Has Social Sector Development Catalysed the Inclusiveness of India’s Economic Growth?

Abstract

The study attempts to examine the impact of social sector development on inclusive growth in India. Ever since Independence, India’s encounter with gnawing poverty and stark deprivation, particularly of the weaker and the marginalised sections of society, cajoled India’s planners to moot the development policies with the sole objective of exacerbating growth with equity. That is why since the beginning of the planning era, the stress had been laid on strengthening and expanding the social sectors with the premise that it would boost the inclusive growth agenda, manifesting in equal access to employment and economic opportunities; equal participation in decision-making and reduction in poverty and inequality. In conformity with the objective of our study, we intend to gauge the effect and contribution of different components of social sector development in India, on inclusiveness of growth using time series data for the period of 1985–1986 to 2015–2016. It was found that the expenditure related to ‘social security and welfare’ contributes significantly towards inclusive growth in India while the expenditure incurred on ‘welfare of marginalised class’ and ‘rural development’ exudes negative association with inclusiveness of growth.

Keywords

Social services poverty and inequality health education and welfare economics equality and inclusion human development

JEL Codes: I30, H5, O11

1. Introduction

Elimination of poverty and employment generation has remained the central pivot around which the development policy has revolved in India. Since Independence, India ardently intended its growth spectrum to be inclusive so that the people even at the grass-roots level could be associated in the development process. This process of development would lead to elimination of hunger, disease, deprivation and reining in inequality. To achieve this, growth had to be inclusive and broad based. Even when India based its development policy on Nehruvian tenets involving extensive investment, heavy industrialisation, import substitution, indigenisation of goods and services, even then employment generation and participation was not side tracked. India intended to evolve a people-centric approach to development (Balkrishnan, 2013). Elimination of monopolies, prevention of concentration of economic power, socio-economic justice and participative growth have all veered round the poorest of the poor. The failure of the trickle-down approach further strengthened India’s resolve to reinforce the inclusiveness of growth. Following Sen’s approach, social sector development had been given precedence to build the capacity and capability of the people (Fukuda-Parr, 2003).

Social sectors comprise social and economic services. ‘Social services’ include education, art and culture, medical and public health, family welfare, water supply and sanitation, housing and urban development, welfare of the underprivileged classes, labour and employment, social security and welfare and other social services. On the other hand, ‘economic services’ relate to rural development and food storage and warehousing. All this social sector expenditure is geared towards the achievement of a broader objective of expanding social opportunities and improving the social indicators of education, health and nutritional standards of the general population (Dev & Mooij, 2004). Social sector development focused on greater livelihood opportunities, modern amenities and services for decent living in rural areas as well as in urban areas, which would increase the opportunities for equal economic participation and thus promote inclusiveness. The social sector development approach of growth policy hinges on human development and human welfare and therefore exacerbates inclusive context of growth (Government of India, 2011).

Government of India (2008) shifted its conventional strategy of high investment and growth to new strategy of achieving faster growth with inclusiveness. India’s commitment to planned economic development was to improve the economic conditions of the people and to achieve the broad-based development for higher living standards of people. The growth must be widely spread so that its outcome in terms of income and employment would be shared by the poor and deprived sections of the society.

Both social sector development and inclusive growth are synergetic in nature. Increasing expenditure on social sector would improve the conditions of the deprived and marginalised sections of society. By improving their income and equal access to employment and productive activities, the deprived sections would be empowered to become equal partners in the growth process. Social sector development would mitigate poverty, inequality, deprivation and diseases. This would promote the inclusiveness of development.

This study is an attempt to gauge the impact of social sector development on inclusive growth in India. The paper is divided into four sections. After introducing the theme in Section 1, Section 2 elaborates data sources used and lays down the conceptual framework and model estimation to arrive at findings and inferences. Section 3 is devoted to the analysis of the impact of different components of social sector development on a composite index of inclusive growth and digs out findings and inferences. Section 4 puts forth main policy implications emanating from the analysis and concludes the paper.

II. Database and Methodology

The importance of the study lies in both theoretical and practical aspects. On the theoretical aspect, the study is expected to analyse the relevance of social sector development in India for rendering development inclusive. On the practical side, the analysis of the relative significance of the different components of social sector in effecting change in inclusive growth can be indicative for policy makers to focus on those social sectors that would exacerbate inclusive growth.

Data for the present study have been taken from various reports of the World Bank, the Planning Commission, Economic Surveys, Annual Budgets, Statistical Abstracts and so on. Time reference of the data and analysis is the 1985–2016 period. All the monetary values have been expressed in constant prices of 2004–2005.

To measure the impact of different components of social sector development on the inclusiveness of India’s growth, a composite index of inclusive growth was constructed by deploying the principal component analysis approach.

Conceptualising Inclusive Growth: Components and Indicators

Deciphering the concept of inclusive growth, precisely it implies the reduction of poverty and inequality, and equal access for all in social and economic opportunities. In other words, inclusive growth will manifest in employment generation and participative decision-making both at micro village level and the macro level of the whole economy. In the present study, the concept of inclusive growth is represented by three components, namely, (a) poverty reduction, (b) equal access to employment opportunity and (c) social and economic inclusion in the growth process (Ramos et al., 2013). These three components of inclusive growth are further represented each by a subset of indicators to generate the index of inclusive growth as depicted in Figure 1.

Constructing Composite Index of Inclusive Growth: The Model

A composite index of inclusive growth is generated based on a scoring methodology and a weighing scheme. The present study employed principal component analysis to the nine indicators Figure 1 for constructing a composite index of inclusive growth. Principal component analysis is a useful technique for transforming a large number of variables into a smaller set of uncorrelated factors the principal components. It attempts to identify the underlying variables or components that explain the pattern of correlation within a set of observed variables (Anthony & Rao, 2007).

Figure 1.

Source: 1. Ramos et al. (2013).

In the present study,

X₁ = ß₁₍₀₎+ ß₁₍₁₎ F₁+ ß₁₍₂₎ F₂+ e₁

X₂ = ß₂₍₀₎+ ß₂₍₁₎ F₁+ ß₂₍₂₎ F₂+ e₂

… … …

X₉ = ß₉₍₀₎+ ß₉₍₁₎ F₁+ ß₉₍₂₎ F₂+ e₉

where X_1,X_{2, … ,} X₉ are the nine variables representing inclusiveness; F₁ and F₂ are the two components extracted; ß₁₍₁₎ is the loading of variable X₁ on factor F₁ and so on; and e₁, e₂,…, e₉ indicate that the hypothesised relationships are not accurate.

The data set is further aggregated into the composite variables by providing weights known as ‘component loadings’. Component loadings are equivalent to the correlation between components and variables when only a single common component is involved. High loadings provide meaning and interpretation of factors. By applying principal component analysis, as many principal components can be obtained as the number of indicators. The first principal component explains the maximum variance in the set of standardised indicators, while the second component explains the maximum in the residual variance, that is, the variance not explained by the first principal component and so on. The second factor represents the majority of the residual variance, subject to being not related with the first factor (Pradhan & Puttuswamaiah, 2005).

A combined weighted inclusive growth index was calculated using the proportion of these percentages as weights on principal component score coefficients (Sekhar et al., 1991). It was formulated using the following formula:

Non-standardised inclusive growth index = (73.243/84.125) × (PC1 score) + (10.943/84.125) × (PC2 score)

The values of the index thus constructed come out to be positive or negative. Therefore, in order to make an easy interpretation, a standardised index was developed, the value of which ranges from 0 to 1 (Anthony & Rao, 2007), using the following formula:

\begin{array}{l} Standardised inclusive growth index \\ = \frac{Non-standardised inclusive growth index - Minimum value non-standardised inclusive growth index}{Non-standardised inclusive growth index - Minimum value non-standardised inclusive growth index} \end{array}

Components of Social Sector as Independent Variables

The following components of social sector have been chosen, representing independent variables in our regression model, to infer about the significance of each in influencing inclusive growth in India.

Expenditure incurred on education, sports, arts and culture as a proportion of total central government plan outlay (lnPOED);

Expenditure on medical and public health and family welfare as a proportion of total central government plan outlay (lnPOHFW);

Expenditure on water supply and sanitation as a proportion of total central government plan outlay (lnPOWSS);

Expenditure on housing and urban development as a proportion of total central government plan outlay (lnPOHUD);

Expenditure on welfare of under privileged classes as a proportion of total central government plan outlay (lnPOWPC);

Expenditure on labour and employment as a proportion of total central government plan outlay (lnPOLE);

Expenditure on social security and welfare as a proportion of total central government plan outlay (lnPOSSW);

Expenditure on rural development as a proportion of total central government plan outlay (lnPORD);

Expenditure on food storage and warehousing as a proportion of total central government plan outlay (lnPOFSW).

The secondary data have been compiled from various data sources. The data series used in the study along with the respective data sources is provided in Table 1.

Table 1

Variables with Respective Data Sources

S. No.	Variables	Data Source
1	Declining absolute poverty	Government of India (2009), World Bank (Datt & Ravillion, 2010) and Planning Commission
2	Dissipating income inequalities	World Bank (Datt, 1999) and United Nation Development Programme (Radhakrishna & Panda, 2006)
3	Rising female-to-male ratio in labour	Government of India, 2015; World Bank Open Data, 2016, and various rounds of NSS quinquennial data on employment and unemployment situation in India
4	Universal employment generation	Ministry of Rural Development, 2016
5	Ascending share of employment of SC and ST classes	Various series of NSS quinquennial data on employment and unemployment situation among social groups in India and Handbook on Social Welfare Statistics
6	Growing gender parity index	Various reports of Education Statistics at a Glance
7	Increasing female-to-male ratio in Parliament	Human Development Reports by UNDP and Electoral Statistics Pocketbook of Elections Commission of India
8	Rising gross enrolment of SC/ST classes	Various reports of Handbook on Social Welfare Statistics
9	Growing political participation of reserved category	Elections Commission of India issued the data about political participation of SCs and STs in Electoral Statistics Pocketbook
10	Various expenditure on components of social sector development	Annual reports of Planning Commission (various issues), Economic Surveys (various issues) and Outlays Budget (various issues)

Source: Author’s own configuration.

The Model to Estimate the Impact of Social Sector Development on India’s Growth Inclusiveness

After constructing a composite index of inclusive growth, backward stepwise multiple regression analysis is deployed by regressing expenditure incurred on various components of social sector in India, on inclusive growth index as the dependent variable. In the backward elimination method, variables are successively removed, which contribute less in the model. It is a systematic mode of removing the multilinear variables from the model based on their statistical significance in regression. This method discontinues when no variable significantly improves the model (Gujarati, 2003). Backward stepwise multiple regression is challenging if there are a large number of candidate variables, and impossible if the number of candidate variables is larger than the number of observations (Smith, 2018). This technique also has few limitations, such as the presence of biasness in parameter estimation, inconsistencies among model selection algorithms, and problem of multiple hypotheses testing (Whittingham et al., 2006). Because of these limitations, all the assumptions of the regression model used in the present study were checked.

Natural log transformations of the variables are preferred to avoid the problem of heteroscedasticity and cumbersomeness while framing the model. This transformation compresses the scale to a twofold difference from tenfold differences between the values in which variables are measured (Gujarati, 2003). Anand et al. (2013) constructed a model to measure the inclusive growth by incorporating indicators of economic growth performance. In that model, inclusive growth was considered as a dependent variable and indicators of economic growth as independent variables by taking log differences of both independent and dependent variables.

In the present study, the formulation of the model is based on log differences of both independent and dependent variables. Regression estimations have been carried out using SPSS. The model is shown as follows:

lnIGI* = a + b₁₁ lnPOED* + b₁₂ lnPOHF* + b₁₃ lnPOWSS* + b₁₄ lnPOHUD* + b₁₅ lnPOWPC* + b₁₆ lnPOLE* + b₁₇ lnPOSSW* + b₁₈ lnPORD* + b₁₉ lnPOFSW* + u_1t

Here,

ln = natural logarithm

lnIGI = inclusive growth index

lnIGI* = ln (IGI_i,t –IGI_i,t−1)

lnPOED* = b₁ln (POED_i,t − POED_i,t−1) and so on.

a = constant;

b₁₁, b_12, b_13, b₁₄, b_15, b_16, b_17, b₁₈ and b₁₉= regression coefficients;

u_t = white-noise disturbance term.

III. Results and Discussions

India’s Composite Index of Inclusive Growth

The idea for constructing an inclusive growth index for India during the period 1985–2016 as per the methodology delineated above is to aggregate these multiple variables into a single index which best represented the set of information. This reduces the data from a complex, multidimensional frame to a single dimension which is easier to interpret. Table 2 explains the descriptive statistics of the variables after standardisation of data. The data set for all the variables was standardised in order to avoid the influence of one variable upon principal components. The variables after standardising had zero mean and unit variance (Pradhan & Puttuswamaiah, 2005).

The next step is to assess the strength of the relationship between the variables by examining the partial correlations through the Kaiser–Meyer–Olkin measure and to study the coefficient correlations using Bartlett’s Test of Sphericity. Table 3 shows the Kaiser–Meyer–Olkin (KMO) measure at 0.829. It indicates a good measure, and Bartlett’s test of sphericity shows a significance level of 0.000, a value that is small enough to reject the null hypothesis.

The total variance explained how the variance is divided among the possible components. For the present data set, the first two principal components explained more than 80 per cent of the variation (Leech et al., 2014). Table 4 projects that the first principal component explains 73.243 per cent of the variation, and the second principal component explained 10.943 per cent of the variation which shows the importance of principal components and leads to the construction of inclusive growth index.

The rotated component matrix in Table 5 explains the plot of component loadings in which ‘universal employment generation’ (0.951), ‘rising gross enrolment of Scheduled Castes (SC) and Scheduled Tribes (ST)’ (0.917), ‘increasing female-to-male ratio in Parliament’ (0.845), ‘dissipating income inequalities’ (0.837) and ‘growing gender parity index’ (0.830) all of them load high, but ‘universal employment generation’ (0.951) has most high and positive loading on the first component. ‘The per cent share of SC and ST classes in employment’ (0.882) has a loading near zero on the first component but loads high on the second.

Table 2

Descriptive Statistics of the Variables after Standardisation of Data

S. No.	Variables	Minimum	Maximum	Mean	SD
1	Declining absolute poverty	−1.32	2.23	0.00	1.00
2	Dissipating income inequalities	−1.82	1.84	0.00	1.00
3	Rising female-to-male ratio in labour	−1.69	1.24	0.00	1.00
4	Universal employment generation	−0.71	2.04	0.00	1.00
5	Ascending share of employment of SC and ST classes	−1.82	2.52	0.00	1.00
6	Growing gender parity index	−1.39	1.74	0.00	1.00
7	Increasing female-to-male ratio in Parliament	−1.11	2.09	0.00	1.00
8	Rising gross enrolment of SC/ST classes	−1.18	1.64	0.00	1.00
9	Growing political participation of reserved category	−2.78	1.92	0.00	1.00

Source: The authors.

Table 3

KMO Measure of Sampling Adequacy and Bartlett’s Test of Sphericity

KMO Measure of Sampling Adequacy	Bartlett’s Test of Sphericity
KMO Measure of Sampling Adequacy	Chi-Square	Df	Sig.
0.829	330.625	45	0.000

Source: Authors’ calculations using IBM SPSS Statistics 21.

Table 4

Principal Components, Eigenvalues and Total Variance

Component	Initial Eigenvalues
Component	Total	% of Variance	Cumulative %
1	7.324	73.243	73.243
2	1.094	10.943	84.125
3	0.590	5.900	90.085
4	0.384	3.845	93.930
5	0.329	3.292	97.222
6	0.136	1.361	98.583
7	0.069	0.691	99.274
8	0.041	0.406	99.680
9	0.032	0.321	100.00

Source: Authors’ calculation using IBM SPSS Statistics 21.

Note: The significance of bold values, has been explained in the first paragraph on page 8.

Table 5

Varimax Rotated Matrix: Values of Components Loadings after Rotation

Components	Variables	Component 1	Component 2
1	Declining absolute poverty	−0.461	−0.715
2	Dissipating income inequalities	0.837	0.128
3	Rising female-to-male ratio in labour	−0.913	−0.173
4	Universal employment generation	0.951	0.123
5	Ascending share of employment of SC and ST classes	0.029	0.882
6	Growing gender parity index	0.830	0.517
7	Increasing female-to-male ratio in Parliament	0.845	0.401
8	Rising gross enrolment of SC/ST classes	0.917	0.339
9	Growing political participation of reserved category	0.635	0.546

Source: Authors’ calculation using IBM SPSS Statistics 21.

Note: 1. Principal component analysis was used as an extraction method.

2. Rotation method: VARIMAX rotation with Kaiser normalisation.

3. Rotation convergence in three iterations.

4. The significance of bold values has been explained in the second paragraph on page 8.

By using the SPSS procedure, component scores were carried out, which were saved as variables in calculating the index value. Using the proportion of these percentages as weights on the component score coefficients, a non-standardised index was developed for two extracted components. The values of the index thus constructed come out to be positive or negative; therefore, in order to make an easy interpretation, a standardised index was developed (Krishnan, 2010).

In spite of the positive trend of gross domestic product (GDP) in earlier plan periods, still the Indian economy lagged behind in the development because it was not perceived as being sufficiently inclusive for many groups, especially SCs, STs and minorities. Gender in equality also remained a pervasive problem. The lack of inclusiveness was proven in earlier years (Government of India, 2008).

Table 6 reveals the composite index of inclusive growth in India. Non-standardised index of inclusive growth indicates negative values of inclusiveness until the year 2007. Non-standardised index shows the positive value of 0.433 to 1.267 from 2008 to 2015, while the highest value is 1.588 in 2009–2010.

Table 6

Composite Index of India’s Inclusive Growth

Years	Component Scores		Inclusive Growth Index
Years	Component 1	Component 2	Non-standardised Index	Standardised Index
1985–86	−0.297	−1.989	−0.517	0.160
1986–87	−3.067	−2.053	−0.534	0.154
1987–88	−0.316	−2.120	−0.551	0.147
1988–89	−0.326	−2.188	−0.569	0.139
1989–90	−0.337	−2.256	−0.586	0.132
1990–91	−0.347	−2.323	−0.604	0.125
1991–92	−0.240	−1.825	−0.447	0.189
1992–93	−0.515	−1.333	−0.623	0.119
1993–94	−0.795	−1.246	−0.855	0.026
1994–95	−0.907	−0.407	−0.843	0.031
1995–96	−0.717	−0.654	−0.709	0.084
1996–97	−0.877	−0.377	−0.813	0.043
1997–98	−0.965	−0.594	−0.918	0.001
1998–99	−0.673	−0.173	−0.609	0.124
1999–2000	−0.627	0.376	−0.498	0.169
2000–01	−0.265	−0.642	−0.314	0.242
2001–02	−1.023	1.015	−0.759	0.064
2002–03	−0.783	0.419	−0.627	0.117
2003–04	−0.898	1.293	−0.613	0.122
2004–05	−0.097	−0.304	−0.124	0.318
2005–06	−0.466	0.814	−0.300	0.248
2006–07	−0.610	2.252	−0.239	0.272
2007–08	0.426	0.476	0.433	0.540
2008–09	1.321	−0.966	1.025	0.776
2009–10	2.044	−1.475	1.588	1.001
2010–11	1.441	0.304	1.295	0.884
2011–12	1.311	0.506	1.208	0.849
2012–13	1.286	0.572	1.194	0.844
2013–14	1.329	0.939	1.280	0.878
2014–15	1.301	1.031	1.267	0.873
2015–16	1.670	1.409	1.637	1.020

Source: Authors’ calculations using IBM SPSS Statistics 21.

Note: Principal component analysis was used as an extraction method for component scores.

Government of India (2008) explained that though the percentage of the poor below poverty line had decreased, the rate of decline in poverty had not accelerated along with the growth in GDP. The incidence of poverty had hardly declined among the marginalised groups. The chasm between the rich and the poor had been nudging rapidly. The indicators of human development had shown the lagged, but steady improvement. Total employment in the economy has improved, but the labour force has grown even faster that led to an increase in the unemployment rate. The standardised inclusiveness index of growth reveals increasing trend initially till 2009–2010, when it starts showing decreasing trend again. The analysis shows a slower trend growth rate of inclusiveness in India.

India has experienced rapid growth over the past decade, but the income inequality had also inflated. There exists vast disparity in education and health conditions across different strata of the population. Inequality of opportunities among different sections of society hinders the growth inclusiveness and thus should be surmounted. Equalising of opportunities will improve the equality in access to benefits that would strengthen the livelihood strategies through empowerment. With the implementation of policy of National Rural Health Mission in 2005, there has been some improvement in the growth phase. The National Rural Health Mission (NRHM) focused on providing equitable, affordable health care facilities to the rural population, specifically the vulnerable groups. Government of India reported that NRHM reduced infant mortality rate at higher rate than earlier (during 2003–2006), increased institutional deliveries, raised the figures of full immunisation, constituted Rogi Kalyan Samitis, appointed and trained accredited social health activists (ASHAs), constituted village health committees, created village health and nutrition days, provided mobile medical units and co-located Ayurveda, Yoga and Naturopathy, Unani, Siddha and Homoeopathy (AYUSH) in a number of health facilities (Sharma, 2014). With the launching of numerous programmes, appropriate steps must be taken for implementation and governance problems to bolster inclusiveness (Government of India, 2013).

Impact of Components of Social Sector on Inclusive Growth

The inclusive growth has remained the prime agenda for India’s development policy, and social sector development reinforces this mission. In fact, social sector development and inclusive growth are both interdependent and synergetic in nature. Social sector development stimulates the achievement of inclusive growth, while the realisation of inclusive growth manifests in social sector development. Since Independence, emphasis had been laid on the development of education, health, and rural sectors with the objective of eliminating poverty and hunger and empowering the weaker section of society. Continuing the same endeavour, in the Eleventh Plan, new priorities were set related to revitalising dynamism in agriculture and building an appropriate infrastructure in rural areas. Similarly, it emphasised undertaking programmes for improving living conditions for the marginalised sections, and improving their access to economic opportunities for social and economic empowerment. Likewise, greater access to education and health was much needed for the removal of chronic poverty, ignorance, disease and inequality. Although the target of multifaceted inclusive growth may be difficult to achieve in a heterogeneous society like India, yet it remains a mandated goal of India’s economic policy (Government of India, 2008).

The Regression Results

To estimate the impact of various components of social sector development on inclusive growth in India, the backward approach of multiple stepwise regressions is deployed. The first step is to check the minimum and maximum values of the independent variables, so that implausible values (outliers) can be detected. Table 7 shows the minimum, maximum, mean and standard deviation values of the independent variables. The average values of lnPOED, lnPOHF, lnPOWSS, lnPOHUD, lnPOWPC, lnPOLE, lnPOSSW, lnPORD and lnPOFSW are 0.303, 0.076, −0.013, 0.180, 0.123, −0.024, 0.100, 0.067 and 0.003, respectively. The values of standard deviation of lnPOED, lnPOHF, lnPOWSS, lnPOHUD, lnPOWPC, lnPOLE, lnPOSSW, lnPORD and lnPOFSW are 0.524, 0.668, 0.410, 0.584, 0.590, 0.166, 0.235, 1.230 and 0.210, respectively. None of these components has any missing values with no presence of outliers in the data.

Table 8 shows that in the first model, all the variables are entered. After that, it removes the insignificant variable step by step and further comes with three variables which are significant that entered in the model. Here, ‘Expenditure incurred on welfare of under privileged classes’ (lnPOWPC), ‘Expenditure on social security and welfare’ (lnPOSSW) and ‘Expenditure on rural development’ (lnPORD) contributed towards the inclusive growth of India in the given model.

Table 7

Descriptive Statistics of the Independent Variables

S. No.	Variables	Minimum	Maximum	Mean	SD
1	lnPOED	−0.851	1.898	0.303	0.524
2	lnPOHW	−1.882	1.725	0.076	0.668
3	lnPOWSS	−0.866	1.099	−0.013	0.410
4	lnPOHUD	−0.618	2.794	0.180	0.584
5	LnPOWPC	0.067	0.301	0.123	0.590
6	LnPOLE	−0.860	0.141	−0.024	0.166
7	LnPOSSW	−0.879	0.336	0.100	0.235
8	lnPORD	−3.166	3.492	0.067	1.230
9	lnPOFSW	−0.064	0.040	0.003	0.210

Source: The authors.

Table 8

Variable Entered/Removed as per the Backward Elimination Method in Model

Model	Variables Entered	Variables Removed	Method
1	lnPOED, lnPOHFW, lnPOWSS, lnPOHUD, lnPOWPC, lnPOLE, lnPOSSW, lnPORD, lnPOFSW		Enter
2		lnPOHUD	Backward (criterion: probability of F-to-remove ≥ 0.100).
3		lnPOHFW	Backward (criterion: probability of F-to-remove ≥ 0.100).
4		lnPOED	Backward (criterion: probability of F-to-remove ≥ 0.100).
5		lnPOSSW	Backward (criterion: probability of F-to-remove ≥ 0.100).
6		lnPOWSS	Backward (criterion: probability of F-to-remove ≥ 0.100).
7		lnPOLE	Backward (criterion: probability of F-to-remove ≥ 0.100).

Source: The authors.

Table 9 summarises the important constituents of the model, that is, the value of R, R² and adjusted R². In model 1, expenditure incurred on the selected components of social sector individually is used as the predictor. The correlation between the components and inclusiveness is found to be 0.821, that is, 82.1 per cent, while the variability accounted in this model, represented by R², is 0.674. However, in model 2, when the variable ‘Expenditure incurred on housing and development’ is removed, then R and R² both remained the same while adjusted R² has increased to 0.555 per cent. The final model, that is, model 7 consists of three variables entered with the values of R, R² and adjusted R² that explain 79.9, 63.8 and 59.8 per cent, respectively, of the variation in inclusiveness. One of the assumptions of the regression model is that the residuals do not have autocorrelation. The Durbin–Watson (DW) test should be between 1.5 and 2.5. Here, the DW statistic is 2.081, as shown in Table 9, and therefore the variables are not autocorrelated.

Table 9

Summary of the Model

Model	R	R-square	Adjusted R-square	Std. Error of the Estimate	Durbin–Watson
1	0.821	0.674	0.534	0.984	2.081
2	0.821	0.674	0.555	0.962
3	0.820	0.672	0.573	0.942
4	0.819	0.671	0.588	0.925
5	0.815	0.664	0.597	0.915
6	0.811	0.658	0.606	0.905
7	0.799	0.638	0.598	0.914

Source: The authors.

Table 10

ANOVA Test of the Model

Model		Sum of Squares	Df	Mean Square	F	Sig.
1	Regression	42.058	9	4.673	4.825	0.001
	Residual	20.340	21	0.969
	Total	62.398	30
2	Regression	42.030	8	5.254	5.675	0.001
	Residual	20.368	22	0.926
	Total	62.398	30
3	Regression	41.953	7	5.993	6.742	0.000
	Residual	20.445	23	0.889
	Total	62.398	30
4	Regression	41.844	6	6.974	8.143	0.000
	Residual	20.554	24	0.856
	Total	62.398	30
5	Regression	41.423	5	8.285	9.874	0.000
	Residual	20.975	25	0.839
	Total	62.398	30
6	Regression	41.076	4	10.269	12.522	0.000
	Residual	21.322	26	0.820
	Total	62.398	30
7	Regression	39.795	3	13.265	15.845	0.000
	Residual	22.604	27	0.837
	Total	62.398	30

Source: The authors.

Table 10 explains the analysis of the variance (ANOVA). If the improvement in prediction to fit the regression is much greater than the inaccuracy within the model, then the value of F will be greater than 1. In the present case, the F ratio is 4.825 in the first model, which increases to 15.845 in the seventh model, which is very unlikely to happen by chance. The p-value is also highly significant.

The next outcome of the present model is the coefficients which represent the parameters of the model in Table 11. The β value indicates the contribution of the predictor to the model and has an associated standard error explaining whether or not the β value differs significantly from zero using the t-statistic. Therefore, if the t test associated with the β value is significant, then the predictor is significantly contributing to the model. Table 11 explains the results of stepwise regression using the backward elimination method. It shows the β values along with the p-values of the variables. It clearly indicates the three variables which contribute significantly. ‘Expenditure incurred on welfare of marginalised class’ (lnPOWPC) and ‘Expenditure on rural development’ (lnPORD) have shown negative beta values while the variable ‘Social security and welfare’ (lnPOSSW) contributes positively to the enhancement of inclusiveness. Values of these three significant variables have been presented in bold numerals in Table 11.

Table 11

Results of the Stepwise Regression-Backward Elimination Method of the Model

Variables	Steps
Variables	1	2	3	4	5	6	7
Constant	2.827	3.019	3.118	4.612	5.388	5.234	4.179
lnPOED	0.272(0.652)	0.236(0.668)	0.167(0.730)	–	–	–	–
lnPOHF	−0.089(0.783)	−0.090(0.775)	–	–	–	–	–
lnPOWSS	−0.064(0.731)	−0.071(0.689)	−0.095(0.532)	−0.833(0.413)	−0.082(0.526)	-	-
lnPOHUD	−0.067(0.867)	–	–	–	–	–	–
lnPOWPC*	−2.164* (0.001)	−2.142* (0.001)	−2.079* (0.000)	−4.319* (0.000)	−1.951* (0.000)	−1.954* (0.000)	−2.032* (0.000)
lnPOLE	0.116(0.473)	0.118(0.454)	0.119(0.439)	1.119(0.274)	0.154(0.237)	0.157(0.222)	–
lnPOSSW*	2.650* (0.003)	2.597* (0.001)	2.542* (0.001)	5.180* (0.000)	2.513* (0.000)	2.523* (0.000)	2.646* (0.000)
lnPORD*	−0.256 (0.181)	−0.256 (0.171)	−0.275 (0.171)	−1.899 (0.070)	−0.284 (0.073)	−0.322 (0.028)	−0.302* (0.038)
lnPOFSW	−0.115(0.521)	−0.104(0.523)	−0.110(0.523)	−0.701(0.490)	–	–	–

Source: The authors.

Note: (1) Figures in parentheses are p-values; *signify that coefficient is significant at the 5 per cent level of significance.

(2) Robustness of regression model has been thoroughly tested by applying various tests as elaborated in appendix I.

For the β values of these variables, it may be inferred that a 1 unit increase in ‘Expenditure on welfare of marginalised classes’ (lnPOWPC) and ‘Expenditure on rural development’ (lnPORD) will cause a decline of 2.032 and 0.30 units in inclusiveness, respectively. The results ironically exude a decline in inclusiveness due to an increase in expenditure on ‘welfare of marginalised section’ (lnPOWPC) and ‘rural development’ (lnPORD) despite augmentation of welfare expenditure. This is so because the benefits fail to percolate down to the grass-roots level due to various leakages and corrupt practices, as well as due to lack of awareness and publicity, which results in many targeted underprivileged beneficiaries remaining bereft of the benefits of the schemes (Fan et al., 2000; Thorat & Dubey, 2013). That is why despite increasing expenditure on the welfare of marginalised sections and on rural development, inclusiveness is at bay. Also, a 1 unit increase in ‘Expenditure on social security and welfare’ (lnPOSSW) will augment 2.646 units in inclusiveness, that is, inclusiveness will more than double due to an increase in ‘Expenditure incurred on social security and welfare’ (lnPOSSW).

The regression model uses the un-standardised coefficients depicted in Table 12. The model is:

Inclusive growth index = 4.179 - 4.726 (lnPOWPC) + 6.003 (lnPOSSW) - 2.528 (lnPORD)

This regression equation shows that an increase in spending on the ‘welfare of marginalised sections’ has not induced inclusiveness in India’s growth. Although the sign of coefficient is contrary to what was expected, it presents the true reality of welfare schemes launched by the government.

Table 12

Coefficients: The Final Model

Model	Un-standardised Coefficients		Standardised Coefficients
Model	Beta	Std. Error	Beta	t-Value	Sig.
Constant	4.179	2.305	–	1.813	0.081
lnPOWPC	−4.726	1.028	−2.032	−4.596	0.000
lnPOSSW	6.003	1.006	2.646	5.965	0.000
lnPORD	−2.528	1.161	−0.302	−2.177	0.038

Source: The authors.

Table 13

Collinearity Statistics and Diagnostics of Entered Variables in the Model

Variables Entered	Collinearity Statistics		Collinearity Diagnostics
Variables Entered	Tolerance Level	VIF	Eigenvalue	Condition Index
lnPOWPC	0.831	1.203	1.367	1.000
lnPOSSW	0.763	1.310	1.010	1.163
lnPORD	0.660	1.516	0.624	1.480

Source: The authors.

In this model, tolerance is 0.362 (1 − R²) and variance inflation factors (VIF) is 2.762 (1/ tolerance limit); hence, it implies there is no problem of multicollinearity as depicted in Table 13. Even the values of the condition index are also less than 15, indicating that there is no multicollinearity in the final model.

The results of this model are substantiated and corroborated by the fact that social exclusion among Dalits is still high, even though untouchability had been constitutionally removed in India (Sangeeth, 2016). In spite of numerous efforts, still there are 170–200 million SCs/STs at the bottom of the caste system, constituting 17 per cent of the population. Many SCs/STs had been living in extreme poverty without adequate opportunities for employment as well as education. Their vulnerability further accentuates due to discrimination and violence. Only a small minority of this class is getting benefits under the programmes and policies for the welfare of marginalised class (Panda, 2016). Divakar (2017) highlighted that the entire schemes for SCs and STs had been reduced from 294 to 256, and 307 to 261, respectively, in 2016. The allocation for SC amounted to ₹91,863 crore, that is, 16.6 per cent of the plan outlay, but only 4.62 per cent was actually disbursed. Similarly, only ₹47,276 crore, that is, 13.76 per cent, was apportioned but only 2.32 per cent was actually distributed during 2015–2016.

In our model, ‘social security and welfare expenditure’ has been found to be positively associated with inclusive growth. There is evidence that the scheme has been reaching its intended beneficiaries. Almost 76 per cent of beneficiaries were receiving their pension amount in the post office and bank accounts. In 2015, three new schemes were simultaneously launched at 115 locations in the country. These schemes were aimed at mainly for the unorganised sector and reached to 10 crore people in the country in a very short span of time (Shivangi, 2017).

In our study, the relationship between the expenditure incurred on ‘rural development’ and inclusive growth has found to be negative and significant. Fan et al. (2000) and Jha and Acharya (2011) also explained that the expenditure incurred on the growth for the rural economy in India reduced noticeably. In the early 1990s, 10.9 per cent was dispensed for rural development, which was reduced to 8 per cent in 2016. The neoliberal economic framework had turned conditions much worse as state governments had been compelled to reduce their rural development expenditure in order to reduce the budget deficit. Programmes exclusively for rural development had helped to reduce rural poverty, but its impact was negligible. This restricted the impact of rural development contribution in promoting inclusiveness.

Despite numerous efforts made by the government for the welfare of marginalised sections, they continue to suffer social and financial exclusion. Inequalities had also adversely affected the poverty eradication process, since equal access to employment and income-generating opportunities were denied to the underprivileged. Benefits of growth have not been evenly distributed. Inclusive development must encompass social inclusion as well as financial inclusion. Major dimensions of an egalitarian welfare society, such as eradication of poverty, productive employment generation, better healthcare, access to education, social protection and social welfare, have to be evenly distributed among all segments of the population (Government of India, 2012; Mukhopadhaya & Saha, 2005; Shukla & Mishra, 2013; Thorat & Dubey, 2013).

IV. Conclusion and Policy Implications

The concept of inclusive growth constituted the basic foundation of India’s economic growth policy. India has been envisaging the strategies which focused on fostering a socialistic pattern of society oriented with self-reliance, social justice and reduction of poverty. Therefore, Indian planners contemplated to jack up social sector allocations to induce inclusive growth. This paper has attempted to assess the impact of components of social sector development on inclusive growth for the period from 1985–1986 to 2015–2016. The analysis reveals a sluggish progress towards rendering the growth inclusive. Further, on measuring the impact of social sector development on growth inclusiveness, three components—‘expenditure incurred on welfare of underprivileged classes’, ‘expenditure incurred on social security and welfare’ and ‘expenditure incurred on rural development’—are found to contribute towards inclusive growth. Two out of these, viz. ‘expenditure incurred on welfare of underprivileged classes’ and ‘expenditure on rural development’ were expected to ameliorate the marginalised sections and contribute to rural development for inclusive growth. However, our results show a negative relation of these two variables with inclusiveness. The probable reasons are leakages in the process of disbursements, corruption, lack of proper management and awareness, lack of monitoring of developmental programmes and shortcomings in implementation of schemes such as favouritism in selection of beneficiaries, allocation of schemes without looking into basic needs and lack of follow-up. Earmarking of funds needs to be done effectively so as to plug the gaps in allocations and actual utilisation and circumvent squandering of scarce resources. Dev (2016) suggested that the expenditures incurred on the social sector needed to be increased. The delivery systems must be improved for social sector components. The role of states for judicious allocation of funds is considered equally important, particularly in infrastructure, health and education and agriculture, to make the growth more inclusive. Although literacy, education, health, maternal and infant mortality rates have shown improvement, the pace has been sluggish.

The development of rural India is considered imperative for the growth to be equitable and inclusive. The Government of India runs various programmes to mitigate poverty and to render growth inclusive. Social sector development programmes, such as Pradhan Mantri Gram Sadak Yojana, Indira Gandhi Awas Yojana, National Rural Drinking Water Programme, Accelerated Irrigation Benefit Programme, National Rural Employment Guarantee Programme, NRHM, Integrated Child Development Programme (ICDS), Sarva Shiksha Abhiyaan, Mid-day Meal, Total Sanitation Campaign and so on, ensured better employment opportunities with an improvement in basic amenities in both rural and urban India. However, despite such efforts, India needs to buckle up so that desired results of inclusiveness and equity are achieved. There is a lack of public awareness about programmes. Raising the level of community participation, programme design modifications, community base with strong community participation, social mobilisation through advocacy, information and communication need to be reinforced (Dhaked & Gupta, 2016).

Committees and subcommittees should be formed to evaluate the implementation of schemes for achieving the desired results. The whole ground of policy framework, which has been astutely designed by policy makers for an inclusive growth, must be clearly monitored for effective implementation. Strict monitoring and overseeing of implementation and devolution of funds to the targeted beneficiaries must be vigilantly effected. While the various measures enumerated above will help reduce the likelihood of corruption, they need to be supplemented by creating stronger mechanisms for investigating, inspecting and punishing those found to be involved in corrupt practices.

Appendix 1. Check of Robustness of Regression Model

To substantiate the results, some additional tests have been carried out. The results reveal that the selected model has no serial correlation, no heteroscedasticity and is normally distributed and well specified. These additional tests were as follows:

1. Breusch–Godfrey Serial Correlation LM Test, LM stands for Lagrange Multiplier which is Test Statistic nR2, used to Check Serial Correlation in the Model

One diagnostic test is the Breusch–Godfrey serial correlation Lagrange Multiplier (LM) test, which flogs or rejects the presence of serial correlation in the residual. The null hypothesis of the model is that there is no serial correlation up to the specified order of lag. Thus, if the model fails to reject the null hypothesis up to the lag value, then there is no autocorrelation (Kolawole, 2016).

In Table 14, it failed to reject the null hypothesis because the p-value is more than 0.05 up to the specified lag of 1. Hence, there is no autocorrelation in the present model.

Table 14

Breusch–Godfrey Serial Correlation LM Test of Model 1

F-statistic	0.998469	Prob. F(1, 20)	0.3926
Obs*R-squared	1.474038	Prob. Chi-square (1)	0.2247
Test equation:
Dependent variable: RESID
Method: Least squares
Included observation: 30
Presample missing value lagged residuals set to zero.
Variable	Coefficient	Std. Error	t-stats (Prob.)
RESID-1	−0.304355	0.304588	−0.99923 (0.3296)
R-squared	0.047550	Mean dependent var	−1.15E−16
Adjusted R-squared	−0.428676	S.D. dependent var	0.829734
S.E. of regression	0.991758	Alaike info criterion	3.092747
Sum squared resid	19.67168	Schwarz criterion	3.601582
Log likelihood	−36.93759	Hannan–Quinn criterion	3.258615
F-statistic	0.099847	Durbin–Watson statistic	2.043279
Prob(F-statistic)	0.999661

Source: Authors’ calculation using EViews 10.

2. Breusch–Pagan–Godfrey Test

To check the heteroscedasticity in the model, the Breusch–Pagan–Godfrey test is employed. The null hypothesis of this test is that homoscedasticity is present in the series. So, if this test fails to reject the null hypothesis, then there is absence of heteroscedasticity in the residuals of the model (Kolawole, 2016).

Table 15 highlights the result of the heteroscedasticity test by Breusch–Pagan–Godfrey. The null hypothesis of the model is that the variables are homoscedastic in nature. The result of heteroscedasticity test confirmed that there is absence of heteroscedasticity in the residuals of the model. Homoscedasticity is present in this model as p-values are greater than 0.05.

Table 15

Heteroscedasticity Test: Breusch–Pagan–Godfrey of Model 1

F-statistic	1.424987	Prob. F(9, 21)	0.2451
Obs*R-squared	12.09338	Prob. Chi-square(9)	0.2081
Scaled explained SS	27.95779	Prob. Chi-square(9)	0.0010
Test equation:
Dependent variable: RESID²
Method: Least squares
Included observation: 30
R-squared	0.390109	Mean dependent var	0.666250
Adjusted R-squared	0.128727	S.D. dependent var	2.149773
S.E. of regression	2.006641	Alaike info criterion	4.486498
Sum squared resid	84.55879	Schwarz criterion	4.949075
Log likelihood	−59.54072	Hannan–Quinn criterion	4.637287
F-statistic	1.492487	Durbin–Watson statistic	2.502738
Prob(F-statistic)	0.214529

Source: Authors’ calculation using EViews 10.

3. Ramsey RESET Test

The Ramsey RESET test is carried out to test specification errors and to check the linearity of the model. This test also suggested that the model is well specified. The null hypothesis of this test is that the model is correctly specified. The RESET F-statistic has a p-value of 0.3225, explaining that there is no need to reject the null hypothesis. Hence, the functional form of the model is correctly specified (EViews 10, 2017).

There is an absence of specification error as the test statistics were not significantly different from zero at the 5 per cent level as shown in Table 16.

Table 16

Ramsey RESET Test of Model 1

Specification: lnIGI C lnPOED-1 lnPOHF-1 lnPOWSS-1 lnPOHUD-1 lnPOWPC-1 lnPOLE-1 lnPOSSW-1 lnPORD-1 lnPOFSW-1

Omitted variables: Squares of fitted values

	Value	Df	Probability
t-statistic	0.839758	20	0.3225
F-statistic	0.763802	(1, 20)	0.3225
Likelihood ratio	1.161846	1	0.2811

Source: Authors’ calculation using EViews 10.

4. Jarque–Bera Test

The null hypothesis of residuals as per the Jarque-Bera (JB) test is that the residuals are non-normally distributed.

Figure 2 shows that values of the JB test and p-value at 0.5120 and 0.7741, respectively, which rejects the null hypothesis as the p-value is more than the assumed level of significance. Hence, the residuals are normally distributed. If the residuals are normally distributed, then the JB normality test should not be significant and the histogram is bell-shaped (EViews 10, 2017).

Figure 2.

Source: Author’s own calculation.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Ishu Chadda

References

Anand

Mishra

Peiris

S. J.

(2013). Inclusive growth revisited: Measurement and determinants. Economic Premise, No. 122.

Anthony

G. M.

Rao

V. K.

(2007). A composite index to explain variations in poverty, health, nutritional status and standard of living: Use of multivariate statistical methods. Public Health , 121, 578–587.

Balkrishnan

(2013). A brief history of poverty counting in India. The Hindu , July 25.

Datt

(1999). Has poverty declined since economic reforms? Economic and Political Weekly , 34(50), 3516–3518.

Datt

Ravillion

(2010). Shining for the poor too? Economic and Political Weekly , 45, 55–60.

Dev

S. M.

(2016). Increase social sector expenditure. The Hindu , February 3.

Dev

S. M.

Mooij

(2004). Patterns in social sector expenditures: Pre and post reform period . Centre for Economic and Social Studies.

Dhaked

Gupta

C. M.

(2016). A critical review of malnutrition and its management. International Journal of Medical and Health Research , 2(4), 43–47.

Divakar

P. N.

(2017). The 2017 budget is taking SC/ST welfare backwards. https://thewire.in/rights/budget-2017-sc-st-welfare

10.

EViews 10. (2017). User’s guide II (2nd ed.). IHS Markit.

11.

Fan

Hazell

P. B. R.

Thorat

S. K.

(2000). Government spending, growth and poverty in rural India. American Journal of Agricultural Economics , 82(4), 1038–1051.

12.

Fukuda-Parr

(2003). The human development paradigm: Operationalizing Sen’s ideas on capabilities. Feminist Economics , 9(2–3), 301–317. https://doi.org/10.1080/1354570022000077980.

13.

Government of India. (2008). Eleventh five year plan (2007–12), Vol II. Social sector . Oxford University Press.

14.

Government of India. (2009). Report of the expert group to review the methodology for estimation of poverty . Planning Commission.

15.

Government of India. (2011). Indian human development report: Towards social inclusion . Planning Commission.

16.

Government of India (2012). Scenarios: Shaping India’s future . New Delhi: Planning Commission.

17.

Government of India. (2013). Twelfth five-year plan (2012–17), Vol. III. Social Sector . SAGE Publications.

18.

Government of India. (2015). Key indicators of the labour market . Ministry of Labour and Employment.

19.

Government of India (various issues). Annual reports . Planning Commission.

20.

Government of India (various issues). Economic survey . Ministry of Finance.

21.

Government of India (various reports). Educational statistics at a glance . Ministry of Human Resource Development.

22.

Government of India (various reports). Electoral statistics pocketbook . Elections Commission of India.

23.

Government of India (various reports). Handbook on social welfare statistics . Ministry of Social Justice and Empowerment.

24.

Government of India (various rounds). NSS reports on employment and unemployment situation in India . Ministry of Statistics and Programme Implementation.

25.

Government of India (various rounds). NSS reports on employment and unemployment situation among social groups in India . Ministry of Statistics and Programme Implementation.

26.

Government of India (various issues). Outlays Budget (Vol. 1). Ministry of Finance.

27.

Gujarati

D. N.

(2003). Basic econometrics . McGraw-Hill.

28.

Jha

Acharya

(2011). Expenditure on the rural economy in India’s budgets since the 1950s. Review of Agrarian Studies , 1, 2.

29.

Kolawole

B. O.

(2016). Government spending and inclusive-growth relationship in Nigeria: An empirical investigation. Zagreb International Review of Economics and Business , 19(2), 33–56. https://doi.org/10.1515/zireb-2016-0007

30.

Krishnan

(2010). Constructing an area-based socioeconomic index: A principal components analysis approach. Early Child Development Mapping Project, Community.

31.

Leech

N. L.

Barrett

K. C.

Morgan

G. A.

(2014). SPSS for intermediate statistics: Use and interpretation . Lawrence Erlbaum Associates Publishers.

32.

Mukhopadhaya

Saha

(2005). Poverty alleviation programmes in India . The Independent Commission on Development and Health in India, VHAI, New Delhi.

33.

Panda

R. K.

(2016). Socially exclusion and inequality: Opportunities in agenda 2030. European Union, World Vision and Welthungerhilfe , www.sustainabledevelopment.un.org/content/documents/11145Social%20exclusion%20and%20Inequality-Study%20by%20GCAP%20India%20.pdf

34.

Pradhan

J. P.

Puttuswamaiah

(2005). Trends and pattern of technology acquisition in Indian organized manufacturing: An inter-industry exploration . Gujarat Institute of Development Research. https;//ideas.repec.org/p/ess/wpaper/id452.html

35.

Radhakrishna

Panda

(2006). Macroeconomics of poverty reduction: India case study . Indira Gandhi Institute of Development Research.

36.

Ramos

R. A.

Ranieri

Lammens

J. W.

(2013). Mapping inclusive growth . UNDP, International Policy Centre for Inclusive Growth, Working Paper 105.

37.

Sangeeth

(2016). Inclusive development of marginalized population through social policy initiatives – Reflections to future development. Paripex—Indian Journal of Research , 5(3), 278–282.

38.

Sekhar

C. C.

Indrayan

Gupta

S. M.

(1991). Development of an index of need for health resources for Indian states using factor analysis. International Journal of Epidemiology , 20(1), 246–250.

39.

Sharma

A. D.

(2014). Understanding the social sector, economic growth, social development and economic development: Interrelationship and linkages. Economic Affairs , 59(4), 585–590.

40.

Shivangi

(2017). Impact of social security on economic growth in India . Symbiosis Law School. http://crsgpp.nujs.edu/wp-content/uploads/2017/02/Article-10-Shilpa.pdf

41.

Shukla

S. S.

Mishra

(2013). Employment generation and poverty alleviation in developing countries: Challenges and opportunities special reference to India. IOSR Journal of Business and Management , 11(4), 18–23.

42.

Smith

(2018). Step away from stepwise. Journal of Big Data , 5(32), 1–12.

43.

Thorat

Dubey

(2013). How inclusive has growth been in India during 1993/94–2009/10? Implications for XII plan strategy . United Nations Development Programme.

44.

United Nation Development Programme (various reports). Human development report . Oxford University Press.

45.

Whittingham

Stephens

Bradbury

Freckleton

(2006). Why do we still use stepwise modelling in ecology and behaviour? The Journal of Animal Ecology , 75, 1182–1189.

46.

World Bank. (2016). Labour force participation rate. World development indicators . https://data.worldbank.org/indicator/SL.TLF.CACT.ZS?loc=IN.