Testing a Measurement Model of Financial Capability Among Youth in Ghana
Financial capability refers to youth’s financial knowledge and skills and having access to formal financial services, such as savings accounts with local banks or credit unions (Sherraden, 2010). To promote financial capability, there is widespread interest in financial education for youth (Council for Economic Education, 2009; MasterCard Foundation, Microfinance Opportunities, & Genesis Analytics, 2011) and efforts to help youth use formal financial services to save and build assets (Deshpande & Zimmerman, 2010; Storm, Porter, & McCaulay, 2010). However, the lack of consistent and standardized measures of youth financial knowledge, attitudes, and behaviors makes it difficult to draw meaningful conclusions from intervention studies in this field (Schuchardt et al, 2009). The purpose of this study was to assess the psychometric properties – including group invariance - of a set of youth financial capability measures.
This study used baseline data from the YouthSave Questionnaire for Youth (YSQ-Y), which was administered to middle school students (N=6,252) in eight out of 10 regions of Ghana in May 2011 by researchers from the Institute of Statistical Social and Economic Research of the University of Ghana as part of the impact assessment of YouthSave, a youth savings demonstration. Exploratory factor analyses (EFA) using polychoric correlations and Weighted Least Squares (WLS) were conducted to assess the factor structure of 18 financial capability items from the YSQ-Y. This was followed by confirmatory factor analyses (CFA) to test a measurement model and assess group invariance by gender using Mplus 6.1 (Muthen & Muthen, 2010). Schools were used as a cluster variable and WLS was selected as the estimator for use with mostly categorical and non-normal data.
After dropping three items with weak factor loadings, a three factor solution with 15 observed indicators emerged. Three latent variables – money management, awareness of financial services, and financial services knowledge – were tested in a measurement model using CFA. Initial model fit was poor (χ2=1576, p<.0001, 87 df; RMSEA=.052 [90% CI .050, 055]; CFI=.816 TLI=.778). Using modification indices, an alternative model was specified in which the measurement errors of three pairs of indicators with similar item construction (and common sources of error) were correlated had very good fit (χ2=351, p<.0001, 84 df; RMSEA=.023 [90% CI .020, 025]; CFI=.967 TLI=.959). All latent variable factor loadings were statistically significant. Finally, using the DIFFTEST command in Mplus, there was no statistically significant χ2 between models with and without latent variable factor loadings that were constrained to be equal for girls and boys, suggesting measurement model invariance by gender.
Conclusions and Implications: Financial capability subscales of the YSQ-Y demonstrated good construct validity among a sample of middle school students in Ghana. These subscales may be good instruments to use and assess self-reported financial knowledge, attitudes, and behavior in studies evaluating the outcomes of financial education and inclusion interventions. However, these subscales should be cross-culturally validated before being used with youth in other countries.