Mediating Effects on Poverty Reduction in India Through Mahatma Gandhi National Rural Employment Guarantee Programme

The employment opportunities in rural areas have signifi cantly decreased for the last few decades in India. Th erefore, Government of India introduced Mahatma Gandhi National Rural Employment Guarantee Programme (MGNREGP) to create employment opportunities for ru ral people. Th e Programme is considered as a “silver bullet” for eradicating rural poverty and unemployment in India. Th e purpose of this empirical research study is to develop a new model for poverty reduction in rural India through this Programme. Th e novelty of this article is an attempt to develop an empirical research model that assists the Mahatma Gandhi National Rural Employment Guarantee Programme when mapping the level of economics service quality and thereby enhance the same. Th is Programme provides an alternative source of livelihood, which will have an impact on reducing migration, restricting child labor, alleviating poverty, and making villages self-sustaining through productive assets creation such as road construction, cleaning up of water tanks, soil and water conservation work, etc. for which it has been considered as the largest anti-poverty programme in the world. Th e paper critically examines the implementation process of this Programme and its impact on tribal livelihoods. Th e following research methodology is used in the article: the data were collected using a structured questionnaire. Th e sampling procedure used for this study is stratifi ed random sampling. Th e stratifi cation is done based on the Taluks are Kumbakonam (Th anjavur District), Keeranur (Pudukottai District) and Nagappatinam (Nagappatinam District) of Tamilnadu state of South-India for the nature of region South, East, Centre, West and North while selecting the MGNREGP Employees from each category, non-probabilistic convenience and judgmental sampling technique is used. Th e fi ndings and conclusions of the study reveal that millions of rural poor with the inclusion of new works under this Programme could able to get some employment which supports their livelihoods. Eff orts are exerted to improve more transparency and accountability in implementing this programme to ensure that the benefi ts reach out to the poor and the needy villagers. Th e regression analysis revealed that the Poverty Eradication on the various dimensions of Economic Development, infl uenced Economic Development followed by Social Development. Th e visual representation of results suggest that the relationships between the dimensions of Economic Development, Social development resulted in a signifi cant impact on the mediated factor ‘Poverty Eradication’. Th e paper suggests the policy framework for the stakeholders in eff ective implementation of the Programme.


MEDIATING EFFECTS ON POVERTY REDUCTION IN INDIA THROUGH MAHATMA GANDHI NATIONAL RURAL EMPLOYMENT GUARANTEE PROGRAMME
Th e employment opportunities in rural areas have signifi cantly decreased for the last few decades in India. Th erefore, Government of India introduced Mahatma Gandhi National Rural Employment Guarantee Programme (MGNREGP) to create employment opportunities for ru ral people. Th e Programme is considered as a "silver bullet" for eradicating rural poverty and unemployment in India. Th e purpose of this empirical research study is to develop a new model for poverty reduction in rural India through this Programme. Th e novelty of this article is an attempt to develop an empirical research model that assists the Mahatma Gandhi National Rural Employment Guarantee Programme when mapping the level of economics service quality and thereby enhance the same. Th is Programme provides an alternative source of livelihood, which will have an impact on reducing migration, restricting child labor, alleviating poverty, and making villages self-sustaining through productive assets creation such as road construction, cleaning up of water tanks, soil and water conservation work, etc. for which it has been considered as the largest anti-poverty programme in the world. Th e paper critically examines the implementation process of this Programme and its impact on tribal livelihoods. Th e following research methodology is used in the article: the data were collected using a structured questionnaire. Th e sampling procedure used for this study is stratifi ed random sampling. Th e stratifi cation is done based on the Taluks are Kumbakonam (Th anjavur District), Keeranur (Pudukottai District) A. ARULRAJ, R. RENA dence in 1947. Aft er the successful enactment of this programme, there was a victory for the fully-fledged right to employment in rural India, particularly for women and aged men. Th e programme aims at fulfi lling the needs of the rural masses by employing at least one member of the family. Th e main objective of MGNREGA was the creation of durable assets and strengthening the livelihood resource based on the rural poor. Some of the encouraging features of this programme are as follows: (i) generation of the slew of female employment (ii); enhancing greater economic security, higher farm wages, lower migration and building of infrastructure and so on.
The Mahatma Gandhi National Rural Employment Guarantee Scheme (MGNREGS) primary objective is to augment the wage employment and its auxiliary objective is to strengthen natural resource management through works that address causes of chronic poverty like drought, deforestation, soil erosion, etc., and thus encourage sustainable development.
Analysis of recent studies and publications. Since the date of its implementation, many social scientists attempted to study the impact of the Programme and also its implementation procedures. Sen et al. (2009) attempted to measure the outcome of good governance practised by Gram Panchayats (GPs) of the West Medinipur district of West Bengal through the employment generated under NREGS. Data regarding diff erent parameters such as transparency, accountability, eff ectiveness and effi ciency, equity was taken into consideration in this study. Khan, Ullah and Salluja (2007) have discussed the direct and the indirect eff ects of NREGS on employment generation and poverty reduction in a local area. For this, a detailed survey was done in a poor agricultural village with 400 households, nearly 2500 people.
Dey and Bedi (2010) studied the functioning of the NREGA between February 2006 and July 2009 in Birubham district, West Bengal. Th is study shows that, in Birubham, there was a great awareness about the NREGA in rural areas.
Nayak, Behera, and Mishra (2008) conducted their study in 2 districts of Orissa, mainly Mayurbhanj and Balasore. NREGA programme was fi rst introduced in 200 most underdeveloped districts of the country. Mathur (2007) alluded that a system of a regular and continuous fl ow of authoritative information is essential for NREGS government officials, panchayat functionaries, elected representatives, NGOs and community groups. Mathur (2009) states that in a social audit undertaken in Andhra Pradesh, it was found that in certain villages, some people revealed that they had not been paid for the work done, which led to fi nancial irregularities of Rs. 50,000.
Th e scientifi c novelty of the article. It is the biggest poverty alleviation programme in the world, which started with an initial outlay of Rs. 11,300 crore in 2006-2007 and then increased to Rs. 40,000 crore in 2010-2011. Th e Act provides a legal guarantee of employment for 100 days in every fi nancial year to adult members (at least one in a family) of any rural household that will do public work-related unskilled manual work at the statutory minimum wage. Th is minimum wage varies from state to state, in some states it is Rs. 80 whereas in others it is Rs. 125 or Rs. 120. According to the Act, the minimum wage cannot be less than Rs. 60. Th e 100 days of work was estimated because the agricultu ral sector does not employ people throughout the year. Th erefore, unskilled workers have no alternative source of income for the remainder of the year.
Th e other key attributes of this scheme are a time-bound guarantee, laborintensive work, decentralised participatory planning, women's empowerment, worksite facilities and, above all, transparency and accountability through the provision of social audits and right to information. Th ere is an eff ort to separate payment agencies from implementing agencies, thereby preventing embezzlement of wages (Vanaik and Siddhartha, 2008).
MGNREGP, which has been adopted in India as a strategy of inclusive growth, is alleged to have several shortcomings. Th ough its eff ective and fair implementation at the grass-root level may bring social equity and strengthen the income resource base of the poor, the fruits of its implementation are yet to be realized. Th e benefi ts of the program to the non-poor instead of the poor can exercise the former's economic power and contribute to the campaign funds of the major political parties in exchange for preferential treatment in welfare services (Jha et al., 2009).
Research methods and data. Nine dimensions are framed for this study. Th ese are: i) Providing Employment Opportunity, ii) Plan Working and Ad ministrative Management, iii) Minimum Wage Awareness, iv) Social Development, v) Banking Services and Utility, vi) Health and Insurance, vii) Poverty Eradication, viii) Children Education and Health, and ix) Economic Development. Here Demographic variables Providing Employment Opportunity, Plan Working and Administrative Management, Minimum Wage Awareness, Social Development, Banking Services, Health and Insurance, Children Education and Health are independent variables, and Poverty Eradication and Economic Development are the dependent variable. Th e article investigates how and to what extent the independent variables make changes in the dependent variable. Th e proposed conceptual research model shows the process of research as follows: (Fig. 1).
Th e research employed a cross-sectional methodological approach. Th e methodology described as cross-sectional "is one used to collect data on all re levant variables at one time" (O'Sullivan and Rassel, 1999). Th is approach is particularly useful for studies designed to collect information on attitudes and behaviors of large geographically diverse populations (O'Sullivan and Rassel, 1999).
Surveys generally fall into one of two categories, descriptive or relational. Descriptive surveys are designed to provide a snapshot of the current state of aff airs, while relational surveys are designed to empirically examine relationships among two or more constructs either in an exploratory or in a confi rmatory manner. Th e current study is a relational survey that seeks to explore the relation- Finally, demographic profi le information was given in the ten questions about respondents. All the dimensions were presented as statements on the questionnaire, with the same rating scale used throughout and measured on a seven-point, Likert-type scale that varied from 1 highly dissatisfi ed to 7 highly satisfi ed and Strongly Disagree to Strongly Agree. For conducting an empirical study, data were collected from respondents in the Wetland area and Dryland area in Rural India. Assurance was given to the respondents that the information collected from them will be kept confidential and will be used only for academic research purposes.
Stratifi ed sampling was the sampling procedure used for the study. Th e stratifi cation has been done based on the Taluks are Kumbakonam (Th anjavur District), Keeranur (Pudukottai District) and Nagappatinam (Nagappatinam District) for the nature of region south, east, centre, west and north while selecting the MGNREGP Employees from each category, non-probabilistic convenience and judgmental sampling technique was used. However, within such District, the respondents were selected by stratifi ed random sampling. Th e data collected were analyzed for the entire sample (also see IIT, Madras, 2009).
Procedure for Data Analysis. Th e data collected were analyzed for the entire sample. Data analyses were performed with Statistical Package for Social Sciences (SPSS) using techniques that included descriptive statistics, Correlation analysis and Analysis of Moment Structures (AMOS) package for Structural Equation Modelling and Bayesian estimation and testing.

Structural Equation Modelling
. Th e main study used Structural Equation Modelling because of two advantages: "(1) Estimation of Multiple and Inter related Dependence Relationships, and (2) Th e Ability to Represent Unobserved Concepts in Th ese Relationships and Account for Measurement Error in the Estimation Process" (Hair et al., 1998). Th erefore, multiple regressions were simultaneously estimated; the direct and indirect eff ects were identifi ed (Tate, 1998). However, a series of separate multiple regressions had to be established based on "theory, prior experience, and the research objectives to distinguish which independent variables predict each dependent variable" (Hair et al., 1998). In addition, because SEM considers a measurement error, the reliability of the predictor variable was improved. AMOS 7.0 (Arbuckle and Wothke, 2006), a computer program for formulating, fi tting and testing Structural Equation Models (SEM) to observed data, was used for SEM, and the data preparation was conducted with SPSS 13.0.
Linear Structural Equation Models (SEMs) are widely used in sociology, econometrics, management, biology, and other sciences. A SEM (without free parameters) has two parts: a probability distribution (in the Normal case specified by a set of linear structural equations and a covariance matrix among the "error" or "disturbance" terms), and an associated path diagram corresponding to the causal relations among variables specifi ed by the structural equations and the correlations among the error terms. It is oft en thought that the path diagram is nothing more than a heuristic device for illustrating the assumptions of the model. However, in this research study, we try to show how path diagrams can be used to solve several important problems in structural equation modelling.
Structural Equation Models with latent variables are more and more oft en used to analyze relationships among variables in marketing and consumer research (see for instance Bollen, 1989;Bentler, 1995;Schumacker and Lomax, 1996; Batista-Foguet and Coenders, 2000, for an introduction and Bagozzi, 1994 for applications to marketing research). Some reasons for the widespread use of these models are their parsimony (they belong to the family of linear models), their ability to model complex systems (where simultaneous and reciprocal relationships may be present, such as the relationship between quality and satisfaction), and their ability to model relationships among non-observable variables while taking measurement errors into account (which are usually sizeable in questionnaire data and can result in biased estimates if ignored).
As is usually recommended, a Confi rmatory Factor Analysis (CFA) model is fi rst specifi ed to account for the measurement relationships from latent to observable variables. In our case, the latent variables are the four perception dimensions and the observed variables the 30 perception items. Th e relationships among latent variables cannot be tested until a well-fi tting CFA model has been reached. In our case, the relationships among Mahatma Gandhi National Rural Employment Guarantee Programme (MGNREP), the mediating impact of Poverty Eradication with the E, AM, WA, SD, BS, HI, CEH and ED dimensions are of interest. Th is modelling sequence stresses the importance of the goodness of fi t assessment. As a combination of regression, path and factor analyses, in SEM, each predictor is used with its associated uncontrolled error and, unlike regression analyses, predictor multi-collinearity does not aff ect the model results.

Bayesian Estimation and Testing in SEM.
With modern computers and software, a Bayesian approach to structural equation modelling is now possible. Posterior distributions over the parameters of a structural equation model can be approximated to arbitrary precision with AMOS, even for small samples. Being able to compute the posterior over the parameters allows us to address several issues of practical interest. First, prior knowledge about the parameters may be incorporated into the modelling process in AMOS. Second, we need not rely on asymptotic theory when the sample size is small, a practice that is misleading for inference and goodness-of-fi t tests in SEM (Boomsma, 1983). Th ird, the class of models that can be handled is no longer restricted to just identifi ed or overidentifi ed models. Whereas each identifying assumption must be taken as given in the classical approach, in a Bayesian approach some of these assumptions can be specifi ed with perhaps more realistic uncertainty.
Regression Model of the "MGNREGP QUAL" Mediated Structural Model. In hierarchical regression, the predictor variables are entered in sets of variables according to a predetermined order that may infer some causal or potentially mediating relationships between the predictors and the dependent variable (Francis, 2003). Such situations are frequently of interest in the social sciences. Th e logic involved in hypothesizing mediating relationships is that "Th e Independent Variable Infl uences the Mediator Which, In Turn, Infl uences the Outcome" (Holmbeck, 1997). However, an important precondition for examining mediated relationships is that the independent variable is signifi cantly associated with the dependent variable before testing any model for mediating variables (Holmbeck, 1997). Of interest is the extent to which the introduction of the hypothesized mediating variable reduces the magnitude of any direct infl uence of the independent variable on the dependent variable (Baron and Kenny, 1986).
Hence the research report empirically tested the hierarchical regression for the model conceptualized in Figure 3 within the AMOS 20.0 graphics environment.
With the analyses conducted, the parameter estimates are then viewed within AMOS graphics and it displays the standardized parameter estimates. Th e regres-   Figure 4 shows the posterior frequency polygon of the distribution of the parameters across the 78 000 samples. Th e Bayesian MCMC diagnostic plots reveals that for all the fi gures the normality is achieved, so the structural equation model fi t is accurately estimated.
Th e trace plot also called as time-series plot shows the sampled values of a parameter over time. Th is plot helps to judge how quickly the MCMC procedure converges in distribution. Figure 5 shows the trace plot of the mediated MGNREGP QUAL model for the mediated factor Poverty Eradication Economic Development dimension across samples. If we mentally break up this plot into a few horizontal sections, the trace within any section would not look much different from the trace in any other section. Th is indicates that the convergence in    To determine how long it takes for the correlations among the samples to die down, an autocorrelation plot which is the estimated correlation between the sampled value at any iteration and the sampled value k iterations later for k = 1, 2, 3,…. is analyzed for the MGNREGP QUAL regression model. Figure 6 shows the correlation plot of the MGNREGP QUAL model for the mediated factor Poverty Eradication to Economic Development dimension across samples. Th e fi gure exhibits that at lag 100 and beyond, the correlation is eff ectively 0. Th is indicates that by 90 iterations, the MCMC procedure has essentially forgotten its starting position. Forgetting the starting position is equivalent to convergence in distribution. Hence it is ensured that convergence in distribution was attained and that the analysed samples are indeed samples from the true posterior distribution.
Even though marginal posterior distributions are very important, they do not reveal relationships that may exist among the two parameters. Th e frequency polygons given in the fi gure 4 describes only the marginal posterior distributions of the parameters. Figure 7 displays the two-dimensional plot of the bivariate posterior density across 78 000 samples. Ranging from dark to light, the three shades of gray represent 50 %, 90 % and 95 % credible regions, respectively. From the fi gure it is revealed that the sample respondent's responses are normally distributed.
Th e various diagnostic plots featured from Figure 4 to Figure 7 of the Bayesian estimation of convergence of the MCMC algorithm confi rms the fact that the convergence takes place and the normality is attained. Hence, the absolute fi t of the MGNREGP QUAL Regression Model. From the MGNREGP QUAL regression model, which is empirically tested with mediated factor Poverty Eradication with dimensions of Social Development, and the Economic Development (ED),

Fig. 7. Shows AMOS path diagram output for the overall MGNREGP QUAL Structural Equation Model
Source: Authors' own creation.
it is evident that the Mahatma Gandhi National Rural Employment Guarantee Programme should concentrate on the Poverty Eradication (PE) as the mandatory aspect of Mahatma Gandhi National Rural Employment Guarantee Programme in the Wetland area and Dryland area in Rural India, India. Th e overall regressed model statistically proved that the poverty reduction is mediated eff ects on the economic development process in the above-stated programme in India. Structural Equation Modeling of "MGNREGP QUAL" Mediated Model. Since the Economic Development in MGNREGP is a theoretical construct, the research study has defi ned its dimension based on the setting used to explore the construct. If Mediated "MGNREGP QUAL" Model is to be applicable in the Indian context, the dimensions and the sub-dimensions on Economic Development have to be reliable and valid in measuring Economic Development in the MGNREGP. Th e model examines the relative importance of dimensions of Economic Development and Poverty Eradication in Mahatma Gandhi National Rural Employment Guarantee Programme in the Wetland area and Dryland area in Rural India. consists of 4 sub-dimensions, and Economic Development (ED) dimension consists of 6 sub-dimensions. Th e RMSEA fi t statistics for the model was 0.05, which was considered as a best-fi t model (Browne and Cudeck, 1993;Diamantopoulos and Siguaw, 2000). Th e sampling distribution for the RMSEA can be derived, which makes it possible to compute confi dence intervals. Th ese intervals allow research reports to test for a close fit and not only for an exact fit, as the X2 does. If both extremes of the confi dence interval are below 0.05, then the hypothesis of close fit is rejected in favour of the hypothesis of better than close fit. If both extremes of the confi dence interval are above 0.05, then the hypothesis of close fi t is rejected in favour of the hypothesis of bad fi t (Senthilkumar and Arulraj, 2011). Th e path diagram shows the Poverty Eradication is the Mediating factor for Economic Development. Th e regression coeffi cient -0.48 signifi es the impact of mediating factor Poverty Eradication (PE) on the on the other dimensions towards Economic Development (ED) the Mahatma Gandhi National Rural Employment Guarantee Programme.
Main results of the research. Th is research has adopted the procedure of assessing convergence of MCMC algorithm of maximum likelihood. To estimate the MCMC convergence, the research report has adopted two methods, namely convergence in distribution, the convergence of posterior summaries. Th e values of the posterior mean accurately estimate the MGNGREGP QUAL mediated SEM model. Th e highest value of Convergence Statistics (C.S) is 1.001 which is less than the 1.002 conservative measures (Gelman et al., 2004).
Th e trace plot shows the sampled values of a parameter over time. Th is plot helps to judge how quickly the MCMC procedure converges in distribution. Th e following fi gure 9 shows the trace plot of the mediated MGNREGP QUAL model for the mediated factor Poverty Eradication with Economic Development dimension across 57000 samples. If we mentally break up this plot into a few horizontal sections, the trace within any section would not look much diff erent from the trace in any other section. Th is indicates that the convergence in distri bution takes place rapidly. Hence, the mediated MGNREGP QUAL MCMC procedure very quickly forgets its starting values.
To determine how long it takes for the correlations among the samples to die down, an autocorrelation plot which is the estimated correlation between the sampled value at any iteration and the sampled value k iterations later for k = 1, 2, 3,…. is analysed for the MGNREGP QUAL regression model. Figure 10 shows the correlation plot of the MGNREGP QUAL model for the mediated factor Poverty Eradication with Economic Development dimension across 57 000 samples. Th e fi gure exhibits that at lag 100 and beyond, the correlation is eff ectively 0. Th is indicates that by 90 iterations, the MCMC procedure has essentially forgotten its starting position. Forgetting the starting position is equivalent to convergence in distribution. Hence it is ensured that convergence in distribution was attained and that the analysis samples are indeed samples from the true posterior distribution.
Economical Implication. From this empirical analysis, the research identifi ed that Poverty Reduction and Economic Development is the Moderating or In India, the Mahatma Gandhi National Rural Employment Guarantee Programme faces the challenge of employing a broad range of benefi ciaries. Th us,  what appears from the critical examination of some of the poverty allevia tion programmes is that the various rural development programmes have achieved expected development and may have been very little progress. It is a continuous process. Although rural poverty cannot be eliminated immediately, continuous eff orts should be made to develop the rural areas on a priority basis. Th e crux of the problem is that the rural poor have to be further organized and motivated on an on-going basis to improve their living conditions. See fi gure 11 for more understanding.
Conclusion. In this article we provided a conceptualized research model that can off er a framework on how to maintain sustainable development in rural India through the MGNREGP. MGNREGP is the ultimate employment generating scheme for reducing the poverty in a rural area and sustainability common resource property in a village of India. Hence, the MGNREGA is considered as a "Silver Bullet" for eradicating rural poverty and unemployment, by way of generating demand for a productive labor force in villages. MGNREGA is distinctive for its unique vision to redefi ne avenues of providing employment opportunities to the deprived in rural India. But the possibility and effi cient chances of employment largely come with a better level of awareness as it marks the level of accessibility. It thus necessitates suffi cient awareness amongst the intended benefi ciaries regarding provisions like guaranteed days of employment, unemployment allowance, minimum wages, availability of complaint register, etc. As most of the worker respondents are illiterate and belong to the economically poor class, the extent of awareness about NREGA has emerged out to be a major concern in all the hamlets. Th e procedural and implementation aspects of NREGA have never been free from confronting some basic challenges like general awareness, understanding policy nitty-gritty, suffi cient access etc.
Given the socioeconomic background of the respondents, the structural issues such as transparency, maintenance of documents and accountability were diffi cult things to actualize from the workers' point of view. Above stated models is more correlated to economic development in the study area. Nevertheless, poverty could be reduced by providing employment opportunities for rural India.
MGNREGA is a landmark legislation in the history of social security legislation in India aft er independence. Enacted aft er a successful struggle for a comprehensive employment guarantee law, this legislation is a partial victory towards a full-fl edged right to employment. Th ough MGNREGA is well-thought-out legislation, a powerful tool in the hands of the common people to get their basic livelihood, its poor execution deprives the rural poor of their basic rights. Th e study reveals that despite numerous problems, MGNREGA is a program that has begun to make a diff erence in the lives of women. For example, women have started asserting their voices in family matters and the nature of spending money. Th ough awareness continues to be a stiff challenge, women in the study area have become pro-active learners and participants in the schemes.