 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 Regarding scale efficiency, majority of the farms (81%) exhibited showed Increasing Return to Scale (IRS). The study also inferred that there is excess use for all inputs especially for fertilizers like potash, phosphorus and farm yard manure among the sample farms. Bingjun and Xiaoxiao (2018) [10] analyzed rice produc- tion efficiency based on DEA-Malmquist Indices in Henan Province of China. The results showed that from the time dimension (2006-2016), the comprehensive TE change, the technological progress change, the pure TE change, the scale efficiency change and the TFP change have not shown much improvement. However, from the perspec- tive of spatial dimension, the TFP of rice in all provinces is less than one, mainly because the production technol- ogy was not fully utilized in each area. So, they suggested strengthening of research and development, dissemination of advanced production technology, proper allocation of production factors etc., should deserve special attention to ensure efficiency improvement and thereby, food security of the country. Joseph et al. (2018) [11] employed DEA to measure TE of rice production in the Centre region of Cameroon con- sidering both CRS and VRS assumptions. The findings re- vealed that the mean TE score is 0.67 and 0.95 at the CRS and VRS respectively and with a mean scale efficiency of 0.70. Shamsudeen et al. (2018) [4] employed input-oriented DEA to analyze the TE of rice production in northern Ghana for the 2011-2012 cropping season. The mean TE score was 77 percent implying the farmers employed higher doses of inputs viz., chemical fertilizer, seed, weedicides and hired labour than their prescribed opti- mum. Around 84.4 of the sample farms experienced IRS, while 5.6 per cent experienced Decreasing Returns to Scale (DRS). Nazir and Abdur (2022) [12] analysed the TFP of cash crops viz., sugarcane, cotton, and rice in Pakistan by em- ploying Malmquist productivity index. The study decom- posed the TFP of cash crops into technical change and TE change. The findings showed an increase in the TFP of selected cash crops in Pakistan by 2.2 percent and this is mainly attributed to technical change. So, the researchers emphasized on increasing both research and extension investments to provide better seed varieties, better infra- structure, and timely credit facilities. 3. Analytical Framework and Methodology This study uses a two-step approach. In the first step, the DEA model was employed to measure TE of DMUs as an explicit function of discretionary variables pertaining to Kharif season, 2021-2022. In the second step, DEA-based Malmquist Index was used to analyze the trends in TE of rice productivity during Kharif season across the DMUs over the reference period, 2019-2020 to 2021-2022. This study considered all the 32 DMUs in Telangana consider- ing output variable (rice productivity) and input variables (seed rate, fertilizer doses (NPK), water applied during crop growth period and organic manure). The secondary data on these variables are collected from respective Joint Director of Agriculture Offices at DMU level. 3.1 DEA This linear programming tool was employed to meas- ure the TE of rice productivity in Telangana considering input-oriented-CRS model [13-15]. In this model, there are 32 DMUs and each DMU uses four inputs (K) and pro- duces one output (M). For the ith DMU, these are repre- sented by the vectors xi and yi, respectively. The selected inputs and output are represented by a K × N input matrix denoted by X, and M × N output matrix denoted by Y respectively. For the ith DMU, the efficiency score θ is ob- tained by solving the linear programming as follows: minθλ θ st -yi + Y λ > 0 θxi - Xλ > 0 λ > 0 Here, θ indicates the TE score of input-oriented CRS of the DMU under evaluation. If the value of θ = 1, it implies the DMU is functioning on the production frontier with 100 per cent of efficiency and hence, there is no need for changing the level of resources employed in the produc- tion. On the contrary, if θ < 1, it implies the DMU under consideration is relatively inefficient and thus, it could reduce the level of inputs usage without affecting the out- put [9]. 3.2 Malmquist TFP Index: Input Oriented, CRS This index based on DEA is employed to study the trends in TE, technological change, Pure TE change, scale efficiency change and changes in TFP of rice productiv- ity during 2019-2020 to 2021-2022 across the selected 32 DMUs. So, the average values of the selected output and input variables during this reference period are subjected to DEA-based Malmquist Index analysis. The change in productivity from the period t to t + 1 is calculated using the following formula [9,16]: M y x y x D y x D y x D yt t t t t t t t t t 1 1 1 1 1 1 1 1 + + + + ( , , , )= Xt+1 t+1 ( , ) ( , ) ( tt t t t t t t t x D y x M y x y x D + + + +       1 1 1 1 1 1 1 2, ) ( , ) / ( , , , )=t+1 t+1 11 1 1 1 1 1 1 1 1 1 1 1 t t t t t t t t t t t t y x D y x D y x D y x + + + + + + + + ( , ) ( , ) ( , ) ( , ) ** DD y x D y x D y x t t t t t t I t t t 1 1 1 1 1 2( , ) ( , ) / ( , ) + −                 = miin min min θλ θλ θ θ θD y x D y x I t t t I t t t + + + − + −   =   = 1 1 1 1 1 1 ( , ) ( , ) λλ θλ θ θD y xI t t t( , )+ + −   =1 1 1 min (1) where, M1 = Malmquist Productivity Change Index 4 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 D1 = Input distance functions [15] y = the level of output(s) x = the level of input(s); and t = time Equation (1) is decomposed as:M y x y x D y x D y x D yt t t t t t t t t t 1 1 1 1 1 1 1 1 + + + + ( , , , )= Xt+1 t+1 ( , ) ( , ) ( tt t t t t t t t x D y x M y x y x D + + + +       1 1 1 1 1 1 1 2, ) ( , ) / ( , , , )=t+1 t+1 11 1 1 1 1 1 1 1 1 1 1 1 t t t t t t t t t t t t y x D y x D y x D y x + + + + + + + + ( , ) ( , ) ( , ) ( , ) ** DD y x D y x D y x t t t t t t I t t t 1 1 1 1 1 2( , ) ( , ) / ( , ) + −                 = miin min min θλ θλ θ θ θD y x D y x I t t t I t t t + + + − + −   =   = 1 1 1 1 1 1 ( , ) ( , ) λλ θλ θ θD y xI t t t( , )+ + −   =1 1 1 min (2) The first term on the RHS of the above equation in- dicates the change in input-based TE between the years t and t + 1, while the second term indicate the change in technology between the selected periods. From the above Equation (2), it can be inferred that the product of change in TE and technological change gives a measure of change in TFP. If the TFP is > 1, it implies the TFP is increasing during the selected periods (t and t + 1) and vice versa and if the TFP = 1, it implies no change [15]. To obtain the change in Malmquist Indices, the following series of Lin- ear Programing Problems (LPPs) are to be solved [16]: 1 ( , )tI t tD y x minθλθ −   =  (3) st -yit + Yt λ > 0 θxit - Xtλ > 0 λ > 0 11 1 1( , ) t I t tD y x minθλθ −+ + +  =  (4) st -yi,t+1 + Yt+1 λ > 0 θxi,t+1 - Xt+1λ > 0 λ > 0 11( , )tI t tD y x minθλθ −+  =  (5) st -yit + Yt+1 λ > 0 θxit - Xt+1λ > 0 λ > 0 1 1 1( , ) t I t tD y x minθλθ − + +  =  (6) st -yi,t+1 + Yt λ > 0 θxi,t+1 - Xtλ > 0 λ > 0 These LPPs are solved for each firm in the sample. Therefore, given the number of periods (T) and number of observations (N), [N × (3T - 2)] problems are to be solved. This study considered all the 32 districts (as the DMUs) in Telangana and the relevant secondary data are obtained from respective Joint Director of Agriculture Offices. Rice yield (kg/ha) is considered as the output, whereas seed rate, fertilizer doses (NPK), annual rainfall received (mm) and organic manure are considered as inputs. The aver- age values of the output and input variables (2019-2020 to 2021-2022) are collected for the DMUs and subjected to DEA and DEA-based Malmquist TFP Index analysis for estimating the TE and change in TE respectively. The efficiency analysis and Malmquist Index for efficiency change over time has been done using the DEAP version 2.1 program developed by Coelli, 1996 [15]. 3.3 Sample Adequacy Test According to Cooper et al., 2007 [17], the thumb rules for sample size acceptable for conducting DEA should be either greater than or equal to the product of inputs (X) and outputs (Y) or the sample size should be at least three times the sum of the number of X and Y variables. So, considering X = 4 and Y = 1, the sample size of 32 DMUs in Telangana confirms the sample adequacy for conduct- ing DEA. 4. Results and Discussion 4.1 Summary Statistics of Output and Input Vari- ables Table 1 shows that the average productivity of rice in Tel- angana was estimated as 3288.28 kg/ha with maximum and minimum productivity levels of 3705 kg/ha and 2720 kg/ha respectively with the estimated Coefficient of Variation (CV) of 59.928 percent. There exist larger variations across the DMUs in terms of inputs usage viz., fertilizer doses, seed rate, water applied and organic manure. Re- garding the quantity of fertilizers (NPK) applied, it ranged from 110 kg/ha to 350 kg/ha with an average value of 263.37 kg/ha and CV of 55.798 percent. The application of chemical fertilizers is on the higher side among all the DMUs compared to the recommended dosages (NPK @ 120:40:40 kg ha-1 for short duration varieties; NPK @ 150:50:60 kg ha-1 for medium duration varieties and NPK @ 150:50:80 kg ha-1 for long duration varieties). Similarly, average quantity of water applied was 1190.01 mm with minimum and maximum values of 780 mm and 1670 mm respectively and with a CV of 41.579 percent. For majority of the DMUs (87%), the actual quantity of water applied is higher than the scientific recommenda- tion of 1200 mm to 1250 mm. The quantity of seed used pitches between 17 kg/ha and 28 kg/ha with a mean value of 23.47 kg/ha and with a CV of 38.508 percent. A close examination of the data collected, the actual seed used by all the DMUs is considerably higher compared to the recommended level of 20 kg/ha. However, the CV is 5 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 slightly lower with respect to organic manure applied for rice cultivation (24.617%) and across the DMUs it varied between 2 t/ha to 12 t/ha with an average of 8.37 t/ha. The higher CVs of inputs is an indicative of presence of technical inefficiency in contributing to the productivity of rice across the DMUs in Telangana. Again for major- ity of the DMUs, the quantity of organic manure applied is higher compared to the recommended dosage of 8 t/ha to 10 t/ha. Though the application of this input is on the higher side, it is heartening that the farmers realized the importance of organic farming in producing both cost- effective and quality output. 4.2 DEA-Input-oriented CRS The results of CRS TE scores (θ) along with bench- marking DMUs and peer lambda weights (λj) for the DMUs under evaluation are presented through Table 2. The findings revealed that only nine out of 32 DMUs namely, Karimnagar, Jogulamba Gadwal, Kamareddy, Khammam, Mahabubnagar, Medak, Medchal-Malkajgiri, Narayanpet and Suryapet received TE score of ‘1’. This implies they are the best performing DMUs in Telangana, as they are operating on the efficiency frontier in the peer group. For the remaining 23 DMUs, the TE scores are less than one ranging between 0.592 (Warangal-Rural) to 0.931 (Jagtial) with a mean TE score of 0.806. This implies pres- ence of relative technical inefficiency in rice productivity, as these 23 DMUs are operating below the efficiency fron- tier. So, these 23 DMUs could reduce current level inputs to the tune of 19.4 per cent without affecting the rice pro- ductivity. The overall mean TE score for all the 32 DMUs was estimated as 0.860 indicating relative technical inef- ficiency is to the extent of 14 percent. This means that, on an average, the DMUs can check over-use of current level input resources to the tune of 14 percent without affecting the rice productivity in the State. The DMU, Warangal- Rural is with the lowest TE score of 0.592 followed by Vikarabad (0.611), Mulugu (0.661), Mancherial (0.717) etc., and all are lying at the bottom of the performance ladder (Table 3). So, these DMUs could reduce the cur- rent level of input usage by 40.80, 38.90, 33.90 and 28.30 percents respectively without affecting their correspond- ing rice productivity levels. For the inefficient DMUs (θ < 1), the benchmarking DMUs are given in Column 4 and it will guide the former to reduce their inputs us- age corresponding to the benchmarking DMUs [9,10]. For example, Suryapet and Kamareddy are the benchmarking DMUs for Adilabad with respective lambda (λj) weights of 0.903 and 0.023. With the λj weights, the benchmark- ing DMUs form linear combinations with the inefficient DMUs in terms of efficiency perspective. For the efficient DMUs (with TE score of 1.000), the benchmarking DMUs are peer of themselves with λj weights of ‘one’. The comparative picture of efficient and inefficient DMUs in terms of TE scores (Figure 1) indicate that the dark color bars represent the DMUs (9) operating on the efficiency frontier (with TE scores of ‘1’) and the light color bars denote the DMUs (23) lying below the efficien- cy frontier (with TE scores of ‘<1’). So, the vertical gap between efficient and inefficient DMUs indicate the extent of technical inefficiencies of 23 DMUs. 4.3 Determining Optimal Level of Inputs Utiliza- tion from the CRS Model From Table 2, it was inferred that there are nine techni- cally efficient DMUs and 23 technically inefficient DMUs. Accordingly, DMU-wise projected input quantities and possible reductions across inefficient DMUs was comput- ed [14,15] to realize higher TE scores without affecting their current level of rice productivity (Table 4). The projected input quantities indicate the minimum quantities of select- ed inputs required across the DMUs to produce technical- ly efficient output on the production frontier. So, the dif- ference between actual and projected quantities of inputs (obtained from the one-stage DEA) indicate the possible input quantity reductions. For example, the actual use of fertilizers, seed rate, water applied and organic manure for the DMU, Adilabad are 205.935 kg/ha, 32.67 kg/ha, Table 1. Summary Statistics of output and input variables (2021-2022) Item Minimum Maximum Mean Std. Deviation CV Rice productivity (kg/ha) 2720 3705 3288.28 1970.60 59.928 Fertilizer Use (NPK) (kg/ha) 110 350 263.37 146.96 55.798 Seed rate (kg/ha) 17 28 23.47 9.04 38.508 Water applied (mm) 780 1670 1190.01 494.79 41.579 Organic manure (t/ha) 2 12 8.37 2.06 24.617 6 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 Table 2. Results of Input-oriented CRS Sl. No. Districts CRS Technical Efficiency (θ) Benchmarking Districts Peer Weights (λj) in Order of Benchmarking Districts 1 Adilabad 0.828 Suryapet, Kamareddy 0.903, 0.023 2 Bhadradri Kothagudem 0.875 Medak, Karimnagar, Khammam 0.179, 0.631, 0.214 3 Karimnagar 1.000 Karimnagar 1.000 4 Jagtial 0.931 Kamareddy 0.920 5 Jangaon 0.803 Karimnagar, Medchal-Malkajgiri, Narayanpet 0.668, 0.028, 0.288 6 Jayashankar Bhupalpally 0.858 Suryapet, Kamareddy 0.334, 0.566 7 Jogulamba Gadwal 1.000 Jogulamba Gadwal 1.000 8 Kamareddy 1.000 Kamareddy 1.000 9 Khammam 1.000 Khammam 1.000 10 Kumuram Bheem 0.812 Khammam, Karimnagar, Suryapet 0.558, 0.403, 0.009 11 Mahabubabad 0.868 Kamareddy Karimnagar, Mahabubnagar, Bhadradri Kothagudem 0.357, 0.371, 0.137, 0.214 12 Mahabubnagar 1.000 Mahabubnagar 1.000 13 Mancherial 0.717 Karimnagar, Kamareddy, Suryapet 0.343, 0.505, 0.035 14 Medak 1.000 Medak 1.000 15 Medchal-Malkajgiri 1.000 Medchal-Malkajgiri 1.000 16 Mulugu 0.661 Khammam, Karimnagar, Suryapet 0.255, 0.469, 0.183 17 Nagarkurnool 0.889 Narayanpet, Mahabubnagar 0.604, 0.365 18 Nalgonda 0.834 Narayanpet, Jogulamba Gadwal, Suryapet 0.631, 0.120, 0.196 19 Narayanpet 1.000 Narayanpet 1.000 20 Nirmal 0.724 Suryapet, Narayanpet, Mahabubnagar, Kamareddy 0.594, 0.036, 0.094, 0.245 21 Nizamabad 0.848 Suryapet, Karimnagar, Kamareddy 0.077, 0.523, 0.356 22 Peddapalli 0.838 Karimnagar, Narayanpet, Kamareddy Suryapet 0.028, 0.319, 0.488, 0.226 23 Rajanna Sircilla 0.836 Karimnagar, Mahabubnagar, Kamareddy, Narayanpet 0.583, 0.115, 0.161, 0.136 24 Rangareddy 0.869 Karimnagar, Medchal-Malkajgiri, Narayanpet 0.174, 0.089, 0.694 25 Sangareddy 0.775 Karimnagar, Narayanpet, Mahabubnagar 0.396, 0.456, 0.205 26 Siddipet 0.819 Karimnagar, Medak, Narayanpet, Suryapet 0.323 0.01,1 0.059, 0.408 27 Suryapet 1.000 Suryapet 1.000 28 Vikarabad 0.611 Suryapet, Narayanpet, Jogulamba Gadwal 0.101, 0.524, 0.211 29 Wanaparthy 0.917 Narayanpet 0.947 30 Warangal (Rural) 0.592 Suryapet, Narayanpet, Kamareddy, Mahabubnagar 0.021, 0.602, 0.224, 0.030 31 Warangal (Urban) 0.804 Kamareddy, Mahabubnagar, Suryapet 0.195, 0.533, 0.201 32 Yadadri Bhuvanagiri 0.819 Suryapet, Narayanpet, Jogulamba Gadwal 0.017, 0.895, 0.173 Average of all districts 0.860 Source: Authors’ estimation from DEAP version 2.1 [15] 7 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 Table 3. Frequency distribution and summary statistics on overall TE, pure TE and Scale efficiency measures of select- ed DMUs Efficiency level No. of DMUs Per cent DMUs 0.501-0.600 1 3.12 Warangal (rural) 0.601-0.700 2 6.25 Mulugu, Vikarabad 0.701-0.800 3 9.38 Mancherial, Niirmal, Sangareddy 0.801-0.900 15 46.88 A d i l a b a d , B h a d r a d r i K o t h a g u d e m , J a n g a o n , J a y a s h a n k a r Bhupalpally, Kumuram Bheem, Mahabubabad, Nagarkurnool, Nalgonda, Nizamabad, Peddapalli, Rajanna Siricilla, Rangareddy, Siddipet, Warangal (urban), Yadadri Bhuvanagiri 0.901-0.999 2 6.25 Jagtial, Wanaparthy 1.000 9 28.13 Karimnagar, Jogulamba Gadwal, Kamareddy, Khammam, Mahbubnagar, Medak, Medchal-Malkajgiri, Narayanpet, Suryapet Total 32 100.00 Minimum 0.592 Maximum 1.000 Mean 0.860 Source: Authors’ estimation from DEAP version 2.1 [15] Figure 1. Position of the DMUs in relation to TE scores 8 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 1511.301 mm and 10.215 t/ha respectively, whereas the projected input values obtained from the model for main- taining the same productivity (3124.73 kg/ha) are 145.395 kg/ha, 27.04 kg/ha, 1276.821 mm and 6.455 t/ha respec- tively. So, the estimated differences between the actual and projected input values (fertilizers 60.54 kg/ha; seed use 5.63 kg/ha; water applied 234.48 mm and organic manure 3.76 t/ha) indicate their excess use in rice produc- tion. Hence, this excess use of inputs should be reduced for Adilabad without affecting rice productivity. The same explanation can be offered for other technically inefficient DMUs. However, for the efficient DMUs with TE score 1.000, the gap between actual and projected input usage is around zero, as they are already operating on the produc- tion frontier (the best performing DMUs) and hence, there is no scope for reduction in the existing level of inputs usage. At the pooled (State) level i.e., considering the average of all the DMUs, there is overuse of fertilizers, seed use, water applied and organic manure to the tune of 53.998 kg/ha, 6.528 kg/ha, 86.436 mm and 2.249 t/ha re- spectively, as the production scenario of rice in dominated by technically inefficient DMUs (23) compared to only nine technical efficient DMUs. So, it is felt appropriate to compare the extent of inputs usage between technically efficient DMUs and technically inefficient DMUs in terms of rice productivity in Telan- gana. As shown through Table 5, the efficient DMUs (n = 9) employed on an average of 170.184 kg/ha of fertilizer, 21.667 kg/ha of seed, 1275.986 mm of water applied and 5.000 t/ha of organic manure to produce a yield of 3317 kg/ha of rice. However, for the inefficient DMUs (n = 23), to move up to the production level of the efficient DMUs, they should check excess application of fertilizers by 40.105 kg/ha, seed by 3.724 kg/ha, water use by 36.100 mm and organic manure by 2.870 t/ha in order to boost rice productivity by 778 kg/ha [4]. 4.4 Trends in TE of DMUs - Malmquist TFP In- dex Table 6 portrayed the Malmquist indices for each DMU during the period 2019-2020 to 2021-2022 [18]. The find- ings revealed that with reference to TE change index, 78 percent of the DMUs have made progress (TE change value >1.000) and remaining 22 percent of DMUs have regressed (TE change value <1.000). The top three DMUs that showed progress with reference to TE change in- clude: Nizamabad (48.3%), Nagarkurnool (45.5%) and Sangareddy (43.4%) and the top three DMUs that are regressed in terms of TE change are Kumuram Bheem (30.3%), Jagtial (22.2%) and Khammam (19.5%). It is heartening that the mean score for TE change in Telan- gana is more than 1 (i.e. 1.153) and this shows that the DMUs as a whole have witnessed impressive performance in TE change of rice productivity during the reference pe- riod [9,10,16]. However, it is disappointing that 56% of the DMUs have regressed with reference to technological change during the above reference period and hence, the mean score of technological index in Telangana is less than one (0.983). The top three DMUs that are regressed include: Mulugu, Medak and Narayanpet with 13.6 percent, 12.9 percent and 12.8 percent respectively. It is found interest- ing that majority of the DMUs have showed progress with reference to pure TE change (53%) and scale efficiency change (59%). Further, 75 percent of the DMUs showed progress with reference to TFP change and remaining 25 percent of DMUs have regressed. The top three DMUs viz., Nizamabad, Karimnagar and Sangareddy have enjoyed TFP growth of 42.1 percent, 40.1 percent and 35.2 percent respectively. At the state level, the results are found encouraging with reference to TE change (15.3%), pure TE change (4.2%), Scale efficiency change (0.9%) and TFP change (11.2%). So, on comparing the TE change and technological change, it can be inferred that the pro- gress in TFP change is purely from TE change during the reference period. The break-up of Malmquist indices across the selected periods viz., 2019-2020 to 2021-2022 (Table 7) revealed that TE change has showed increasing trend during from 1.139 (2019-2020) to 1.179 (2021-2022) with mean TE change of 1.153. This shows that there is a gradual pro- gress in terms of TE change for enhancing rice productiv- ity in the State during the overall reference period. On the contrary, the mean technological change was regressed during the reference period with 0.983. Though techno- logical change was marginally progressed (2.7%) during 2021-2022 compared to 2020-2021, the mean technologi- cal change is regressed during the overall reference pe- riod. It is also interesting that the DMUs have marginally progressed in terms of pure TE change (4.2%) and Scale Efficiency change (0.9%) during the reference period. The TFP change has witnessed progress in the State with an average value of 1.112. Considering these trends, it can be inferred that at State level, pure TE change and scale ef- ficiency change have almost remained stagnant and hence, the gain in TFP of rice in Telangana is solely due to TE change of inputs over time. 9 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 Table 4. Results of Input-oriented CRS: Single Stage Calculation S.No Districts Projected Input Quantities Possible Inputs Reduction (Actual - Projected) Fertilizer Use (NPK) (kg/ha) Seed rate (kg/ha) Water applied (mm) Organic manure applied (t/ha) Fertilizer Use (NPK) (kg/ha) Seed rate (kg/ha) Water applied (mm) Organic manure applied (t/ha) 1 Adilabad 145.395 27.040 1276.821 6.455 60.540 5.630 234.480 3.760 2 Bhadradri Kothagudem 181.191 36.177 1399.781 3.501 51.620 5.160 130.776 1.000 3 Karimnagar 145.670 38.000 1232.784 3.000 0.000 0.000 0.000 0.000 4 Jagtial 100.580 34.959 1562.051 5.520 14.840 5.370 42.354 4.960 5 Jangaon 142.383 36.562 1122.035 3.212 69.900 14.100 91.800 0.900 6 Jayashankar Bhupalpally 114.741 31.185 1418.841 5.733 37.860 5.150 129.900 2.540 7 Jogulamba Gadwal 201.000 28.000 871.146 7.000 0.000 0.000 0.000 0.660 8 Kamareddy 109.330 38.000 1697.940 6.000 0.000 0.000 0.000 0.660 9 Khammam 205.670 29.330 1649.358 5.000 0.000 0.000 0.000 0.660 10 Kumuram Bheem 174.908 31.947 1429.609 4.061 80.860 7.390 185.790 1.220 11 Mahabubabad 137.995 40.209 1400.504 5.207 42.020 6.120 71.058 0.920 12 Mahabubnagar 115.000 39.000 1025.550 8.000 0.000 0.000 0.000 -0.660 13 Mancherial 110.722 33.239 1328.369 4.305 87.220 13.090 194.862 2.720 14 Medak 252.330 33.000 1499.022 3.000 0.000 0.000 0.000 -0.660 15 Medchal-Malkajgiri 208.000 53.000 1217.412 2.000 0.000 0.000 0.000 0.000 16 Mulugu 149.843 30.627 1250.676 3.966 153.640 15.710 290.424 3.400 17 Nagarkurnool 124.410 34.599 930.028 5.341 31.180 7.070 38.850 3.320 18 Nalgonda 141.268 30.309 953.896 4.739 56.140 6.020 63.162 5.860 19 Narayanpet 136.330 33.670 918.846 4.000 0.000 0.000 0.000 -0.660 20 Nirmal 136.427 31.388 1359.256 6.520 103.820 11.950 172.374 4.300 21 Nizamabad 127.272 35.636 1354.761 4.242 45.460 6.360 120.462 1.520 22 Peddapalli 136.674 36.894 1466.075 5.869 52.660 7.110 94.134 2.260 23 Rajanna Sircilla 134.357 37.354 1235.797 4.181 52.620 7.310 80.676 1.640 24 Rangareddy 138.404 34.680 960.221 3.475 41.860 6.650 48.396 1.060 25 Sangareddy 143.316 38.366 1116.504 4.648 83.360 12.970 108.246 2.700 26 Siddipet 122.589 26.480 1029.404 4.095 54.160 5.850 75.798 1.820 27 Suryapet 158.330 29.000 1371.816 7.000 0.000 0.000 0.000 0.000 28 Vikarabad 129.762 26.471 803.320 4.276 187.140 16.860 170.538 6.120 29 Wanaparthy 129.065 31.876 869.879 3.787 70.540 5.790 26.154 7.760 30 Warangal (Rural) 113.414 30.574 993.618 4.142 156.500 21.090 228.534 6.380 31 Warangal (Urban) 114.463 34.042 1153.809 6.844 55.740 8.290 93.636 4.320 32 Yadadri Bhuvanagiri 159.532 35.477 996.773 4.913 138.260 7.860 73.548 1.500 Average of all Districts 145.012 33.972 1215.497 4.814 53.998 6.528 86.436 2.249 Source: Authors’ estimation from DEAP version 2.1 [15] Table 5. Comparison of average input use between inefficient and efficient farmers in Telangana Input use Number of DMUs Mean TE score Fertilizer Use (NPK) (kg/ha) Seed rate (kg/ha) Water applied (mm) Organic manure applied (t/ha) Yield (kg/ha) Average of efficient DMUs 9 1.000 170.184 21.667 1275.986 5.000 3317 Average of inefficient DMUs 23 0.806 210.289 25.391 1312.086 7.870 2539 Source: Authors’ estimation from DEAP version 2.1 (Coelli et al., 1996 [15]) 10 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 Table 6. Malmquist Index Summary for District Means Districts TE Change Technological Change Pure TE Change Scale Efficiency Change TFP Change Adilabad 0.879 0.979 0.867 1.013 0.861 Bhadradri Kothagudem 1.217 0.961 0.950 1.070 1.209 Karimnagar 1.410 1.092 1.000 1.010 1.401 Jagtial 0.778 1.042 0.855 0.910 0.811 Jangaon 1.161 0.957 1.115 1.042 1.112 Jayashankar Bhupalpally 1.117 1.048 0.863 0.970 1.108 Jogulamba Gadwal 1.113 0.996 1.000 0.941 1.108 Kamareddy 1.084 1.044 1.055 1.027 1.132 Khammam 0.805 0.979 0.853 0.944 0.788 Kumuram Bheem 0.697 0.918 0.726 0.960 0.640 Mahabubabad 0.826 1.015 1.000 0.826 0.838 Mahabubnagar 1.254 0.972 1.044 1.010 1.211 Mancherial 1.290 0.964 1.417 0.910 1.317 Medak 1.340 0.871 1.280 1.047 1.303 Medchal-Malkajgiri 1.390 1.014 1.044 1.044 1.284 Mulugu 1.113 0.864 1.084 1.026 1.064 Nagarkurnool 1.455 0.968 1.074 0.964 1.002 Nalgonda 1.061 1.010 1.012 1.049 1.072 Narayanpet 0.862 0.872 1.000 0.862 0.752 Nirmal 1.170 0.924 1.000 1.170 1.162 Nizamabad 1.483 1.000 1.265 1.123 1.421 Peddapalli 1.333 0.996 1.186 1.124 1.328 Rajanna Sircilla 1.123 0.952 0.953 0.992 1.048 Rangareddy 1.343 1.002 1.100 1.039 1.345 Sangareddy 1.434 1.015 1.250 0.987 1.352 Siddipet 1.165 1.046 1.068 1.090 1.089 Suryapet 1.026 0.970 1.000 1.026 0.995 Vikarabad 1.043 1.017 0.958 1.088 1.060 Wanaparthy 1.275 1.009 1.036 1.133 1.211 Warangal (Rural) 1.356 0.966 1.202 0.961 1.316 Warangal (Urban) 1.331 1.006 1.151 0.896 1.298 Yadadri Bhuvanagiri 0.954 0.983 0.922 1.035 0.938 Average of all Districts 1.153 0.983 1.042 1.009 1.112 Note: All Malmquist index averages are geometric means Source: Authors’ estimation from DEAP version 2.1 [15] 11 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 5. Summary and Conclusions Input-oriented DEA Model with CRS was employed in this study to analyze the TE in rice productivity in Telan- gana. Out of 32 DMUs considered, only nine DMUs are found technically efficient. The overall TE score for Tel- angana is 0.860 implying that the DMUs, on an average, could reduce their inputs usage by 14 per cent without af- fecting their current level of rice productivity. Compared to technically efficient DMUs, inefficient DMUs has to check the use of inputs viz, fertilizer use by 40.105 kg/ ha, seed use by 3.724 kg/ha, water use by 36.100 mm and organic manure use by 2.870 t/ha in order to boost yield by 778 kg/ha and to reach on the production frontier. Malmquist index analysis concluded that the progress in TFP change during 2019-2020 to 2021-2022 was purely due to TE change only. During this period, on an aver- age, the technological change has regressed and pure TE change and scale efficiency change have almost remained stagnant. 6. Policy Recommendations Policy suggestions from this study include: dissemina- tion of modern production technologies to the farmers, capacity building of farmers on Good Agricultural Prac- tices, supply of quality inputs to farmers at affordable prices etc., should deserve special attention. The poor and marginalized farmers cultivating rice in the State must be encouraged to join Farmer-Producer Organizations (FPOs) for availing need-based assistance, participation in various training programs and benefit from strengthened back- ward linkages to enhance TE of inputs usage. Further, to boost the technological change, the Government should enhance investments both in research and extension. The enabling environment in the State should be conducive to promoting private sector agricultural investments [19]. The coordination between demand-driven research and tech- nology dissemination should also be given priority. Conflict of Interest There is no conflict of interest. References [1] Inauguration address by Ms. Fresco, ADG, FAO on the occasion of the International Year of Rice 2004, (IYR) www.fao.com. (Accessed on 8/7/2022) [2] Rice paddy production in the Asia-Pacific region in 2020, by country, 2022. https://www.statista.com/ statistics/681740/asia-pacific-rice-paddy-produc- tion-by-country/. [3] Agricultural Statistics at a Glance, 2020. Ministry of Agriculture & Farmers Welfare, Department of Ag- riculture, Cooperation & Farmers Welfare, Govern- ment of India. [4] Abdulai, S., Nkegbe, P.K., Donkoh, S.A., 2018. As- sessing the technical efficiency of maize production in northern Ghana: The data envelopment analysis approach. Food Science & Technology, Cogent Food And Agriculture. [5] Mohapatra, T., Nayak, A.K., Raja, R., et al., 2013. Central Rice Research Institute. Cuttack: ICAR-Na- tional Rice Research Institute. Retrieved from http://www.crri.nic.in/ebook_crrivision2050_final_ 16Jan13.pdf. [6] Socio-Economic Outlook, 2022. Planning Depart- ment, Government of Telangana. [7] Tipi, T., Yildiz, N., Nargeleçekenler, M., et al., 2009. Measuring the TE and determinants of efficiency of rice (Oryza sativa) farms in Marmara region, Turkey. New Zealand Journal of Crop and Horticultural Sci- ence. 37, 121-129. [8] Fabio, A., 2015. Madau Technical and Scale Effi- ciency in the Italian Citrus Farming: A Comparison between SFA and DEA Approaches. Agricultural Economics Review. 16(2). [9] Sivasankari, B., Vasanthi, R., Prema, P., 2017. Determination of technical efficiency in Paddy farms Table 7. Malmquist Index Summary of Annual Means Year TE Change Technological Change Pure TE Change Scale Efficiency Change TFP Change 2019-2020 1.139 1.029 1.038 1.019 1.089 2020-2021 1.140 0.947 1.033 0.986 1.120 2021-2022 1.179 0.974 1.055 1.023 1.127 Mean 1.153 0.983 1.042 1.009 1.112 Note: All Malmquist index averages are geometric means Source: Authors’ estimation from DEAP version 2.1 [15] http://www.fao.com http://www.crri.nic.in/ebook_crrivision2050_final_16Jan13.pdf http://www.crri.nic.in/ebook_crrivision2050_final_16Jan13.pdf 12 Research on World Agricultural Economy | Volume 03 | Issue 03 | September 2022 of canal irrigated systems in Tamil Nadu: A data envelopment analysis approach. International Journal of Chemical Studies. 5(5), 33-36. [10] Li, B.J., Zhu, X.X., 2018. Analysis of Maize Produc- tion Efficiency Based on DEA-Malmquist Indexes: A Case Study of Henan Province. Journal of Agricul- tural Chemistry and Environment. 7, 176-187. [11] Joseph Serge Evouna Mbarga, Joël Sotamenou, Mar- tin Paul Jr.Tabe-Ojong and Ernest L. Molua, Tech- nical Efficiency of Maize Production in the Centre Region of Cameroon: A Data Envelopment Analysis (DEA), Developing Country Studies www.iiste.org, Vol.8, No.4, 2018 [12] Khan, N.U., Rehman, A., 2022. Decomposition of Total Factor Productivity of Cash Crops in Pakistan: A Malmquist Data Envelop Analysis. Journal of Eco- nomic Impact. 4(1), 139-144. [13] Banker, R.D., Charnes, A., Cooper, W.W., 1984. Some Models for Estimating Technical and Scale Inefficiencies in Data Envelopment Analysis. Man- agement Science. 30, 1078-1092. [14] Charnes, A., Cooper, W.W., Rhodes, E., 1978. Mea- suring the efficiency of decision making units. Euro- pean Journal of Operations Research. 2, 429-444. [15] Coelli, T.J., 1996. A Guide to DEAP Version 2.1: A Data Envelopment Analysis (Computer) Program (CEPA Working Papers No. 8/96). Armidale: Centre for Efficiency and Productivity Analysis (CEPA), University of New England, Department of Econo- metrics. [16] Benli, Y.K., Degirmen, S., 2013. The Application of Data Envelopment Analysis based Malmquist Total Factor Productivity Index: Empirical Evidence in Turkish Banking Sector. Panoeconomicus. 2(Special Issue), 139-159. [17] Cooper, W.W., Seiford, L.M., Tone, K., 2007. Data Envelopment Analysis: A Comprehensive Text with Models, Applications, References and DEA-Solver Software (Second Edition). New York: Springer Sci- ence + Business Media. [18] Malmquist, S., 1953. Index Numbers and Indiffer- ence Surfaces. Trabajos De Estatistica. 4(2), 209- 242. [19] Kumar, K.N.R., Babu, S.C., 2021. Can a Weath- er-Based Crop Insurance Scheme Increase the Tech- nical Efficiency of Smallholders? A Case Study of Groundnut Farmers in India. Sustainability. 13, 9327. DOI: https://doi.org/10.3390/su13169327 http://www.iiste.org