Adoption of modern weaving technology in the handloom micro-enterprises in Assam: A Double Hurdle approach

Technological Forecasting & Social Change xxx (2015) xxx–xxx

TFS-18303; No of Pages 13

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

Adoption of modernweaving technology in the handloommicro-enterprises in Assam:A Double Hurdle approach

Bhabesh Hazarika a, Madhurjya Prashad Bezbaruah b, Kishor Goswami a,⁎a Department of Humanities & Social Sciences, Indian Institute of Technology Kharagpur, West Bengal 721 302, Indiab Department of Economics, Gauhati University, Gauhati, Assam 781 014, India

⁎ Corresponding author.E-mail addresses: [email protected] (B. H

[email protected] (M.P. Bezbaruah), [email protected][email protected] (K. Goswami).

http://dx.doi.org/10.1016/j.techfore.2015.08.0090040-1625/© 2015 Elsevier Inc. All rights reserved.

Please cite this article as: Hazarika, B., et al.,Hurdle approach, Technol. Forecast. Soc. Cha

a b s t r a c t
a r t i c l e i n f o
Article history:Received 6 February 2015Received in revised form 12 August 2015Accepted 13 August 2015Available online xxxx

Keywords:Technology adoptionExtent of deploymentMicro-enterpriseHandloomDouble Hurdle

Technological uplift is imperative for enterprises to achieve and sustain competitiveness in terms of both cost andquality of products. While the literature on technology adoption is voluminous, studies focused on adoption re-lated issues concerning rural, nonfarm, and informal micro-entrepreneurs in developing economies are few andfar between. In view of significance of these enterprises in employment and income generation at the lower endof income distribution in developing countries, a study of factors influencing adoption of modern technology insuch enterprises assumes importance. Using firm-level data collected through a primary survey, the presentpaper analyses the determinants of adoption and extent of deployment of weaving technologies in the handloommicro-enterprises in rural areas of Assam. The results of the Cragg's Double Hurdle model reveal the significanceof financial inclusion, availability of family labor, and social network on adoption and extent of deployment ofweaving technologies. The presence of proper market linkages also appears to be crucial for adoption and useof such technologies in the rural areas. The study urges for a comprehensive policy framework to tackle theexisting bottlenecks related to access to credit/capital, market linkages, and extension services to promote thetechnology adoption among the rural micro-entrepreneurs.

© 2015 Elsevier Inc. All rights reserved.

1. Introduction

Technology acquisition and adaptation are crucial for an enterprisefor sustaining cost effectiveness and quality improvement of its prod-ucts, which is imperative for its survival and growth in a competitivemarket situation (Fu et al., 2011; Tripathi et al., 2013). Phasing out ofthe Multi Fiber Agreement (MFA) by the first day of January 2005 hasopened up new opportunities while posing new challenge of moreopen competition in textile industries across countries such as China,India, Bangladesh, Sri Lanka, Vietnam, and others (Ministry of Textile,2015; Tewari, 2006). Given this environment, adoption ofmodern tech-nology in handloom segment of the textile sector has assumed addedimportance as the segment have been providing income and employ-ment to a sizable population in the lower end of income distributionin many developing counties including India (Bortamuly et al., 2013;NCAER, 2010; Ministry of Textile, 2015).

Indeed a significant response of the Indian handloom industry tointensified market competition has been in the form of adoption ofmodern weaving technologies (NCAER, 2010; Ministry of Textile,2015; Bortamuly and Goswami, 2015). A decade down the line since

azarika),iitkgp.ernet.in,

Adoption of modern weavinnge (2015), http://dx.doi.org

phasing out ofMFA, it is now instructive to probe how the handloom in-dustry has fared in adapting itself to the new business environment. Aparticular point of interest in this context is the extent to which thehandloom enterprises, which are typically small-scale and disadvan-taged in accessing market and finance, have succeeded in standing upto the challenges of adopting and deploying modern technologies.

The literature on technology adoption in general is voluminous.Even the segment of technology adoption and its impact on perfor-mance and development of micro, small, and medium enterprises(MSMEs) in India in particular is quite substantial (Bailey, 1993; Lal,1999; Dangayach and Deshmukh, 2005; Subrahmanya, 2006; Toddand Javalgi, 2007; Beddig, 2008; Gomez and Vargas, 2012; Kannabiranand Dharmalingam, 2012). However, available studies mostly coverthe enterprises in formal and organized sector and explain how IndianformalMSMEs initiated themodernization process through the innova-tion and adoption of technology especially the information technologyto meet the market challenges. According to these studies, technologyadoption by the Indian MSMEs is influenced by attitude towards inno-vative activities, size of operations, market share, skill intensity, experi-ence, and infrastructure (Lal, 1999; Subrahmanya, 2006; Kannabiranand Dharmalingam, 2012). On the other hand, credit constraint, lackof awareness, lack of human capital, isolation from technology hubs,and associated risk and uncertainty appear as substantial hurdles intechnology adoption in Indian MSMEs and thus need proper policy in-terventions (Tripathi et al., 2013).

g technology in the handloom micro-enterprises in Assam: A Double/10.1016/j.techfore.2015.08.009

http://dx.doi.org/10.1016/j.techfore.2015.08.009

mailto:[email protected]




Journal logo


http://www.sciencedirect.com/science/journal/00401625


2 B. Hazarika et al. / Technological Forecasting & Social Change xxx (2015) xxx–xxx

Studieswhich have tried to address the issue of technology adoptionin thedecentralized, informal, and rural nonfarm silk (textile) industries(Bortamuly and Goswami, 2015; Latif, 1988; Varukolu, 2007; Rajesh,2012) are relatively fewer in number. These studies brought the impor-tance of credit availability in the process of technology adoption anddiffusion. It is found that higher cost on modern technology may forcethe marginal and financially unsound micro-entrepreneurs in ruralareas to stay with traditional and obsolete technology (Bortamuly andGoswami, 2015). The size of operation does also play a critical role inmodern technology adoption resulting economies of scale (Varukolu,2007; Rajesh, 2012). There are some mixed results with respect to de-mographic factors such as age, educational attainment, and experiencein the Indian context (Bortamuly and Goswami, 2015; Varukolu, 2007;Rajesh, 2012). However, the issue of social capital was not addressedin these studies. Social capital results in social learning that speed upthe adoption–diffusion process through information and knowledgesharing. The recent works of Bortamuly and Goswami (2015), focusedonly on adoption decision, lacks a conceptual/operational frameworkaddressing the impact of social capital, family capital, and financialinclusion, and a rigorous econometric analysis. This is particularly im-portant in the rural areas where the access to information, awareness,and knowledge is limited which may hamper the technology adop-tion–diffusion process. There is need to bring into analysis the impactof these factors on adoption and use of modern technology. Moreover,as many micro-entrepreneurs have adopted modern production tech-nology, effective deployment of these technologies in their work pre-mises appears to have remained limited. Hence, there is a need tounderstand not just what determine the adoption of modern technolo-gy but also what restrict/promotes the extent of its deployment in therural micro-enterprises. The present study was induced by this necessi-ty of a fuller analysis of what influences technology adoption and extentof deployment in the context of rural, nonfarm, and informal sector in adeveloping country.

For operational focus, the study concentrated on adoption of weav-ing technologies by micro-entrepreneurs in the handloom industry inAssam, a state in the geographically and economically peripheral butstrategically significant northeast region of India. Apart from the choiceof the infrequently studied location, the novelty of the present study liesin (a) analyzing not only the factors related to the adoption decision, butalso those impacting extent of deployment of modernweaving technol-ogies in the handloom micro-enterprises; (b) bringing in the contribu-tion of family labor towards fostering technology adoption in theseenterprises; and (c) accounting for the role of context specific socialcapital/network in technology adoption.

2. Technology adoption in the Indian handloom industry

The handloom industry has a unique place in Indian economy facil-itating the second largest employment after agriculture (Bortamulyet al., 2013; NCAER, 2010; Ministry of Textile, 2015; Goswami, 2009;Hazarika and Goswami, 2014). The industry is dispersed, decentralized,labor intensive, and the production has been taking place mostly in therural areas. Therefore, adoption of modern weaving technologies toachieve competitiveness, cost effectiveness, and quality production isimportant not only for the growth of the industry but also for the localeconomic development. Among the available modern handloom tech-nologies in India, the use of high-speed jacquard, dobby machines, pitlooms, sophisticated reelingmachines, network drafting, pattern weav-ing, newand blended rawmaterials, newdesigns, newproduction tech-niques, improved management practices, etc., are frequently used. Afew traditional technologies include throw-shuttle loom, fly-shuttleloom, loin loom, hand-operated spinning/reeling instruments, small-size drum, punching plate, etc. However, there exists technologicalbackwardness in the industry across the country and a major segmentof the handloom households still operates with obsolete technologies.

Please cite this article as: Hazarika, B., et al., Adoption of modern weavinHurdle approach, Technol. Forecast. Soc. Change (2015), http://dx.doi.org

In order to bring a technological uplift in the textile industry, thegovernment of India introduced Technology Upgradation Fund Scheme(TUFS) in 1999. The scheme aimed at providing financial assistance fortechnology upgradation in the textile units to enhance the viabilityand competitiveness in the markets. TUFS had facilitated improvedproductivity and quality, and helped in reducing cost and waste acrossthe value chain but with uneven benefits distribution across the textilesegments. In order to achieve more balanced growth in value chainacross the segments, the scheme was restructured in 2011 (Ministryof Textile, 2015). Despite such efforts, technology adoption and deploy-ment in the handloom industry still remains poor. Out of the 2.38million looms in the country in 2009–2010, only 19% were installedwith dobby/jacquard machine (NCAER, 2010). It indicates that the in-dustry is lagging behind in sustaining cost effectiveness and quality im-provement of its production in themarket competition compared to thesophisticated mill/powerloom industry. From the market share point ofview, the industry has fallen far behind its rival powerloom sector in thelast few years. The share of the handloom industry in total cloth produc-tion stood at 11% against the powerloom industry's 59% in 2014–2015(Ministry of Textile, 2015).

Assam, known as the reservoir of Indian handloom activities, oc-cupies a unique place in Indian handloom industry by producing allfour varieties of natural silks such as Muga, Tassar, Mulberry, and Eri(Goswami, 2009; Hazarika and Goswami, 2014). The state accountedfor nearly 1.24million (44.30%) handloom households and 1.11 million(46.87%) looms in 2009–2010. The industry is a rural industry asmost ofthe production activities take place in rural areas. Almost 98.50% of theworking looms are found in the rural areas. In addition, the handloomactivities are mostly unorganized, informal, and operated in smallscale (Bortamuly et al., 2013; NCAER, 2010; Bortamuly and Goswami,2015; Goswami, 2009; Hazarika and Goswami, 2014; Bortamuly andGoswami, 2012). Despite this, the industry is providing employmentand income to a significant segment of the rural population. In thematter of technology adoption in the weaving sector of the state, how-ever, the status has not been quite impressive (NCAER, 2010; Bortamulyand Goswami, 2015). In terms of types of looms, the percentage of pitloom is very less (0.34%) compared to rest of India (74%)which is most-ly used with weaving machines (NCAER, 2010). Efforts have been initi-ated at bothmicro andmacro levels for inducing adoption and diffusionof handloom technology, and perhaps in response to such initiatives,up-gradation of production technology in handloom micro enterpriseshas picked up in the recent years. Yet, the industry continues to be byand large traditional and with most enterprises primarily saddled withthe traditional technology (Bortamuly and Goswami, 2015; Beddig,2008; Tewari, 2006). Thus, the question remainswhy theweaving tech-nology adoption is poor and how different factors affect adoption ofsuch technologies in the state.

Adoption and diffusion of weaving technology in the state seems tobe influenced by a set of macro characteristics such as governmentpolicies, market competition, cultural and social values, and microcharacteristics such as experiences, access to capital, risk attitude, etc.Lack of market linkage, access to credit, exposure, training and skill,awareness and knowledge about modern technologies are the prevail-ing barriers towards technology adoption in the handloom micro-enterprise (Beddig, 2008; Tewari, 2006). Understanding the key charac-teristics and obstacles in handloom technology adoption anddiffusion isimportant from policy perspective. It is of particular interest due to itsimpact on performance and growth of the industry and thereby theoverall local economic development.

3. Sampling strategy and sources of data

The study is based on primary data collected from 328 handloommicro-entrepreneurs spread over six districts namely Kokrajhar, Baksa,Kamrup, Udalguri, Lakhimpur, and Dhemaji of Assam during January2013 to June 2013 (Fig. 1). The study used a multi-stage sampling



Fig. 1. Study area showing the sample districts in Assam.

3B. Hazarika et al. / Technological Forecasting & Social Change xxx (2015) xxx–xxx

technique. At the first stage, the districts of Assamwere selected follow-ing stratified samplingmethod to capture the different aspects of spatialdiversities. Using the data from Statistical Handbook of Assam 2010, allthe 27 districts were first distributed into three strata based on theproportion of full-time handloom households to the total handloomhouseholds for the state (Directorate of Economics and Statistics,2011). The first stratum included the districts with a higher proportionthan the state average figure of 10.5%. The second stratum included thedistricts with a proportion close to the state figure, and the last stratumincluded the districts with a lower proportion than the state figure.Considering the probable constraints during the data collection process,two districts from each stratum were purposively selected. At the sec-ond stage, two blocks from each district and a minimum of two villagesfrom each block were selected purposively based on concentration ofcommercial handloom activities.

Before collecting the data at individual level, a list of micro-entrepreneurs was prepared in each village at the third stage. Foroperational purpose, a handloommicro-entrepreneur has been definedas an individual who owns a handloom enterprise with not more than10 paid employees or working looms in the survey year. Those newenterprises which were yet to complete even one year of operationhave not been considered for inclusion into the sample. From thesample frame so defined, about 10% of the total handloom micro-entrepreneurs from each selected villagewere chosen using the randomnumber1 as ultimate sample units. A total of 328 respondents thus se-lectedwere interviewed for an average of 45min for generatingprimarydata ofwhich 31.4%weremale and68.6%were female. The primary datawere collected through a semi-structured interview schedule at the

1 The random number for choosing sample unit was generated in generated inMicrosoft Excel. For example, in order to get randomnumbers from1 to 500, one can enterthe following formula: = INT(500 ∗ RAND()) + 1. The INT eliminates the digits after thedecimal, the 500∗ creates the range to be covered, and the +1 sets the lowest numberin the range.


work-shed of the respondents covering different aspect of adoption ofhandloom technologies.

In the sample, 34.2% of the total respondents were found withpositive responses for adopting at least one dobby or jacquard machinein their work-premises. Table 1 presents the distribution of the respon-dents in the sample districts with respect to adoption and non-adoptionof themodern technologies. The micro-entrepreneurs in the rural areasof Dhemaji and Lakhimpur districts mostly use traditional weaving tech-nologies and are far behind the other four districts inmodernization thehandloom activities. Adoption rate was found to be the highest in Baksadistrict (59.4%) followed by Udalguri (54.0%), Kamrup (45.3%), andKokrajhar districts (34.2%).

Though the figures on adoption of weaving technologies are consid-erable in a few sample districts, the limited effective deployment ofthese technologies in the handloom micro-enterprises is a matter ofconcern. Most of the adopters (55.4%) are working with one or twoweaving machines. Only 10.71% of the adopters are found to operatewith five or more numbers of weaving machines. Table 2 presents theextent of deployment of modern technologies across the sample dis-tricts. It is found that only 27.68% of the adopters have achieved fullmechanization.

Table 3 presents the distribution of the technology adopters with re-spect to a few sample characteristics. The table reveals that those whoare financially included, trained, and have access to extension serviceshave gone for higher level of technology deployment. Adopters withless risk aversion are also more likely to expand their activities andhave a higher ratio of mechanization in their work premises.

4. Operational framework

4.1. Conceptual framework

Technology adoption can be defined as a stage of making a choiceas the best course of action towards using a technology/innovation



Table 1Distribution of the respondents across the types of looms and the sampled districts.

Type of looms Baksa (32) Dhemaji (54) Kamrup (53) Kokrajhar (79) Lakhimpur (34) Udalguri (76) Total (328)

Technology adoption Yes 19 0 24 27 1 41 112No 13 54 29 52 33 35 216

Traditional loom Throw shuttle 4 54 1 5 33 5 102Fly-shuttle 9 0 23 43 0 29 104Frame loom 0 0 8 11 0 14 33Only traditional 13 54 29 52 33 35 216

Modern loom Frame loom with dobby machine 18 0 24 27 1 40 110Frame loom with jacquard machine 1 0 0 0 0 2 3Only modern 19 0 23 24 1 32 99

Mix – 0 0 1 3 0 9 13

Table 2Distribution of the respondents with respect to the extent of modern handloom technol-ogy deployment across the sampled districts.


available in the production process (Feder et al., 1985; Rogers, 2003). Inthe present study, technology adoption is conceptualized through in-stallation of dobby2/jacquard3 machines in handlooms. Both these twomachines increase the productivity facilitating different designsthrough network drafting.4 At individual level, it indicates the decisionof the micro-entrepreneurs whether or not to adopt such machineries.Apart from technology adoption decision, the present study also ana-lyzes what influences the extent of deployment of such machineries.The extent of technology deployment is measured by the ratio of thenumber of installed machines to the total number of working looms in-dicating the level ofmechanization of the handloomweaving activities.5

It can be noted that a positive outcome for technology adoption can berealized only after an individual crosses the two decisions viz., whetheror not to adopt and if adopt, then to what extent. In the subsequentparts of the paper, the term technology adoption will refer using ofdobby/jacquard weaving machines.

The conceptual model of technology adoption in handloommicro-enterprises proposed here is based on the existing literature(Bortamuly and Goswami, 2015; Rogers, 2003; Adesina and Zinnah,1993; Noltze et al., 2012; Tambo and Abdoulaye, 2012; Choudhuryand Goswami, 2013; Kassie et al., 2013; Goswami and Choudhury,2015; Hayami and Ruttan, 1971). For understanding the drivers oftechnology adoption–diffusion in the industry, the proposed modelis guided by three key theoretical models viz., innovation-diffusionmodel (Rogers, 2003), economic constraint model (Hayami andRuttan, 1971), and adoption perception model (Adesina and Zinnah,1993). Existing adoption literature reveals that it is difficult to explainindividuals' adoption behavior independently through a single gen-eral model (Noltze et al., 2012; Tambo and Abdoulaye, 2012; Kassieet al., 2013; Bandiera and Rasul, 2006; Marenya and Barrett, 2007;Mazvimavi and Twomlow, 2009). Within the three key models, thepresent study, therefore, considers different contextual factors thoseaffect technology adoption–diffusion and are classified into four catego-ries such as social factors, technological and economic factors, institu-tional factors, and personal factors (Fig. 2).

The proposed model emphasizes on the roles of three key issuessuch as social network, family labor contribution, andfinancial inclusionin analyzing the technology adoption behavior of the rural handloom

2 A device that controls all thewarp threads in a frame loom or pit loom. It increases thedesign capability of the weaver. Thus, dobbymachines expand a weaver's capabilities andremove some of the tedious work involve in designing and producing fabric. Many newcloth design techniques such as network drafting can only reach their full potential on adobby machine.

3 An improved version of dobbymachines withmuch greater versatility in theweavingprocess and offers the highest level of warp yarn control. This mechanism is probably oneof the most important weaving inventions as Jacquard shedding made possible the auto-matic production of unlimited varieties of pattern weaving.

4 Network drafting is a method for creating a weaving draft to express a design in theconsistent structure.

5 The definition is based on Langyintuo and Mungoma (2008), Shiferaw et al. (2008),and Tambo and Abdoulaye (2012).


micro-entrepreneurs. Within the social capital dimension, it focuseson micro-entrepreneurs' social network and the contribution of familylabor in production activities. Though intangible, social network playsa crucial role in the technology adoption–diffusion process. The knowl-edge and information sharing through social network provides themicro-entrepreneurs a platform for social learning and results in spill-over effects that fosters the technology adoption–diffusion process(Feder et al., 1985; Kassie et al., 2013; Bandiera and Rasul, 2006). Inthe context of handloommicro-entrepreneurship, knowing other tech-nology adoptersmakes a potential adopter aware of the advantages anddisadvantages ofweaving technologies. The interaction and informationsharing through one's network is likely to promote technology adoptionamong the potential adopters.

The proposed model also emphasized on the contribution of familylabor in production activities. In adoption literature, the role of familylabor contribution in technology adoption and expansion decision getsrelatively less attention (Goswami and Choudhury, 2015; Marenyaand Barrett, 2007; Mazvimavi and Twomlow, 2009). Given the laborintensive and home-based nature of the industry, family labor contribu-tion is crucial in adoption of weaving technology especially amongthe rural micro-entrepreneurs. Availability of family labor helps inmitigating the risk of labor shortage, reduction of labor cost, and possi-ble moral hazard problem associated with hired labor (Noltze et al.,2012; Marenya and Barrett, 2007; Mazvimavi and Twomlow, 2009).

Within the economic and institutional perspectives, the creditunavailability or financial exclusion emerges as one of the binding con-straints in technology adoption and its deployment in the SMEs (Rogers,2003; Shiferaw et al., 2008; Noltze et al., 2012; Beltran et al., 2013).Given the context of rural areas and informal sector, the handloommicro-enterprises are more vulnerable to the capital/credit constraints.Indeed, the adoption process is assumed to be hampered if the initialinvestments in technologies are high. Micro-entrepreneurs may not be

Extent ofdeployment

Baksa Dhemaji Kamrup Kokrajhar Lakhimpur Udalguri Total

0.00 13 54 29 52 33 35 2160.33 0 0 0 0 0 2 20.50 2 0 0 4 0 13 190.60 1 0 3 1 0 2 70.67 4 0 2 6 0 9 210.75 0 0 3 2 0 5 100.80 4 0 4 3 0 1 120.83 0 0 2 0 0 1 30.86 1 0 1 1 0 0 30.88 0 0 1 0 0 0 10.89 1 0 0 0 0 0 10.90 0 0 1 1 0 0 21.00 6 0 7 9 1 8 31Total 32 54 53 79 34 76 328



Table 3Distribution of the modern technology adopters with respect to a few sample characteristics.

Sample characteristics Extent of technology deployment

Level 0.33 0.50 0.60 0.67 0.75 0.80 0.83 0.86 0.88 0.89 0.90 1.00 Total Fisher'sexact test

Technology awareness Yes 2 18 6 19 10 11 3 3 1 1 2 29 105 0.947No 0 1 1 2 0 1 0 0 0 0 0 2 7

Extension services Yes 1 13 7 18 9 11 3 3 1 1 2 30 99 0.198No 1 6 0 3 1 1 0 0 0 0 0 1 13

Financial inclusion Yes 2 19 2 20 2 4 3 0 0 0 0 22 74 0.001No 0 0 5 1 8 8 0 3 1 1 2 9 38

Financial semi-inclusion Yes 2 19 7 18 8 8 0 3 1 1 2 13 82 0.001No 0 0 0 3 2 4 3 0 0 0 0 18 30

Training availed Yes 0 13 6 12 10 9 2 2 1 1 2 25 83 0.137No 2 6 1 9 0 3 1 1 0 0 0 6 29

Risk averse⁎ 0 0 3 0 6 1 6 1 2 1 0 1 11 32 0.1011 1 4 3 10 8 4 0 1 0 1 1 10 432 1 10 4 5 1 2 2 0 0 0 0 8 333 0 2 0 0 0 0 0 0 0 0 0 2 44 0 0 0 0 0 0 0 0 0 0 0 0 0

⁎ The scale of risk averse is taken as 0 for less risk averse to 4 for highly risk averse.


able to invest considerably in expansion of technology usagewithout fi-nancial provisions thatmay further affect the development of the indus-try (Noltze et al., 2012; Beltran et al., 2013). In such an environment,financial inclusionmay boostmechanization in the handloom activities.Financial inclusion of handloom micro-entrepreneurs through banksand other formal non-banking credit societies, non-governmental orga-nizations (NGOs), and self-help groups (SHGs) may be regarded as acatalyst for increasing the rate of weaving technology adoption and itsextent of deployment.

Micro-entrepreneurs in the rural areas, especially the femaleoften face imperfection in credit and labor markets (Shiferaw et al.,2008; Noltze et al., 2012). In such situations, personal characteristicssuch as age, education, experience, and risk attitude of the micro-entrepreneurs appears crucial in adoption as well as deployment oftechnology. Individuals' age and experience tend to helps in assessingthe cost and benefits of weaving technologies usage, and hence in-creases the probability of the adoption (Bortamuly and Goswami,2015). However, the preference for traditional technology andrisk averse attitude of older micro-entrepreneurs may result in a nega-tive association between age and adoption and expansion decision(Beltran et al., 2013). Literature greatly acknowledges the role of riskand uncertainty in the process of technology adoption and diffusion

Fig. 2. Conceptual model of adoption and extent of


(Choudhury and Goswami, 2013; Goswami and Choudhury, 2015).Adoption of new technology may be perceived as a risky investmentas a micro-entrepreneur needs to learn about the new practices(Choudhury and Goswami, 2013; Goswami and Choudhury, 2015;Peltier et al., 2012). Education also tends to create favorable attitudefor technology adoption facilitating better access and acquisition ofinformation, knowledge, and better understanding of advantages ofweaving technologies. In order to identify the obstacles of technologyadoption and expansion, policy issues such as dissemination of infor-mation and awareness about technology, provision of institutionaltraining, extension services, and market distance are also concep-tualized in the technology adoption–diffusion models (Bortamulyand Goswami, 2015; Latif, 1988; Choudhury and Goswami, 2013;Goswami and Choudhury, 2015; Mazvimavi and Twomlow, 2009;Goswami et al., 2012). Regarding the relationship between firm-sizeand technology adoption, scholars found that scale of operation posi-tively influences both the adoption and the extent of deployment ofnew technology (Feder et al., 1985; Langyintuo and Mungoma, 2008;Peltier et al., 2012; Hall and Khan, 2003; Zhu et al., 2009). The possibleeconomies of scale in the larger firm often encourage technology adop-tion and its extent of deployment. However, the relationship betweenthe firm size and adoption decisionsmay be affected by labor and credit

deployment of modern handloom technology.



6 See Wooldridge (2010).


market imperfections and risk attitude (Feder et al., 1985; Langyintuoand Mungoma, 2008; Mazvimavi and Twomlow, 2009).

Thus, both the technology adoption and extent of deploymentdecisions can be conceptualized as a function of a set of contextualfactors. However, there is every possibility that the set of factors thoseinfluence individuals' adoption decision may not be the same as thoseinfluence the extent of deployment decision. In addition, though apositive outcome for technology adoption is witnessed only afterpassing the adoption and extent of deployment decisions, both the de-cisions may be independently determined (Cragg, 1971; Jones, 1989;Burke, 2010). It means there may be an exclusion of initial factors or in-clusion of a few new factors into the extent of deployment equation ormay have different directions for explanatory factors in both the deci-sions (Aristei et al., 2008). The issue is discussed in detail in the nextsection.

4.2. Analytical model

Over the last few decades, the focus of adoption literatures haschanged from the importance of information spreading in adoption–diffusion (Feder et al., 1985; Rogers, 2003) to a rational decision ofeconomic agent whether or not to adopt (Gomez and Vargas, 2012;Bandiera and Rasul, 2006; Koellinger, 2008). Often, the adoption is nota simple binary (yes or no) response (Noltze et al., 2012). For example,a handloommicro-entrepreneur may decide to adopt modern weavingtechnology, but may not be in a state of full mechanization. Onemay beinterested to install weaving machines into a few of the working loomsi.e., production in the other looms involves traditional technologies.While most of the adoption literatures use binary choice models,there are studies which analyzes the extent of deployment with contin-uous models (Noltze et al., 2012; Tambo and Abdoulaye, 2012). Inthe context of handloom industry, the present study hypothesizestechnology adoption in two stages viz., whether or not to adopt tech-nology andwhat proportion ofmachines to the total number ofworkinglooms per micro-enterprise shall be installed. Literature suggeststhat the technology adoption and the extent of deployment decisionsare the outcomes of optimization that depend on resource endow-ments and different constraints such as budget, information, creditaccess, and availability of technology (Gomez and Vargas, 2012;Rogers, 2003; Noltze et al., 2012; Bandiera and Rasul, 2006;Koellinger, 2008). Since resource endowments and constraints tendto be heterogeneous, the extent of deployment too can be heteroge-neous across the micro-enterprises.

Adoption of a new technology involves risk. Hence the adoption de-cision needs to be analyzed in the framework of choice under uncertain-ty. As per the expected utility hypothesis, the ith micro-entrepreneurwill adopt themodern technologies if the expected utilitywith adoption(EU1i) is greater than the expected utilitywithout adoption (EU0i) or thelatent random variablewi

⁎=(EU1i− EU0i) N 0. The observable outcomeof this choice process can be recorded as the following.

wi ¼ 1; if EU0i b EU1i the modern technology is adoptedð Þ0; if EU0i ≥ EU1i the modern technology is not adoptedð Þ

�ð1Þ

The decision to adopt or not can be estimated through a standardbinary choicemodel. However, associatedwith the question of whetherto adopt or not, micro-entrepreneurs also need to decide the extent towhich the new technology has to be deployed. Thus, technology adop-tion decision is better analyzed with a continuous model rather than abinary choice model. The quasi linear utility model outlined below hasbeen found to be more suitable in the present context (44).

In order to describe the situation, consider the micro-entrepreneuras a rational economic agent whomaximizes his/her utility by consum-ing a bundle of goods (c) and adopting weaving technology (y) in the


production process. FollowingWooldridge (2010), a quasi-linear utilityfunction for handloom technology adoption is given as below.

Ui c; yð Þ ¼ ci þ αi log 1þ yið Þ ð2Þ

where ci is the annual consumption, yi is extent of the technology de-ployment, and αi is themarginal utility of the technology. Maximizationof Eq. (2) subject to a budget constraint (ci + piyi ≤ mi) and non-negative constraints (c, y ≥ 0), where pi is the price of technology andmi is the income level, results in

yi ¼0 if

αi

pi≤1

αi

pi−1

� �if

αi

piN1

8>><>>: : ð3Þ

If αi= Exp(Ziδ+ υi), where Zi is a vector of explanatory factors, δ pa-rameterizes the partial utility function, and υi follows a normal distribu-tion, then from Eq. (3), the extent of technology deployment for themicro-entrepreneurs can be determined by the following equation.6

log 1þ yið Þ ¼ max 0; Ziδ− log pið Þ þ υif g½ � ð4Þ

Eq. (4) thus involves amaximization that undertakes a lower thresh-old zero. It can be further redefined in reduced from as given below.

y� ¼ Ziδþ υi; where υi │Zi∽N 0; σ2� � ð5Þ

where y* is the extent of technology deployment which is latent innature and can be operationalized by its observable counterpart y(Wooldridge, 2010). In other words,

y ¼ y� if y�≥0 and wi ¼ 1 the extent of technology adoptionð Þ0 otherwise zero level of technology adoptionð Þ

�: ð6Þ

Eqs. (5) and (6) allow investigating two decisions of the handloommicro-entrepreneurs in concern to technology adoption i.e., whetheror not to adopt weaving technology and the extent of technology de-ployment if the initial decision was made in favor of adoption.

4.3. Empirical model

The present study uses two concepts for econometric modeling oftechnology adoption viz., a) whether or not a micro-entrepreneuradopts handloom weaving machines (a dichotomous choice model)and b) the extent of technology deployment or the ratio of the numberof installed weaving machines to the total number of working loomsonce adopted indicating the level of mechanization (a continuouschoice model).

It is found that not all the micro-entrepreneurs adopt weaving tech-nologies and thus results in some observation being zero. In otherwords, given the constraints, the optimal choice for the non-adopterappears to be a corner solution. In such situation, the standard Tobitmodel (Tobin, 1958) has been conventionally used for estimation(Choudhury and Goswami, 2013; Wooldridge, 2010). However, Tobitmodel can be restrictive both in economic and statistical reasons(Burke, 2010; Wooldridge, 2010). From economic point of view, zeroadoption observations exist due to corner solution given the resourceendowment and cost constraints. From statistical point of view, themodel assumes both the adoption and extent of deployment decisionsare being jointly made and determined by the same set of factors. Inother words, the choice of dependent variable ‘w’ (technology adop-tion) and the value of ‘y’ (extent of technology deployment) given



7 While using of stata command craggit instead of separate probit and truncated regres-sion makes estimation more coherent, it will not change results. The primary benefit ofusing craggit is its ability to facilitate postestimation analysis (average partial effects)and interpretation (Burke, 2010).

8 The Likelihood Ratio (LR) test compares the log likelihood values of the two nestedmodels assessing for significant differences. The test statistics is given by− 2 ln λ, whereλ is the ratio of likelihoods ofModel 1 toModel 2 (subject toModel 1 is nested inModel 2).The statistic has a χ2 distribution with degrees of freedom equal to the difference in thenumber of estimated parameters of the two models (Wooldridge, 2010; Greene, 2009).

9 The Vuong test is used for comparison between two non-nested models. The test isbased on the standard likelihood ratio test statistic Z0 ¼ ΔLRffiffi

np �wn

, which follows a standardnormal distribution. In the present study, ΔLR = {(LP + LTR) − LH}, where LT, LP, and LTRare likelihood of the Tobit, Probit and Truncated Regression models respectively. The teststatistic uses the transformation of log-likelihood values through wn ¼ fð1nÞðΔLRÞ2g−fð1nÞΔLRg

2. If the value of the test statistic is greater in absolute value than a critical valuefrom the standard normal distribution, then the double hurdlemodel fits these data betterthan the Heckman model (Vuong, 1989).


that y N 0, are determined by same underlying process. Thus, it assumesthat the coefficients in the uncensored and censored portions of thelikelihood function have the same directions and magnitudes (Jones,1989; Burke, 2010; Wooldridge, 2010; Greene, 2009). However, itmay not be true in every circumstance. In the context of handloom tech-nology adoption, a micro-entrepreneur in one location may be morelikely to adopt technology but his/her extent of technology deploymentmay be less than the micro-entrepreneurs in other locations due someissues such as lack of credit availability, extension service, and distanthandloommarket. In order to avoid such limitations, in the recent liter-atures, a bivariate generalization of Tobit model is given stress that hy-pothesizes that both the decisions are different andmay be determinedby two different set of factors (Mazvimavi and Twomlow, 2009; Cragg,1971; Jones, 1989; Greene, 2009).

Following Cragg's Double Hurdle model (independent DoubleHurdle Model or IDS, thereafter), the present study hypothesizes thata micro-entrepreneur must pass two separate hurdles before observinga positive value for weaving technology adoption i.e., whether or not toadopt the modern technology or not and the extent of deployment(Cragg, 1971; Jones, 1989; Burke, 2010). It also argues that both the de-cisions are structured differently and the factors that influence theadoption decision are independent to that of the extent of deploymentdecision (Aristei et al., 2008). Thus, separating the two decisions, theDouble Hurdle model can be specified as follows.

w�i ¼ Xiβ þ εi εi � N 0; σ2

ε� �

Adoption modelð Þ ð7Þ

where

wi ¼ 1; if w�i N 0

0; otherwise

�

and,

y�i ¼ Ziδþ υi υi � N 0; σ2υ

� �Extent of adoption modelð Þ ð8Þ

where

yi ¼ y�i ; if y�i N 0 and wi ¼ 10; otherwise

�

where wi⁎ is the latent variable indicating the micro-entrepreneurs'

decision to adopt and wi is the observed micro-entrepreneur's decisionto adopt, wi takes a value 1 if a micro-entrepreneur adopts weavingtechnology and 0, otherwise, yi⁎ is the latent variable indicating the ex-tent of deployment and yi is the observed response on the extent oftechnology deployment i.e., yi takes a value 0 for non-adopter andsome positive values for the adopters. β and δ are the coefficients tobe estimated, Xi and Zi are vector of factors those influence the technol-ogy adoption and the extent of deployment decisions respectively, andεi and υi are the respective error terms following normal distributionwhich are assumed to be independent (Cragg, 1971; Burke, 2010;Wooldridge, 2010). The assumption of conditional independence of dis-tributions of εi and υi i.e.,D(y*|w, x)=D(y*|w) is important for unbiasedestimation (Cragg, 1971; Jones, 1989; Wooldridge, 2010).

Several studies relax the assumption and estimated a DependentDouble Hurdle (DDH) model allowing correlation between the two de-cisionswhere the assumptions of independence between the two errorscould not be rejected (Jones, 1989; Moffatt, 2005). These studies foundthat the results are similarwhen the independency assumptionwas andwas not held. Another alternative to the independency assumption,Heckman Selection (HS) model (Heckman, 1979) assumes that theerror terms of the two decision equations are correlated where theadoption decision dominates the extent of deployment decision. Ac-cording to HSmodel, some positive values for the extent of deploymentdecision are only observed once a micro-entrepreneur decides to adopt


technology. Only the micro-entrepreneurs with positive extent ofdeployment are included in the second hurdle, as zeros are not generat-ed by the extent of deployment decision (Aristei et al., 2008). Thus,it differs from IDH with respect to the sources of zeros. While HSmodel rules out the possibility of adoption by the non-adopters underany circumstances, the IDH model considers non-adopters as a cornersolution in a utility-maximizing model (Cragg, 1971). In the case ofweaving technologies, non-adopters may be motivated towards adop-tion due to the changes in price structure of the respective technologies,provision of training and extension programs, etc. Thus, HS model alsoappears to be restrictive in the present circumstances.

Following Cragg (1971), Burke (2010), Ricker-Gilbert et al. (2011),Tambo and Abdoulaye (2012), Eqs. (7) and (8) are assumed to be inde-pendent and the joint likelihood function of the IDH model is given asbelow.

f w; yjX; Zð Þ

¼ 1− Φ Xiβð Þf g1 w¼0ð Þ Φ Xiβð Þ 2πð Þ−12σ−1 exp

− y−Ziδð Þ22σ2

( )=Φ

Ziδσ

� �" #1 w¼1ð Þ

ð9Þ

wherew is a binary variable equal to 1 if w is positive and 0 otherwise. yis continuous variable for non-censored portion which is observed onlywhen w = 1. The function shows that the probability of w N 0 and thevalue of y, given y N 0, may be determined by different mechanisms(vectors β and δ respectively). It shows that there are no restrictionson the elements of X and Z, implying that each decisionmay even be ex-plained by a different vector of factors altogether (Burke, 2010). Further,it can be noted that the joint likelihood function gives the distribution ofTobit model if X = Z and γ = β/σ. Given the independency in errorterms, the present study uses stata command craggit7 to estimate IDHlikelihood function that uses a binary probit model for the first decisionand a truncated normal regression for the second decision simulta-neously rather than separately for maximum likelihood (ML) estimates(42).

In order to justify the use of Cragg's IDH model, it can be testedagainst models with other specifications. Since the Tobit and DDHmodels are nested in IDH model, they are tested against IDH using alikelihood ratio test.8 The same cannot be applied for a comparison ofHS model with IDH as it is not nested in a double hurdle framework.For such non-nested models, Vuong (1989) provided a conventionaland adjusted likelihood ratio test known as Vuong test.9

4.4. Descriptions of explanatory factors and descriptive statistics

Based on the literature and conceptual model, a set of explanatoryfactors are derived for analyzing the adoption and extent of deploymentdecisions. Table 4 presents description of the factors, hypothesized rela-tion with technology adoption, and the rationale behind choosing theparticular factor.



Table 4Descriptions of the explanatory factors and their hypothesized relation with adoption of handloom technology.

Factors Description Measure Hypothesizedrelation

Rationale

Gender Gender of the micro-entrepreneur 1 for female and 0 for male − Female perceived themselves to be more constrainedNew firm Micro-enterprise established within three

years prior to survey1 for new and 0 for older + Fresh firm likely to adopt technology considering present

market situationAge Age of the respondents Years + Becomes more experienced in entrepreneurial activitiesEducation Number of years spent in school Years + Gives exposure to better access to and acquisition of

technology related information and knowledgeRisk-averse Risk bearing attitude of the micro-entrepreneurs 1 for less risk-averse to

5 for highly risk-averse− More risk averse people are unlikely to adopt technologies

Working experience Working experience with handloom technologies Years + Aware of the benefits of weaving technologiesTraining availed Availed handloom training from Govt./NGOs 1 for yes and 0 for no + With training, one becomes skilled and efficientTechnology awareness Aware of handloom modern technologies 1 for yes and 0 for no + Becomes aware of benefits of technology adoptionExtension service Having access to technical help while

operating with handloom technologies1 for yes and 0 for no + Helps in mitigating technical disturbances during

operationsFinancial inclusion Financial inclusion through commercial banks 1 for yes and 0 for no + Provides financial strengthFinancialsemi-inclusion

Financial inclusion through NGOs, SHGs, andmicro-finance institutions

1 for yes and 0 for no + Provides financial strength

Social network Knowing other entrepreneurs who arealready technology adopter

Numbers + Speeds up the dissemination of knowledge and information,generates perception of associated benefits and costs, andhave spillover effect towards technology adoption

Family capital Ratio of family full-time labor to total laborin the micro-enterprise

Ratio + Reduces the labor cost and moral hazard problem andprovide psychological support

Firm size Number of working looms in handloomenterprise

Numbers + Benefited from economies of scale

Distance Distance to the nearest handloom market Kilometers − Hampers due to communication and marketing gapsand increases the transaction cost


Table 5 presents the descriptive statistics of the exploratory factorswith mean difference test across adopters and non-adopters. As the in-dustry is female intensive, it is obvious that female outnumbers male inmicro-entrepreneurial activities. However, adoption of weaving tech-nologies is found to be limited in female owned micro-enterprises.Only 27.6% of the female appear to be the adopter ofmodern technologyagainst the 48.5% of their male counterpart. New micro-entrepreneurs(started micro-enterprise within three years prior to the survey) aremore likely to adopt modern weaving technologies indicating that thenew generation is well aware of the importance of modern technologyfor sustaining cost effectiveness and quality production in a competitivemarket environment. While education is crucial for technology adop-tion, lower levels of educational attainments are found for both theadopters and non-adopters. It limits the information dissemination

Table 5Descriptive statistics of the explanatory factors used in the technology adoption model.

Factors All (328) Adopter (112)

Mean Std. dev. Mean St

Gender 0.686 0.465 0.554 0.4New firm 0.159 0.366 0.196 0.3Age 36.668 8.998 35.938 8.8Education 8.027 4.140 8.446 3.3Risk averse 1.518 1.052 1.080 0.8Working experience 3.530 4.455 5.991 5.4Training availed 0.207 0.406 0.259 0.4Technology awareness 0.616 0.487 0.938 0.2Extension services 0.230 0.459 0.116 0.3Financial inclusion 0.159 0.366 0.339 0.4Financial semi-inclusion 0.168 0.374 0.268 0.4Social network 1.421 1.544 3.063 1.1Family capital 0.660 0.304 0.640 0.3Firm size 2.695 1.817 3.589 1.9Distance 21.912 18.842 8.875 9.3

Notes: Figures in the parentheses represent sample size.⁎ Significant at 10%.

⁎⁎⁎ Significant at 1%.


and thus hampers the process of technology adoption–diffusion. Italso explains why the rural micro-entrepreneurs especially the femalesfail to read the market signals which further affect their performances.

The working experience with handloom technologies plays an im-portant role in adoption and extent of using of such technologies. Sig-nificant mean difference is witnesses between the two categories ofthe respondents. On an average, the adopters had six years of workingexperience while the non-adopters had only two years of such experi-ence. In relation to the risk and uncertainty issue, adopters werefound to be less risk-averse than that of the non-adopters in the sample.The social network was found statistically different between the twocategories of the respondents. On an average, an adopter knows threemicro-entrepreneurs who had already installed weaving technologieswhile it is only one for the non-adopters. The higher the level of social

Non-adopter (216) Pearson Chi2 t-test

d. dev. Mean Std. dev.

99 0.755 0.431 13.841⁎⁎⁎ –99 0.139 0.347 1.831 –49 37.046 9.071 – −1.05998 7.810 4.469 – 1.32150 1.745 1.076 40.464⁎⁎⁎ –04 2.255 3.213 – 7.841⁎⁎⁎

40 0.181 0.386 2.757⁎ –43 0.449 0.499 74.374⁎⁎⁎ –22 0.394 0.490 27.099⁎⁎⁎ –76 0.065 0.247 41.651⁎⁎⁎

45 0.116 0.321 12.223⁎⁎⁎

33 0.569 0.912 – 21.559⁎⁎⁎

08 0.670 0.302 −0.86452 2.231 1.556 6.854⁎⁎⁎

01 28.671 18.995 – 10.396⁎⁎⁎




network, the more will be the technology diffusion as rural peoplehave a tendency to learn from others through demonstration. Theadopters were also found to be less constrained with respect to accessto credit or capital. While the proportion of the adopters under fi-nancial inclusion of banks was 33.9%, only 6.5% of the non-adopterswere financially included by the banks. Similarly, a notable differencewas also observed between the two categories with respect to financialinclusion by formal credit institutions other than banks. Other impor-tant aspects towards technology adoption are the awareness about thetechnology and extension services. While 61.6% of the respondentswere aware of the handloom technologies, the extension services interms of availability of technically sound person such as demonstratorsandmaster weavers are found to be less in the state. Only 23% of the re-spondents stated that they had the access to extension services. This de-mands policy initiative for better access to extension services as thesecan minimize production and market risk and thus promote the extentof technology deployment.

5. Results and discussion

5.1. Model specification

Adoption of modern weaving technology in the present study ismodeled in two decision stages i.e., the adoption and extent of deploy-ment. The hypothesis of independency between the two decisions isfirstly tested as the existence of issue of independency may lead to in-consistent estimators (Cragg, 1971). Similar to the findings of Jones(Cragg, 1971), Moffat (Moffatt, 2005), and Aristei and Pieroni (2008),the LR test (ρ = 0.435; χ2 = 1.59; p = 0.207) disregarded the depen-dency in two decisions in the present study. The model specificationtests for a correct specification of the adoption decision are presentedin Table 6. The LR test (χ2 = 272.485; p = 0.001) strongly rejects theTobit specification against Cragg's IDH and confirms the independencyin errors and separate structures for adoption and extent of deploymentdecisions. Though the LR test between DDH and IDH results no prefer-ences for one model over another, the IDH specification is preferredbased on the insignificant value of rho (ρ = 0.070; Z = 1.46; p =0.144). Similarly, the Vuong test favors IDH model over HS model. Theinformation criterion tests i.e., the Akaike Information Criterion (AIC)and Bayesian Information Criterion (BIC) also indicate better fit for theIDH model over the Tobit and DDH models. Thus, the Cragg's IDHmodel is suitable for understanding the adoption behavior of thehandloom micro-entrepreneurs and in the subsequent analysis, the re-sults of the IDH model are considered for interpretation.

5.2. Factors influencing adoption and extent of deployment of modernweaving technologies

Table 7 presents the maximum likelihood results of Cragg's IDHmodel that reports the marginal effects (MEs) and average partial ef-fects (APEs) of the explanatory factors. APEs are calculated to measurethe overall level of technology adoption. APEs are calculated on the un-conditional probability of being an adopter and on the conditional prob-ability of technology adoption. The corresponding standard error andsignificance levels are obtained via bootstrapping of 100 replicationsfollowing Burke's approach (Burke, 2010). The Cragg's IDH model is

Table 6Model specification tests.

Models Test type

Tobit vs. Cragg's Independent Double Hurdle Likelihood RatioCragg's Independent Double Hurdle vs. Cragg'sDependent Double Hurdle

Likelihood Ratio

Heckman Selection vs. Cragg's Double Hurdle Vuong


found to be significant at 1% level with a Wald Chi2 value of 67.44. Forinterpretation purpose, the present study uses the marginal effects.

The results of 1st hurdle (adoption of technology) of the Cragg'smodel show that the handloom micro-entrepreneurs in the rural areasare influenced significantly towards weaving technology adoption bytheir age, weaving experience, risk attitude, awareness, financial inclu-sion through formal credit institution both banking and non-banking,social network, family labor contribution, and distance to the nearesthandloom market. However, the present study fails to provide evi-dences for significant influences of a few other factors such as gender,new firm, education, and training. The results of the 2nd hurdle showthat gender, weaving experience, risk attitude, financial inclusionthrough formal credit institutions both banking and non-banking, firmsize, social network, family labor contribution, and market distance sig-nificantly influence the extent of deployment of weaving technology(Table 7).

The influence of gender, though not significant on adoption ofweaving technology, is negative on the extent of deployment of suchtechnology in the handloom micro-enterprises. The finding is in linewith the findings of Mazvimavi and Twomlow (2009) indicating thatless mechanization in the female owned micro-enterprise may beexplained by their less access to information and credit services, andrisk-averse attitude. Moreover, male are operating in a better organizedmanner that increases the demand for new technology by the maleowned firms.

The influence of age of the respondents is found to significant butappears to be negative in adoption model. It indicates that, with in-creases in age, handloom micro-entrepreneurs are less likely to adoptmodernweaving technologies. Young entrepreneurs have a higher aspi-ration for a better life and are less risk averse. On the other hand, agedmicro-entrepreneurs tend to better acquainted with the traditionaltechnologies and are relatively higher risk averse. Moreover, workingwith modern technologies requires investments, skill, maintenance,and availability of skilled labor which increases associated risk. Thus,they prefer to work with traditional rather than modern weavingtechnologies or prefer a lower extent of mechanization. It is perceivedthat the less risk-averse micro-entrepreneurs have a positive attitudetowards changes in entrepreneurial activities and willing to accept thechallenges under uncertainty (Rajesh, 2012; Choudhury and Goswami,2013; Peltier et al., 2012). The present study also suggests that the riskaverse attitude of the micro-entrepreneurs not only hampers the tech-nology adoption, but also limits the extent of adoption. The presentstudy also suggests that, the risk averse of the micro-entrepreneursnot only hampers adoption, but also limits the extent of deploymentof weaving technology.

Working experience as a source of informal learning about the tech-nology appears to be important in adoption and extent of deployment ofsuch technologies.With one year additional experiences, the probabilityof adoption and extent of deployment increase by 1.4% and 0.5% respec-tively. The adoption rate will be more if the micro-entrepreneurs areaware of the benefits and inherent characteristics of weaving technolo-gies. Ceteris paribus, the probability of adopting weaving technology is18.4% more for the handloom micro-entrepreneurs who are awareabout the technology than the others. Supporting the findings of Pelteret al. (Peltier et al., 2012), the present study argues that awareness,knowledge, and information not only help in understanding the relative

Test-value p-value Preferred choice

272.485 0.001 Cragg's Independent Double Hurdle2.120 0.999 Indifferent

9.46 0.001 Cragg's Independent Double Hurdle



Table 7Double hurdle estimates of factors influencing the technology adoption and the extent of deployment in the handloom micro-enterprises.

Factors 1st Hurdle: adoption 2nd Hurdle: extent of deployment

Marginal effects p-value APE BSE Marginal effects p-value APE BSE

Gender −0.016 0.796 −0.009 0.038 −0.061 0.011 −0.061 0.030New firm 0.038 0.478 0.019 0.031 – – – –Age −0.006 0.059 0.001 0.002 0.001 0.895 −0.003 0.002Education 0.004 0.536 0.002 0.004 −0.005 0.204 −0.005 0.004Working experience 0.014 0.007 0.008 0.003 0.005 0.058 0.005 0.003Risk averse −0.120 0.000 −0.066 0.016 −0.035 0.020 −0.035 0.015Financial inclusion 0.187 0.077 0.074 0.047 0.237 0.001 0.132 0.041Financial semi-inclusion −0.101 0.020 −0.083 0.035 0.258 0.001 0.030 0.032Training availed −0.061 0.130 0.017 0.034 0.017 0.553 0.017 0.034Technology awareness 0.184 0.002 0.117 0.045 – – – –Extension services – – – – 0.013 0.788 0.013 0.056Firm size – – – – −0.022 0.056 −0.022 0.013Family capital 0.199 0.000 0.095 0.048 0.039 0.015 0.174 0.046Social capital 0.171 0.030 0.110 0.016 0.174 0.001 0.039 0.020Distance −0.008 0.001 −0.004 0.001 0.004 0.005 −0.002 0.001Sigma 0.110 0.001Wald Chi2(13) 67.440p-value 0.001Sample 328Assumption Test Chi2(1) p-valueHeteroskedasticity Breusch-Pagan/Cook-Weisberg test 2.570 0.109Normality Pr(Skewness) – 0.636

Pr(Kurtosis) – 0.962Adjusted Chi2(2) 0.230 0.893

Note: APE and BSE stand for average partial effect and bootstrapped standard error.Figures in the parentheses represent degrees of freedom.

10 Apart from SHG's, bilateral transfer of capital is exercised among the handloommicro-entrepreneurs within a group in most of the sample areas. For example, micro-entrepreneurs in Kamrup district of Assam form such group popularly known as Santha.


advantages of adoption, but also change the subjective perception of themicro-entrepreneurs on the cost of adoption that play a crucial role inthe extent of deployment of modern technology. Therefore, invest-ments towards awareness and knowledge dissemination among therural micro-entrepreneurs should be a priority involving the NGO'sand SHG's.

Access to extension services in terms of availability of technicallysound personals such as handloom demonstrator and master weaveris important to tackle the disturbances in theweaving technologies dur-ing theweaving process. The present study hypothesized that the accessto extension services are more important in explaining the extent ofweaving technology deployment rather than the awareness whichusually appears to be crucial in the initial adoption decision. The influ-ence of extension services is found to be positive for the extent of adop-tion of the handloom technologies. However, the present study fails toprovide enough evidences for a significant result of the relationship be-tween the two.

While financial inclusion under the banking increases the likelihoodof adoption and extent of deployment, financial assistance through for-mal non-banking credit societies, NGOs, and SHGs, which is termed asfinancial semi-inclusion, is found surprisingly to have a negative influ-ence on individuals' adoption decision in the present study. This unex-pected result might be due to the higher cost of borrowing. As foundin FGDs, micro-entrepreneurs faces annual interest rates ranging from36–60% for the loans obtained from formal non-banking credit societies.Since investment in technology adoption is long term phenomena, sucha high interest rate over long run may be unaffordable for the ruralmicro-entrepreneurs. In addition, diversion of credit into other activitiesdue to poor economic conditionmight results in a negative influence onadoption decision (Bortamuly and Goswami, 2015). In contrast, the sig-nificant and positive influences of both the types of financial inclusionsin second hurdle indicate that the extension of financial services is cru-cial in the extent of technology deployment rather than just the adop-tion decision. Access to credit enables handloom micro-entrepreneursto meet the expenses of establishment, and operation andmaintenancecost of using of weaving technologies (Bortamuly and Goswami, 2015;Goswami and Choudhury, 2015). Appendix 1 indicates that the APEs


of financial inclusions vary across locations. APE of financial inclusionis found to be the highest in Udalguri district which perhaps explainedby the location of the micro-enterprises nearer to the business centers.Female micro-entrepreneurs, in particular are facing discriminationsin accessing the credit from both banking and non-banking financialinstitutions. From policy perspective, to encourage greater extent of de-ployment ofweaving technologies in rural areas, government should di-rect the formal credit institutions to extend financial services to coverfinancially unsound micro-enterprises especially those with moderntechnologies.

The influence of handloom social network, measured in terms ofnumber of adopter of handloom technologies connected with a micro-entrepreneur, is found to be positive and significant in both the adop-tion and extent of deployment. Ceteris paribus, the probabilities ofadoption and the extent of deployment for themicro-entrepreneurs in-crease by 121.6% and 4.9% respectively with an increase in the size ofone's network. Through social network, micro-entrepreneurs accessinformation about the types of weaving technologies, the associatedprices, input requirements, optimal use of inputs, and profitability asso-ciated with new handloom technologies such as dobby and jacquardmachines. Thus, it not only generates learning externalities and spill-over effects (Bandiera and Rasul, 2006), but also helps the rural micro-entrepreneurs in accessing credit/finance (mostly informal). Whilemissing financial market appears as binding constraints for ruralmicro-entrepreneurship development, micro-entrepreneurs often usehandloom networks for bilateral capital transfer10 to meet investmentrequirement.

The handloom micro-entrepreneurship in rural areas represents apicture of homebased or family based businesswhich is labor intensive.Therefore, the availability of more family member as full-time laborwould be an advantage for reduction in the labor costs and dependencyon hired labor. The influence of the proportion of full-time family laborto the total firm labor is found to be positive and significant on both the




adoption ofmodernweaving technologies and its extent of deployment.Availability of more family labor on full-time basis not only addressesthe financial constraints in hiring wage laborers, but also addressesthe possible problem of moral hazard which could have resulted inhigher real cost of hiring non-family labor beyond the observed wagerate (Noltze et al., 2012; Mazvimavi and Twomlow, 2009).

Surprisingly, the influence of firm size is found to be negative on theextent of technology deployment in the industry. There are evidencesfor such unexpected results of negative correlation between firm sizeand technology adoption (Langyintuo and Mungoma, 2008; Hall andKhan, 2003; Zhu et al., 2009). It indicates that an increase in the numberof looms results in more dependence on traditional technologies. Theparticular barriers faced by the larger firms in extending technologydeployment are greater complexity in enterprise structure, lack ofresources, risk of labor shortage, and lack of extension services. Incontrast, small firms enjoy a greater control over the handloom opera-tions and can intensively use the technologies, and thereby reap morebenefits.

Distance to the nearest handloommarkets reflects the remoteness ofa firm and the problems in accessing market information, inputs, tech-nology, and other accessories, and thus can be a proxy for infrastructurebottlenecks. A distant market results in higher transaction cost for mar-keting associated while dealing with handloom inputs and output andthus hampers the adoption and extent of technology deployment. Theresults indicate that, further away a handloom micro-enterprise fromthe nearest handloom market, the less likely will be the technologyadoption and extent of deployment. A distant market with limitedmar-ket information forces the rural micro-entrepreneurs to rely mostly onthe local market where there exists higher demand for some productswhich can easily be produced using traditional weaving technologies.Appendix 1 indicates that existences of nearest market not only in-creases the adoption rate but also gets higher access to credit. Fromgender perspective, nearest markets have important implications forfemales in reducing the transaction cost as they have higher familyobligations. Therefore, developing handloom market infrastructureshould be brought into the policy prescription to promote the adoptionof handloom technologies.

Compared to some earlier adoption studies (Bortamuly andGoswami, 2015; Gomez and Vargas, 2012; Koellinger, 2008), thepresent study fails to produce significant results for education on tech-nology adoption. The study also fails to provide evidences for a sig-nificant result for institutional training. It might be because of the factthat training alone cannot boost adoption of modern technologieswithout addressing the constraints related to infrastructure, accessto credit/capital, extension services, and market linkages. Even if amicro-entrepreneur is trained to use modern technologies, these con-straints will force him/her to continuewith the traditional technologies.Therefore, the present study urges for a more comprehensive policyagenda for the development of the industry alongwith giving special at-tention to the adoption of weaving technologies in the rural areas of thestate.

5.3. Robustness check

For efficient and robust maximum likelihood estimation, the as-sumptions of normality and homoscedasticity in the data were tested.The present data set does not violate the two assumptions (Table 6).The VIF values deny the problem of severe multicollinearity amongthe explanatory factors (Mean VIF 1.48). The robustness of the IDH re-sults is assessed by comparing the results to that of a few competingmodels such as Tobit, HS, and DDH models (Appendix 2). The coeffi-cients in the HSmodel and DDH show similar directions and the signif-icant level to that the IDH implying the stability of the IDH results. Theresults of the key factors like social network, family labor, and financialinclusions in different model specifications appear to be stable indicat-ing that the IDH results are robust.


Another IDH model is estimated with inclusion of the square termsof age and education, and an interaction term for extension serviceand distance to the nearest handloom market in both the adoptionand extension of adoption equations (estimates not reported). All thecoefficients appear to be similar in direction, effect, and significancelevel as appeared in the original model. The squared terms of age andeducation are positive but are appeared to be statistically insignificant.Similarly, the effect of the interaction term appears to be insignificantbut with a negative effect. Further, introduction of districts dummy var-iables for location fixed effect brought no changes in terms of the direc-tion and significance level except for distance to the nearest handloommarket. The influence of distance to the nearest market which wasfound to be significant in the model without location fixed effect be-come insignificant in the second hurdle with location fixed effect. Over-all, the results of IDH models carry stability and thus the results areconsidered to be robust.

6. Conclusion

Technology adoption in micro-enterprises has assumed added im-portance as such enterprises generate employment and income for peo-ple in the lower segment of income distribution in the developingcountries. Given this context, thepresent study attempts to bring a com-prehensive analysis of what influences adoption and extent of deploy-ment of weaving technologies in the handloom industry in Assam. Itemphasizes on the role of social capital, family labor, andfinancial inclu-sion in adoption ofmodern technologywhich are still under-researchedin the current adoption literature. The study reveals that, though thestatus of adoption of modern weaving technologies is quite consider-able in a few sample districts, the low extent of its deployment is a mat-ter of concern. The results of the Cragg's independent double hurdlemodel show that the weaving experience, attitude towards risk, aware-ness about technology, financial inclusion, social network, family laborcontribution, and market distance significantly influence both adoptionand extent of deployment of handloom weaving technologies in therural areas of Assam.

The knowledge and information dissemination through micro-entrepreneur's social network is found to be effective in adoption anddiffusion of modern weaving technologies. Therefore, efforts should bemade to provide platforms for more interaction, training, workshop,and exhibition to strengthen the handloom social network. The contri-bution of family labor emerges to be crucial towards technology adop-tion addressing issues of financial liquidity and possible problem ofmoral hazard associated with hired labors. The present study alsoshows importance of financial inclusion in both technology adoptionand its extent of deployment among the micro-entrepreneurs in therural, nonfarm, and informal sector. Therefore, policy initiatives shouldbe made for extending financial inclusion programs to cover marginaland financially unsound micro-entrepreneurs especially in the ruralareas. The institutional training, to be effective on technology adoption,should beprovidedwith a better policy framework to tackle the existingbottlenecks related to infrastructure, access to credit/capital, marketinglinkages, etc. It is evident that the longer distant handloom marketcauses a higher transaction cost especially for the female micro-entrepreneurs and adversely affects the adoption and extent of deploy-ment of modern technologies. Therefore, developing marketing link-ages should be a policy priority to promote technology adoptionamong the rural micro-enterprises.

The study has a few limitations. It is cross-sectional in nature andfails to provide information on the trend of technology adoption overa period. Further investigations are needed related to the impacts ofdifferent types of social network and the firm size on technology adop-tion. Nevertheless, the present study contributes to the adoption litera-ture by providing important policy insights in an under-researchedhandloom industry which can be applied to other non-farm, informal,and rural micro-entrepreneurship in the developing countries.




Appendix 1. Variation in average partial effects (APEs) of financial inclusion and market distance on technology usage across location andgender

Characteristics

DBDKKLaUT

GM

GNAEWRFiFiTTEFiSoFaDC/srhWpAB

Please cite this artHurdle approach, T

Financial inclusion

icle as: Hazarika, B., et al., Adoption of modechnol. Forecast. Soc. Change (2015), http:/

Financial semi-inclusion

ern weaving technology in the handloom m/dx.doi.org/10.1016/j.techfore.2015.08.009

Market distance

Probability
Conditional Mean Probability Conditional Mean Probability
icro-enterprise

Conditional

s in Assam: A Do

Mean

istricts
aksa 0.217 0.130 0.188 0.022 −0.144 0.188 −0.004 −0.008 31.28 hemaji 0.015 0.018 0.259 −0.010 −0.020 0.000 −0.001 −0.001 24.85 amrup 0.122 0.019 0.302 0.100 −0.022 0.094 0.001 −0.001 14.00 okrajhar 0.134 0.079 0.089 0.040 −0.088 0.152 −0.002 −0.005 16.00 khimpur 0.012 0.010 0.000 −0.006 −0.011 0.000 0.001 −0.001 63.29 dalguri 0.236 0.152 0.118 0.018 −0.170 0.421 −0.005 −0.009 9.03 otal 0.132 0.074 0.159 0.030 −0.083 0.168 −0.002 −0.004 21.91
ender
ale 0.149 0.046 0.272 0.086 −0.052 0.175 0.001 −0.003 19.84 male 0.124 0.087 0.107 0.004 −0.097 0.164 −0.003 −0.005 22.86 Fe
Appendix 2. Maximum Likelihood estimates for different model specifications

Factors
Tobit Cragg's dependent Double Hurdle Heckman selection Cragg's independent Double Hurdle
1st Hurdle
2nd Hurdle 1st Hurdle 2nd Hurdle 1st Hurdle 2nd Hurdle
Coeff.
Sig. Coeff. Sig. Coeff. Sig. Coeff. Sig. Coeff. Sig. Coeff. Sig. Coeff. Sig.
ender
−0.022 0.753 −0.094 0.805 −0.061 0.020 −0.066 0.862 −0.061 0.022 −0.094 0.796 −0.061 0.011 ew firm 0.011 0.875 0.209 0.597 0.212 0.582 0.209 0.478 ge −0.008 0.026 −0.035 0.043 −0.001 0.664 −0.031 0.074 0.000 0.726 −0.035 0.059 0.001 0.895 ducation 0.005 0.587 0.024 0.596 −0.004 0.312 0.025 0.574 −0.004 0.309 0.024 0.536 −0.005 0.204 orking experience 0.026 0.001 0.087 0.014 0.006 0.020 0.074 0.050 0.006 0.028 0.087 0.007 0.005 0.058 isk averse −0.210 0.001 −0.729 0.001 −0.045 0.005 −0.702 0.000 −0.042 0.007 −0.729 0.000 −0.035 0.020 nancial inclusion 0.117 0.196 0.816 0.229 0.251 0.001 0.777 0.241 0.248 0.001 0.816 0.077 0.237 0.001 nancial semi-inclusion 0.020 0.829 −0.910 0.013 0.251 0.001 −0.948 0.008 0.253 0.001 −0.910 0.020 0.258 0.001 raining availed 0.011 0.887 −0.440 0.161 0.008 0.772 −0.446 0.160 0.011 0.695 −0.440 0.130 0.017 0.553 echnology awareness 0.330 0.001 1.289 0.007 1.300 0.006 1.289 0.002 xtension services 0.009 0.923 0.010 0.785 0.012 0.741 0.013 0.788 rm size 0.252 0.001 −0.025 0.018 −0.024 0.023 −0.022 0.056 cial capital −0.013 0.599 1.216 0.001 0.052 0.001 1.243 0.000 0.048 0.001 1.216 0.000 0.039 0.015 mily capital 0.355 0.003 1.045 0.044 0.182 0.001 1.107 0.032 0.179 0.001 1.045 0.030 0.174 0.001 istance −0.010 0.002 −0.048 0.001 0.003 0.025 −0.047 0.001 0.003 0.016 −0.048 0.000 −0.004 0.005 onstant −0.320 0.160 −1.467 0.256 0.471 0.001 −1.742 0.173 0.484 0.001 −1.467 0.248 0.513 0.001 igma 0.379 0.109 0.001 0.112 0.110 0.001 o 0.070 0.144 0.435 ald or LR Chi2 340.290 257.653 315.05 67.440 -value 0.001 0.001 0.001 0.001 IC 235.239 −13.365 −12.838 −13.246 IC 299.720 100.425 100.952 96.752 ample 328 328 328 328 S
Note: * The values of Tobit and Heckman selection model are LR Chi2.

References

Adesina, A.A., Zinnah, M.M., 1993. Technology characteristics, farmers' perceptions andadoption decisions: a Tobit model application in Sierra Leone. Agric. Econ. 9, 297–311.

Aristei, D., Pieroni, L., 2008. A double-hurdle approach to modelling tobacco consumptionin Italy. Appl. Econ. 40, 2463–2476.

Aristei, D., Perali, F., Pieroni, L., 2008. Cohort, age and time effects in alcohol consumptionby Italian households: a double-hurdle approach. Empir. Econ. 35, 29–61.

Bailey, T., 1993. Organizational innovation in the apparel industry. Ind. Relat. 31 (1), 30–48.Bandiera, O., Rasul, I., 2006. Social networks and technology adoption in Northern

Mozambique. Econ. J. 116 (514), 869–902.Beddig, C., 2008. Cluster development policy rooted in the collective efficiency approach:

an effective poverty alleviation tool in the Indian handloom sector? case studies:The Varanasi and Chanderi handloom clusters (2007–08). https://unido.org/fileadmin/user_media/Services/PSD/Clusters_and_Networks/publications/Cluster%20Development%20Policy%20in%20Indian%20Handloom%20Cl%20BeddigDec%2009.pdf (accessed on 14 September, 2013).

Beltran, J.C., White, B., Burton, M., Doole, G.J., Pannell, D.J., 2013. Determinants of herbicideuse in rice production in the Philippines. Agric. Econ. 44, 45–55.

Bortamuly, A., Goswami, K., 2012. Factors influencing wage structure of the handloomworkers in Assam: an assessment from gender perspective. J. Rural. Dev. 31 (2),139–150.

Bortamuly, A.B., Goswami, K., 2015. Determinants of the adoption ofmodern technology inthe handloom industry in Assam. Technol. Forecast. Soc. Chang. 90, 400–409 (Part B).

Bortamuly, A.B., Goswami, K., Hazarika, B., 2013. Determinants of occupational choice ofworkers in handloom industry in Assam. Int. J. Soc. Econ. 40 (12), 1041–1057.

Burke, W.J., 2010. Fitting and interpreting Cragg's tobit alternative using Stata. Stata J. 9(4), 584–592.

Choudhury, H.K., Goswami, K., 2013. Determinants of expansion of area under Jatrophaplantation in North East India: a Tobit analysis. Forest Policy Econ. 30, 46–52.

Cragg, J.G., 1971. Some statistical models for limited dependent variables with applicationto the demand for durable goods. Econometrica 39, 829–844.

Dangayach, G.S., Deshmukh, S.G., 2005. Advancedmanufacturing technology implementa-tion: evidence from Indian small and medium enterprises (SMEs). J. Manuf. Technol.Manag. 16 (5), 483–496.

Directorate of Economics and Statistics, 2011. Statistical Handbook of Assam 2010. Gov-ernment of Assam.

Feder, G., Just, R.E., Zilberman, D., 1985. Adoption of agricultural innovations in developingcountries: a survey. Econ. Dev. Cult. Chang. 33 (2), 255–298.

Fu, X., Pietrobelli, C., Soete, L., 2011. The role of foreign technology and indigenous innova-tion in the emerging economies: technological change and catching-up. World Dev.39 (7), 1204–1212.

Gomez, J., Vargas, P., 2012. Intangible resources and technology adoption in manufactur-ing firms. Res. Policy 41 (9), 1607–1619.

uble

http://refhub.elsevier.com/S0040-1625(15)00256-5/rf0005









https://unido.org/fileadmin/user_media/Services/PSD/Clusters_and_Networks/publications/Cluster%20Development%20Policy%20in%20Indian%20Handloom%20Cl%20BeddigDec%2009.%20pdf

































Goswami, K., 2009. Globalisation of silk trade: a comparative analysis between China andIndia. Indian J. Reg. Sci. 41 (1), 122–129.

Goswami, K., Choudhury, H.K., 2015. To grow or not to grow? Factors influencing theadoption of and continuation with Jatropha in North East India. Renew. Energy 81,627–638.

Goswami, K., Choudhury, H.K., Saikia, J., 2012. Factors influencing farmers' adoption ofslash and burn agriculture in North East India. Forest Policy Econ. 15, 146–151.

Greene, W.H., 2009. Econometric Analysis. 7th eds. Prentice Hall, New Jersey.Hall, B.H., Khan, B., 2003. Adoption of new technology. In: Jones, D.C. (Ed.), New Economy

Handbook. Academic Press, San Diego, CA.Hayami, Y., Ruttan, V., 1971. Agricultural Development: An International Perspective. John

Hopkins University Press, Baltimore, MD.Hazarika, B., Goswami, K., 2014. Rural non-farm micro-entrepreneurship or not: gender

issue in decision making. 6th Bolivian Conference on Development Economics(BCDE2014) (Available at http://www.inesad.edu.bo/bcde2014/papers).

Heckman, J., 1979. Sample selection bias as a specification error. Econometrica 47,153–161.

Jones, A.M., 1989. A double-hurdle model of cigarette consumption. J. Appl. Econ. 4 (1),23–39.

Kannabiran, G., Dharmalingam, P., 2012. Enablers and inhibitors of advanced informationtechnologies adoption by SMEs: an empirical study of auto ancillaries in India.J. Enterp. Inf. Manag. 25 (2), 186–209.

Kassie, M., Jaleta, M., Shiferaw, B., Mmbando, F., Mekuria, M., 2013. Adoption of interrelat-ed sustainable agricultural practices in smallholder systems: evidence from ruralTanzania. Technol. Forecast. Soc. Chang. 80 (3), 525–540.

Koellinger, P., 2008. The relationship between technology, innovation, and firm perfor-mance: empirical evidence from e-business in Europe. Res. Policy 37 (8), 1317–1328.

Lal, K., 1999. Determinants of the adoption of information technology: a case study of elec-trical and electronic goods manufacturing firms in India. Res. Policy 28 (7), 667–680.

Langyintuo, A.S., Mungoma, C., 2008. The effect of household wealth on the adoption ofimproved maize varieties in Zambia. Food Policy 33 (6), 550–559.

Latif, M.A., 1988. Towards technological, transition in the Bangladesh handloom industry:a project proposal on replacement of the ordinary fly-shuttle loom by ChittaranjanLoom. Bangladesh Institute of Development Studies, Bangladesh.

Marenya, P.P., Barrett, C.B., 2007. Household-level determinants of adoption of improvednatural resources management practices among smallholder farmers in westernKenya. Food Policy 32, 515–536.

Mazvimavi, K., Twomlow, S., 2009. Socioeconomic and institutional factors influencingadoption of conservation farming by vulnerable households in Zimbabwe. Agric.Syst. 101 (1–2), 20–29.

Ministry of Textile, 2015. Financial Year-wise, Sector-wise Production of Cloth. http://texmin.nic.in/ermiu/pdata/prod_sec_cloth.asp (accessed on 07 May, 2015).

Moffatt, P.G., 2005. Hurdle models of loan default. J. Oper. Res. Soc. 56, 1063–1071.NCAER, 2010. HandloomCensus of India 2009–10.Ministry of Textiles, Government of India

(http://www.ncaer.org/downloads/Reports/HandloomCensusReport.pdf accessed onJune, 2012).

Noltze, M., Schwarze, S., Qaim, M., 2012. Understanding the adoption of system technol-ogies in smallholder agriculture: the system of rice intensification (SRI) in TimorLeste. Agric. Syst. 108, 64–73.

Peltier, J.W., Zhao, Y., Schibrowsky, J.A., 2012. Technology adoption by small businesses:an exploratory study of the interrelationships of owner and environmental factors.Int. Small Bus. J. 30 (4), 406–431.


Rajesh, G.K., 2012. Diffusion of agricultural innovations in India the case of Bivoltine hy-brid technology in South Indian sericulture. Int. J. Agron. Plant Prod. 3 (10), 374–384.

Ricker-Gilbert, J., Jayne, T.S., Chirwa, E., 2011. Subsidies and crowding out: a double-hurdle model of fertilizer demand in Malawi. Am. J. Agric. Econ. 93, 26–42.

Rogers, E.M., 2003. Diffusion of Innovations. 5th eds. Free Press, New York.Shiferaw, B., Kebede, T., You, L., 2008. Technology adoption under seed access constraints

and the economic impacts of improved pigeonpea varieties in Tanzania. Agric. Econ.39 (3), 309–323.

Subrahmanya, M.H.B., 2006. Technological innovations in Indian small and medium en-terprises (SMEs) sector: does firm size matter? Int. J. Innov. Learn. 3 (5), 499–517.

Tambo, J.A., Abdoulaye, T., 2012. Climate change and agricultural technology adoption:the case of drought tolerant maize in rural Nigeria. Mitig. Adapt. Strateg. Glob.Chang. 17, 277–292.

Tewari, M., 2006. Adjustment in India's textile and apparel industry: reworking historicallegacies in a post-MFA world. Environ. Plan. 38, 2325–2344.

Tobin, J., 1958. Estimation of relationships for limited dependent variables. Econometrica26, 24–36.

Todd, P.R., Javalgi, R., 2007. Internationalization of SMEs in India: fostering entrepreneur-ship by leveraging information technology. Int. J. Emerg. Mark. 2 (2), 160–188.

Tripathi, R., Shastri, R.K., Agarwal, S., 2013. Survival and growth strategies for small andmedium scale enterprises in India: a key for sustainable development. In:Subrahmanya, M.H.B., Balachandra, P., Srinivasan, R. (Eds.), Driving the Economythrough Innovation and Entrepreneurship. Springer, New Delhi, pp. 163–174.

Varukolu, V., 2007. Technology Adoption of IndianGarmentManufacturing FirmsMS The-sis, Louisiana State University, USA (from etd.lsu.edu/docs/available/etd-07012007-180241/unrestricted/Thesis_venu_varukolu.pdf accessed on 20 July, 2013).

Vuong, Q.H., 1989. Likelihood ratio tests for model selection and non-nested hypotheses.Econometrica 57, 307–333.

Wooldridge, J.M., 2010. Econometric Analysis of Cross Section and Panel Data. 2nd eds.MIT Press.

Zhu, K., Xu, S., Kraemer, K.L., Dedrick, J., 2009. Global convergence and local divergence ine-commerce: cross-country analyses. In: Kraemer, K.L., Dedrick, J., Melville, N.P., Zhu,K. (Eds.), Global e-Commerce: Impacts of National Environment and Policy.Cambridge University Press, pp. 345–384.

Bhabesh Hazarika is a Ph. D. student of Economics in the Department of Humanities andSocial Sciences, Indian Institute of Technology Kharagpur, India. His research interests arein the areas of economics of choice, entrepreneurship, risk and uncertainty, and genderand income inequality.

Madhurjya Prashad Bezbaruah is a Professor in the Department of Economics, GauhatiUniversity, India. He did his Masters from Delhi School of Economics and received a doc-toral degree from Gauhati University of India in 1990. His research interests are in theareas of agricultural economics, development problems and issues.

Kishor Goswami is an Associate Professor of Economics in the Department of Humanities& Social Sciences, Indian Institute of Technology Kharagpur, India. His research focuses onDevelopment Economics, Agricultural Economics, and Economics of Biofuels.














http://www.inesad.edu.bo/bcde2014/papers


























http://texmin.nic.in/ermiu/pdata/prod_sec_cloth.asp

http://texmin.nic.in/ermiu/pdata/prod_sec_cloth.asp


http://www.ncaer.org/downloads/Reports/HandloomCensusReport.pdf






























http://etd.lsu.edu/docs/available/etd-07012007-180241/unrestricted/Thesis_venu_varukolu.pdf

http://etd.lsu.edu/docs/available/etd-07012007-180241/unrestricted/Thesis_venu_varukolu.pdf










Adoption of modern weaving technology in the handloom micro-enterprises in Assam: A Double Hurdle approach

Documents