The general theory of economic voting posits that voters reward or punish the incumbent government depending on the performance of the national economy. An almost infinite number of empirical studies support this proposition across different time periods and world regions. Two lines of research have gained traction in recent years. The first explores how individual-level characteristics such as political knowledge and contextual-level factors including political institutions moderate the effect of economic conditions on incumbent support. The second analyses how voters adopt different international or historical reference points to gauge relative economic performance. The present paper combines insights from these two strands with theories of political socialization. It argues that economic crises in the years of political socialization shift voters' reference points to assess current economic performance and thus change how they respond to economic circumstances during election years. I test this argument by combining individual-level data from the Latin American Public Opinion Project (LAPOP) with macro-economic performance data and comparative data on economic crises (e.g. banking crisis, sovereign debt default) from the Behavioral Finance and Financial Stability (BFFS) project. Latin America, with its various economic crises from the 1980s onward, seems well suited to test this argument. The empirical results partly corroborate the argument and indicate that economic crises during political socialization moderate the effect of current economic performance on incumbent support. These results have important implications for explaining voting behavior and the impact of economic conditions on political representation.