![Sample our Urban Studies journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110826/TOC-Subject-Sample-2021-Urban-Studies.jpg"})
Abstract
Mahadevan R. and Suardi S. Regional differences pose challenges for food security policy: a case study of India, Regional Studies. This paper examines factors affecting the calorie gap by considering the risk of calorie inadequacy or excess at the tails of food intakes. Non-linear estimations accounting for rural/urban differences in more and less developed states allow for policy-making on two levels. First, the calorie gap was found to respond differently depending on the calorie status of the individual, to various socio-economic characteristics, social assistance programmes, as well as caste and religion. Second, these impacts depended on rural/urban differences and at other times on the development of the states. These findings pose significant challenges towards achieving a balanced regional food security policy.
Mahadevan R. and Suardi S. 区域差异对粮食安全政策所带来的挑战：印度的案例研究，区域研究。本文藉由考量粮食摄入端卡路里热量不足与过剩的风险，检视影响卡路里热量差距的因素。用以解说大至为已开发国家城乡差异的非线性评估，在以下两大层面考虑到政策的制定：首先，热量摄取的差异对于各种社会经济条件、社会支持计划，以及种性和宗教有着不同的反应，端看个人的热量摄取情况而定。再者，这些冲击取决于城乡差距，有时亦取决于国家的发展状况。这些研究成果，对达成更为平衡的区域粮食安全政策，提出了重大的挑战。
Mahadevan R. et Suardi S. Les écarts régionaux posent un défi pour la politique de sécurité alimentaire: étude de cas de l'Inde, Regional Studies. Cet article cherche à examiner les facteurs qui influencent la différence dans la nutrition en considérant le risque d'une insuffisance ou d'un excès de calories aux limites des besoins alimentaires. Des estimations non-linéaires, qui expliquent les écarts ruralo-urbains dans des états plus ou moins développés, permettent l'élaboration des politiques à deux niveaux. Primo, il s'est avéré qu'en fonction de l'état nutritionnel de l'individu, la limitation calorique réagit différemment à diverses caractéristiques socioéconomiques, à des programmes d'assistance sociale, aussi bien qu'aux systèmes de castes et à la religion, suivant le régime nutritionnel de l'individu. Secundo, ces impacts dépendent des écarts ruralo-urbains et, d'autres fois, du développement des états. Ces résultats posent d'importants défis pour la réalisation d'une politique en faveur de la sécurité alimentaire régionale équilibrée.
Mahadevan R. und Suardi S. Regionale Unterschiede erschweren die Politik für Nahrungssicherheit: eine Fallstudie von Indien, Regional Studies. In diesem Beitrag untersuchen wir die Faktoren, die sich auf die Kalorienlücke auswirken, indem wir das Risiko eines Kalorienmangels oder -überschusses am Ende der Nahrungsversorgungskette analysieren. Nichtlineare Schätzungen unter Beachtung der Unterschiede zwischen ländlichen und städtischen Gebieten in mehr oder weniger entwickelten Bundesstaaten ermöglichen die Berücksichtigung der Politik auf zwei Ebenen. Erstens reagiert die Kalorienlücke den Ergebnissen zufolge je nach Kalorienstatus der Einzelperson unterschiedlich auf verschiedene sozioökonomische Merkmale und Sozialhilfeprogramme sowie auf Kaste und Religion. Zweitens hängen diese Auswirkungen von den Unterschieden zwischen ländlichen und städtischen Gebieten und zu anderen Zeitpunkten vom Entwicklungsstand der Bundesstaaten ab. Diese Ergebnisse verdeutlichen die erheblichen Schwierigkeiten bei der Verwirklichung einer Politik für ausgeglichene regionale Nahrungssicherheit.
Mahadevan R. y Suardi S. Diferencias regionales dificultan la política de seguridad alimentaria: estudio del caso de India, Regional Studies. En este artículo analizamos los factores que afectan a la brecha calórica al analizar el riesgo de una insuficiencia o un exceso de calorías en la cola de ingesta de alimentos. Las estimaciones no lineales para las diferencias rurales/urbanas en estados más y menos desarrollados permiten crear políticas a dos niveles. En primer lugar, se observó que la brecha calórica respondía de modo diferente, en función del estado calórico de la persona, a diferentes características socioeconómicas y programas de asistencia social, así como a la casta y la religión. En segundo lugar, estos efectos dependían de las diferencias rurales/urbanas y otras veces del nivel de desarrollo de los estados. Estos resultados plantean retos importantes a la hora de lograr una política de seguridad alimentaria equilibrada en un ámbito regional.
Notes
1. These refer to the Food for Work programme and the Public Distribution System, which are detailed in later sections.
2. While an extensive survey is provided by Strauss and Thomas Citation(1995), more recent studies in this area include those by Salois et al. Citation(2010), Babatunde et al. Citation(2010), Skoufias et al. Citation(2009) and focus on India by studies such as those of Deaton and Dreze Citation(2009), Meenakshi and Viswanathan Citation(2005), and Subramanian and Deaton Citation(1996), amongst others.
3. On this note, the present study also documents the role of rural areas in regional development, which overlaps with some of the issues raised in the Regional Studies special issue on ‘Placing the Rural in Regional Development’ (Ward and Brown, Citation2009). Specifically, the present finds the need to adopt a paradigm shift when tackling the problem of food security at the regional level, where efforts should be made to improve the coordination of regional policies within central governments and to decentralize policy administration.
4. This weighted index captures three indicators: the proportion of undernourished as a percentage of the population; the prevalence of underweight in children under the age of five; and the mortality rate of children under the age of five (IFPRI, Citation2008). The calorie cut-off used for the index was 1632 kilocalories (kcal) per person per day.
5. For instance, even though calorie intakes may be low, nutrients can be better utilized if healthcare, clean water, and sanitation are more advanced and readily available.
6. It is not possible to do so because there is no single survey in India that captures the measures of all the three elements of food security. For instance, food absorption is proxied by body mass index (BMI) and the survey that has these data is the National Family Health Survey, while the measure for food availability (as considered in this paper) comes from the National Sample Survey Organisation (NSSO) data.
7. The FAO, on the other hand, uses 1890 kcal to represent the minimum amount of calories necessary for good health as opposed to the average requirement norm. The National Sample Survey (NSS) (Citation2001) has devised 2700 kcal per consumer unit as another norm based on age–gender-specific conversions, but it was decided here not to use this norm as it does not distinguish between rural and urban areas.
8. The authors thank the referee for highlighting overpopulation as another reason for the existence of a high proportion of undernourished people in India.
9. Another important variable is income inequality, but it is not possible to incorporate directly this macro variable, which is computed at the state level, into the microeconomic framework using the household data. One potential method is to use a multilevel hierarchical model that requires panel data. First, the set up of the NSSO data is such that the identification of household is not possible over the years. While it is possible to construct a pseudo-panel data model, the exercise is complex due to the need to identify a relevant non-time-varying variable to anchor the observations (Bourguignon et al., Citation2004). But more importantly, using panel data for the analysis detracts from the objectives set out in the present paper.
10. These include the inefficient delivery of services; errors in targeting; a bias towards rural or urban areas; and discriminating against disadvantaged castes (Thorat and Lee, Citation2010; Swaminathan, Citation2009; Suryanarayana and Silva, Citation2007).
11. Caste in India is based on a traditional Hindu hierarchical system that ascribes an occupation to people in different groups. From bottom to top there are four broad categories: Scheduled Castes, or the so-called ‘untouchables’, Backward Caste and Forward Caste, plus the Scheduled Tribes, which are outside the traditional hierarchy but often placed below Scheduled Castes. Economic and social power is generally concentrated in the hands of the Forward Caste.
12. Muslims make up 13.4% of the population and in urban areas they comprise 17.3% compared with 12% in rural India (GOI, Citation2006).
13. Price data for food are not directly available in the NSS, but following Ray Citation(2007) the present authors constructed the price index of various cereals (which is the single largest source of calories) based on the expenditure on these cereals and the quantity consumed of these cereals. These were then used in the regression analysis and found to be insignificant and hence dropped. Results are available from the authors upon request.
14. See the Human Development Reports of the Planning Commission of India and the Economic Survey 2009–10 (GOI, Citation2010) for state-wide data.
15. These differences could however be due to the macro-factors stated in note 1.
16. This is open to all who are in need of wage employment and volunteer to do unskilled work. Under this scheme, a household is said to be participating in this programme when at least one member has been employed for at least sixty days over the last year. Often labour is engaged for development works and a part of their wages is given as food grains.
17. The Breusch and Pagan Citation(1979) test of heteroskedasticity is performed on calories' gap data for the different regions. The squared residuals from the regression are regressed on the explanatory variables and their coefficients are jointly tested for their statistical significance. The test statistics for forward urban, forward rural, backward urban and backward rural states are 35.53 (0.04), 38.04 (0.03), 43.76 (0.01), and 39.94 (0.02), respectively. (Figures in parenthesis are p-values.) Based on the null hypothesis of homoskedasticity, the p-value shows that the test comfortably rejects the null in favour of heteroskedasticity in the data.
18. Although quantile effects can be accommodated in standard regressions by categorizing the dependent variable into the lowest 10%, lowest 25%, middle 50%, and highest 25% among other variations, the problem of different error distributions within each category may still persist, thereby leading to inaccurate estimates.
19. It is important to note that a positive (negative) calories gap suggests that a household has achieved a better (worse) than expected caloric norm. Thus, in terms of interpreting the estimates for the continuous independent variables, for the under (over) nourished or negative (positive) calories gap household, a positive (negative) coefficient is good as it means that the group can close (widen) the gap.
20. The child–adult ratio that is often used to control for household composition is not included as the NSS data are quite robust to alternative assumptions about equivalence scales (Gang et al., Citation2008; Meenakshi and Ray, Citation2002) and the caloric norms have an inbuilt correction for demographics that corresponds to Indian population characteristics (Gosh and Guha-Khasnobis, Citation2006).
21. This is because expenditure and calories consumed share a similar basis of calculation in terms of quantities of food consumed and any unobserved factor that affects food consumed will affect them both.
22. Household expenditure data are used instead of income as the former are said to have a smoothing effect in the long run.
23. For instance, price indices, variety of food (food substitutes), composition of diet, and level of physical activity among other variables are important but have not been included in the equation given that the focus of this study is not on obtaining elasticities based on a calorie demand equation. Another problem may exist in the form of endogeneity raised above, but the authors share the sceptical view of Subramaniam and Deaton (1996), Bliss and Stern Citation(1978), Mirrlees Citation(1975), and Stiglitz Citation(1976) that behaviour in nonlinear models (in this case, the quantile regression) cannot be appropriately accommodated by the routine application of instrumental variables and two-stage least squares. While there has been some work to deal with this problem within the quantile framework (Lee, Citation2007; Chernozhukov and Hansen, Citation2006; Abadie et al., Citation2002), these have yet to be widely used due to the lack of ease in application.
24. See note 5.
25. Also, intra-household differences are not considered in this paper as there are no data available for the analysis from the NSSO.
26. This is partially supported by Gosh and Guha-Khasnobis's (Citation2006) evidence of an apparent mismatch in the targeting of this programme towards food insecurity in Madhya Pradesh and Orissa which are included in the backward states in this paper.
27. In fact, Nandakumar et al. Citation(2010) provided evidence of the employment scheme having raised minimum wages for four of the backward states in this study. Farrington et al. Citation(2007), on the other hand, warned that the programme should be designed to be attractive only to the needy. For instance, the scheme should not be for those who own land or other productive assets and wages need to be kept low so that only those with a low opportunity cost of time will present themselves for jobs.
28. To check the robustness of this result, an alternative dummy variable denoting a 1 for households whose grain consumption from PDS makes up 50% of total grain consumption was used. The results were quite similar, showing that PDS was generally ineffective in raising the calorie gap.
29. The universal PDS in Kerala and Tamil Nadu has proven to work well.
30. The authors thank the anonymous referee for highlighting this point.