The importance of the comparative benchmark for measuring composite financial literacy with survey data.

Introduction: Using survey data to calculate composite financial literacy (CFL), existed studies do not consider the geographical difference of the means of objectively-measured financial literacy and subjectively-perceived financial literacy, i.e., comparative benchmark. Methods: Taking the survey data of National Financial Capability Study (NFCS) for example, we explain why it is more reasonable to use the within-state average rather than the national average of financial literacy as the comparative benchmark to measure CFL. Then we use NFCS 2009, 2012, 2015 and 2018 dataset to comparatively analyze the difference between CFL measured with the two benchmarks. Results: The results of statistical analysis show that there is a great difference among the four groups of CFL measured with the two benchmarks, and 10.7% of respondents are categorized as a particular group of CFL incorrectly for all datasets. Additionally, the findings of spatial distribution analysis unveils that 36, 19, 15, and 6 states have respondents miscategorized in the four groups of CFL for 2009, 2012, 2015, and 2018 respectively, in which the highest proportion of the population miscategorized in a state is up to 49.91%. Finally, we find that several groups of CFL measured with the two benchmarks have significantly different effects on stock market participation behavior. Discussion: Using the national average as a benchmark to determine all the respondents' relative financial literacy levels for different states is not meaningful, and will lose the practical appeal to tackle the regional inequalities of financial literacy among the households. Therefore, we suggest that the within-state average of financial literacy, not the national average, should be taken as the comparative benchmark for identifying the more precise groups of CFL in survey.