![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
ABSTRACT
Poor rural households in developing countries often endure many-faceted burdens including monetary poverty, nutrition deficiency and energy shortage due to reliance on limited local natural resources with low utilisation efficiency. We investigate a sustained pathway in rural China to escape the vicious circle between three important dimensions of poverty – deficiency of income, malnutrition and a low energy consumption profile in terms of reliance on firewood. By exploiting household panel data and a dynamic and recursive multi-equation mixed mode, we identify inter-locking deprivations in income, nutrition and energy consumption. Firewood plantations only offer short-term solutions to break them through income effects, while the sustained pathways in the long-term are increasing agricultural labour productivity and provision of agricultural loans.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. The ‘energy ladder’ hypothesis postulates a complete transition away from traditional fuels such as crop residuals and firewood to more sophisticated fuels as their economic status improves (for example, an early empirical study for Zimbabwe in Hosier & Dowd, Citation1988). Using national longitudinal data for 175 countries, Burke and Dundas (Citation2015) do not find insignificant association between higher income and reduction in use of biomass energy.
2. Typically, an upward transition along the energy ladder is to use more efficient fuels like liquefied petroleum to replace biomass fuels like firewood and agricultural residuals (for example, Pachauri & Jiang, Citation2008 for India and China; and Malla, Citation2013 for Nepal), as the former generates more heating and energy for daily livelihood.
3. See for example, the case of Shaanxi province in China (Olivia, Gibson, Rozelle, Huang, & Deng, Citation2011) and in Peru (in Swinton & Quiroz, Citation2003).
4. See Jha, Gaiha, and Sharma (Citation2009) for empirical evidence on income-nutrition poverty traps in India.
5. It is acknowledged that other dimensions such as education and health are crucial in wellbeing, but including them into our multi-equation system would add much complexity in modelling sequence and feedbacks in multidimensional decision-making and would undermine achieving identification. The present study thus focuses on more ‘root’ dimensions such as income and nutrition that are likely to lead to other dimensions and relate them to sustainable livelihood such as energy use and production.
6. It is worth noting that the sample period over 2000–2004 might be limited in its capacity to capture the current poverty situation in rural China given the rapid economic development across the country. Yet, considering the persistently low income levels in Gansu and Inner Mongolia, the results based on our dataset do have wider implications for low- or lower middle-income countries experiencing similarly low economic wellbeing and facing similar difficulties in sustainable and multidimensional development as our sample areas. For example, according to the World Development Indicators at the World Bank, Nepal, Vietnam, Zimbabwe and Honduras showed the rates of access to commercial fuels for cooking of 26.1 per cent, 50.9 per cent, 31.3 per cent and 48.1 per cent respectively in 2014, which are similar to the rates of 34–36 per cent in our data. The household annual per capita expenditure in 2004 was US$ 604.8 in rural Gansu and US$ 860.8 in rural Inner Mongolia, measured by 2010 constant US$, which are also similar to those of Nepal (US$ 554.1), Vietnam (US$ 1121 US$), Zimbabwe (US$ 605.5) and Honduras (US$ 1806) in 2015, measured by 2010 constant prices.
7. It should be noted that nutrient intake is likely to be over-estimated in our data. Sample households’ food records might be less precise when time elapsed, given the diary method (on the monthly basis) lasting for four years. This might result in under-estimated nutrient intake and thus, higher nutrition poverty rate. By comparison, another nation-wide survey, the China Health and Nutrition Survey from 1989 to 2011, adopted a 24-h recall diary on three consecutive days for household food consumption. It indicates lower rural nutrition poverty rates in eight provinces (but without Gansu and Inner Mongolia) between 15 per cent and 20 per cent over the period 2000 and 2004 (You et al., Citation2016). Nevertheless, the increasing trend of nutrition poverty rates is very consistent between our data and the CHNS. It would thus be better to treat our estimated coefficients of nutrition poverty incidence y2it in Equations (1), (3) and (4) as lower bounds of its genuine effects.
8. The Chinese government initiated the SLCP in 1999 and implemented it all over the country in 2002 for the purposes of ecological restoration as well as economic sustainability of land. The SLCP requires rural households to convert their cropland with a slope equal to or greater than 25 degrees to forests or grasslands. The converted forests could be either ecological (for example, timber-producing) or economic forests. The sample households in our data were affected by the SLCP only in 2003 and 2004. Section 4.3 discusses the robustness of our results to the introduction of SLCP.
9. It is worth noting that binary indicators only capture significant changes, while the poor might gradually improve their livelihood. Ideally one should use continuous income and nutrition for robustness checks, though in our case the empirical model using such variables does not yield convergence and entails considerable multicollinearity. The reason might be attributed to sluggish changes in income and nutrient intake in our data (with the latter effect being more likely). Household per capita net income per annum was 592.1 yuan in 2000. It increased to 680.7 yuan in 2003 and decreased to 548.8 yuan in 2004. Nutrient intake was 1408.1 kcal per day in 2000 and remained roughly stable over the sample period ending at 1393.1 kcal in 2004. Poverty gap in both deprivations only dropped in 2001, while became ‘stable’ between 2002 and 2004. See You, Wang, and Roope (Citation2017) for more detailed poverty profiles.
10. Land allocation is made on a per capita basis within the village. Villages determine whether and when to reallocate land. The frequency of reallocation across regions and villages varies considerably, despite the 15-year rotation period suggested by the central government (Brandt, Huang, Li, & Rozelle, Citation2002). The right of using forest land was not allowed to be transferred until 2003 when pilot property right reforms on forest land began in Fujian and Jiangxi provinces. The nation-wide reform has begun since 8 June 2008 when the State Council issued ‘Guidance of Comprehensively Promoting Reforms on Collective Forest Land Property Right’. The provinces in our sample were not subject to structural changes in land allocation in the study period. Nevertheless, the sample provinces were subject to the national Sloping Lands Conversion Programme (SLCP) promoting conversion of cultivated land to forests since 2003. Given the high eligibility standards for including land in the SLCP, the proportion of forest areas in the household’s total land holdings only increased marginally by about one percentage point. Section 4.3 will further discuss robustness of our results to the SLCP.
11. This actually makes the joint distribution of household poverty and energy behaviour in subsequent time periods the same as their exogenously initial joint distribution in 2000 (Wooldridge, Citation2005). This implicit assumption simplifies substantially the maximisation process, and could be justified if there were no structural changes in households’ behaviour or preferences between 2000 and 2001.
12. We also estimated each of Equations (1)–(4) independently by standard two-step least square estimation (2SLS) with their own excluded instruments. The first-stage F-statistic cannot reject the null hypothesis of joint significance of instruments. The over-identification tests (Hansen J statistic) yield p-values between 0.149 and 0.622. Our selection of excluded instruments appears to be valid.
13. See You et al. (Citation2017) for another empirical application of the mixed equation model in multidimensional poverty.
14. Our data suggest that an average non-poor household (living above the US$1.25-a-day line) used 85.4kg firewood per annum as opposed to 13.2kg used by its poor counterpart. The non-poor invested 45 per cent more in fixed productive assets per annum compared to poor households, typically, four percentage points’ increases in irrigated cropland out of total cropland and 2.8 percentage points’ increases in the proportion of harvests using machines out of all harvests. The non-poor also had better access to tap water, sanitation and housing in terms of using concrete rather than wood or clay compared with the poor, at 1–5 per cent statistical significance levels.
15. This long-term impact can be understood through the income effect. As mentioned in Section 2, Chinese rural households’ nutrient intake declines as income rises, and higher income is associated with using commercial fuels as indicated by 0.587 in Column 3 of Table 2.
16. Using the size of firewood plantations gives qualitatively the same results. In the short-term, the estimated coefficients of the size (measured by the Chinese unit mu) of firewood plantations on income and nutrition poverty are −0.095 (at the 5% significant level) and −0.352 (at the 1% significance level), respectively. In the long-term, the lagged size of firewood plantations has positive impact on nutrition poverty. The estimated coefficient is 0.136 at the 10 per cent significance level.
17. One may also notice that more education may also slightly increase the likelihood of nutrition poverty (Columns 2 and 6 of ). As mentioned in Section 2, this can be explained by the dietary changes towards more fat and oil but less calories when Chinese get richer, which is usually related with more education. The form of education also matters. See Shimokawa (Citation2013) for further investigation into the role of education in Chinese nutrient intake.
18. This echoes Christiaensen et al. (Citation2013) who use the same dataset as is used in this paper and find labour productivity increases in agriculture are the most effective pathway out of poverty. This is also consistent with Imai and You (Citation2014) who use the CHNS (1989–2009) dataset and find both agriculture and out-migration are effective means to enable households to escape poverty, and agricultural income helps prevent households from returning into poverty again.
19. Unfortunately, we cannot test this in our data. The indicator of whether the village has access to electricity jumped approximately 16 fold in 2003 compared to average values in previous years and dropped back in 2004. There may well be substantial measurement errors in 2003.
20. In rural China, within each administrative village, there are several ‘natural villages’ defined as household clusters living nearby. The Grade IV road is also known as the ‘township road’, which is the lowest grade in the Chinese road system and provides basic traffic between villages and towns. The Grade IV roads are 3.5m wide and have speed limits at 20–40km per hour, depending on the slope of the roads.
21. Relevant average marginal effects based on Equation (7) are 10.3 percentage increases in the probability of using fossil fuels and 0.4 per cent decreases in the proportion of firewood plantations.
22. The positive correlation between village income growth and the incidence of nutritional poverty can be explained by the wealth effect on household dietary changes and nutrition intake, which has been discussed in Section 2. See You et al. (Citation2016) for more investigations.
23. The existing literature also suggests a nonlinear link between resource extraction and income (López-Feldman, Citation2014). For example, firewood collection in Nepal rises rather than decreases with the growing local economy (Baland, Libois, & Mookherjee, Citation2013).