I estimate the treatment effect of Battle Against Poverty, a place-based poverty alleviation program in China using government reported measure and neutrally observed measure in economic development. The empirical evidence from this paper suggests while the program increases aggregate GDP of poverty stricken region by 4.6%, the aggregate night light decreases by 11.1%. The diverging pattern of treatment effect can be partially explained by sector-specific growth patterns and not due to data manipulation. I also provide evidence that the change in phasing out evaluation makes the counties more likely to phase out, which increases the offsetting phasing-out effect. Eventually, neither GDP nor night light measures suggest the treatment effect is persistent.