Rajesh, G.K Gandhigram Rural Institute-Deemed University, India

Transcription

1 1 Factors Influencing the Partial versus Complete Adoption of Component technologies in Bivoltine Hybrid Technology Package by farmers in Karnataka, India Abstract Rajesh, G.K Gandhigram Rural Institute-Deemed University, India Bivoltine Hybrid Technology is a HYV technology in Indian sericulture, individual component technologies of which are adopted by farmers at a differential rate. The Partial versus Complete Adoption of Component technologies in Bivoltine Hybrid Technology Package in Karnataka, India is analysed using a Poisson Count Regression model to identify the factors determining selection of technology components of the HYV package. This model is expected to give a better and truer picture of sericulture technology adoption in sericulture, as against the usual practice of classifying farmers into adopters and non adopters based on cut-off percentages of inputs / technologies purchased. The results reveal a few socio economic factors to be significant determinants of farmer s choice of adoption of individual technologies from the package of technologies, some of them going against the findings of already existing adoption literature. Introduction In spite of its small volume in global textile production 1, silk has importance in developing economies primarily because of its favourable socio economic consequences. The development of sericulture 2 had been the states priority in all practicing countries. The Indian sericulture industry is constrained by low productivity, low quality of rawsilk, price instability and import competition. With the advent of sophisticated power looms and relaxation in exim policies large quantities of high quality silk is being imported at prices lower to local silk. These constraints stem from the poor quantitative and qualitative performance of the prevalent, traditional low yielding silkworm varieties. To ameliorate the traditional variety silkworms, a superior hybrid namely Bivoltine Hybrid (BV) and HYV technology namely Bivoltine Hybrid Technology (BVHT) was evolved as early as 1970s and systematic popularisation efforts began during 1990s. But the hybrid technology has not diffused well. The Current percentage of Bivoltine hybrid silk production in India is below 10% only. This paper attempts to study the determinants of farmer s decision on BVHT package adoption, based on the level of adoption of individual technologies (of the BVHT package); employing Poisson Count Model, as specified by Octavio et al., The aim of the paper is to generate a clearer understanding of farmer level decision making on adoption of component technologies from a given package and the impact of differential adoption of component technologies in yield. Indian sericulture industry- importance, issues 1 Silk has a miniscule percentage of the global textile fibre market, less than 0.2%. This figure can be an under estimation, as the actual trading value of silk and silk products is much more impressive. The unit price for raw silk is roughly twenty times that of raw cotton. The annual turnover of the China National Silk Import and Export Corporation alone was US$ billion (ITC Silk review, 2001). 2 Sericulture is an activity comprising of cultivation of mulberry leaf which is fed to silkworms, reared to produce silk cocoons (Haumappa and Erappa, 1988).

2 2 India is world s second largest silk producer. It is also the largest consumer and importer of silk and silk goods (UN Comtrade data 2007). Sericulture is important to Indian economy as a cottage industry spread over villages employing nearly 56 lakhs people (Central silk Board data base, 2007). As a labour intensive activity practiced throughout the year it is identified as a means for rural employment generation and as a remedy for seasonal unemployment (Jayaram et al. 1998). The other merits of sericulture as an agro-industry are: its short gestation period to establish, potential for regular returns to the farmers, reelers and weavers, environment friendly production and processing technologies, potential for farm diversification, cash flow from rich to the poor, sustainability as a rural based activity involving family labour and women and high value addition to the end products with potential export markets (Benchamin and Giridhar, 2005). The Indian sericulture industry is currently faced with the problems of stagnation in production, low productivity, poor quality of produce, high cost of production and competition from cheap raw-silk imports. The sericulture industry is built upon two living organisms: an insect namely silkworm and its food plant namely mulberry 3. Thus the quality and quantity of raw-silk output are primarily dependent on the genetic potential of mulberry and the silkworm breeds. Almost 95% of silk produced in India is from traditional low yielding indigenous multivoltine silkworm varieties or cross breeds 4 which are relatively poor yielders (CSB database 2007). The cocoons produced by them are unsuitable for reeling in sophisticated reeling machines and the raw-silk produced from which is characterised by lower filament length and, lesser tensile strength leading to breakages making it unfit for high speed power loom weaving (Kumaresan et al., 2002). Thus the power-loom industry is heavily dependent on imported Chinese raw-silk which is of superior quality (Vasumathi, 2000 and Thomas et.al, 2005a).The indigenous raw silk is largely consumed by the handloom sector and partly by the power loom sector as weft 5 (Vasumathi, 2000). The import price of raw silk has been lower than the domestic raw silk, as the cost of production of Indian silk is high (Kumaresan, 2002). Moreover as shown by Naik and Babu (1993), the price of imported Chinese raw-silk is dependent on the prevailing prices of Indian raw-silk, though the causative nature of indigenous raw-silk price has not been clearly elucidated. This has affected the indigenous raw-silk prices and in turn the domestic cocoon prices (Tikku, 1999). This could probably be one of the reasons for large scale uprooting of mulberry plantations which resulted in considerable labour displacement in the farm sector (Central silk Board data base, 2007). There is a growing demand supply gap of raw silk in the domestic industry. Naik and Babu (1993) estimated that the total high quality silk production in India could meet at the most 60% of the estimated demand. 3 Though there are four species of silkworms the mulberry silkworm Bombyx mori accounts for the lion share of global silk production and this study is exclusively on mulberry silk. 4 Crossbreed (CB) is a hybrid between Pure Mysore (an indigenous multivoltine race known for its hardiness), and NB4D2, a bivoltine breed developed in India. CB is comparatively easy to rear but yield relatively poor quality silk. 5 Weft is the yarn running breadth wise in the fabric, the mechanical tension on which is lower as compared to that on the warp, which run length wise. As warp is stretched under tension, if the yarn is not strong enough it is liable to break, rendering the weaving process difficult. The tension on warp is more in power-looms as compared to hand-looms

3 3 A solution to the qualitative and quantitative problems of Indian silk industry is popularization of high yielding silkworm hybrids that can also yield better quality silk. The bivoltine silkworm races prevalent in the temperate countries are characterized by high productivity ( kg cocoons / hectare of mulberry) and high quality silk as compared to multivoltine races of tropical countries ( kg cocoon / ha. of mulberry) (Jayaswal etal, 2001). The comparative performance of Bivoltine hybrid vis-a-vis Cross breed furnished in table 1 clearly establishes the superiority of Bivoltine hybrid. The superiority of Bivoltine hybrids is not confined to the quality of silk they produce. The hybrid yields significantly higher quantities of cocoons and thus supposed to be more remunerative to farmers. The comparative data furnished in table 2 is illustrative of this. Table 1. Comparative performance of BV hybrids and Cross Breeds Hybrid Cross Breed (PM X NB 4D 2) Bivoltine hybrid (CSR 2 X CSR 4) Colour Yellow White Silk quality 6 B 2A to 4A Renditta Filament length m m. Yield per 40,000 larvae 50 kg. 70 kg. Survival % 70% 53% Cocoon price per kg. (Rs.) Source: Dandin, S.B; H.K. Basavaraja and N. Suresh Kumar (2005) Table.2. Comparison of bivoltine and cross breed cocoon production (per annum) Items BV Hybrids Cross Breeds Rs / acre % Rs / acre % Leaf cost The international quality standards prescribe grading of raw silk from A to D, A being the higher quality. Above A grade a further classification in the ascending order 2A, 3A etc. is done. 7 Renditta is the measure that indicates the quantity of cocoons required to produce one kilogram of raw silk, for the crossbreed it is above 8. This means that an average of 8 kg cocoons are required to produce one kg of raw silk. On the other hand, the new bivoltine hybrids have the renditta less than six. Hence, the silk production can be improved by 30 per cent by merely switching over to bivoltine raw silk production. 8 Filament length is the length of the continuous filament that could be recovered from the cocoon.

4 4 Silkworm seed Disinfectants and materials Labour Depreciation on fixed capital Other costs Total cost Revenue Net return per annum B:C ratio Source:Kumaresan, 2002 Considering this the Tropical Sericultural Technology was developed in India during 1970 s and a National Sericulture Project (NSP) was launched in 1990 with World Bank support (World Bank, 1997). The major thrust of these projects was development of bivoltine silkworm hybrids and appropriate agronomic practices for rearing them and development of sophisticated technology for processing cocoon and silk. Against the 1000 tons per annum target of BV hybrid cocoon production under NSP, only 400 tons was realised (World Bank, 1997). Statement of the problem The efforts to popularise Bivoltine hybrids in India met with limited success at the adoption level (Ramakrishnan, 2001 and Kumaresan, 2002). It is seen that at present bivoltine silk forms below 10% of total raw-silk production, the remaining being produced from traditional inferior breeds and cross breeds (CSB, 2011). This indicates that only below 10% of the farmers have adopted bivoltine hybrids in the country and the remaining are with conventional cross breeds or other inferior breeds, the silk produced out of which is of low quality suitable for handlooms only. Chart 1 illustrates the diffusion of bivoltine hybrid in major silk producing states which contribute to 90% of Indian silk, for the period from 1990 to Except Tamilnadu and Maharashtra, no other states have shown any encouraging trend in diffusion. Chart 2 gives the contribution of various states to total cocoon production. It shows that the major contributories are Karnataka and Andhra Pradesh (where diffusion of BV-hybrid is very low) and Tamilnadu and Maharashtra, where BV-hybrid diffusion is comparatively faster, have very small shares in total cocoon production. However, from Chart 3 it can be seen that Tamilnadu has the greatest share (33%) of total BV-hybrid production in the country. Thus the diffusion of BV-hybrid in Tamilnadu is a commendable achievement, which makes the case of Tamilnadu, a worthy research topic to pursue. Nevertheless this paper is confined to the BV-hybrid diffusion status of Karnataka, the largest contributor to silk production in the country.

5 % Diffusion of BV hybrid in states 5 55 Chart: 1 Diffusion of BV Hybrid in various states in India 1990 to 2012 (22 yrs) 45 BV% Maharashtra BV% TN 15 5 BV% AP BV% India BV % KAR BV% WB -5 Source: CSB data base 2012

6 6 Chart 2 % contribution to total cocoon production J&K TN 1 % 7 % WB 12 % Maharashtra 1 % Others 2 % Karnataka 40 % AP 37 % Source: CSB data base 2012

7 7 Chart-3 % contribution to BV hybrid production Others 19 % Karnataka 20 % Maharashtra 5 % WB 0 % AP 15 % J&K 8 % TN 33 % Source: CSB data base 2012 Huge investments made towards developing suitable bivoltine hybrids, developing appropriate agronomic practices and extension efforts have not resulted in matching diffusion of the bivoltine hybrid in the country. It is established that without producing bivoltine silk in sufficient quantities, India cannot hold its ground in the domestic silk market, let alone compete in the global market. Considering the fact that the domestic sericulture and silk industry is undergoing a struggle for existence in the post liberalisation era, facing tough competition from cheap imports of raw silk and silk products mainly from China (Directorate General of Anti-Dumping and allied Duties, 2005), the issue of slow diffusion of bivoltine hybrid silkworm assumes importance. This issue has not been subjected much to systematic economic investigation 9. Bivoltine Hybrid Technology (BVHT) is a HYV package consisting of a number of component technologies including agronomic practices for both high yielding mulberry varieties and Bivoltine Hybrid silkworms (Dandin et al. 2005). At the centre of the technology package is the BV hybrid silkworm which has the potential to produce higher quantities of high quality silk in comparison with 9 The various reports and studies available are mostly departmental studies undertaken by central and various state governments. Most of the economic investigations undertaken by western scholars were restricted to the period up to 1930 may be the period up to when there was active western interest in silk. Sinha (1989) reported that.within a substantial body of literature on silk production systematic information on the socio economic dimensions of the activity is lacking

8 8 traditional varieties. This qualitative and quantitative edge is realised at the cost of robustness, the ability of the insect to withstand high temperature, humidity, sub-optimal quality food and pathogens. Hence a large number of technologies form part of the package, designed to provide optimum, disease free micro-environment and quality feed. Even though there is adequate information available free of cost to farmers, few, in practice adopt all these technologies. The available, limited numbers of studies on this topic were conducted by specialists in agriculture extension. Their studies dealt either with the problem of differential acceptance as a function of status, role and motivation or with the problem of communication of innovations. Most of the technology adoption studies are based on categorising farmers into adopters or non adopters of BVHT based on cut off percentages of purchase of Bivoltine Hybrid seed. Usually such studies consider those farmers whose 50% or higher percentage of annual seed consumption is BV hybrid, as BVHT adopters and others as non adopters. As discussed under the theoretical and conceptual framework, this assumption could go wrong, as there could be vast difference among farmers so defined as adopters in the adoption of individual component technologies in the BVHT package. The BVHT package contains at-least 22 component technologies, specially developed for the hybrid. This paper attempts to study the determinants of farmer s decision on BVHT package adoption, based on the level of adoption of individual technologies (of the BVHT package); employing Poisson Count Model, as specified by Octavio et al., The aim of the paper is to generate a clearer understanding of farmer level decision making on adoption of component technologies from a given package and the impact of differential adoption of component technologies in yield. Theoretical and conceptual framework: Complete versus Partial adoption of technology packages: In the vast and diverse literature on agricultural technology innovation, two major research lines are discernible: research on innovations generation and research on the adoption and use of innovations (Sunding and Zilberman, 2001). Much of the agricultural adoption literature was developed to explain adoption patterns of highyield seed varieties (HYV). Two distinct patterns of adoption of HYV technologies observed in literature are based on land allocation (among HYVs and non-hyvs) and choice of components (of the technology package) that is extent vs intensity. HYV technologies are not fully adopted by farmers in the sense that farmers allocate only part of their land to HYV while continuing to allocate land to traditional technologies (Sunding and Zilberman, 2001). Feder (1982) categorises agricultural technologies into two types namely scale neutral and lumpy technologies 10. According to him HYVs are scale neutral and adopted by all farmers, devoting a larger or a smaller portion of the land to it while lumpy innovations are adopted by only larger farms. HYV technologies are technology packages containing a number of component technologies out of which farmers pick and use technologies as they deem fit 11. Feder et al. (1985) observes that while the components of a package may complement each other, some of them can be adopted independently. The terminology used in adoption literature to refer 10 Scale neutral and lumpy technologies 11 A few of the recent studies include: Rauniyar and Goode (1992), Chaves and Riely (2001), Fernandez-Crnejo et al. (2001), Lohr and Park (2002), Cooper (2003), Lambert et al., (2007), Isgin et, al. (2008) and Sharma et al. (2010)

9 9 to this is diverse. A few are: portfolio selection (Holland and Oakley, 2007), multiple technology selection, adoption intensity, adoption frequency. (Schutjer and Van der Veen, 1977) Differentiating between divisible and non-divisible innovations, Feder et al. (1985) observes that while the adoption of non-divisible technology in a given period is dichotomous, adoption of the divisible technology should be expressed by the share of farm area utilising the technology or by the per hectare quantity of input used. Thus as observed by Mann, 1978, farmer may face several distinct technological options of adopting complete package or its subsets, leading to several simultaneous adoption processes with specific (and predictable) sequential patterns. As the socio economic forces driving the choice could be different, they are to be analysed separately. This paper focuses on the second set of choices namely choice of components within the technology package. According to Schutjer and Van der Veen (1977), the majority of technology issues relate to the extent and intensity of use at individual farm level rather than to the initial decision to adopt a new practice or not. Therefore a dichotomous qualitative variable is incompetent to capture adoption of such technologies. Many existing studies model technology adoption using a dichotomous variable (adopt or not), where determinants of this choice are assessed econometrically (Fernandez-Cornejo et al., 2001) However in a number of cases it is not appropriate to model technology adoption as a simple dichotomous choice, as it is the combination of technologies employed that matters. In situations where a large number of techniques are available to farmers, technology adoption is more appropriately modelled as a multiple technology selection problem. Multiple technology adoption issue is widely discussed in literature 12. Sharma et. Al. (2010) finds from a study among UK cereal farms that only 22% of farms adopt more than 50% of the 18 technologies considered. A Rauniyar and Goode (1992) shows that maize farmers in Switzerland adopt specific sets of technologies (out of seven) and argues that in relation to extension activities, emphasis should be on the adoption of a package of practices and not specific practices or technologies in isolation. Lohr and Park (2002) in the case of insect management portfolios by organic farmers and Isgin et, al. (2008) in the case of precision farming technology adoption, demonstrate that adoption of isolated technologies do not preclude the adoption of other technologies; thereby defying path dependency arguments (Eg. Cowen and Gunby, 1996 ). Sharma et al. (2009) argues in their paper on pest control strategies among UK cereal farmers that there is no limit to the number of technologies adopted by a farmer except as it relates to profitability. A brief review of literature on agricultural technology adoption Rogers (1995) identifies four key aspects of communication behaviour that encourage the adoption of innovations: (1) greater social participation, (2) a high level of interconnectedness, (3) being more cosmopolitan and 4) opinion leadership. Adoption literature identifies age as a determinant of adoption of innovations (Shetty, 1966; Subrahmaniam et.al., 1982). In the diffusion literature experience is considered as an important 12 Rauniyar and Goode (1992), Chaves and Riely (2001), Fernandez-Crnejo et al. (2001), Lohr and Park (2002), Cooper (2003), Lambert et al., (2007), Isgin et, al. (2008) and Sharma et al. (2010)

10 10 determinant of adoption and diffusion. The accumulation of experience of an innovation is found to have positive externalities on its adoption as the experience gained by the early adopters affects the perceptions of other farmers (Feder and Omara, 1981). As more experience is gained uncertainty regarding the performance of the innovation is reduced. Some authors showed that the efficiency of a new technology increases with experience (learning by doing) (Black man, 1999). The profit differential often will increase with experience because of learning by using; that is farmers will get more yield and save cost with more experience in the use of the new technology. Sidhu (1976) who studied the factors determining agricultural yields in the early stages of the Green Revolution in the Punjab found that farmers' education has some positive effect on yields. However According to Feder, Just and Zilberman (1985) Formal schooling play a more important role in determining allocative ability than worker ability. Lin (1991) studied the role of education in adoption of hybrid rice in China and found that a household head's education has positive and statistically significant effects on the household's probability and intensity of adopting hybrid seed. The early economic modeling of the 1970s emphasized the impact of information and knowledge on the adoption process and the time lag between awareness and actual adoption (Kislev and Shchori- Bachrach, 1973; Hiebert, 1974). Differences in adoption rates were also attributed to endogenous factors such as differences in skills (Kislev and Shchori Bachrach, 1973), risk aversion (Hiebert, 1974) and prior beliefs (Feder and O Mara, 1982). In the case of risk neutrality, differences in the adoption rates were attributed to differences in prior beliefs about the new technology (Feder and O Mara, 1981). The stock of information on a technology was recognized to be a determinant of agricultural technology diffusion (Hiebert, 1974). The probability of adoption was expected to increases as the stock of information pertaining to modern production increases say, through extension efforts. Grilliches (1957) in his analysis of hybrid corn diffusion suggested that the "advertising" activities of the extension agencies and private seed companies could have influenced the rate of acceptance of hybrid corn in the United States. Feder (1980) found that better information dissemination regarding new technologies through extension agents can reduce the level of subjective uncertainty, and increase adoption of agricultural innovations. Jamison and Lau (1982) analyzed adoption of chemical inputs in Thailand and found similar positive relationship between the likelihood of adoption and extension activity. Munshi (2004) supports this by the example of the Training and Visit extension system in India which was key to diffusion of the high yielding wheat varieties. Feder (1980) suggested that a reduction in uncertainty through extension services can even circumvent a binding credit constraint and will induce higher adoption of farm technologies. Thus the extension efforts from the supply side are expected to create a positive influence on hybrid adoption. The empirical investigations on risk and uncertainty are rare due to the difficulty in measurement. Feder, Just and Zilberman (1985) had opined that more exposure to information through various communication channels reduces subjective uncertainty. On this ground they have cited proxy variables to represent risk bearing capacity of farmers, such as whether the farmer was visited by extension agents or whether he attended demonstrations organized by the extension service or other agencies. The relationship between relative risk aversion and income was hypothesized to be a determinant of agricultural technology adoption (Feder, 1980). Just and Zilberman (1983) showed that the intensity

11 11 of modern technology use depended on whether the modern inputs are risk reducing or risk increasing and on whether relative risk aversion is increasing or decreasing. According to Feder, Just and Zilberman (1985) The relationship of farm size to adoption depends on such factors as fixed adoption costs, risk preferences, human capital, credit constraints, labor requirements, tenure arrangements, and so on. Binswanger (1978) have found a strong positive relationship between farm size and adoption of tractor power in south Asia. Parthasarathy and Prasad (1978) found a significant positive relationship between size and HYV seed adoption in an Andhra-Pradesh village in Thus, the majority of evidence indicates that the incidence of adoption of HYVs is positively related to farm size. Farm size is understood as a determinant of technology adoption. Farm size is also considered as a surrogate for a large number of factors such as access to credit, capacity to bear risk, access to inputs, wealth, and access to information (Feder, 1980; Feder and O Mara, 1981; Just and Zilberman, 1983). However a number of theoretical studies show that variable inputs use could be higher on smaller farms even when uncertainty prevails (Srinivasan 1972). The relative importance of a particular crop within the farmer s enterprise was first used as a variable for diffusion research by Griliches (1957). A case of relative importance of maize in farmer s total land holding determining adoption level of new technologies is discussed by Zeller et.al. (1997) and Sain and Martinez (1999).Credit constraint has been identified as an impediment to technology adoption in developing economies (Feder et al., 1985). Farmers will allocate land to the new technology up to the point where credit is binding and this will result in partial adoption. Pedersen (1970) found that government support guarantees, loans, or subsidies reduce threshold levels of acceptance of innovations. According to Feder and Omara (1981) subsidies restricted to small farmers reduce the fixed-cost element associated with non adoption of technologies. Though they support giving subsidy to early innovators for enhancing adoption, they are apprehensive about the possibility that the early adopters happen to be the higher income farmers temporarily worsening income distribution. Feder (1982) says that when credit is not a constraint, variable input subsidies will enhance per hectare application of inputs. However under credit constraint it may not work. At the same time output price subsidies may reduce the per hectare intensity of variable input use with effective credit constraint. However when credit is scarce subsidies will stimulate adoption of the scale neutral innovations while discouraging the lumpy ones. It is expected that the influence of subsidies on technology adoption is positive. Labour availability or constraint is identified to be an important variable determining new technology adoption decision (Feder, Just and Zilberman, 1985). HYV technology generally requires more labor inputs, so labor shortages may prevent adoption. Moreover, new technologies may increase the seasonal demand of labor, so that adoption is less attractive for those with limited family labor or those operating in areas with less access to labor markets. On the other hand uncertainty regarding the availability of labor in peak seasons can explain adoption of new laborsaving technology (Feder, Just and Zilberman, 1985). Hicks and Johnson (1974) shows that higher rural labour supply leads to greater adoption of labour-intensive rice varieties in Taiwan. There are evidences for shortages of family labour explaining non adoption of HYVs in India (Harriss, 1972).

12 12 Another factor found to be influencing farmers adoption decision of new technologies is the presence or absence of tenacy. According to Feder et.al. (1985), the results are rather conflicting. While some people argue that tenants had a lower tendency to adopt HYVs than owners (Parthasarathy and Prasad, 1978). Feder et.al. (1985) cites studies referring to HYV wheat adoption in India to show that tenants are not only as innovative as landowners but sometimes used more fertilizer per hectare than did owners. A number of studies have found that lack of credit significantly limit adoption of HYV technology. Bhalla (1979) in a study of Indian agriculture reported that lack of credit was a major constraint for 48% of small farms and for only 6% of large farms. Similarly Wills (1972) have found that a majority of small farms reported shortage of funds as a major constraint on adoption of divisible technology such as fertilizer use. The role of economic incentives and profitability in hybrid adoption decision by farmers has been established by Grilliches himself and many other economists who followed the economic perspective of technology diffusion. (Grilliches, 1957, 1958, 1962). Econometric modelling of partial adoption The most commonly applied econometric methods, bivariate statistical analysis and multiple ordinary least squares (OLS) models (Current et al., 1995; Gomez, 1995; Melgar, 1995 and Quiros, 1993) doesn t permit effective quantification of the relationship between bio-physical, socioeconomic and institutional variables and levels of technology adoption when the dependent variables are discrete (Octavio, 2000) as is the case with developing countries (Perez, 1996; Melgar, 1995). Judge et al. (1985) and Octavio et al. (2000) report that sometimes researchers artificially lump adoption levels into two categories (1 for full adoption and 0 for no adoption) in order to use Binomial Probit or Logit models, thereby inducing statistically undesirable measurement errors. Cameron and Trivedi (1986) recommend use of count data models in such cases. Octavio et al. (2000) in a study addressing adoption of agricultural and natural resource management technologies by small farmers in Central American countries claim to have introduced the use of Poisson Count Regression model in technology adoption studies. According to them this tool allows for a statistically efficient and sound evaluation of adoption when the dependent variable is an integervalued gradient and the explanatory variables being farmer s socio economic characteristics and bio physical characteristics of their farm. Another parametric specification employed in the existing count data literature on technology adoption is the Negative Binomial. A criticism on these two models is about their narrow theoretical base. As an alternative, non-parametric methods as specified in Racine and Li (2004) are used by some researchers who concurrently admit these techniques to be computationally burdensome (Octavio et al 2000.) Source of data Both primary and secondary data were used for the study. The study was based on the micro data generated from a sample survey conducted in Sreerangapattanam taluk, Mandya district, Karnataka state. The study location was selected after carefully evaluating the status of sericulture

13 13 production 13. Out of the five sericulture ranges in the taluk, two ranges (comprising of 22 villages and 665 farmers) were selected by random sampling method. From the entire list of farmers of these ranges 71 farmers were selected at random. The data was collected through direct interview method by using a pre-tested schedule. Variable definition Dependent variables: A list of twenty two technologies was drawn considering their degree of impact on crop success (Dandin et al. 2005). These technologies were further classified into three sets as follows. i. Crucial technologies (10 no.): These technologies are crucial for ensuring a successful crop. They are: possession of separate silkworm rearing house, proper ventilation & exhaust in the rearing house, twice disinfection per crop- one before and one after the crop, maintaining air-tightness after disinfection for effectiveness of the disinfectant action on pathogens, use of recommended quantity of Calcium Oxide disinfectant, scientific care during moulting of the worms 14, recommended number of bed cleanings, ensuring hygiene & sanitation within the rearing premises, maintenance of recommended temperature & humidity with the help of thermometer, Provision of adequate Rearing bed area ii. Quality enhancing technologies (7 no.): These technologies are recommended for realising optimum quality of cocoon, the hybrid variety is capable of producing. They are: adoption of HYV mulberry for nutritious feed supply, meshed doors & windows against parasitic uzifly, use of chawky reared worms for healthy brood, separate spinning hall for provision of optimum spinning conditions, use of improved mountage 15, timely harvest and disinfection of mountage to prevent mortality due to post spinning infection iii. Labour saving and advanced technologies (5 no.): These technologies help the farmer cut expenses on labour and reduce drudgery and enhance crop production and 13 Karnataka, Andhra Pradesh West Bengal and Tamilnadu together account for 93.2% of commercial mulberry cocoon production in India. The major contributory is Karnataka state with a share of 42.9%. Karnataka s share in the total bivoltine hybrid cocoon production is 51.7%, being the largest. In Karnataka, 28 districts practice sericulture. Four major sericultural districts of Karnataka namely Kolar, Bangalore Rural, Mandya and Tumkur, together account for 73.9 % of mulberry area, 80.5 % of silkworm seed intake and 88.6 % of total cocoon production. Kolar accounts for 36.2% of the total mulberry plantation and 37.8% of total cocoon production of the state. Bangalore Rural district follows with 20.79% of total plantation and 30% of cocoon production. Mandya district accounts for 13.2% of mulberry area and 16.59% of cocoon production in the state. However Mandya s cocoon productivity (yield per 100 layings) is high at 75.2% as compared to 53.9% of Kolar and 43.9% of Bangalore Rural districts. This probably is indicative of a higher percentage adoption of technologies in Mandya district. Apart from the above reasons the Mandya district is selected for the current study considering the convenience to reach and find contiguous farm units with large number of both BV rearers and CB rearers providing sufficient variability in the sample. 14 Moulting is the periodical casting of skin by worms, which they do four times in life at the end of each growing stages. Worms under moult don t feed and are particularly vulnerable to disease infection, pests and predators. 15 Mountage is a gadget used to mount spinning worms for spinning shapely and clean cocoons with minimum wastage of silk. The traditional bamboo mountages are to be replaced with plastic mountages or improved rotary mountages for ensuring quality BV cocoons.

14 14 protection. They are: adoption of shoot rearing 16, use of phyto-ecdysteroid 17, use of chemical weedicides, use of plant growth regulators (GR) for better quality leaf yield and use of uzi-trap/ uzicide against parasitic uzi fly. Table 1: Percent adoption levels of 22 technologies, categorised into three technology sets Sl. No. Technologies Percentage adopted SET I: Crucial technologies (10) 1 Posession of Separate Rearing House 45 2 Proper ventillation & exhaust 52 3 Twice disinfection per crop 59 4 Airtightness after disinfection 33 5 Use of recommended Cao 6kg/100 dfl 63 6 Scientific moulting care 72 7 Recommended bed cleaning 21 8 Hygene & snitation 42 9 Temp & Humdity control in rearing hose 7 10 Adequate Rearing bed area 64 SET II: 7 Quality enhancing technologies (7) 11 HYV Mulberry adoption Meshed doors & windows against uzifly Use of chawky reared worms Separate spinning hall 7 15 Use of improved mountage 7 16 Timely harvest Disinfection of mountage 16 SET III: 5 Labour saving & advanced technologies (5) 18 Shoot Rearing Adoption Use of Phytoecdysteroid Use of chemical weedicides 0 21 Use of plant GRs Use of uzi-trap/ uzicide 38 Explanatory variables: Fifteen socio economic variables including four dummy variables (with values 1 for yes and 0 for no) were used as explanatory variables. These variables are defined as follows 1. Family labour availability (number): Labour availability to the farmer is captured in the variable as the family labour available in the household. The farmers were asked how many members of the family actively participate in the farm operations namely mulberry cultivation and silkworm rearing. Only the number of hands with considerable participation is counted and recorded. 2. Farmers experience in sericulture (no. of years): The total number of years of experience of the farmer in sericulture 16 In shoot rearing, individual leaf harvesting is replaced by harvesting of entire shoot. This saves up to 30% labour (Dandin et al., 20005), keeps leaves fresh for a longer time and makes cleaning easy 17 Phyto ecdysteroids are hormones enhancing maturity of worms into spinning stage. Hasten early and uniform maturity and helps when there is a shortage of leaf.

15 15 3. Age (years): Age in years as reported by the farmer 4. Education (no. of years of schooling): The number of years of formal education received by the farmer. 5. Ratio of area under mulberry to total land: Ratio of mulberry acreage to total land holding is used as a proxy for the relative importance attached by the farmer to sericulture. When the relative importance of sericulture is more it is expected that the farmer s dependency on it for his livelihood is more and it is more probable that the farmer adopt the new technology. Here the total landholding is inclusive of own and leased land. 6. Crop failure ratio: The ratio of number of eggs failed to the total number of silkworm eggs purchased in a year, expressed in percentage is a strong measure of profitability, independent of prices. 7. Number of days of training received: The farmers were asked whether they received any technology related training and if so how many days of training did they undergo 8. Average selling price per kg of BV cocoon (Rs.): The farmers were asked to recall price per kg of BV cocoons they received in every crop for the previous year 9. Total amount of subsidies received (Rs.): Farmers were asked whether they received any subsidy as cash from government in their entire history of sericulture and if so how much? 10. Average yield of BV hybrid (Kg/Yr.): Average yield is a common measure used in sericulture literature for profitability. It is the quantity of cocoon realised (for BV hybrid here) per 40,000 eggs (also called 100 layings) in kg. 11. Visits of govt. extension agent per crop (number): Farmers were asked how many times the government extension worker visited their crop for every crop during previous year. The average is worked out per crop. 12. Participation in mass contact programs (1=Yes, 0=No): The farmers were asked whether they participate in any mass contact programmes related to sericulture 13. Access to credit (1=Yes, 0=No): Farmers were asked whether they borrowed any amount from bank any time in their sericulture history 14. Possession of separate rearing house (1=Yes, 0=No): Whether the farmer possesses a separate building constructed exclusively for the purpose of silkworm rearing 15. Status as BV hybrid starter (1=Yes, 0=No): The farmer was asked whether he/ she started out as a BV hybrid rearer or not. This variable is expected to capture pure BV rearing experience and save the farmer from exposure to un-scientific practices by previous experience with traditional varieties. Table 2: Explanatory variables used in the study Sl no. Variables and units Average values 1 Family labour available (no.) 4 2 Experience in sericulture (no. of years) 15 3 Farmer s age (Years) 43 4 Education level (no. of years of schooling) 5 5 Ratio: Area under mulberry to total land Crop failure ratio Training received by farmer (no. of days) 4 8 Av. selling price per kg. of BV cocoon (Rs) Total subsidies received (Rs.) 7604

16 16 10 Average yield of BV hybrid (Kg/Yr.) Visits of govt. extension agent per crop (no.) 2 12 Participation in mass contact programs (1=Yes, 0=No) Yes: 31 % 13 Access to credit (1=Yes, 0=No) Yes: 8 % 14 Possession of separate rearing house (1=Yes, 0=No) Yes: 45 % 15 Status as BV hybrid starter (1=Yes, 0=No) Yes: 10 % The Poisson Count Model The Poisson Count Model is an Event Count Duration Regression (ECDR) model (King 1989) useful in analysing adoption data from developing countries where farmers seldom adopt all the practices in a HVY package, but modify them according to their means and perceived needs (Nelson; Quiros), which render their measurements the form of categorically ordered variables undertaking values such as none, low, average, high and total (Octavio et al 2000). These models assume that the dependent variable results from a counting of events using positive integer numbers. This process implies an ordering scheme like that observed when measuring adoption. The model predicts expected level of adoption by a farmer, given the type of extension program in which the farmer participated and his/ her socio economic profile. Quantifying the impact of each independent variable on the level of adoption is also straight forward. In the Poisson Even Count Model (King, 1989b): Where E[Yi]is the expected value of the dependent variable for the i th observation, exp- the exponential function, β- is a 1 by k vector of parameters, Xi is a k by 1 vector with the values of the k independent variables in the i th observation and n is the number of observations. Equation (1) can be used to predict the expected level of adoption given the value taken by the vector of independent variables Xi Equation (1) can also be expressed as Where j can take any one value from 1 to k and identifies a specific explanatory variable and Ci is a constant representing the product of the remaining exponential terms in (2). For dichotomous explanatory variables, if Xji =0, E [Yi] = Cj and when Xji =1, E [Yi] = exp [βi] Cj. Therefore: Calculates the percentage change on E [Y] wen Xj goes from zero to one, for all observations (i). In general, for independent variables those take several integer- values, the percentage change in the expected level of adoption when Xj goes from Xj1 to Xj2 can be calculated as:

17 17 Results and discussion Table 1 and presents the 22 technologies under three sets and the percentage of farmers adopted them. Table 2 presents the 15 variables used as explanatory variables and their average values in the sample. The four technology sets (including one with all 22 technologies put together) were used as independent variables against 15 independent explanatory variables for estimating four separate Poisson Count Regression models. The results are summarised in table 3. Standard Error Estimates and Significance Levels for the Parameters of the Poisson Count Regression Models of four technology sets are furnished as tables 4 to 7 Table 3: Significance level and probable relative impacts in %- Summary Exlanatory Variables Dependent variables from 4 models 22 TECH 10 TECH 7 TECH 5 TECH Family labour available (no.) 19.1** -19.5** Experience in sericulture (no. of years) 8.0** 20.3** Farmer s age (Years) -3.8* -21.2** -19.5** 21.8** Education level (no. of years of schooling) -10.0** -14.8** -10.8** Ratio: Area under mulberry to total land 7.4** 31.8** 13.1** -12.0* Crop failure ratio Training received by farmer (no. of days) -12.6** -4.8** -6.3** -5.5** Av. selling price per kg. of BV cocoon (Rs) 10.9* 15.0** Total subsidies received (Rs.) Average yield of BV hybrid (Kg/Yr.) 21.1** 30.0** 33.3** Visits of govt. extension agent per crop (no.) 5.9* Participation in mass contact programs (1=Yes, 0=No) 35.5** 36.1** 35.7** 34.9** Access to credit (1=Yes, 0=No) 16.5* 29.8** Possession of separate rearing house (1=Yes, 0=No) 44.6** 73.3** 66.0** Status as BV hybrid starter (1=Yes, 0=No) **Significance at 99 % level *Significance at 95 % level Table 4: Standard Error Estimates and Significance Levels for the Parameters of the Poisson Count Regression Models of Adoption of 22 integral technologies in the Bivoltine Hybrid Technology Package (BVHTP) by sericulture farmers Relative S.E. Chi- P- Sig. Variable Coef. impact! Coef Square value level of X on Y

18 18 Constant Family labour available (no.) Experience in sericulture (no. of years) ** Farmer s age (Years) * Education level (no. of years of schooling) ** Ratio: Area under mulberry to total land ** Crop failure ratio Training received by farmer (no. of days) ** Av. selling price per kg. of BV cocoon (Rs) Total subsidies received (Rs.) Average yield of BV hybrid (Kg/Yr.) ** Visits of govt. extension agent per crop (no.) Participation in mass contact programs (1=Yes, 0=No) ** Access to credit (1=Yes, 0=No) * Possession of separate rearing house (1=Yes, 0=No) ** Status as BV hybrid starter (1=Yes, 0=No) Note: Adjusted R-Sq value: 58.69% **Significance at 99 % level *Significance at 95 % level! Relative impact of continuous X variables on Y are calculated between the minimum & average values of X. Unit change in categorical X variables is expected to elicit the % relative impact on Y Table 5: Standard Error Estimates and Significance Levels for the Parameters of the Poisson Count Regression Models of Adoption of 10 crucial technologies in the Bivoltine Hybrid Technology Package (BVHTP) by sericulture farmers Variable Coef. Relative impact! of X on Y S.E. Coef Chi- Square P- value Constant Sig. level

19 19 Family labour available (no.) Experience in sericulture (no. of years) ** Farmer s age (Years) ** Education level (no. of years of schooling) ** Ratio: Area under mulberry to total land ** Crop failure ratio Training received by farmer (no. of days) ** Av. selling price per kg. of BV cocoon (Rs) * Total subsidies received (Rs.) Average yield of BV hybrid (Kg/Yr.) ** Visits of govt. extension agent per crop (no.) Participation in mass contact programs (1=Yes, 0=No) ** Access to credit (1=Yes, 0=No) ** Possession of separate rearing house (1=Yes, 0=No) ** Status as BV hybrid starter (1=Yes, 0=No) Note: Adjusted R-Sq value: 52.75% **Significance at 99 % level *Significance at 95 % level! Relative impact of continuous X variables on Y are calculated between the minimum & average values of X. Unit change in categorical X variables is expected to elicit the % relative impact on Y Table 6: Standard Error Estimates and Significance Levels for the Parameters of the Poisson Count Regression Models of Adoption of 7 quality enhancing technologies in the Bivoltine Hybrid Technology Package (BVHTP) by sericulture farmers Relative S.E. Chi- P- Sig. Variable Coef. impact! of Coef Square value level X on Y

20 20 Constant Family labour available (no.) ** Experience in sericulture (no. of years) Farmer s age (Years) ** Education level (no. of years of schooling) ** Ratio: Area under mulberry to total land ** Crop failure ratio Training received by farmer (no. of days) ** Av. selling price per kg. of BV cocoon (Rs) ** Total subsidies received (Rs.) Average yield of BV hybrid (Kg/Yr.) ** Visits of govt. extension agent per crop (no.) Participation in mass contact programs (1=Yes, 0=No) ** Access to credit (1=Yes, 0=No) Possession of separate rearing house (1=Yes, 0=No) Status as BV hybrid starter (1=Yes, 0=No) Note: Adjusted R-Sq value: 44.72% **Significance at 99 % level *Significance at 95 % level! Relative impact of continuous X variables on Y are calculated between the minimum & average values of X. Unit change in categorical X variables is expected to elicit the % relative impact on Y Table 7: Standard Error Estimates and Significance Levels for the Parameters of the Poisson Count Regression Models of Adoption of 5 labour saving & advanced technologies in the Bivoltine Hybrid Technology Package (BVHTP) by sericulture farmers Variable Coef. Relative S.E. Chi- P- Sig.