Identifying whether an intervention results in the desired effect is the classical problem in causal inference, often solved by performing randomised experiments. However, to determine effects of public policies that affect entire nations altogether, randomised trials are infeasible and unethical. Non-randomised, observational data on interventions and outcomes are all that we have at our disposal. In this thesis, I explore different methods for identification of causal effects of public policy decisions from panel data. In Chapter 1, I estimate the effect of lockdown restrictions in Bangladesh in April 2021 on the spread of COVID-19. I contrast Bangladesh to its neighbouring country, India, that did not have a lockdown. I compare data from the two countries in retrospect using difference-in-differences methods to establish that in the absence of preventive lockdown measures, reported COVID-19 deaths would have nearly doubled in Bangladesh during the study period. In Chapter 2, I investigate an abrupt shift in birth rates in North America following the announcement of the COVID-19 pandemic in March 2020. I use interrupted time series methods to compare data from North America before and after March 2020 and ascertain that the 2%-3% drop in birth rates was indeed due to reduced international travel. In Chapter 3, I extend difference-in-differences methods to identify treatment effects conditional on the intended value of treatment. Motivated by referendums in the Swiss political system, where voters indicate their intended policy implementation, I found that Swiss cantons that predominantly voted 'Yes' and those that voted 'No' towards the Schengen Agreement had differing effects on trade with neighbouring countries. Finally, in Chapter 4, I tackle the problem of identifying harmful or negative individual treatment effects. Previous work in the domain has focused on problems in binary settings, comparing only two potential outcomes. I demonstrate that these comparative methods do not generalise to settings with multiple treatment options, and can result in inconsistent or unethical medical recommendations.