Financial Resilience, Financial Ignorance, and their impact on financial well-being during the COVID-19 pandemic: evidence from Brazil

Abstract

This study aims to introduce a measure of Financial Resilience for evaluating its influence on financial well-being amidst the COVID-19 pandemic. Financial Resilience, in this context, pertains to an individual's capacity to adapt, cope, and financially recover in the face of the new challenges posed by an economic downturn. We conducted a survey among 591 Brazilian citizens in 2021 and employed Structural Equation Modeling to examine the relationships between various constructs and validate our proposed measure of Financial Resilience. Our findings reveal a negative impact of the COVID-19 pandemic on Financial Resilience, resulting in heightened uncertainty and insecurity concerning individuals' financial future during this period. Notably, those with lower Financial Resilience were found to be the most financially vulnerable during the pandemic. These results provide valuable insights into the measurement of financial contingencies faced by the population and can be useful to policy and governmental strategies aimed at mitigating the social, economic, and financial repercussions of economic downturns. Fostering Financial Resilience and reducing vulnerability among lower-income individuals are critical objectives, particularly in times of economic uncertainty. While the concept of Financial Resilience is of paramount importance, the literature on this subject remains relatively nascent, with ongoing developments in measurement instruments. In this regard, our study contributes by proposing a concise measure of Financial Resilience through two simple questions.

Appendix 1: Methodological procedures for validating the constructs of loss of financial well-being and financial ignorance

The first validated construct pertains to the loss of financial well-being. Upon analyzing the initial measurement model for the FWB loss construct, it was observed that all variables were significant at the 1% level and exhibited satisfactory factor loadings (values exceeding 0.50). However, during model estimation, it became apparent through the analysis of fit indices that the model fit was inadequate, as displayed in Table 10. Thus, the "Initial Model" column presents the results of the model prior to correlation adjustments, while the "Final Model" column indicates the results of fit indices for the final estimation, which includes the inserted correlations. Additionally, Fig. 4 presents the coefficients of the scale items, the correlations, and their respective significances.

Table 10 Adjustment statistics of the loss of financial well-being construct—initial model and final model.

Fig. 4

Final measurement model of the loss of financial well-being construct with the respective standardized coefficients and the significance of each variable. Note(s): ***Significant at the 1% level. ¹z-value not calculated, where the parameter was set to 1, due to the requirements of the model

To achieve a better-fitting model, the chosen improvement strategy was characterized by the gradual inclusion of correlations between the errors of observable variables that made theoretical sense.

The first correlation introduced was between errors e6 and e7. These questions pertain to respondents' perception of future financial security during the pandemic. Therefore, the theoretical rationale for this correlation is that these questions share common measurement errors, as respondents who have difficulty identifying their perception of financial security lasting until the end of their life (Q06) during the pandemic period will also have difficulty perceiving that they are financially secure until the end of their life (Q07).

Next, correlations were introduced between errors e3 and e4 and between errors e4 and e5. Items 3 and 4 also relate to respondents' perception of becoming financially secure during the pandemic (item 3) and the perception that interviewees are securing their financial future (item 4). Respondents' perceptions of securing their financial future during the pandemic (item 4) are also associated with the possibility of achieving the financial goals they have set (item 5).

The next established correlation was between errors e10 and e12. Questions related to these errors concern people's perception that finances control their lives (item 10) and consequently, they cannot enjoy life because they worry too much about money (item 12).

Continuing with the adjustments, correlations were included between errors e3 and e5 and between e11 and e12, and e9 and e10. The relationship between the first two errors is that respondents' perception of "becoming financially secure during the pandemic" (item 3) is related to the possibility of achieving the financial goals they have set (item 5). On the other hand, errors e11 and e12 inquire about people's perception that finances control their lives during the pandemic (item 11), i.e., how much people perceive that they do not have control over their lives is related to their perception of being able to deal with unexpected events that may disrupt their financial control (item 12). Meanwhile, respondents' perception of their ability to keep their financial life in order during the pandemic (item 9) is associated with their perception that their finances control their life (item 10) during this period.

Finally, a correlation was included between errors e8 and e12. The relationship between these errors is based on the assumption that respondents who, due to their financial situation, will never have the things they want in life (item 8) also cannot enjoy life because they worry too much about financial matters (item 12), especially during pandemic periods.

Following these additions, the model exhibited appropriate fit indices, as demonstrated in the "Final Model" column of Table 10. The final measurement model of the construct loss of financial well-being and the correlations established to obtain the adjusted model are shown in Fig. 4.

For the financial ignorance construct, the individual components of the first-order model were initially validated, followed by the validation of the final measurement model of the construct. Therefore, Table 11 displays the fit statistics of the first-order components.

Table 11 Adjustment statistics for the first-order constructs—financial ignorance construct.

From the analysis, it is evident that the Decision Avoidance and Motivated Reasoning constructs are deemed adequate. Conversely, the Avoidance/Evasion of Information construct exhibited significant chi-square statistics, as the chi-square/degrees of freedom ratio exceeded 5 and the RMSEA value surpassed 0.06. Additionally, the Aggregation Bias construct had an AVE value lower than 0.5. Consequently, no adjustments were feasible to enhance these models, given that these constructs consist of only three variables. The unidimensionality and reliability of the constructs were verified in line with the established criteria in the literature.

Subsequent to the validation process, our aim was to evaluate the discriminant validity of the constructs by comparing the AVE with the correlations between the constructs, as evidenced in the correlations presented in Table 12.

Table 12 Discriminant validity for the decision avoidance, information avoidance, aggregation bias and motivated reasoning dimensions.

The correlation values between the constructs are lower than the square root of the AVE (represented by the diagonal values highlighted in italics), which indicates the discriminant validity of the constructs. Additionally, all the correlation values between the constructs were less than 0.85, underscoring that each construct represents distinct dimensions of financial ignorance.

Subsequently, the validation of the second-order model for the financial ignorance construct was conducted (Table 

Table 13 Adjustment statistics for the financial ignorance construct—initial model and final model.

13).

It is observed that the initial model proposed for measuring financial ignorance is inadequate. To obtain an adjusted measurement model, the same strategy presented earlier was employed, involving the insertion of correlations between the errors of the observed variables, which made theoretical sense. In this case, just two correlations were needed for the model to become adequate, as illustrated in Fig. 

Fig. 5

Initial and final measurement model of the financial ignorance scale with the respective standardized coefficients and the significance of each variable. Note(s): *** Significant at the 1% level. ¹z-value not calculated, where the parameter was set to 1, due to the requirements of the model. Source(s): Research results (2023)

5.

Following the inclusion of these correlations, the final measurement model demonstrated suitable fit indices, as depicted in Fig. 3. It is important to highlight that all relationships within the second-order model for the financial ignorance construct were statistically significant at the 1% level. Among these dimensions, Information Avoidance (0.890), Decision Avoidance (0.771), and Aggregation Bias (0.738) had the most substantial impact on financial ignorance.