Ländliche Armut in Indonesien: Indikatoren, Dynamik und Verbindung zur Entwaldung
The analysis of the link between poverty and deforestation uses a relative poverty index that ranks households according to their wealth status. In 2005, two very promising operational poverty assessment models were developed for Central Sulawesi using 15 proxy indicators to predict absolute poverty based on expenditures. In 2007, both models were tested regarding the extent to which these models are robust over time in terms of prediction power and in terms of their indicator composition. In
... dition, an indicator based poverty assessment tool provided by the U.S. Agency for International Development for Indonesia was evaluated here using the data set from Central Sulawesi. Regarding the robustness of indicator-based poverty assessment by proxy-means tests, the sets of indicators derived in 2005 were still capable of detecting the very poor households (those living on less then 1 Dollar purchaising power paritiesper capita and day) in 2007. However, the models tend to over-predict the very poor. For the assessment of absolute poverty in the research area, we recommend the use of 15 easy to verify indicators (Model 2) in combination with the quantile coefficients of these indicators derived from the one-step procedure (estimated in 2005). This is based on the comparison of accuracy performance of all tools in both years. The accuracy levels of the two models tested remain similar when estimated using the 2007 dataset. However, the indicator composition of the tools changed. The nationally calibrated tool provided by USAID is shown to perform poorly in terms of accuracy when applied to our data set. To gain a better understanding of the poverty dynamics in the region, different measures of poverty are compared across the survey years. Additionally, transitions into and out of poverty are obtained. In general severe poverty (less the 1 US$ per capita and day) is shown to have decreased. The headcount index for the severely poor declined insignificantly from 19.3 percent in 2005 to 18.2 percent in 2007. However, people in the research area got poorer over the same period since significantly more people are shown to slipped into expenditures below the 2 US$-poverty line. Moreover, the poverty deficit in 2007 is also shown to be greater than the poverty deficit of 2005 irrespective of which poverty line is used. While 49 percent of the very poor households remain very poor in both survey years, 33 percent of them moved out of severe poverty but still had to live on less than 2 US$ PPP per capita and day. Nonetheless there is also movement in the opposite direction. Twenty-three percent of the households in the category of poor (living on less than 2 US$-poverty line) in 2005, became severely poor Summary iii (less than 1 US$-poverty line) in 2007. To trace the underlying determinants of chronic and transitory poverty, multi-nominal logit regression analyses were applied. Results show that large households are more likely to be chronically and transitorily very poor. They are also shown to have higher probability of being chronically poor. Lack of access to electricity also makes severe chronic and transitory poverty, as well as chronic poverty more probable. Household without access to social capital are similarly more likely to get chronically very poor. Access to credit reduces the probability of becoming chronically very poor and also makes chronic and transient poverty less likely. Household without access to remittances from relatives working away from home are also more prone to (severe) chronic poverty. Finally, lack of opportunities to in engage non-agricultural income generating activities increases the probability to become chronically or transitorily very poor. Results from the study on the linkage between poverty and deforestation suggests that conversion of forest into farm land in the research area is indeed a severe problem. Approximately 52 km 2 of forest area was converted into farm land between 1999 and 2006 by smallholders. While the poorest and the poor mainly replace forests with subsistence crops such as maize and dry rice, the wealthier households mainly grow cocoa. The findings also show that poorer households are more likely to clear forest than their wealthier counterparts. However, most of the area converted is dedicated to cocoa production. Furthermore, households with younger household heads tend to clear more forest area than households with older household heads. Interestingly, access to social capital tends to increase deforestation. Secure property rights, however, tend to reduce deforestation. Additionally, the location of the household plays a crucial role: households living in a sub-district closer at the forest boarder are more likely to clear forest than those located further away. In general we find the results quite satisfactory. The study would, however, have benefited from a larger sample size. For example, a more precise calibration of the poverty assessment tools would have been possible if an out-of-sample test was applied. More rounds of expenditure surveys would also allow for use of the components approach to analyze poverty dynamics. Furthermore, the sample used is only representative of the research area and therefore policy implications are hardly applicable to other parts of Indonesia without further analysis of nationally representative data. However, this situation presents a unique advantage since policy implications derived are suited for direct implementation in the research area. With Summary iv respect to the link between poverty and deforestation, geo-referenced data at plot level would be of great benefit since they could provide more details on the "true" rate of deforestation. However, obtaining such data is often time-consuming and costly. v Content Summary.