Young people are not blameworthy: the generation’s awareness of COVID-19 and behavioral responses

At a time when the COVID-19 pandemic has been ongoing for more than a year, young people have been the subject of vigilant scrutiny and criticism regarding their active engagement in social activities. We posed the question of whether young people's response to COVID-19 was different from that of other generations and analyzed awareness and behavior to investigate this question. Specifically, we examined internet searches for information on COVID-19 and credit card consumption in South Korea among young people in their 20s and compared them to a reference group of people in their 50s. Our research has confirmed that there was no statistically significant difference between young people and the reference group in this regard. Furthermore, in the 25 sub-sectors of industry we examined, young people's consumption activities recovered significantly faster than the reference group in only three sub-sectors. This study demonstrated that young people showed stronger interest than the reference group in their response to COVID-19, and that they cooperated with the government’s social distancing policy by reducing their activities. Through this study, we presented a scientific approach for evaluating young people in regard to their response to COVID-19, offering useful implications for designing appropriate policies for public health.

shows the cumulative confirmed cases and presents the search population composition ratio, to explain differences by age in the total search volume. Table T1 also provides data that suggests the economic hardships young people experienced during the COVID-19 period. Compared to January 2020, the share of the economically active population in all age groups had decreased as of January 2021. However, whereas in the reference group the share decreased by only 1.8% compared to the previous year (1.4%p), it decreased among young people by 5.4% (3.4%p), from 62.8% to 59.4%, confirming that the younger a person was, the greater the impact of the pandemic was on economic activities. 4  TABLE T2. DESCRIPTIVE STATISTICS   Table T2 shows descriptive statistics for the raw data on detailed variables related to the 3 variables  explained in Table T2.  Figure F1 shows the similarity in the trends of COVID-19 searches and new confirmed cases, and three distinct waves of surges in COVID-19 infections. It can be seen that as we approach the third wave, the search volume relatively decreases. For reference, vaccination in South Korea began on February 26, 2021, and therefore we judged that it would have had little effect on trends during the period targeted in this study.   Figure F2 for each group is based on a comparison of credit card use cases in the 4th quarter of 2019, before the outbreak of COVID-19, and in the 4th quarter of 2020, when the third wave of COVID-19 was surging in South Korea: the value was obtained by dividing the latter by the former. This value, which we refer to as "quarterly recovery," indicates the degree to which consumption activity recovered compared to 1 year ago.
As shown in Figure F2, the average quarterly recovery of credit card use among young people was lower than that of the reference group, and this was common across all five sectors. Looking at the differences across sectors, we see that in ICT there was nearly a full recovery to the level of credit card use that existed before COVID-19 (96.7%), while the recovery in the Recreation and Leisure sectors lingered at 56.9% and 60.0%, indicating that the aftereffects of COVID-19 still loomed large. Analyzed by sub-sector, the data shows that within the Recreation sector, movies showed the lowest quarterly recovery at 14%, while in the ICT sector, home appliances and media had a quarterly recovery far exceeding 100%, as consumption increased despite the spread of COVID-19.
Despite the fact that the reference group overall had higher quarterly recovery, there were subsectors in which young people's recovery was higher by a significant margin (a margin of 10% or more), as follows. In the Leisure sector, young people had distinctly higher quarterly recovery in tourist transportation and accommodation. In each of the other sectors, there was only one sub-sector in which young people's quarterly recovery was higher by margin of 10% or more than the previous year: entertainment under Recreation sector, media under ICT sector, public transportation under Daily Living sector and bars (pubs) under Restaurant/Fashion sector.
The quarterly recovery data discussed above helped us to intuitively judge the scale of recovery in credit card use by age group before and after the COVID-19 outbreak. However, it was difficult to draw conclusions with strong statistical significance, and another limitation was that changes in periods other than the 4th quarter could not be tracked. Therefore, in Results Section, we performed statistical comparisons of monthly trends during the COVID-19 pandemic to identify sub-sectors exhibiting significant differences. Figure F2. Comparison of quarterly recovery of credit card use before and after the COVID-19 outbreak by sub-sector

FIGURE F3. MONTHLY CREDIT CARD USAGE COMPARISON BY AGE GROUP
As explained in Method Section, we used the term "credit card use intensity" to refer to the standardized value obtained by dividing the monthly credit card use cases in 2020 in each age group by the average value in the 4th quarter of 2019 (before COVID-19) in each age group; the monthly flow of this value is presented in Figure F3. Figure F3. Credit card use intensity comparison by sector and age group Here, the relatively high credit card use intensity does not mean that one group used credit cards more than the other groups, because this value was standardized to a different value for each group; what it means is that the group's use intensity was higher than that of the other groups compared to the 4th quarter just before the COVID-19 outbreak. Therefore, the data indicates that in the Leisure sector, young people often showed a relatively faster recovery in consumption than the reference group. In the Recreation sector, the recovery of the reference group was higher or similar to that of young people in most of the periods, and in the remaining three sectors (ICT, Daily Living, Restaurant/Fashion), the recovery of the reference group was significantly higher than that of young people. Looking at the flow of all groups, we noted that the ICT and Daily Living sectors did not show any great change, while the remaining three sectors (Leisure, Recreation, Restaurant/Fashion) showed a slight decrease overall. Another common feature in these three sectors was the sharp and temporary decline in September: beginning on August 19th, the level of social distancing required in the Seoul metropolitan area was raised (Level 2), and finally, on August 30, a reinforced social distancing measure (Level 2.5) was implemented. 8 These events occurred in the second wave section shown in Table 2, and the peak of new confirmed cases was August 26th. As the peak of the new confirmed cases passed, the social distancing level was de-escalated on September 14th and extended to the September 27th (Level 2). 9 The impact of this series of measures is reflected clearly in the three sectors. In Figure F3, there is also one trend that is common in all sectors, although there is a difference in degree: in all sectors, credit card usage increased in May compared to April (this can be regarded as a temporary phenomenon if we exclude Leisure sector). May was the period in which the National Disaster Relief Fund, which was issued only once in South Korea, began to be distributed. 10 Figure F4-1 presents a comparison of the intensity of monthly credit card use by sub-sectors within the Leisure sector and age group. Other than golf courses, these sub-sectors exhibited poor performance with a generally decreasing trend, confirming that this is an industry that has been hit relatively hard by COVID-19. The data demonstrated that the relative prospering of golf courses was led by the reference group, and in tourist accommodations, young people reduced their consumption relatively less despite the depression. impacted. There was a sudden rise in credit card usage in the hobbies sector in May, when the South Korean government issued Disaster Relief Funds to the public, and this confirms that the hobbies sector was a beneficiary of that funding. Notably, this indicates that young people spent more of the relief fund in the hobbies sector. In addition, Figure F4-2 shows that there were significant differences in the entertainment and cultural facilities sectors, which were supported by consumption by different age groups. Figure F4-2. Credit card use intensity comparison by age group (sub-sectors of Recreation sector) Figure F4-3 compares the intensity of monthly credit card use by sub-sectors in the ICT sector and by age group. Unlike in other sectors, the impact of COVID-19 is barely noticeable as there was no change in credit card usage trends, and sub-sectors such as home appliances and media even showed an increase. In particular, home appliances and mobiles experienced a temporary increase due to the Disaster Relief Fund in May. Analyzing by age group, we found that the reference group led the strong activity in sub-sectors within the ICT sector. Figure F4-3. Credit card use intensity comparison by age group (sub-sectors of ICT sector) Figure F4-4 compares the intensity of monthly credit card use by sub-sectors within the Daily Living sector and by age group. With the exception of gas stations and public transportation, overall, the level of credit card usage showed a tendency to remain steady or rise slightly, though initially there had been slight decreases in usage. This may be because the Daily Living sector has the character of industries essential for survival, while the depression experienced by gas stations and public transportation may have been caused by social distancing and working from home. In the Daily Living sector, the food and interior design sub-sectors received a temporary boost from the Disaster Relief Fund in May, and the reference group used relatively more of this fund on interior design. Analyzing by group, we found that the reference group led the sub-sectors that had an upward trend and by contrast, young people were not able to cut their spending in public transportation, a sector which showed a downtrend.   To perform a Bayesian paired sample test, one must first define a prior distribution (probability) for hypothesis testing. 11 In this study, a normal distribution approximated to the monthly distribution of the sub-sectors in each sector was used as the prior distribution. Table T4-1 shows the descriptive statistics of the differences in paired samples for each sector and also presents the result of approximating this to 100,000 counterfactual normal distributions by simulation. We assumed the normal distribution (mean and variance) of the sample approximated by the simulation presented in Table T4-1 to be the prior distribution and proceeded to perform Bayesian inference.  Table T4-2 shows the results of the Bayesian paired sample test. These results can be regarded as indicators of whether there were any significant differences after taking account of the distribution reflecting the possibility that credit card recovery may be slower in young people than in the reference group. In Table 5 and Table T4-1, it should be noted that in the Leisure sector, the recovery of young people was in fact quicker, and this trend was reflected in our results. There were 8 sub-sectors for which we determined there was substantial evidence because the Bayesian inference results were significant and the Bayesian factor was also less than 1/3. 12 According to Bayesian inference, media and bars (pubs) in addition to tourist accommodation, entertainment, and public transportation were sub-sectors with higher credit card consumption recovery among young people compared to the overall trend of the sector, while golf courses, food, and cafes were sub-sectors in which the reference group was significantly higher. Note: *: p < 0.05, **: p < 0.01, ***: p < 0.001.