Statistical learning models to measure the impact of COVID-19 on financial fragility (2024)

Box 1

Survey (English translation): main questions.

• D1-Gender: M; F; other/prefer not to respond

• D2-Age: in years

• D3-Family members: 1; 2; 3; 4; or more

• D4-Children: 0; 1; 2; 3; 4; or more

• D5-Co-habitants/Family structure status: I live alone; young couplewith no kids; elderly couple with no kids; I live with young children(<18 years); I live witholder children (>18years); I live with my parents; I live with my child's family; I live with other people

• D6-Marital status of the respondent: single; cohabitant; married; widow; separated;divorced

  See Also

  3.10 Access to mental health services

• D7-Postal code

• D8-Educational level of the respondent: Post-graduate degree; Bachelor degree or equivalent;Upper secondary school; Junior secondary school; Elementary; None

• D9-Is the house you live in currently living in ...? Property; Rented; Owned by third parties(non-cohabitants) in free use

• D9.1-Do you have a loan for the purchase of the residence? Yes; No

• D9.2-How much is the installment your family pays in a month for the mortgage or mortgages? Up to 200 euros per month; 201–400; 401–600; 601–600; 801; 1,000; 1,001–2,000; more than2,000 euros per month

• D10-currently in the family you are paying the installments of a loan (consumer credit, personal loan, salary-backed loan) [for all questions, answers: yes; no]

  • - D10.1-for the purchase of furnishings; furniture or household appliances (TV, washing machines, computers, ...)?

  • - D10.2-for the purchase of cars, motorcycles, other vehicles?

  • - D10.3-If you answered “yes” to one of the two previous questions, on average how much is the monthly payment that your family pays in a month for non-mortgage loans (consumer credit, personal loan, salary-backed loan)? Up to 200 euros permonth; 201–400; 401–600; 601–600; 801; 1,000; 1,001-2,000; more than 2,000Euros per month

• D11-Now let's refer to your situation before and after Covid. For your family, the monthly payment related to the previous application represented an expense/burden that you incurred with ... [March 2019–February 2020] & [March 2020–June 2021]: much ease; easily; some difficulty;difficulty; much difficulty

• D12-Have you requested a loan from banks or financial companies starting from March 2020 (start of lockdown measures)? Yes; no

  • - D12.1-Financing has been granted? Yes; no; partly

  • - D12.2-Which entity did you contact to request the loan? bank; consumer loanor other specialized lending institution; crowdfunding or peer-to-peer lendingplatform; other

  • - D12.3-Through what channel? branch; online; other

• D13-Since March 2020, someone in the family has benefited from income support benefits—e.g., Unemployment benefits, mobility, family unit allowances, etc. or financial support also in the form of services? Yes, by state / state institutions; Yes, by private or religiousinstitutions (foundations, private entities, companies, etc.); No, we didn't benefited fromit

• D14-Now let's refer to your situation before and after Covid. At that time, the monthly income allowed you to reach the end of the month with ...? [March 2019–February 2020] & [March 2020–June 2021] A lot of ease, manages to save enough; Ease, it also manages to savesomething; Some difficulties, we just manage to make ends meet; Difficulty, we had to dentsavings; Much difficulty, we had to ask for help

• D15-Now let's refer to your situation before and after Covid. In that period, would you have been able to incur an unexpected expense of 500 Euros? [March 2019–February 2020] & [March 2020–June 2021] Yes, very easily; Yes, with ease; Yes, but with difficulty; Yes, but with greatdifficulty; No

• D16-With reference to the period that started in March 2020 [for all questions, answers: yes, asin the past; yes, for the first time; no]

  • - did you have difficulty in supporting daily expenses (food, supermarket shopping)?

  • - have you had difficulty paying utility bills (electricity, gas, telephone, etc.)?

  • - have you had difficulty paying the rent or the home loan?

  • - have you had difficulty repaying the loan installments (other than the mortgage if present)?

  • - did you have to forgo a specialist medical examination or therapeutic treatment for economic reasons?

  • - have you had to renounce for economic reasons to take out non-compulsory insurance policies (e.g., accidents, illness, and life/death cases)?

• D17-With reference to the period that started in March 2020, have you postponed investments or expenses already planned for economic reasons? yes, car or motorbike purchase; yes, mainhouse purchase; yes, other real estate purchase; Yes, different or non-essential goods orservices purchase (TV, computer, holidays, sports, personal training courses ...);No

• D18-Since March 2020, have you lost your job (or any of your cohabiting family members lost it)? Yes; no; prefers not to answer

  • - D18.1-If you answered yes to the previous question, the loss of income due to the dismissal has had an impact: very high; high; medium; relatively small; small

• D19-Since March 2020, has an illness involved you or your family members leading to an unexpected reduction in income or an unexpected increase in family expenses? yes;no

• D20-Since March 2020, has a death involved you or your family members resulting in an unexpected reduction in income or an unexpected increase in household expenses? yes;no

• D21-Current profession of the respondent: student; Houseperson; Retired; Unemployed;Manager, officer, middle manager; Office worker, teacher, military; Worker, clerk, apprentice;Entrepreneur, freelancer; Merchant, craftsman, farmer; Atypical work (occasional collaboration, project work, etc.)

  • - D21.1-If unemployed: what activities did you carry out before being unemployed? Student; Houseperson; Retired; Manager, officer, middle manager; Office worker, teacher, military; Worker, clerk, apprentice; Entrepreneur, freelancer; Merchant, craftsman, farmer; Atypical work (occasional collaboration, project work, etc.)

• D22-Approximately what range does your household's total monthly net income fall into? up to 600 Euros; 601–1,200 Euros; 1,201–2,000 Euros; 2,001–3,000 Euros;3,001–4,000 Euros; 4,001–5,000 Euros; more than 5,000 Euros; don't know; prefers not toanswer

• D23-I use financial instruments and products [for all questions, answers: yes; no]

  • - D23.1-Do you own a current account?

  • - D23.2-Do you own at least one credit card?

  • - D23.3-Do you own any government bonds or other bonds?

  • - D23.4-Do you own shares or funds/SICAVs?

• D24-Imagine that you have a debt of 1,000 Euros on your credit card and that the interest rate applied is 20% (compound annual rate). If you pay nothing, at this interest rate, how many years would it take to double the amount owed? 2 years; < 5 years; 5–10 years; don'tknow

• D25-You have a debt of 3,000 Euros for the use of your credit card. Pay an installment of 30 Euros each month. At an annual rate of 12% (or 1% monthly), how many years would it take to eliminate credit card debt if you don't make any further purchases? lessthan 5 years; 5–10 years; 10–15 years; Never, will you continue to be in debt; don'tknow

• D26-You buy a product that costs 1,000 euros. To pay, you have the following two options: (a) pay 12 monthly installments, each of 100 Euros; (b) borrow at an annual interest rate of 20% and pay back 1,200 euros in a year. Which option is more advantageous? option a; option b;indifferent; don't know

• D27-On a scale of 1 to 5, where 1 means very low and 5 means very high, how would you rate your knowledge of financial topics? 1(Very low); 2; 3; 4; 5(Very high)

• D28-On a scale of 1 to 5, how well do you feel you are able to make good decisions? 1(not goodat all); 2; 3; 4; 5(very good)

E(θi|X,Y)=∑k=1KE(θj|gk,Y)p(gk|Y),

(1)

4.1 Data

The survey was conducted between June 2021 and June 2022, and 58 valid responses were collected during this period¹. The gender distribution of the respondents is relatively equal, with 52% identifying as male, 47% as female, and one respondent not providing their gender.

The majority of respondents are under 30 years old. Answers are based on family decisions, so age is not considered a relevant variable. The surveyed households mostly consist of families with four members. Around 67% of households consist of three or more members. Approximately 21% of households are comprised of three members, 5% of households have two members, and the remaining 7% are single-member households. In terms of assets, most households own their homes, with around one-third having an ongoing mortgage. The monthly installment amount varies, but the majority pay less than 600 euros per month. A few families also have consumer or personal loans.

4.2 Results of the survey

The COVID-19 outbreak has affected households' ability to manage expenses and make mortgage and loan payments. Figure 1 illustrates that during the COVID-19 period, the number of families experiencing difficulty paying their monthly installments has increased. Specifically, there has been an increase in the number of families finding it “somewhat difficult” and an increase from 0 to 8 families finding it “difficult.” The families at the extremes of the distribution, whether very easy or very difficult, did not experience any change (see Figure 1).

When assessing whether families can make ends meet, we observe an increase in the number of families experiencing financial difficulties and a corresponding decrease in those able to save (Figure 2). Similarly, the number of families capable of comfortably handling an unexpected expense of 500 euros decreased after the COVID outbreak. Some families were forced to forgo purchases related to daily necessities such as food (6), utilities (10), health or medical expenses (7), or insurance (6). Approximately 30% have postponed non-essential expenses such as TV, entertainment, or car/motorbike purchases.

4.3 Financial vulnerability factors

A statistical analysis was conducted to elucidate the factors influencing financial vulnerability during the pandemic. The study specifically investigates the potential roles of education level and financial literacy in mitigating the economic impact of pandemics on households.

The questionnaire answers are categorical variables from a statistical perspective. Some indicators of financial distress include the ease of making ends meet, while predictive variables can include self-evaluation of financial knowledge. The definition, type and source of the variables used in this empirical study are reported in Table 1.

We, thus, apply the methodology presented in Section 3 to combine the available information on household features and behaviors in a “financial vulnerability” index, which should reflect the household's exposure to the potential financial impact of pandemics. The potential weakness factors that we consider are related to low educational level, low financial literacy and low confidence with financial instruments. Specifically, starting from the variables described in Table 1, we build the following binary indicators:

• binarized education level: a binary variable which is equal to 1 if the respondent has an education level lower than degree, 0 otherwise;

• binarized financial literacy: a binary variable which is equal to 1 if the respondent has given less < 2 correct answers, 0 otherwise;

• binarized confidence with financial instruments: a binary variable which is equal to 1 if the respondent holds no more than one instrument among credit cards/stocks/government or other bonds, 0 if the held instruments are more than one.

Then, using the variable “Capability of affording an unexpected expense” (see Table 1), we define our target variable as a binary one, which is set equal to 1 if the respondent had a worsening in the capability of affording unexpected expenses with respect to the pre-pandemic period. Based on the collected data, nearly 34% of respondents had a worsening in the spending capacity due to the pandemic.

For each of the three predictors, using the methodology presented in Section 3, we calculate a weight, based on the variable's accuracy in predicting the target, i.e., the possible deterioration in the respondent's spending capacity. The estimated weights are reported in Table 2.

The weights are then used to calculate the financial vulnerability index. For example, if an individual has a score of 0 in “educational level,” 1 in “financial literacy” and 1 in “confidence with financial instruments,” the financial vulnerability index will be calculated as: 0*0.22 + 1*0.23 + 1*0.55 = 0.78. Preliminary analyses—not shown here for brevity—revealed a positive relationship between each of the three dummy variables and the target variable. Thus, the individual will be considered as more financially vulnerable than, for instance, an individual which was assigned an index of 0.7. As explained in Section 3, the index can be interpreted as a superscore combining different aspects related to financial fragility.

Similar results (with weights equal to 21%, 21%, and 58%) were obtained by using, as the target variable, deterioration in the ease of reaching the end of the month, a binary variable set equal to 1 if there was a worsening in the “Capability of making ends meet” variable (see Table 1). Indeed, nearly 31% of respondents declared a worsening in the capability of making ends meet in the pandemic and post-pandemic period.

¹To preserve the privacy of respondents, data are anonymously collected and are available upon request from the author only at an aggregated level.

• Agosto A., Cerchiello P., Giudici P. (2023a). Bayesian learning models to measure the relative impact of esg factors on credit ratings. Int. J. Data Sci. Anal. 2023, 1–12. 10.1007/s41060-023-00405-9 [CrossRef] [Google Scholar]
• Agosto A., Giudici P., Tanda A. (2023b). How to combine ESG scores? A proposal based on credit rating prediction. Corpor. Soc. Respons. Environ. Manag. 30, 3222–3230. 10.1002/csr.2548 [CrossRef] [Google Scholar]
• Albert C., Caggese A., González B., Martin-Sanchez V. (2023). Income inequality and entrepreneurship: lessons from the 2020 COVID-19 recession. J. Bank. Finan. 149:106779. 10.1016/j.jbankfin.2023.106779 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Baker S. R., Farrokhnia R. A., Meyer S., Pagel M., Yannelis C. (2020). How does household spending respond to an epidemic? Consumption during the 2020 COVID-19 pandemic. Rev. Asset Pric. Stud. 10, 834–862. 10.3386/w26949 [CrossRef] [Google Scholar]
• Banca d'Italia (2021). The 2020 Annual Report at a Glance. Annual Report for 2020. [Google Scholar]
• Cerchiello P., Giudici P. (2014). Bayesian credit ratings. Commun. Stat. Theory Methods43, 867–878. 10.1080/03610926.2013.810264 [CrossRef] [Google Scholar]
• Giudici P., Mezzetti M., Muliere P. (2003). Mixtures of products of Dirichlet processes for variable selection in survival analysis. J. Stat. Plan. inference111, 101–115. 10.1016/S0378-3758(02)00291-4 [CrossRef] [Google Scholar]
• Jia H., Fan S., Xia M. (2022). Impact of the COVID-19 pandemic on household financial asset allocation: a China population study. Front. Psychol. 13:4871. 10.3389/fpsyg.2022.990610 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Jung H., Kim J. H., Hong G. (2023). Impacts of the COVID-19 crisis on single-person households in South Korea. J. Asian Econ. 84:101557. 10.1016/j.asieco.2022.101557 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Kansiime M. K., Tambo J. A., Mugambi I., Bundi M., Kara A., Owuor C. (2021). COVID-19 implications on household income and food security in Kenya and Uganda: findings from a rapid assessment. World Dev. 137:105199. 10.1016/j.worlddev.2020.105199 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Lecoutere E., van den Berg M., de Brauw A. (2023). Changes in women's empowerment in the household, women's diet diversity, and their relationship against the background of COVID-19 in Southern Bangladesh. J. Asian Econ. 84:101559. 10.1016/j.asieco.2022.101559 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Mallawaarachchi T., Rahut D. B. (2023). Realising rural economic transformation: pathways to inclusive and sustainable prosperity in post-COVID-19 Asia. Econ. Anal. Policy77, 1076–1082. 10.1016/j.eap.2023.01.009 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Padhan R., Prabheesh K. (2021). The economics of COVID-19 pandemic: a survey. Econ. Anal. Policy70, 220–237. 10.1016/j.eap.2021.02.012 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Rathnayaka I. W., Khanam R., Rahman M. M. (2023). The economics of COVID-19: a systematic literature review. J. Econ. Stud. 50, 49–72. 10.1108/JES-05-2022-0257 [CrossRef] [Google Scholar]
• Yazdanparast A., Alhenawi Y. (2022). Impact of COVID-19 pandemic on household financial decisions: a consumer vulnerability perspective. J. Cons. Behav. 21, 806–827. 10.1002/cb.2038 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Yue P., Gizem Korkmaz A., Zhou H. (2020). Household financial decision making amidst the COVID-19 pandemic. Emer. Markets Fin. Trade56, 2363–2377. 10.1080/1540496X.2020.1784717 [CrossRef] [Google Scholar]