Abstract
This paper examines the impact of various socio-economic factors on consumption in rural West Bengal, an eastern state of India, using a regression-based technique reformulated in a spatial framework. The difference of incidences of poverty (head count ratios) in two parts (North and South) of rural West Bengal is then decomposed using the familiar Oaxaca decomposition methodology into an aggregate characteristics effect, which is interpreted as a resource effect and an aggregate coefficients effect, which is interpreted as an efficiency effect. An important observation from the present analysis is that the poorer North Bengal has a scarcity in terms of availability of characteristics (resources) compared to that in South Bengal and the resource scarcity in North Bengal is the dominant factor causing the poverty gap between the two parts of West Bengal. Thus, attention needs to be paid to North Bengal with respect to enhancement of important policy variables like education level, Government aid, and employment opportunities.
Similar content being viewed by others
Notes
For comparability of results across districts, the issue of spatial variation in prices, i.e., the cost of living has been ignored here.
See Coudouel et al. (2002) for a discussion about the World Bank method using linear regression in analyzing the determinants of poverty. This method has been used also in Bhaumik et al. (2006) for studying difference in poverty incidences between Serbians and Albanians in Kosovo using Living Standard Measurement Survey. This methodology has also been used by Gang et al. (2008) to analyze the determinants of rural poverty in India, contrasting the situation of scheduled caste (SC) and scheduled tribe (ST) households with the general population.
The present paper is, in fact, a spatial reformulation of the paper by Chattopadhyay (2011a).
Following (Foster et al. 1984), the FGT \((\alpha )\) measure is given by \(F_\alpha =\frac{1}{N}\sum \nolimits _{y_i \le z}{\left( {\frac{z-y_i }{z}} \right) ^{\alpha }}\); \(y_i \) and \(N \) being the income of the ith person and the number of persons in the society, respectively, and z being the poverty line. FGT0 signifies the HCR, i.e., proportion of people below the poverty line.
Two other sources of spatial autocorrelation are the spatial heterogeneity in parameters (Anselin 1992) and functional form heterogeneity (Darmofal 2006). While the former takes into account the particular features of each location (spatial unit) by explicitly considering varying parameters, random coefficients, or various forms of structural change in place of ordinary regression analysis; the latter deals with the Footnote 6 Continued issue of spatial heterogeneity in functional form, where different functional forms are valid in different spatially indexed subsets of the data. A third form of spatial heterogeneity is spatial heterogeneity in error variance, leading to spatial heteroskedasticity (Anselin and Griffith 1988). These, however, require time series/panel data and have been avoided in this paper.
Application of spatial analysis in the context of poverty may be found in (Ayadi and Amara 2008). In their analysis of poverty in Tunisia, they find that spatial models with spatially correlated and unobserved spatial heterogeneity are preferred to the traditional non-spatial regression model, and give a better approximation of the Tunisian poverty map.
A completely non-spatial analysis with the same data set, however, showed that the poorer North Bengal has a deficiency both in terms of availability and utilization of resources compared to that in South Bengal (Chattopadhyay 2011a).
The definition of neighborhood in the present context is given in Sect. 3 of this paper.
The nature of spatial dependence could possibly have been captured by additionally introducing neighborhood dummies to accommodate aggregate characteristics of the neighborhood of residence of a household. However, since in our context the neighborhoods are overlapping (please see definition in Sect. 3 and Tables 1, 2), this has not been attempted.
Anselin (2003) refers to (3) and (4) as a model with “spatial externalities in both modeled and unmodeled effects” (p. 161) because its reduced form applies a spatial multiplier to both the independent variable and the errors. He also points out that it is constrained by postulating a single multiplier matrix for both (Small and Steimetz 2009). Given, \(W\iota =\iota , \left( {I-\rho W}\right) ^{-1}\iota =\left( {1-\rho }\right) ^{-1}\iota \) (see Kim et al. 2003), i.e., the sum of each row of the inverse of the matrix \(\left( {I-\rho W}\right) \) sum to \(\frac{1}{\left( {1-\rho }\right) }\) which is the spatial multiplier.
See Appendix 1 for the derivation.
\(({\text{ MX }})_i\) denotes the ith row of the matrix \(({\text{ MX }})=X^{*}\).
Standard terminology cannot be used due to the presence of the heteroscedasticity and spatial weight adjusted transformed variables \(X^{**}\). The paragraph following Eq. (10) provides the interpretation.
- $$\begin{aligned}&R_{M_1 } =\overline{\varPhi \left( {X_\mathrm{A}^{**}\hat{{\beta }}_\mathrm{A}^{*}}\right) }-\varPhi \left( {\overline{X_\mathrm{A}^{**} }\hat{{\beta }}_\mathrm{A}^{*}}\right) ; \quad R_{M_2 } =\overline{\varPhi \left( {X_\mathrm{B}^{**}\hat{{\beta }}_\mathrm{A}^{*}}\right) } -\varPhi \left( {\overline{X_\mathrm{B}^{**} } \hat{{\beta }}_\mathrm{A}^{*}}\right) \quad \\&R_{T_1 } =\left\{ {\varPhi \left( {\overline{X_\mathrm{A}^{**} } \hat{{\beta }}_\mathrm{A} ^{*}} \right) } \right. -\left. {\varPhi \left( {\overline{X_\mathrm{B}^{**} } \hat{{\beta }}_\mathrm{B}^{*}} \right) } \right\} -\left\{ { \left( {\overline{X_\mathrm{A}^{**}} -\overline{X_\mathrm{B}^{**} } } \right) \beta _\mathrm{A}^{*}}\right\} \phi \left( {\overline{X_\mathrm{A}^{**}}\hat{{\beta }}_\mathrm{A}^{*}}\right) . \end{aligned}$$
See Appendix 2 for the derivation.
Consider the linear regression model that is estimated separately for groups g = (A, B): \(Y_{ig} =X_{ig} \beta _g +\varepsilon _{ig} ;i=1,2,\ldots ,N_g.\)
Following (Oaxaca 1973), the difference in mean outcome level between A and B can be decomposed as: \(\overline{Y_\mathrm{A}} -\overline{Y_\mathrm{B}} =\left( {\overline{X_\mathrm{A}} -\overline{X_\mathrm{B}}}\right) \hat{{\beta }}_\mathrm{A} +\overline{X_\mathrm{B}} \left( {\hat{{\beta }}_\mathrm{A} -\hat{{\beta }}_\mathrm{B}}\right) \), where the first term in bracket is the aggregate characteristics effect (C) and the second term in bracket is the aggregate coefficients effect (D).
The application of Oaxaca decomposition technique has been made in Bhaumik and Chakrabarty (2009) to examine the difference in average (log) earnings between two religious groups in Indian and in Bargain et al. (2009) to examine the earnings difference between Indian and Chinese wage earners.
Strictly speaking, \(C_\mathrm{con}^1 \) is however not the conventional characteristics effect as it involves terms containing \(\sum _\mathrm{A}^{-1}\) and \(\sum _\mathrm{B}^{-1} \), where \(\sum {^{\prime }}\)s are the spatial MLE estimates.
As mentioned earlier, in a strict sense it is not the conventional coefficients effect as it involves the term \(\sum _\mathrm{B}^{-1}{X_\mathrm{B}; \sum _\mathrm{B}}\) being the spatial MLE estimate.
It may be mentioned here that the decompositions in (9), (10), (12) and (13) have been used only to illustrate that the decomposition of \(\delta _\mathrm{pov}\) into C and D as in (7) is actually a spatial generalization of the corresponding Oaxaca decomposition. The individual components have not been estimated in the above form. However, a simpler version (in share form) has been estimated in the following part.
The idea of opposite signs of C and \(C^{1}\) is not quite appealing intuitively as this means a situation where the effect of the term \(C_R\) is more prominent (in absolute terms) compared to C. This is an undesirable situation given the objective of the decomposition analysis and thus not considered. In the present context, however, C and \(C^{1}\) have the same signs.
See Appendix 3 for finding the statistical precisions of C, D, \(C_k{^{\prime }}\)s and \(D_k{^{\prime }}\)s.
Alternative weight functions can be defined subject to the availability of data. Since the household identification code does not provide information at sub-district level, this definition of weight has been adopted. It may be pointed out that recently Beck et al (2006) constructed a spatial weight matrix based on non-geographic notion of space. They argued in favor of considering political economy notions of distance, such as relative trade or common dyad membership in situations of spatial analysis involving trade and democracy.
The sampled households of rural West Bengal are obtained using a two-stage stratified sampling design. The first stage strata are the districts and the three second stage strata are as follows:
-
SSS 1:
relatively affluent households
-
SSS 2:
households not belonging to SSS1and having principal earning from non-agricultural activity
-
SSS 3:
other households spatial weight being equal to zero, if otherwise.
-
SSS 1:
The regression analysis has been done using the command “SPATREG” of the STATA software. It may be mentioned that since an equation for a household involves all households in its neighborhood, the household-specific sampling weights (multipliers) could not be incorporated.
Educational levels considered are: not literate, literate without formal schooling, literate but below primary, primary, middle, secondary, higher secondary, diploma/certificate course, graduate, post graduate, and above. The average educational level of each household is obtained as the average over codes assigned to different educational levels (in increasing order), starting from zero for the illiterate to the maximum for the category: post graduate and above.
The parameters of the education variables, PSECEDU and PTERTEDU are supposed to reflect something like the marginal returns to skills in terms of higher (per capita) household income or consumption expenditures. Defining these variables as shares of the household members in specific skill groups in all household members would imply that, conditional on the education level (secondary or tertiary), each household member contributes to household income to the same extent, irrespective of whether he/she is active in the labor market (earns money) or not. This may not be true. The education variables will consequently be misspecified, and the inference drawn from their parameters will be unreliable. We thank an anonymous referee for pointing this out.
It may be mentioned that in the literature the direction of causality between poverty and education linkages is shown to flow in both ways. On the one hand, poverty acts as a factor that makes education inaccessible. On the other hand, those with education are considered to be at lower risk of poverty (Oxaal 1997). However, according to Appleton (1997), each year of primary schooling is associated with a 2.5 % fall in the risk of poverty and lower secondary schooling has roughly twice this effect.
These domestic duties are different from those listed under PDOMO. Other duties listed under PDOMO include free collection of goods (vegetables, roots, firewood, cattle feed, etc.), sewing, tailoring, weaving, etc., for household use.
This is because by increasing \(\hat{{\beta }}_k\) in South Bengal, poverty will decrease in South Bengal and the poverty gap will be widened, because \(H_\mathrm{A}-H_\mathrm{B}\) is positive.
North Bengal is, in fact, having higher coefficients attached to (1-DEPRAT), D_FEMH, PLAND, and D_GOVAID.
Column 3 of Tables 5, 6, showing the mean levels of explanatory variables (resources) actually give the severity of deficiency in a non-spatial framework (Chattopadhyay 2011a). Comparing these values with those in Table 12, we see that the severity of deficiency in terms of availability of resources for North Bengal compared to that in South Bengal is much pronounced in the spatial analysis.
Concerning PLAND, it is to be mentioned here that as a result of the “land reform policy” implemented under the aegis of the Left Front government in West Bengal, average rural people gained access to agricultural land holdings, which contributed to lower economic inequalities and a higher level of consumption among them. The Development and Planning Department, Government of West Bengal (2004) reports that in the area of land relations the recent increase in landlessness needs to be analyzed in terms of its causes and the viability of peasant cultivation needs to be strengthened. Also the recognition and granting of land rights to women needs to be further extended. In education, certain spatial and social pockets of illiteracy need to be addressed. In occupational terms agricultural labor households (and specially females in these households) require special policy attentions. Scheduled tribe households also require a focused drive to increase literacy. It is also stressed that employment generation has to be a critical focus of future policy.
The variance of any parameter which is a function of random variables can be approximated using the delta method (see Powell 2007; Seber 1982). Suppose \(G\) is the parameter, which is a function of the random variables: \(\left( {X_1,X_2,\ldots ,X_N } \right) ;\) i.e., \(G=f\left( {X_1,X_2,\ldots ,X_N } \right) \) the variance of \(G\) is given by \(var\left( G \right) =\sum \nolimits _{i=1}^N {var\left( {X_i } \right) } \left[ {\frac{\partial f}{\partial X_i }} \right] ^{2}+2\sum {\sum {{\text{ cov }}\left( {X_i,X{ }_j} \right) \left[ {\left( {\frac{\partial f}{\partial X_i }} \right) \left( {\frac{\partial f}{\partial X_j }} \right) } \right] } } \), where \(\frac{\partial f}{\partial X_i}\) is the partial derivative of \(G\) with respect to \(X_i \).
The explicit forms of \(\sigma _\mathrm{C}^2, \sigma _\mathrm{D}^2, \sigma _{C_k }^2\) and \(\sigma _{D_k }^2\) have not been given in this paper. Detailed derivations would be given to the interested readers on request.
References
Anselin L (1992) Space stat tutorial: a workbook for using space stat in the analysis of spatial data. University of Illinois, Urbana
Anselin L (1999) Spatial econometrics. https://csiss.ncgia.ucsb.edu/aboutus/presentations/files/baltchap.pdf. Accessed 09 Mar 2013
Anselin L (1988) Spatial econometrics: methods and models. Kluwer, Dordrecht
Anselin L (2003) Spatial externalities, spatial multipliers and spatial econometrics. Int Reg Sci Rev 26(2):152–166
Anselin L, Bera AK (1998) Spatial dependence in linear regression models with an introduction to spatial econometrics. In: Ullah A, Giles DE (eds) Handbook of applied economic statistics. Marcel Dekker, New York, pp 237–289
Anselin L, Griffith D (1988) Do spatial effects really matter in regression analysis? Pap Reg Sci Assoc 65:11–34
Appleton S (1997) Leaping into the ark: some reflections on free primary education in Uganda. University of Oxford, Oxford
Ayadi M, Amara M (2008) Spatial patterns and geographic determinants of welfare and poverty in Tunisia. In: Equity and economic development, ERF 15th annual conference, 23–25 November 2008
Bargain O, Bhaumik SK, Chakrabarty M, Zhao Z (2009) Earnings differences between chinese and indian wage earners, 1987–2004. Rev Income Wealth 55(s1):562–587
Barma S (2007) Socio-political movements in North Bengal (a sub-himalayan tract). Global Vision Publishing House, New Delhi
Beck N, Gleditsch KS, Beardsley, K (2006) Space is more than geography: using spatial econometrics in the study of political economy. Int Stud Q 50(1):27–44
Becker SO, Ichino A (2002) Estimation of average treatment effects based on propensity scores. Stata J 2(4):358–377
Bhaumik SK, Chakrabarty M (2009) Is education the panacea for economic deprivation of Muslims?: evidence from wage earners in India, 1987–2005. J Asian Econ 20(2):137–149
Bhaumik SK, Gang IN, Yun MS (2006) Note on decomposing differences in poverty incidence using regression estimates: algorithm and example. IZA Discussion Paper no. 2262. IZA, Bonn
Bigman D, Srinivasan PV (2002) Geographical targeting of poverty alleviation programs: methodology and applications in rural India. J Policy Model 24(3):237–255
Bokosi FK (2007) Household poverty dynamics in Malawi: a bivariate probit analysis. J Appl Sci 7(2): 258–262
Burger R, Van Der Berg S, Van Der Walt S, Yu D (2004) Geography as destiny: considering the spatial dimensions of poverty and deprivation in south africa. Cornell University Conference on African Development and Poverty Reduction, 13 to 15 October
Caliendo M, Kopeinig S (2008) Some practical guidance for the implementation of propensity score matching. J Econ Surv 22(1):31–72
Chattopadhyay S (2012) Analysis of poverty and efficiency: an earnings frontier approach. Bulletin of Economic Research. doi:10.1111/j.1467-8586.2011.00426.x
Chattopadhyay S (2011b) Earnings efficiency and poverty dominance analysis: a spatial approach. Econ Bull 31(3):2298–2318
Chattopadhyay S (2011a) Inter-regional poverty comparisons: case of West Bengal. J Quant Econ 9(2): 104–122
Coudouel A, Hentschel JS, Wodon QT (2002) Using linear regressions for analyzing the determinants of poverty. In: Klugman J (ed) A sourcebook for poverty reduction strategies (2-Volume set) (Vol. 2, p. Annex A.8). World Bank, New York
Cushing B, Zheng B (1999) Locational difference in the underlying characteristics of poverty. Paper presented at the Southern Regional Science Association conference
Darmofal D (2006) Spatial econometrics and political science. Presented at the Annual Meeting of the Southern Political Association, Atlanta, GA
Development and Planning Department, Government of West Bengal (2004) West Bengal Human Development Report
Foster J, Greer J, Thorbecke E (1984) A class of decomposable poverty measures. Econometrica 52(3): 761–766
Gang IN, Sen K, Yun MS (2008) Caste, ethnicity and poverty in rural India. Rev Income Wealth 54(1):50–70
Ganguly R (2005) Poverty, malgovernance and ethnopolitical mobilization: Gorkha nationalism and the Gorkhaland agitation in India. Nationalism Ethn Politics 11(4):467–502
Geary RC (1954) The contiguity ratio and statistical mapping. Incorporated Statistician 5:115–145
Geda A, Jong N, Mwabu G, Kimenyi M (2001) Determinants of poverty in Kenya: household-level analysis. KIPPRA Discussion Paper no. 9. Kenya Institute For Public Policy Research and Analysis, Nairobi
Gibson J, Olivia S (2007) Spatial autocorrelation and non-farm rural enterprises in Indonesia. In: 51st Conference, Australian Agricultural and Resource Economics Society, February 13–16
Kim CW, Phipps TT, Anselin L (2003) Measuring the benefits air quality improvement: a spatial hedonic approach. J Environ Econ Manag 45(1):24–39
Lesage JP, Charles JS (2008) Using home buyers’ revealed preferences to define the urban–rural fringe. J Geograph Syst 10:1–21
Manski C (1993) Identification of endogenous social effects: the reflection problem. Rev Econ Stud 60(3):531–542
Moran P (1948) The interpretation of statistical maps. J R Stat Soc B 10:243–251
Oaxaca R (1973) Male–female wage differentials in urban labor markets. Int Econ Rev 14:693–709
Oxaal Z (1997) Education and poverty:a gender analysis. Gender Equality Unit, Swedish International Development Cooperation Agency
Patton M, Mcerlean S (2005) Spatial effects within the agricultural land market in Northern Ireland. J Agric Econ 54(1):35–54
Petrucci A, Salvati N, Seghieri C (2004) Autologistic regression model for poverty mapping and analysis. Metodološki zvezki 1(1):225–234
Pinkse J., Slade M.E. (2009) The future of spatial econometrics. In: Prepared for the 50th anniversary symposium of the Journal of Regional Science held at the New York Federal Reserve
Powell LA (2007) Approximating variance of demographic parameters using the delta method: a reference for avian biologists. Condor 109(4):949–954
Seber GA (1982) The estimation of animal abundance and related parameters. Griffin, London
Small KA, Steimetz SS (2009) Spatial hedonics and the willingness to pay for residential amenities. Economics Working Paper no. 05-06-31. University of California, Irvine
Tobler W (1970) A computer movie simulating urban growth in the Detroit region. Econ Geogr 46(2): 234–240
Voss PR, Long DD, Hammer RB, Friedman S (2006) County child poverty rates in the US: a spatial regression approach. Popul Res Policy Rev 25:369–391
Wan G, Zhang Y (2008) Explaining the poverty difference between inland and coastal China: a regression-based decomposition approach. Rev Dev Econ 12(2):455–467
World Bank (2005) Development in practice: priorities and strategies for education. World Bank, Washington, DC
Yun MS (2004) Decomposing differences in the first moment. Econ Lett 82(2):275–280
Yun MS (2005) Hypothesis tests when decomposing differences in the first moment. J Econ Soc Meas 30:295–304
Acknowledgments
The authors are grateful to the anonymous referees for their meticulous checking and very insightful comments which have greatly enriched the paper. The authors owe a debt of gratitude to Prof. Arup Bose of Indian Statistical Institute, Kolkata for his invaluable suggestions towards remodeling this paper. Many thanks are due to Prof. Babulal Seal of the Burdwan University for some helpful discussions about the Delta method.
Author information
Authors and Affiliations
Corresponding author
Additional information
S. Chattopadhyay—Formerly affiliated to Indian Statistical Institute, Kolkata.
Appendices
Appendix 1
1.1 Finding the incidence of poverty
From (4)
where \(M=\left( {I-\rho W} \right) ^{-1}\) is the Leontief inverse, \(X^{*}= \text{ MX }, \varepsilon ^{*}= M\varepsilon \).
Let \(p_{_i}\) be the probability of the ith household being poor.
Then,
The incidence of poverty for any region (R) will be asymptotically equal to the sample average of \(p_{_i} {^{\prime }}\)s.
Thus, \(H_R =\frac{1}{n^{R}}\sum _{i=1}^{n^{R}} {\varPhi \left( {X_{i_{R}}^{**} \hat{{\beta }}_{R }^*} \right) } \); R = A, B, where \(n^{R}\) is the number of households in region R.
Appendix 2
Proof
Substituting \(X^{**}=\sum {^{-1}}\text{ MX }\) from (21) and (22) with \(\sum {{=}diag\left( {\sigma _1,\sigma _2 ,\ldots ,\sigma _n} \right) }\), where \(\sigma _{i }^2 ={\text{ var }}\left( {\varepsilon _i^*} \right) \), \(C^{1}\) can be written as
[Higher ordered terms involving \(\rho W\) are ignored in empirical exercises (Lesage and Charles 2008)]
\(\square \)
Proof
Writing \( X^{**}=\sum {^{-1}} \text{ MX },\, D^{1}\) can be written as
[Higher ordered terms involving \(\rho W\) are ignored in empirical exercises (Lesage and Charles 2008)]
\(\square \)
Appendix 3
1.1 Statistical precision of \(C, D, C_k\), and \(D_k \)
For finding the precision of aggregate (C and D) as well as individual \((C_k\,\text{ and }\,D_k)\) effects, asymptotic variances of C, D, \(C_k\), and \(D_k\) are found following Yun (2005) and Chattopadhyay (2012) as.
1.2 Asymptotic variance of \(C, D, C_k\), and \(D_k\)
The aggregate characteristics effect, C is a function of the estimated coefficients of region A and the autoregressive coefficient of region B i.e.,
\(C=f\left( {\hat{{\rho }}_\mathrm{A},\hat{{\beta }}_\mathrm{A}^*,\hat{{\rho }}_\mathrm{B}} \right) ;\) using (9) and (10).
Using the delta method, the asymptotic variance of C \(\left( {{=}\sigma _\mathrm{C}^2 } \right) \) can thus be obtainedFootnote 34 as
where \(\sum {\left( {\rho _\mathrm{A}, \beta _\mathrm{A }^*} \right) }=\) asymptotic variance–covariance matrix of \(\left( {\rho _\mathrm{A}, \beta _\mathrm{A }^*} \right) , \sum {\rho _\mathrm{B}} =\) asymptotic variance of \(\rho _\mathrm{B} \).
Again the aggregate coefficients effect, D is a function of the coefficients \(\left( {\beta ^{*}}\right) \) of both the regions A and B and of the autoregressive coefficient of region B, i.e., \(D=f\left( {\hat{{\beta }}_A^*;\hat{{\rho }}_\mathrm{B},\beta _\mathrm{B}^*} \right) \); using (12) and (13).
Thus using the delta method, the asymptotic variance of coefficients effect, \(D \left( {{=}\sigma _\mathrm{D}^2 } \right) \) can be derived as
where \(\sum {\left( {\beta _\mathrm{A}^*} \right) }=\) asymptotic variance– covariance matrix of \(\beta _\mathrm{B}^*\) and \(\sum {\left( {\rho _\mathrm{B}, \beta _\mathrm{B}^*} \right) } \) = asymptotic variance–covariance matrix of \(\left( {\rho _\mathrm{B}, \beta _\mathrm{B}^*} \right) \).
Now \(\sum {\left( {\rho _\mathrm{A},\beta _\mathrm{A }^*}\right) }, \sum {\left( {\beta _\mathrm{A}^*}\right) }\) and \(\sum {\left( {\rho _\mathrm{B} ,\beta _\mathrm{B}^*}\right) }\) can, respectively, be obtained from \(\sum {\left( {\rho _\mathrm{A}, \beta _\mathrm{A}}\right) }, \sum {\left( {\beta _\mathrm{A} } \right) } \) and \(\sum {\left( {\rho _\mathrm{B}, \beta _\mathrm{B} } \right) } \) by using the delta method, where
Using the same methodology as in above, the asymptotic variance of
\(C_k \left( {=\sigma _{C_k }^2 } \right) \) and the asymptotic variance of \(D_k \left( {{=}\sigma _{D_k }^2 } \right) \) can be obtained as
andFootnote 35
Rights and permissions
About this article
Cite this article
Chattopadhyay, S., Majumder, A. & Jaman, H. Decomposition of inter-regional poverty gap in India: a spatial approach. Empir Econ 46, 65–99 (2014). https://doi.org/10.1007/s00181-013-0683-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00181-013-0683-8