With all this in mind; it is clear that disaster planning and government-led disaster risk reduction (DRR) has been part of the economic planning process of the government of Bangladesh for a long time. Most recently for Cyclone Sidr in 2007, for example, Bangladesh managed to evacuate millions away from the coast and the storm (Paul and Dutt, 2010).1 Bangladesh's successful disaster risk reduction policies are also mentioned in the context of annual flood management. of monsoons (del Ninno et al., 2003). 2 The severity of the 1998 flood is identified in terms of the affected area (affecting two-thirds of the country) and lasted for a long period (from early July to mid-September) in many areas and direct damage was valued at USD 2. billion (Khandker, 2007).
For example, the Department of Disaster Management (DDM) in Bangladesh builds bridges/culverts (up to 12 meters long) under its annual development plan – the main purpose for this infrastructure is DRR rather than development or poverty alleviation more broadly. The disaster risk reduction data on public spending at the local government level was collected from publications of Bangladesh's Ministry of Food (former Ministry of Food and Disaster Management) – the information was collected from the Ministry's web portal where sub-district (upazila) disaster risk reduction and mitigation funding allocation data from FY (fiscal year ) 2010-11 to FY2013-14 were available. Data were aggregated by adding allocations in general and special categories under each DRR program for each of the 483 sub-districts.
Due to its geographical location in the south-eastern part of the Hindu-Kush Himalayan region and being at the confluence of three major rivers - Ganges, Brahmaputra and Meghna, Bangladesh is extremely prone to floods. Population numbers and poverty rates for each sub-district (in each year) were collected from the circular orders of the government's Disaster Management Department. Our proxy for 'economic development' for each sub-district is a composite variable that averages the percentages of the population with access to basic amenities (electricity, safe drinking water and sanitation facilities).
According to the Bangladesh government's Coastal Zone Policy (2005), the zone is divided into 'exposed coast' (the area/upazilas that are directly on the sea, and 'inner coast' (the area/upazilas that are behind the sea). ).
Descriptive Statistics and Model Specification
The first identifies the central sub-district in any given district (in most cases this means a larger population, a higher degree of urbanization and more industrialized). Other binary measures indicate urban sub-districts associated with two megacities in Bangladesh (Dhaka and Chittagong). We do not have a preconceived notion of the types of influences that influence the regional distribution of public spending, but for ZRN spending we assume that these are determined by risk perception, socioeconomic vulnerability, and political and geographic factors.
Some sub-counties receive no funding for some of the DRR programs we investigate during some fiscal years. Because of this data truncation, we use a two-stage Heckman selection model to identify the determinants of public expenditures on disaster risk reduction and mitigation. Expenditure is also a function of population (pop) and poverty (pov) rates in the host sub-district, and measures of socio-economic deprivation (in depth: measured as access to certain assets - see discussion of data earlier in section 5).
This public spending is also a function of a set of characteristics, measured as binary variables (vector D), which political affiliation with the center, presence of ethnic minorities, being a district headquarters, belonging to either of the two major metropolitans, include. Our theoretical precedence is that these determinants should have a positive relationship with sub-district DRR funding allocation. Ceteris paribus, a subdistrict with higher perceived risk, more poverty, less access to assets, more deprivation, more political connections and a coastal location should receive more DRR funding (either allocated or realized).
We are agnostic with respect to several of the other controls, including location as a district headquarters or as part of the two metropolitan areas, and the presence of ethnic minorities. Where, 𝑧𝑖𝑗𝑡 is a latent variable indicating funding and is the dependent variable in the selection equation [2]. The selection variable 𝑧𝑖𝑗𝑡 is binary and we therefore use a probit regression specification to estimate the first stage selection equation [2].
The second stage specifies the outcome (public spending) equation, where public spending (allocated or realized) is the dependent variable. Where 𝑌𝑖𝑗𝑡 is the dependent variable of the outcome equation, 𝑋𝑖𝑗𝑡 is the vector of covariates, 𝛽 is the vector of coefficients, and 𝑢𝑖𝑗𝑡 is the random disturbance term. The selection equation (first stage) includes a population variable that is not included in the outcome equation (second stage).14.
Estimation Results
Perceptions of low and high flood risk variables appear to have a counter-intuitive negative association with the allocation of disaster risk reduction resources with consistent statistical significance. The political connection to the central indicator also has a counterintuitive negative sign, but this estimate is statistically insignificant. Contrary to our selection estimate, the outcome for ethnicity and county seat showed a positive association with the allocation of disaster reduction resources.
However, as in first-stage estimates, political connections and flood risks showed negative correlations with allocated spending patterns. In particular, a one standard deviation increase in high flood risk leads to a 0.33 standard deviation decrease in expected DRR allocated consumption per capita compared to 0.38 s.d. Overall, and particularly this finding on flood risk measures, our findings suggest that there is no obvious logic to the way the Bangladeshi government allocated its DRR funding.
All the columns in these two tables represent the same set of variables with the dependent variable being the realized financing per capita in ZRRF. In particular, we observe a similar pattern for the two variables we highlighted earlier: flood risks. Again, low and high flood risks tend to show negative associations with statistically significant coefficient estimates, while the political connection variable appears to be negatively associated with funding, but statistically significant at 10%.
A one standard deviation increase in high flood risk leads to a 0.39 standard deviation decrease in predicted realized per capita disaster risk reduction spending compared to 0.38 s.d. In Tables 5 and 6, we report separately the results of Heckman's two-stage regression for the climate investment fund. The first two columns in Table 5 show the determinants of public spending on climate change by subdistrict per capita.
As before, the most intriguing of the results reported are negative coefficients for the flood risk measure and the political context variable; however, in this case the coefficients are not always statistically significantly different from zero. The second-stage regression results for the Climate Investment Fund in Table 6 show similar patterns to the previous Table 5. Among the RHS variables; poverty rate, ethnicity and city center show sign consistency without statistical significance.
Conclusion
Establishing links between disaster risk reduction and climate change adaptation in the context of loss and damage. Available at http://indiaenvironmentportal.org.in/files/file/establishing-links-disaster-risk-reduction-n-climate-change-adaptation-in-context-of-loss-n-damage.pdf. Allocated and realized for each safety net program indicates the total (per capita) amount of public funds that have been allocated and the total (per capita) amount of public funds that have been spent out of the total allocation in disaster risk reduction on a rolling basis.
2 TR_ALOCATED Total amount (per capita) of public funds allocated to disaster risk reduction through the pilot aid program. 3 TR_REALIZED Total amount (per capita) of public funds spent from the total allocation for disaster risk reduction through the pilot aid program. 4 FFW_ALOCATED Total amount (per capita) of public funds allocated to disaster risk reduction through the Food For Work program.
5 FFW_REALIZED Total amount (per capita) of public funds spent from the total allocation for disaster risk reduction through the Food For Work program. 6 INFRA_ALOKATED The total amount (per capita) of public funds has been allocated to bridges and canal construction under the Food For Work program. 7 INFRA_REALIZED The total amount (per capita) of the public fund has been spent from the total allocation for bridge and canal construction within the Food for Work program.
8 GR_ALLOCATED The total (per capita) amount of public funds allocated to disaster risk reduction through a free relief program. 9 GR_REALIZED The total (per capita) amount of public funds spent out of the total allocation for disaster risk reduction through the gratuitous relief program. 10 VGF_ALLOCATED The total (per capita) amount of public funds allocated to disaster risk reduction through the feeding program for vulnerable groups.
11 VGF_REALIZED The total (per capita) amount of public funds spent out of the total allocation for disaster risk reduction through a program for feeding vulnerable groups. 12 CIF_ALLOCATED The total (per capita) amount of public funds that have been allocated in the Climate Investment Fund to combat risks caused by climate change. 13 CIF_REALIZED The total (per capita) amount of public funds spent out of the total allocation in the Climate Investment Fund to combat risks caused by climate change.
The number of times each sub-district is likely to be at risk of flooding each year. 19 DISTRICT STABLE Dummy variable; 1 if the sub-district is central (in most cases, the largest population size and main economic center) in any given district, 0 otherwise.
