In analyses of the optimal design of government programs, an important consideration is the extent to which a program can impact labour market outcomes. Typically, such analyses consider one government program in isolation, abstracting away from other roles of government, and thereby ignoring "fiscal externalities," or the effects of the program's labour market impacts on income tax revenues. This paper assesses the importance of fiscal externalities in the context of optimal unemployment insurance. I calibrate and simulate a standard dynamic job search model, and perform a theoretical and numerical analysis of the seminal Baily (1978) two-period model of unemployment, in order to answer the question: how much does the abstraction from other government activities matter for optimal UI calculations? Numerical results indicate that it can matter a lot: with moderate risk-aversion and no effect of UI benefits on subsequent wages, the optimal replacement rate drops from around 40% to zero. Higher risk-aversion moderates this effect, and a large positive effect of UI on wages could significantly increase the optimal benefit level. The theoretical analysis indicates that accounting for fiscal externalities increases the estimated optimal benefit level if and only if higher benefits increase total post-unemployment earnings.