TY - JOUR A1 - Hallerberg, Mark A1 - Gandrud, Christopher T1 - The Measurement of Real-Time Perceptions of Financial Stress: Implications for Political Science T2 - British Journal of Political Science N2 - How do politicians and voters respond to financial market stress, and with what political effects? Previous research addressing these questions lacks a crucial variable: a continuous, real-time indicator of the level of financial market stress that policy makers and voters perceived. We need a measure of actors’contemporary perceptions of financial market conditions to understand why they made a given choice and with what effects. Previous binary crisis measures are constructed post hoc, so tend to be biased towards severe crises and away from circumstances in which governments effectively responded to emerging trouble. As such, they suffer from clear selection bias. Annual post hoc measures do not necessarily capture conditions as they were perceived at the time of events such as elections. As dichotomous indicators, they do not measure crisis severity or how it varies over time. They use ad hoc methods to determine when crises have ended. Previous continuous measures of financial market stress are less common and suffer from other problems. They capture quantities whose importance, measurement, and reporting varies significantly across countries and over time. To overcome these issues, we develop a continuous measure of real-time perceptions of financial market stress with a kernel principal component analysis (KPCA) of detailed qualitative data, namely monthly Economist Intelligence Unit (EIU) reports. We call it the EIU Perceptions of Financial Market Stress Index, or FinStress for short. FinStress enables new political research possibilities. As a continuous measure, it could be used to examine which policies can effectively prevent or reduce extreme stress, and which political conditions are conducive to implementing these policies. As a comparable continuous monthly indicator, FinStress could be used to test hypotheses that rely on sub-annual data and follow the intensity of stress over time. Here we provide examples for studying the impact of financial market stress on voters’choices and on revisions to European Union government budget figures. We thus contribute to the wider methodological toolkit by showing how KPCA can be used to summarize vast quantities of similarly formatted qualitative texts into continuous cross-sectional time-series indicators. Y1 - 2018 UR - https://opus4.kobv.de/opus4-hsog/frontdoor/index/index/docId/2603 SP - 1 EP - 13 ER -