In 2010, Pakistan was hit by the worst flood in the century. A total of 1.6 million homes were damaged and 0.9 million utterly destroyed or washed away (World Bank, 2010), meaning that more than 5 million people were left homeless. This is equivalent to a quarter of the Australian population. The task of rebuilding was, and always is, a formidable one for cash-constrained developing countries like Pakistan. This holds both for individual households who have lost their home and for the government that might wish to help them. Because resources and funding are limited, government must set priorities and often make hard choices. If help is to be offered to the victims, what is to be rebuilt first and what later? Who is to be helped first and whom later? Should the amount of help differ between victims and if so, how?

This paper investigates whether different measures of ex-post poverty and ex-ante vulnerability to poverty provide consistent estimates of poverty and vulnerability to poverty across households. Moreover, if there is some heterogeneity in the identification of households as poor and vulnerable, it investigates the degree of mismatch between measures? The ex-post monetary poverty (MP) and multidimensional poverty (MDP) measures are used to identify poor households. Likewise, the ex-ante vulnerability to monetary poverty (VMP) and vulnerability to multidimensional poverty (VMDP) measures are used to identify vulnerable households. Using a large household survey data-set of about 90,000 households from the Punjab province of Pakistan, we find that most of the vulnerable households are accurately identified by the ex-ante measures of vulnerability to poverty. However, the ex-post measures of poverty identify different households as poor. Our results show that 18 percent households experiencing MDP are not captured by the one-dimensional measure of MP. The important implication of this study is that the choice of measures does matter in ex-post poverty identification, but not as much in the identification of ex-ante vulnerability to poverty.

Empirical studies in different developing countries have investigated the impact of social protection (SP) on ex-post poverty; however, few studies analysed the impact of SP on ex-ante vulnerability to poverty (VtP). This paper contributes to the literature by evaluating the impact of SP on vulnerability to poverty (VtP) and poverty rates among households in the Punjab province of Pakistan. A hierarchical modelling approach is used to analyse the Multiple Indicator Cluster Survey (MICS-2011) data of about 90,000 households. Matching methods and simultaneous endogenous switching regression are used to control for potential selection bias and estimate average treatment effects. Our results show that SP has a positive impact in reducing household poverty and VtP, but this impact is mainly driven by a short-term flood relief cash transfer programme, pension of government employees and households’ regular purchase from utility store network.