Attempts to increase electoral access are generally seen as unambiguously desirable in new democracies. While these policies can increase the size of the electorate, they may also carry costs through changing electorate composition, by differentially enfranchising those who are highly sensitive to costs. Using new administrative data from South Africa and a difference-in-differences design, I show that a 15 year, large scale, expansion of access to voting stations increased national turnout by between 2.3 and 4.7 percentage points. In the context of a natural quasi-experiment, I then use 39,000 survey respondents geo-referenced to their nearest voting station to document that those of high socioeconomic status, and those who are older, are much more sensitive to electoral access than others. This may affect electorate composition in favor of those already economically or politically powerful, who also happen to be more opposed to redistributive public policies and have particular partisan preferences.