 Old age poverty is an important concern in light of shrinking public pension and increasing longevity. And this issue is especially important for elderly women who are at much greater risk of old age poverty than men in almost all OECD countries. And governments have implemented or provided safety nets for pensioners with low income. However, policymakers all face a very important trade-off that is how to provide additional pension benefits without further eroding workers' incentive to work. And this trade-off is again very salient for women because the reason that they experience a low pension income is partly due to the fact they have relatively low wage and also shorter working life. You may ask the question why additional pension benefits increase the incentive for workers' retire earlier or make them exit the labor market earlier. This comes from two perspectives. First, higher pension benefits makes working less attractive because workers have to forego higher pension benefit if they continue working. And secondly, higher pension benefit also gives workers a much higher lifetime wealth so they can afford to retire earlier. So the sign of this effect is unambiguous. Therefore the central question here is the extent to which additional pension benefit will affect workers' retirement timing. So to understand the size of this response is important and helps to inform policymakers about the debate around the design of anti-poverty programs for elderly workers overall and also for women in particular. To measure the extent to which additional pension benefits affect retirement timing is actually understudied in the literature. And this is partly due to the difficulty of isolating the variations in pension benefits from changes in other parameters in the public pension system. For example, recent pension reforms have often increased the statutory retirement age at the same time introducing financial penalties to claim pension early. To address this challenge and isolate the causal impact of additional pension benefits, I explored a pension subsidy program in Germany implemented in 1992. This subsidy program provides additional pension benefits around 90 euros per month to eligible recipients, which is around 15% increase of their pension benefits. Three features of this subsidy program allow me to isolate the causal impact of additional pension benefits. First, it provides a good source of exogenous variations in the benefit level because the subsidy size is predetermined by the recipient's wage before 1992, so before the announcement of the reform. Secondly, the subsidy size has a kinked relationship with the recipient's average wage before 1992. So this allows me to compare very similar workers to the left and to the right side of the kinked point. And lastly, the only variation provided by this subsidy program is the benefit level. So it provides me an environment in which I can causally identify and isolate the impact of additional pension benefits. Using high quality administrative data sets, which covers around 20% of all insured workers in Germany, I provide the first investigation of this subsidy program. Looking at West German female recipients born between 1935 and 1951, I find that an increase of 100 euro of monthly pension benefits induce those female recipients to claim pension earlier by around six months. So the impact on age at exiting employment has similar magnitude but is a bit noisier. It is quite common for workers to not directly transition from employment to pension claim, therefore age at claiming pension is not necessarily the same as age at exiting employment. I also look at transitional behaviors, so I find those female recipients are more likely to use unemployment insurance as a stepping stone to retirement. And they also reduce their time spent in marginal jobs, which are referred to as mini jobs in Germany. I also show that the phasing out of this subsidy program can account for around 16% increase in retirement age for women in West Germany in over the past decades. Going back to the old age poverty question, I find that overall this subsidy program increases the lifetime wealth of the recipients after accounting for the loss in labor earnings due to early claiming. The findings are relevant from two perspectives. First it contributes to the literature by providing a novel identification strategy where I can causally isolate the impact of additional pension benefits from other changes in the pension system. And this is the first application of this subsidy program to study this question. Secondly, the findings also has important policy implications. The empirical findings suggest that this subsidy program is relatively less distortionary. A back of envelope calculation suggests that in order to provide one euro of the subsidy, government needs to raise additional 25 cents due to the behavior response. This measurement of physical externality is relatively small compared to that of other welfare programs such as extending unemployment insurance benefit durations. This insight is important for policy considerations because many governments are facing long-run software issues due to ageing population. There are some open questions about this subsidy program. First, the subsidy might benefit women in wealthy households. This is because the subsidy size is determined by individual earnings rather than household wealth. So in that sense, the program might now be that effective in terms of combating poverty. Secondly, by designing this subsidy program will be phased out gradually. So we might ask the question how does it affect the younger generations? Will that raise some unfairness concerns? In future work, I myself would like to look at the impact of additional pension benefits on health outcomes and mortality outcomes. In addition, also explore the impact on savings and expenditure consumptions of those pensioners.