This dissertation comprises three research papers, each of which addresses a question in public economics.Chapter 1 investigates the response of taxpayers to tax incentives in Ireland before and after the Great Recession. Pre-2009 there is clear evidence of bunching in the income distribution just below thresholds that trigger large tax liabilities. This evidence disappears from 2009 onwards. This suggests that the taxpayer response is weaker during a recession. Much of the difference reflects reduced employment in sectors such as construction that exhibit above-average ability to report tax-advantaged incomes. However, even for people who remained with the same firm over the period the likelihood of reporting a tax-advantaged income fell during the recession.Chapter 2 examines the link between political fragmentation and tax policy. A model of government is presented where an n-member coalition chooses revenue and expenditure policies. I derive the response of tax policy to a change in the number of coalition partners. The model predicts that more fragmentation leads to (i) lower taxes; (ii) lower expenditure; and (iii) lower social security transfers. These results run counter to the conventional wisdom that countries with more fragmented governments have larger public sectors. I test the model on a large panel of developed countries, and all three of the model;;s predictions are supported. I estimate that moving from a two- to three-party legislature lowers tax revenue by 6.7%, expenditure by 9.5%, and transfers by 5.4%.Chapter 3 tests the relationship between crime and the labour market in Ireland during the boom and bust period 2003--2014. Based on detailed county-level panel data on crime and unemployment register figures, higher unemployment is associated with higher crime rates: I estimate a property crime elasticity of about 0.5. This implies that a 10% rise in numbers on the unemployment register increases thefts and burglaries by 5%. To estimate causal effects, I also test the relationship using a Bartik-style instrumental variable. The IV results confirm that a reduction in employment leads to more crime.