KEY MESSAGES THE LONG-LASTING IMPACTS OF COVID-19 The long-lasting impacts of COVID-19 3 • Peru has been one of the countries hardest hit by the COVID-19 pandemic in the last two years. According to the Ministry of Health, between 2020 and 2021, 3.5 million Peruvians were infected with COVID-19. and more than 213,000 died from the disease. By July 2022, Peru had experienced more than 6,000 deaths per million population due to COVID-19, placing it as the country with the highest or second-highest number of deaths per capita in the world. • To prevent the spread of the virus, the government implemented a strict lockdown at the beginning of the pandemic. The strict quarantine measures hindered people from engaging in economic and social activities. As a result, the country suffered significant social and economic losses. • This chapter explores two understudied topics related to the impacts of COVID-19 on the population. First, it develops an incidence analysis of COVID-19 deaths to understand whether richer or poorer people were more affected. Then, it provides an analysis of the potential long-term impacts of COVID-19 on economic and social outcomes, which could serve as a roadmap among policy makers in seeking to prevent scarring effects. • The incidence analysis of COVID-19 deaths is performed by matching national registries of excess deaths and survey data. The findings show that income did not protect Peruvians from COVID-19 deaths, while age and region of residence (departamentos) explained most of the variation in mortality rates. In most regions, urban populations ages more than 65 showed no difference in mortality rates between the lowest expenditure quintile and the highest expenditure quintile. The difference in mortality rates varied more across regions. Thus, the excess deaths in regions such as Lambayeque and Piura were twice those in Cajamarca and Huánuco and four times those in Apurimac and Huancavelica, irrespective of income levels. • The immediate impacts of the pandemic revealed the fragility of the social gains. The COVID-19 shock erased a decade of social progress, as seen by the sharp increase in poverty. Employment was greatly affected, and, although they have almost recovered, the job quality and income levels still lag. • The pandemic affected the process of human capital accumulation. An estimated 1.7 learning-adjusted years of schooling (LAYS) have been lost during the pandemic. In the absence of corrective policies, this generation of students will experience lower levels of human capital and lower future incomes. • The loss in income during the pandemic caused families to experience food insecurity, which has long-term implications in malnutrition and stunting, which may translate into losses in productivity and still lower incomes. 4 The long-lasting impacts of COVID-19 • COVID-19 deaths will also have an effect on the families left behind by the deceased. Estimates derived from national registries show an orphan excess in Peru of 43,500 children in 2020–21 relative to 2018–19. Evidence from past epidemics shows that the death of a parent can cause serious harm to the long-term health and development of the children. • The lower labor market outcomes experienced by youth during the pandemic will be a serious problem in the future, as greater exposure to unemployment is likely to affect long-term labor market fortunes. • The short- and long-term effects of the pandemic call for urgent actions to prevent scarring effects, but also to better prepare for future shocks. The long-lasting impacts of COVID-19 5 2.1. The incidence of COVID-19 in Peru Peru has been one of the countries hardest hit by the COVID-19 pandemic in the last two years. According to the Ministry of Health, between 2020 and 2021, 3.5 million Peruvians were infected with COVID-19, and more than 213,000 died from the disease. By July 2022, Peru had experienced more than 6,000 deaths per million due to COVID-19, placing it as the country with the highest or second- highest deaths per capita in the world.1 Within Latin America, it was followed by Brazil, which showed half that rate, that is, around 3,000 deaths per million were recorded due to COVID-19. To prevent the spread of the virus, the government implemented a strict lockdown at the beginning of the pandemic. The lockdown was initially announced as a seven-day policy, but continued to be extended for a total of 54 consecutive days from mid-March to mid-May. Subsequently, there was a gradual relaxation of restrictions in all regions (departamentos) until July 2020, except in Ancash, Arequipa, Huánuco, Ica, Junín, and San Martín, where the quarantine ended in August. From there, a moderate quarantine was imposed, mainly in the form of curfews that limited the hours people could move freely. At the end of January 2021, when the third wave of COVID-19 hit the country, the strict lockdowns were again introduced for a total of 31 days. Table 1 summarizes the quarantine measures. Table 1. Quarentine measures and length by resolution Length Measures Start Resolution End Resolution (number of days) Strict quarantine 3/16/2020 DS Nº044-2020 7/1/2020 DS Nº116-2020 107 Strict quarantine for some regions 1/ 7/1/2020 DS Nº116-2020 8/12/2020 DS Nº139-2020 42 Moderate quarantine 8/12/2020 DS Nº139-2020 1/29/2022 DS Nº010-2022 535 Strict quarantine for some regions 1/ 1/27/2021 DS Nº008-2021 2/27/2021 DS Nº036-2021 31 1/ Regions: Arequipa, Ica, Junín, Huánuco, San Martín, Madre de Dios y Áncash 2/ Regions: Lima, Provincia Constitucional del Callao, Áncash, Pasco, Huánuco, Junín, Huancavelica and Ica. From February 14th of 2021 onwards following regions were added: Utcubamba, Arequipa, Santa, Abancay, Huamanga, Cutervo, Canchis, y la Convención, Maynas, Ramón Castilla, Ilo, Puno y Tacna Peru’s strict quarantine measures removed people from economic and social activities, and mobility returned to pre-pandemic levels in mid-2022. According to Google Mobility Trends, activities in retail and recreation and the number of people at public transport stations dropped by more than 80 percent the day after the strict lockdown was imposed, compared with the median during the five weeks from January 3 to February 6, 2020. Activity in parks and workplaces fell by 72 percent, and visits to supermarkets and pharmacies fell by 60 percent after the strict lockdown was imposed. Meanwhile, the time spent at home had increased by almost 40 percent after the start of the lockdown (Figure 1). The lockdown and mobility limits were the strictest in the region. Although most mobility had returned to normal by the end of 2020, the third wave of COVID-19 (followed by the return of strict quarantine) pushed back the recovery in activities. Once again, most of the mobility in these activities took a full year to recover. The exception were grocery stores and pharmacies, which had recovered their clientele by July 2021, and public transport stations and workplaces, which had not yet done so by mid-2022. 1. According to Our World in Data, Peru ranks first among the countries with the highest number of deaths per million inhabitants. According to recent estimates in Lancet, Peru ranks second after Bolivia. 6 The long-lasting impacts of COVID-19 60 Figure 1. Changes in mobility by category (%), 2020–22 40 20 0 -20 -40 -60 -80 -100 Retail and recreation Grocery and pharmacy Residential Transit Parks Workplaces Source: Elaboration based on Google Mobility Trends, 2022. The country also suffered significant social While the short-term impacts of the COVID-19 and economic losses because of the pandemic. pandemic are well documented, the long- Peruvian schools closed for a total of 360 school term consequences of these losses are days, or the equivalent of 3.3 semesters or 1.8 unknown. Losses in human lives are expected academic years.2 These numbers compare to have left households without a key source of poorly with the global average of 307 days income (in the form of labor income, pensions, closed. The loss of learning-adjusted years of or cash transfers) and children without a main schooling (LAYS) in Peru due to COVID-19 has or secondary caregiver. Additionally, the loss of been estimated at 1.7 years per pupil, whereas the a family member can be exceptionally traumatic average loss estimated for the region is 1.5 LAYS.3 and have serious consequences for surviving As a result of social distancing, lockdowns, family members.4 Understanding the magnitude reduced mobility, and substantial educational and distribution of COVID deaths across and health impacts, domestic and foreign households will be helpful in assessing the demand contracted. The Peruvian economy potential long-term distributional implications suffered one of the deepest recessions in the of COVID and in designing mitigation policies. world, contracting 11 percent in 2020. Services and industry showed the largest drop in gross This chapter explores the incidence and domestic product (GDP) with growth rates of long-term effects of COVID-19 deaths on negative 10.7 percent and negative 13.3 percent. the well-being of surviving households. Efforts to identify COVID-related deaths based on individual characteristics have been quite limited. Most statistics present mortality at the level of aggregate variables, such as age or sex. The main contribution of this chapter therefore lies in shedding light on the 2. The closure of schools stated in mid-March, 2020, and was announced in the Resolución Viceministerial No. 079-2020. The reopening of schools occurred in March, 2022, which was announced in the Fuente: Resolución Ministerial No. 531-2021. 3. Estimates of World Bank (2022) “Two Years After”, based on Azevedo et al. (2022). 4. See Verdery et al. (2020), PNAS. The long-lasting impacts of COVID-19 7 incidence of COVID-19 deaths along the income The variables reported in SINADEF, along distribution by combining several datasets on with other socioeconomic characteristics, the socioeconomic profiles of people who have including household income, can be found died from COVID-19. This is important for policy in the National Household Survey (Encuesta makers because the results point to a need to Nacional de Hogares, ENAHO). Following a improve social protection and the quality of methodology thoroughly explained in Appendix services among the entire population to prepare A, both datasets are combined to estimate the for future shocks. Moreover, this chapter also incidence of deaths along the household income contributes by highlighting the main challenges distribution. The results follow. Peru will face in the future as a result of COVID-19 deaths and the associated economic impact. COVID-19 affected disproportionately the The findings can be used as a roadmap for elderly in urban areas. Death rates among the policy makers to guide medium-term plans and urban population more than doubled in 2020 and investments. The rest of the chapter is organized 2021 relative to 2019 (from 0.35 percent to 0.76 as follows. Section 2.2 presents results on who percent). As has been shown, COVID-19 spreads was most affected by COVID-19, and section mainly between people who are in close contact, 2.3 and 2.4 explores the potential short and which explains why urban areas have a higher long-term impacts of COVID-19, respectively. incidence of deaths. Mortality rates in rural and Section 2.5 conclude with recommendations. semirural areas increased by a half or a third of the rate in urban areas in 2020.5 In semiurban 2.2. Who was most impacted by areas, mortality rates went from 0.30 percent in COVID-19 deaths? 2019 to 0.46 percent in 2020, and, in rural areas, they went from 0.33 percent to 0.44 percent. To calculate the distributional incidence of the However, by 2021, both rural and semirural areas impacts of COVID-19 (to determine whether had almost doubled the death rates as well, the rich or the poor are the most impacted), suggesting that the disease took longer to reach it has been necessary to combine datasets the more remote areas of the country. Still, a higher from various sources. Deaths are reported rate of mortality persisted in urban areas. In 2021, in a national registry, the Sistema Informático urban areas registered excess deaths of 116,00 Nacional de Defunciones (SINADEF), along people in a population of 25.7 million, or 4,500 with socioeconomic characteristics such excess deaths per million (Figure 2). Rural areas as date of death, sex, place of residence, registered 10,000 excess deaths in 4.4 million in and education level. However, there is no 2021, or 2,400 excess deaths per million (Figure 2). information on the income or expenditure of the household of the deceased: crucial variables in calculating distributional incidence. 5. Urban, semiurban, and rural areas are defined following the district typology of the National Statistics and Informatics institute (INEI), the Sistema de Datos Micro Regionales. Urban areas are located in the national metropolis, in regional metropolises, or areas with some type of city or populated center with more than 2,000 inhabitants. Rural areas are populated centers with 2,000 or fewer people with different degrees of access to the district capitals. And semiurban areas, an intermediate category between the two, corresponds to a type of district with a minority of the population settled in populated centers of more than 2,000 inhabitants. A detailed explanation of the typology can be found at Sistema de Datos Micro Regionales (2020), https://sdmr.inei.gob.pe/cms/multimedia/home/menuSect-2-78. 8 The long-lasting impacts of COVID-19 Peruvians ages 40–65 were three times more Figure 3. Excess deaths per million, by age-group likely to die during the pandemic than in a 21,231 normal year, and Peruvians ages more than 65 21,000 19,362 were two times more likely to die during the 18,000 pandemic. In 2019, the mortality rate among those 15,000 ages 40–65 was 0.26 percent, while the mortality 12,000 9,000 rate among elderly individuals was 1.95 percent. In 6,000 5,069 2021, the mortality rate among those ages 40–65 3,849 3,000 rose to 0.77 percent, and among those ages 65 or - more rose to 4.07 percent. To contextualize these 40 - 65 years old 65+ years old numbers, 24,000 deaths were reported in the 2020 2021 40–65 age-group before the pandemic, while, in 2021, close to 72,000 deaths were reported in the Source: Elaboration based on SINADEF and ENAHO data. same group. This implies 48,000 excess deaths in a population of 9.3 million or 5,000 excess deaths According to these numbers, in Peru, the per million (Figure 3). Among the 65+ age-group, second year of the pandemic (2021)—which 72,000 deaths were reported annually, before coincided with the third wave of COVID-19 the pandemic, but, in 2021, there were a total cases—was deadlier than the first. In 2021, of 151,000 deaths reported. This implies 79,000 22,000 more Peruvians died relative to the excess deaths in a population of 3.7 million or number in 2020. At the peak of the pandemic, 21,000 excess deaths per million (Figure 3). 1,273 deaths were reported in one day, the highest number in the world. This level of devastation Figure 2. Excess deaths per million, by area during the second year of the pandemic was not 5,000 the norm in all regions. Daily excess deaths in 4,530 3,989 2020 and 2021 can be seen in Figure 4, where 4,000 daily deaths in 2020 and 2021 (red lines) are 3,000 compared to daily deaths in 2019 (blue lines). 2,353 2,406 The difference between the red and the blue 2,000 1,602 1,124 lines is a measure of excess deaths per day. 1,000 Note that the beginning of 2020 does not report - differences in daily death compared with 2019, Urban Semi-urban Rural as the virus had not yet arrived to Peru. The first 2020 2021 COVID-19 death was reported in March 2020. Source: Elaboration based on SINADEF and ENAHO data. Note: Area definitions are derived from the classification of the National Statistics and Informatics institute (INEI) of urban (A), rural (B), and semiurban (AB) areas at the district level. Urban districts are located in the national metropolis, in regional metropolises, or areas with some type of city or populated center with more than 2,000 inhabitants; rural districts are populated centers with 2,000 or fewer people with different degrees of access to the district capitals; and semiurban areas, an inter- mediate category between the two, correspond to a type of district with a minority of the local population settled in populated centers of more than 2,000 inhabitants. A detailed explanation of the definition can be found in Sistema de Datos Micro Regionales (2020). The long-lasting impacts of COVID-19 9 Figure 4. Excess deaths in 2020 and 2021 (compared with 2019) 1400 2020 2021 1200 1000 Daily deaths 800 600 400 1097 deaths 1273 deaths 200 0 Pandemic (2020 - 2021) Pre-pandemic (2019) Source: Elaboration based on SINADEF. Note: The figure compares 2019 daily deaths reported by SINADEF (blue lines) with 2021 and 2022 daily deaths reported by SINA- DEF (red lines). The excess deaths are the differences between the red and blue lines. Age and area of residence explained most of Figure 5. ANOVA Analysis the variation in mortality rates. An analysis of 21.5% 100% variance (ANOVA) in mortality rates by individual 90% characteristics shows that age and the region 80% 7.4% 0.1% of residence are the main determinants of the 70% 57.5% 0.6% probability of dying from COVID-19. Figure 5 60% shows the decomposition of the variance in excess 50% 40% mortality by variables. Almost 60 percent of the 30% variance is explained by the age of 65 or more. 20% Another 7 percent is explained by the regions 10% 0.7% 0.6% where individuals live. Education and sex had 0% little explanatory power. Household income did Residual 40-64 yo High skilled Region Income Male 65+ not explain any of the variations in mortality rates. Source: Elaboration based on INEI-ENAHO and SINADEF. Mortality rates are homogeneous throughout the income distribution; they vary mainly by region of residence. Initially, a comparison of excess mortality between poor and nonpoor groups seems to indicate that excess mortality is greater among the highest deciles in Peru. 10 The long-lasting impacts of COVID-19 This reflects the fact that the richest population Figure 6. Excess deaths per thousand among 65+ in Peru lives in urban areas and is the longest population by region and expenditure quintiles living. Both characteristics (urban and age) made Elderly and urban Q1 Elderly and urban Q5 them more at risk of dying. On average, among 40 the highest expenditure deciles, 17.3 percent of 35 30 people were ages 65 or more, and 98.7 percent 25 were among the urban population. Meanwhile, 20 the lowest spending decile is represented by only 15 9.9 percent of people ages 65 or more and only 10 28 percent of the urban population. However, 5 0 after controlling for age-group and region of Cajamarca Huancavelica San Martin Lima Moquegua La Libertad Arequipa Tacna Ucayali Ancash Piura Madre de Dios Junin Pasco Tumbes Huanuco Lambayeque Loreto Cuzco Puno Amazonas Ayacucho Callao Apurimac Ica*** residence, income level is no longer significant. For the urban population ages more than 65, the difference in mortality rates between the highest expenditure quintile and the lowest expenditure Source: Elaboration based on SINADEF and ENAHO data. Note: Expenditure quintiles are defined for the entire population from quintile is not significant. In most regions, the ENAHO. Lambayeque is not considered in the sample because no deaths were registered in Lambayeque among the first quintile of urban population ages 65 or more showed no expenditure. In Ica, the comparison is between the second and the difference in mortality rates between the lowest fifth quintiles because there are no individuals in the lowest quintile of expenditure. Area definition comes from INEI’s classification of urban expenditure quintile and the highest expenditure (A), rural (B), and semiurban (AB) areas at the district level. A detailed explanation of the definition can be found in Sistema de Datos Micro quintile (Figure 6). The difference in excess deaths Regionales (2020). varied much more across regions. Thus, the excess deaths in regions such as Lambayeque The finding that mortality rates did not vary and Piura were twice those in Cajamarca and much across socioeconomic groups could Huánuco and four times those in Apurimac and be verified using additional sources of data. Huancavelica, independent of income levels. To verify the robustness of these results, the Income therefore did not protect Peruvians analysis was replicated using data from the 2017 from COVID-19 deaths. The probability of dying population census, which has the advantage depended more on the region of residence. of including data from all districts (the local administrative unit). The 2017 population census includes expenditure data projected based on ENAHO data and used to replicate the analysis.6 The results of the analysis using census data are consistent with those of the ENAHO: excess mortality is greater among people ages 65 or more, the urban population, and people living in certain regions (see Figure 5). After controlling for these characteristics, excess deaths are homogeneous across household spending in most regions, that is, spending is not a good predictor of COVID-19 deaths (see Appendix B for a detailed description of the census analysis). 6. The 2017 population census uses asset data to project expenditure based on parameters from an ENAHO regression of expenditure and assets. INEI performed this regression to create the 2018 poverty map. The long-lasting impacts of COVID-19 11 2.3. The short-term economic Figure 8. Variation in employment by sector, 2020 effects of COVID-19 20.0 18.1 14.8 GDP real percentage change (%) 10.0 In the short term, the pandemic represented 0.0 an economic and social setback, with an 11 -10.0 percent fall in GDP and a 10 percent increase -20.0 -17.0 -11.1 -20.5 in poverty, especially in the commerce and -30.0 -22.5 -26.4 -26.2 services sectors. The economy contracted -40.0 11 percent in 2020, its biggest fall in the last 30 years and the biggest fall registered in any country in Latin America during that year. Poverty Source: ENAHO 2019–20. increased by 10 percentage points, from 20.2 percent to 30.1 percent, eliminating two decades Employment fell in most sectors, and women of progress in poverty reduction in the span of and youth were the most affected. Figure a year. Extreme poverty also increased from 2.9 7 shows that employment fell by between 10 percent in 2019 to 5.1 percent in 2020. In 2020, percent and 26 percent in fishing, mining, GDP in real terms in the construction sector manufacturing, construction, commerce, and fell by 13.3 percent; commerce dropped by 16.0 other services between 2019 and 2020. Mining, percent; and other services fell by 9.9 percent. commerce, and services were the most affected. Women were more concentrated in some of these Figure 7. Variation in GDP by sector, 2021 sectors. By 2019, one woman in every four was working in the commerce sector, and another 5.0 4.2 40 were working in other services. In contrast, GDP real percentage change (%) 1.0 0.0 15 percent of men worked in commerce, and 35 -5.0 percent in other services. -6.1 -10.0 -9.9 Women increasingly left the labor force, -15.0 -13.4 -12.5 -13.3 -16.0 mainly to take care of children and the elderly -20.0 given the closure of schools and the reduced supply of support and care systems. Female labor force participation was 68.7 percent in 2019 and 58.9 percent in 2020, while male labor force Source: BCRP 2021. participation was 82.1 percent in 2019 and 74.4 percent in 2020. This implies that the gap between male and female labor force participation was 13.4 percentage points in 2019; it increased to 15.5 percentage points in 2020. That women were exiting the labor market at a higher rate was partly explained by the disproportionate increase in household work and children accompaniment. According to the World Bank High-Frequency Phone Survey (HFPS) performed during the 12 The long-lasting impacts of COVID-19 pandemic, 31 percent of women reported that The immediate impacts of higher unemployment they experienced an increase in the amount of and lower job quality are clear: loss of the domestic work, compared with 20 percent of main source of income, exposure to stress, and men. Furthermore, 42 percent of women reported unproductive time (that is, without accumulating that they experienced an increase in the amount additional human capital). of childcare work, while only 34 percent of men Figure 10. Youth and adult unemployment, by reported such an increase (Figure 9). trimester, 2018–21 Figure 9. Share of respondents who experienced an increase in the volume of various work indicators 9% 8% 7% 6% 5% Education and schoolwork 51 4% accompaniment of children 39 3% 2% 1% 0% I II III IV I II III IV I II III IV I II III IV Childcare such as feeding, 42 2018 2019 2020 2021 playing with them, and caring for them 34 Youth unemployment Adult unemployment Domestic work, like 31 Figure 11. Youth and adult informality, by trimester, washing, cooking, and cleaning 20 2018–21 0 10 20 30 40 50 60 85% 80% Female Male 75% Source: HFPS 2021. 70% 65% Young people (ages 15–24) also lost jobs at a 60% 55% higher rate. Youth unemployment averaged 6.6 50% percent in 2020, while, among adults, the annual I II III IV I II III IV I II III IV I II III IV average was 3.5 percent that year (Figure 10). 2018 2019 2020 2021 Furthermore, young people showed an increase Youth informality Adult informality in unemployment in the first quarter of 2021, Source: Permament Employment Survey (EPE), 2018–21. while adults did not. Likewise, the informality rate among this group, which was already higher Depressed labor market conditions also led than the rate among adults, increased during to lower labor income and lower per capita the pandemic. In 2019, informality, measured household expenditure. Average labor income as the absence of social security, was almost 8 went down from S/1219 in 2019 to S/1105 in 2020, percentage points higher among young people a 21 percent decrease.7 Average household per than among adults. By the second semester of capita monthly expenditure went from S/664 in 2020, the gap had widened to 15 percentage 2019 to S/580 in2020, a 15.9 percent decrease.8 points. The fall in labor income was more pronounced 7. At constant Lima Metropolitana prices of 2021 (INEI 2022). 8. At constant Lima Metropolitana prices of 2021 (INEI 2022). The long-lasting impacts of COVID-19 13 among those at the lower end of the income distribution. Among the bottom 40 percent of the income distribution (the bottom 40), the decrease in labor income was 22 percent, while, among the top 60 percent (the top 60), the reduction in average monthly labor income was 15 percent. In contrast, the fall in per capita expenditure was almost the same among the bottom 40 (12 percent) and the top 60 (13 percent). These trends in expenditure per capita reflect the government support for the most vulnerable households through transfers. The quality of jobs dropped substantially because of the pandemic. A good-quality job, according to the job quality index (JQI), is defined as a job that gives a worker a sufficient income (above the poverty line), benefits (health care and retirement), stability (a contract), and satisfaction.9 Before the crisis struck, Peru was already among the countries with the lowest JQI in the region compared with, for example, Argentina, Bolivia, Brazil, Costa Rica, Ecuador, and Paraguay (Figure 12).10 With the shock, job quality fell drastically: the share of workers with good jobs declined from 57.0 percent in 2019 to 45.5 in 2020 (Figure 12). The drop was also the biggest among the comparable countries with data available data on 2021. Figure 12. Job quality index, by country, 2019–21 1.00 0.80 0.78 0.80 0.72 0.71 0.72 0.70 0.71 0.64 0.66 0.66 0.66 0.62 0.62 0.63 0.61 0.61 0.59 0.60 0.56 0.57 0.53 0.46 0.40 0.20 0.00 Costa Rica Brasil Argentina Ecuador Paraguay Bolivia Peru 2019 2020 2021 Source: Elaboration based on data from SEDLAC (2019-2021) and methodology from Brummund, Mann, and Rodriguez-Castelan (2018). Note: The JQI was estimated for countries with data available on 2021. To estimate each dimension, the following criteria were followed. On income, the international poverty line was used to estimate the share of workers who earned more than the threshold. On benefits, access to pensions was considered. On security, the presence of a contract was considered. On satisfaction, a proxy was used on whether a worker has a secondary job. 9. Brummund et al. (2018), “Job Quality and Poverty in Latin America.” Poverty and Equity Note 9, World Bank.8. At constant Lima Metropolitana prices of 2021 (INEI 2022). 10. The JQI was estimated for Argentina, Bolivia, Brazil, Costa Rica, Ecuador, Paraguay, and Peru using SEDLAC datasets for 2019, 2020, and 2021. On income, the international poverty line was used to estimate the share of workers who earned more than that threshold. On benefits, access to pensions were considered. On stability, the presence of a contract was considered. On satisfaction, a proxy was used on whether a worker has a secondary job. 14 The long-lasting impacts of COVID-19 Figure 13. Job quality index, by component, 2019–21 27.5 million Peruvians had received two doses of the COVID-19 vaccine (84 percent of the target 2019 2020 2021 population), and 18.1 million had received three Benefits doses (64 percent of the target population). 1.0 0.8 Yet, employment, earnings, and job quality 0.6 have not recovered to pre-pandemic levels. 0.4 0.2 Monthly labor income was, on average, 5 percent Not poor 0.0 Security lower in 2021 than the pre-pandemic level, and expenditure was, on average, 3 percent lower in 2021 than the pre-pandemic level. Differences in income and expenditure with respect to pre- pandemic times were more pronounced among Satisfied higher income and expenditure deciles. While, in Source: Elaboration based on data from SEDLAC (2019-2021) and the lowest decile of the expenditure distribution, methodology from Brummund, Mann, and Rodriguez-Castelan (2018). per capita household expenditure was only 1 Note: The JQI was estimated for countries with data available on 2021. To estimate each dimension, the following criteria were followed. On percent lower in 2021 than in 2019, at the highest income, the international poverty line was used to estimate the share of workers who earned more than the threshold. On benefits, access to decile, expenditures were 9 percent lower than pensions was considered. On security, the presence of a contract was pre-pandemic. The recovery in employment has considered. On satisfaction, a proxy was used on whether a worker has a secondary job. been slower among women than among men. Estimates of job quality show that all The labor participation of men and women, dimensions of job quality decreased because which was 16.5 percentage points in 2019, was of the pandemic. The biggest drop occurred in still 17.8 percentage points in 2021, the same level job security, as expected (Figure 13). However, as in 2016. the drop in job security was larger in Peru than in the rest of the comparable countries in the On quality, between 2019 and 2021, informality region. This was followed by the drop in job went from 72.7 percent to 76.8 percent, benefits, where Peru was also the country that which represents 693,500 new informal regressed the most among the comparison jobs. The share of self-employed workers group. Earnings likewise suffered significantly. increased from 35.8 percent to 36.2 percent of Job satisfaction was the only dimension in which the working population. The share of workers the regression was small and smaller than in in small and micro firms rose from 70.3 percent comparison countries. of the workforce in 2019 to 73.8 percent in 2021. The share of salaried workers, whose working Two years after the beginning of the pandemic, conditions are associated with better earnings economic activity appears to have recovered. and more stability, dropped from 43.8 percent GDP grew at a 13.3 percent rate in 2021 and is to 42.1 percent between 2019 and 2021. The 2021 expected to reach 2.7 percent in 2022, returning JQI shows a similar pattern: the recovery in job to the pre-pandemic rate. Higher vaccination quality was slower in Peru than in other countries rates have made this recovery possible because with available data. In 2021, only 52.1 percent of restrictions were lifted when mass vaccination the jobs in Peru were high quality, a proportion was launched. By the second trimester of 2022, that was below the pre-pandemic level. The long-lasting impacts of COVID-19 15 2.4. The potential long-term 2.4.1. Education impacts of COVID-19 The pandemic affected the human capital In addition to the short-term impacts on accumulation process, and, because of the the economy, the COVID-19 shock affected disruption, many individuals will acquire other areas of human development whose lower levels of human capital. COVID-19 forced potential impacts on the lives of Peruvians will school closures in 188 countries, interrupting the manifest with the passage of time. The mobility learning process among more than 1.7 billion restriction described above also involved the children and young people.11 Children in Latin closure of schools for a prolonged period. As a America and the Caribbean experienced some result, the generation of children that stayed of the longest uninterrupted COVID-19 school out of classrooms will acquire less human closures in the world. On average, students lost capital, which will translate into lower earnings two-thirds of all in-person school days after the if corrective measures are not undertaken. onset of the pandemic.12 Similarly, young adults who did not the labor market because of the economic crisis will have Peru and the rest of the Latin America and a harder time finding quality jobs. Caribbean region already lagged behind developed countries before the COVID-19 Furthermore, the incidence of death from crisis. According to the Comparative and COVID-19 was greater among numerous Explanatory Regional Studies (ERCEs) of the households in which more surviving members United Nations Educational, Scientific, and will experience impacts. The probability of Cultural Organization (UNESCO)most students death has been higher in households with four in the region do not reach the minimum level of or more members (0.66 percent) than in smaller proficiency in fundamental skills. such as reading households (0.61 percent). This is especially true and mathematics.13 In Peru in 2019, between 24 among members ages 40–65 or more. The death percent and 25 percent of 3rd grade students of thousands of adults has left behind orphaned did not meet the minimum proficiency level children, who, if not cared for, will suffer from (MPL) in reading and mathematics, respectively. the scarring effects on their development. The Even more worrying, one young person in three greater incidence of death among the elderly will in secondary school did not reach the level of leave children without an additional source of learning in mathematics required to attend the childcare and households without an additional grade. source of income. The food crisis generated by the loss in income among households will also Nonetheless, relative to other Latin American affect children through stunting. This section of countries, Peru had been making progress the chapter delves more deeply into these issues. in the years before the pandemic. In both mathematics and reading, the share of students below the MPL was lower than the average in the region. Moreover, according to Peru’s 2019 11. OECD (2020), “Education and COVID-19: Focusing on the Long-Term Impact of School Closures.” 12. World Bank (2022), “Two Years After: Saving a Generation.” 13. UNESCO’s Latin American Laboratory for Assessment of the Quality of Education (LLECE) performs a large-scale assessment program in most Spanish-speaking countries in Latin America and the Caribbean. The ERCE has measured standardized outcomes in mathematics, reading, and natural science since 1996 among students in grades 3 and 6. The assessments evaluate students on a scale from 1 to 4, where 1 is the lowest level and 4 the highest. For each subject and grade, a minimum proficiency level (MPL) is established. 16 The long-lasting impacts of COVID-19 national assessment (ECE), only 24 percent Furthermore, the share of third graders below of students in primary school and 9 percent of the MPL is expected to increase from 30 percent students in secondary school reach satisfactory to 50 percent in mathematics and from 60 performance in mathematics and reading. percent to 80 percent in reading.16 Even under Even though its performance was low, Peru the most optimistic scenario, the loss in learning was one of three countries in the region that is expected to be substantial, erasing a decade improved learning outcomes between 2013 of progress. In an optimistic scenario, the LAYS and 2019, along with Brazil and the Dominican with reach 7.7, which is the level of Peru in 2010. Republic. Additionally, according to national assessments, the share of second graders who School closures also affected other scored satisfactorily on the reading assessment dimensions of child and youth development, increased by 4 percentage points between including mental health and nutrition. 2013 and 2019, and, among fourth graders, the According to a joint study of the United Nations share of students who scored satisfactorily on Children’s Fund (UNICEF) and the Ministry the mathematics reading assessment rose 9 of Health of Peru, 73.4 percent of parents or percentage points between 2016 and 2019.14 caregivers consider that staying at home during the COVID-19 quarantine affected the mental The pandemic meant a step back in this health of their sons and daughters. Additionally, progress because Peru was among the for many children in the country, school meals are countries with the longest school closures in the only reliable source of daily food and nutrition. the region. According to the UNESCO school With the closure of schools, many children also closure tracker, Peru, at 34 weeks, ranked 16th lost access to a crucial element of their food with the longest full extent of school closures security. According to the HFPS, households with among 41 countries in the region. It also ranked children in 2020 and 2021 were more exposed to 13th with the longest partial school closure, at 43 the risk of exhausting food supplies, experiencing weeks, among 41 countries. The regional average at least one food insecurity, or living with an adult was around 30 weeks of fully closed schools who did not eat for an entire day (Figure 14). and around 32 weeks of partially closed schools, both below the values in Peru. Peru was also the last country in the region to allow the return to hybrid classes at the end of 2021, when it allowed . the reopening of 20 percent of all schools in the country. The loss in LAYS due to COVID-19 in Peru has been estimated at 1.7 years per pupil. In contrast, the average estimated loss in the region was 1.5 LAYS. The LAYS in Peru declined from 8.6 pre-pandemic to 6.9 after the pandemic, considering an intermediate scenario.15 14. MINEDU (2019), “Evaluaciones nacionales de logros de aprendizaje 2019.” 15. World Bank (2022), based on Azevedo et al. (2022). The intermediate scenario refers to the assumption on the extent of school closures in partially opened systems (50 percent, 25 percent, and 15 percent in the optimistic, intermediate, and pessimistic scenarios, respectively) and the effectiveness of mitigation efforts (high, medium, and low in the optimistic, intermediate, and pessimistic scenarios, respectively). 16. Request data from Azevedo et al. (2022). The long-lasting impacts of COVID-19 17 Figure 14. Food insecurity by presence of children % of households 70% 62% 60% 51% 50% 50% 50% 45% 39% 40% 40% 34% 36% 14 p.p 31% 30% 27% 14 p.p 19% 20% 17% 17% 20% 15% 13% 11% 4 p.p 10% 0% HH w/o HH with children HH w/o HH with children HH w/o HH with children children 0-17 children 0-17 children 0-17 Run out of food Experienced at least one Where an adult did not food insecurity eat for an entire day 2020 2021 - W1 2021 - W2 Source: HFPS 2020–21. To mitigate the negative effects of school education are lagging from pre-pandemic closures, authorities in the Peruvian levels. In primary education, attendance is 0.5 educational system undertook substantial percentage points below the pre-pandemic efforts to provide distance education. The level; in secondary education, it is 1 percentage government created digital content as part of an point below the pre-pandemic level; and, in initiative to roll out a massive remote learning early education, it is 7.7 percentage points below program, Aprendo en Casa. Using a multimodal the pre-pandemic level. Official statistics of the remote learning strategy that included Ministry of Education reveal that more than television, radio, computer, and some in-person 120,000 students dropped out of school in 2021 interactions, the program reached children in because of a lack of digital connectivity, family 18,000 schools across the country.17 The learning issues, or economic constraints.18 Among these experience included a facilitator to introduce the students, 62 percent had been attending public lecture, an expert teacher who would explain schools, and 37.5 percent private schools. the main concepts, and students who would hold discussions. The Ministry of Education also Digital gaps limit the opportunities for poor and launched a website, PeruEduca, to help train vulnerable students to participate in virtual teachers in their new role of remote teaching. education, thereby exacerbating existing inequalities in education outcomes. Recent School enrollments have not recovered to pre- data of the World Bank (2022) show that only one pandemic levels. According to the Statistic Unit household in four in Peru has access to internet at the Ministry of Education (ESCALE), attendance through Wi-Fi at home. For indigenous students, in early education and primary and secondary the ratio is one household in five. Access to digital devices is also limited in the country. Fewer than 17. Barron Rodriguez et al. (2021). 18. Gestion (2022). 18 The long-lasting impacts of COVID-19 30 percent of primary-school students have a are made. This reduction is expected to increase computer at home. Among indigenous students, the poverty rate by 6.5 percent. The authors also the share is 15 percent. According to ENAHO simulate adjustment scenarios. After parental 2021, the ratio of internet access by any source is adjustments, the decrease in earnings would one student in every two, a value that goes down go down to 4.0 percent, and, if government to one student in every three among indigenous adjustments are included, the decrease would and Afro-descendant households and to one be 4.6 percent.19 A combination of both would student in every five among rural households. yield a decrease in earnings of only 2.6 percent. Access to televisions—an important, but less interactive tool in Aprendo en Casa—is more School closures and learning loses are also prevalent throughout the country: 8 households expected to reduce intergenerational mobility. in every 10 have access to a television. The ratio The patterns described above in the unequal among indigenous households is 7 households in capacity to engage in meaningful learning during every 10. school closure will result in unequal losses in human capital formation. As a result, absolute and The compensatory actions of parents also play relative intergenerational mobility will decline. an important role in children’s learning. The Estimates suggest that, in upper-middle-income literature suggests that such actions may have a and high-income countries, absolute educational long-term effect on human capital accumulation. mobility will decline as the share of children with Fuchs-Schündeln et al. (2020) estimate the more years of schooling than their parents will effects of parental compensatory actions (that drop by 5 percent or more.20 Latin America and is, investing time in children’s education) during the Caribbean is expected to show the largest COVID-19 school closures on children’s future decline. As a result, in that region, relative mobility earnings and well-being. They conclude that is expected to decrease by 3 percent.21 In Peru, the parental investments reduce the negative impact decline in relative mobility could reach 11 percent. of school closures, but do not fully offset it. They also find that the negative effects are especially 2.4.2. Health and nutrition severe among children with less well educated, low-income parents. The COVID-19 pandemic also had consequences on stunting and malnutrition. Lower levels of human capital will have long- The reduction in income, the increase in term consequences on earnings and poverty poverty, the food crisis, and the higher rates of among this generation of children and youth. orphanhood caused families to experience food According to a World Bank study by Bracco et insecurity. As a result, the rates of malnutrition, al. (forthcoming), more than a third of the 2020 child stunting, and children born to women with academic year was lost because of school a low body mass index (BMI) have been rising. closures even after considering compensatory A study by Osendarp et al. (2021) published in measures. They conclude that the new generation Nature estimates that, by 2022, there will be an of Peruvian workers will exhibit a decrease of 8 additional 2.6 million stunted children in 118 low- percent in earnings by 2045 if no adjustments and middle-income countries. 19. Parental reactions include assisting in the learning process at home. Government reactions include compensations by school systems for the lack of classes with other technologies and inputs, as defined by Neidhöfer et al. (2021). 20. Azevedo et al. (forthcoming). 21. Relative educational mobility is measured as the extent to which a child’s position in the distribution of educational attainment is independent of the distribution among parents. The long-lasting impacts of COVID-19 19 A competing study of the Fredrick S. Pardee As one of the countries most affected by Center for International Futures (2022) estimates COVID-19, the incidence in Peru of stunting an increase of 1.6 million children experiencing and future income losses should be a priority stunting. Analysis by Osendarp et al. (2021) points concern. Statistics collected by the World Bank to an additional 168,000 deaths among under-5- through the HFPS exemplify the level of food year-olds, 2.1 million cases of maternal anemia, insecurity that families have suffered during the and 2.1 million children born to women with low pandemic. In 2020, at the peak of the pandemic, BMI. There is also evidence of stunting because of 58 percent of households in Peru reported that orphanhood. In areas of Tanzania exposed to the they were experiencing at least one type of food HIV/AIDS pandemic, adults who had experienced insecurity (Figure 16). Although, by June 2021 maternal orphanhood at ages 7–15 experienced (2021-wave 1), the share of households had an average loss of 2 centimeters in final height declined to 46 percent, the drop was not much attained.22 A meta study in 49 low- and-middle- larger by December 2021 (2021-wave 2), to 41 income countries finds that maternal orphanhood percent. This points to a slower recovery relative is associated with a higher risk of stunting, to other, more common and more frequently especially among children whose surviving measured indicators, such as employment. By fathers were not living in the same household.23 December 2021, the share of households that had run out of food was still up by 9 percentage points The potential long-term implications of compared with pre-pandemic times (Figure 15). malnutrition and stunting are losses in Figure 15. Households that ran out of food in the productivity and lower incomes. The literature previous 30 days (%) points to a 21 percent reduction in adult earnings 60% because of undernutrition and stunting in early 50% childhood.24 A study commissioned by the World 40% Bank from Gasparini and Laguinge (forthcoming) 30% estimates the potential reduction in future 20% earnings because of COVID-19–induced stunting 10% 0% in Latin America. The study uses the estimates Bolivia Chile Ecuador Peru Colombia of Osendarp et al. (2021) on excess stunting due Pre-pandemic (Feb - 2020) 2021-W1 2021-W2 to COVID-19 and of Horton and Ross (2003) of Source: Elaboration based on HFPS. future earning reductions. Although the results Figure 16. Households with at least one food point to only a small average drop in income (0.012 insecurity experience (%) percent) due to additional stunting, the reduction is not evenly distributed across households. The 60% 58% 50% 51% reduction in income among the poorest income 49% 40% 41% deciles is around 0.11 percent, while incomes 30% among the richest deciles remain unchanged. 20% 22% 10% 13% 0% 2020 2021-W1 2021-W2 Bolivia Chile Ecuador Peru Colombia Source: Elaboration based on HFPS. 22. Beegle et al. (2010). 23. Finlay et al. (2016). 24. Horton and Ross (2003). 20 The long-lasting impacts of COVID-19 2.4.3. Orphanhood Peru shows a greater incidence of orphans Estimates derived from national registries show an orphan excess in Peru of 43,500 children in 2020–21 relative to 2018–19.25 An indirect effect of deaths due to COVID-19 is experienced by the family members left behind by the deceased: an increase in orphan excess by 20 percent with respect to previous years, representing five additional orphans per 1,000 children, including orphanhood due to the death of a mother, father, or both. Excess orphanhood is higher in urban areas and varies greatly across regions (Figure 17) Figure 17. Excess orphanhood per 1,000 children, 2021 relative to 2019 10 9 8.6 Excess orphans per 1000 children 8 7.2 7.1 7 6.6 6.1 5.8 6 5.2 5.1 4.9 5 4.8 4.5 4.4 4.2 4.1 4.1 4.0 4 3.9 3.8 3 2.7 2.6 2.6 2.0 1.8 1.8 1.8 2 1 0 Piura Ucayali San Martin Cajamarca Ayacucho Ancash Puno Libertad Moquegua Tacna Cusco Apurimac Ica Lima Huancavelica Callao Junin Huanuco Loreto Amazonas Pasco Madre De Dios Arequipa Tumbes Lambayeque Source: World Bank, 2022e. Note: World Bank estimates are derived from the Unique Identification Registry of Natural Persons, which is managed by the independent natio- nal organization responsible for the identification of Peruvians through the National Registry of Identification and Civil Status. Estimates included in the technical note “Orphanhood in Peru: Estimation and proposal of a registration system”, prepared through the project “Social Protection Reforms in Peru for post-COVID-19 recovery” . Recent estimates show that Peru exhibits the By comparison, the second highest rate was 6.4 highest incidence of orphanhood associated in South Africa, and developed countries average with the pandemic. According to a study that a rate of less than 1.0 per 1,000 children. Updated compares 21 countries, Peru had the highest estimates by London Imperial College to July incidence of orphanhood because of COVID-19 2022 also place Peru with the highest incidence (that is, loss of a father, mother, or both), with a of orphanhood. rate of 9.6 orphans per 1,000 children one year after the onset of the pandemic.26 In addition, the Other indicators show that the share of children rates of children who lost primary or secondary living with only one parent has increased in caregivers were also the highest in Peru, at 14.1 recent years. Although orphanhood is difficult per 1,000 children. to capture in household surveys given the limited 25. Estimates are derived from the Unique Registry of Identification of Natural Persons, which is managed by an independent national organization, the National Registry of Identification and Civil Status. 26. Hills et al. 2021. The long-lasting impacts of COVID-19 21 panel sample of the surveys, indicators allow estimates of an upper bound in orphanhood.27 which translates into less time and resources For example, the share of children (ages under 18) devoted to the child’s accumulation of human living with only one parent rose from an average capital. For instance, two studies in Tanzania of 16.0 percent pre-pandemic to 19.9 percent show that orphanhood had a permanent impact in 2020. The share increased an additional 1.3 equivalent to one year of educational attainment percentage points in 2021. Although there may and delayed school attendance among girls.28 be several reasons why a child lives in a single A study of orphanhood in Indonesia shows that parent household, the rise is partly explained by the child of a parent who has recently died is, an increase in orphanhood. on average, two times more likely to drop out of Figure 18. Percentage of children living with only one school than children living with both parents.29 parent, 2018–21 Moreover, the death of a parent reduces a household’s ability to pay school fees, which 25.0% 21.1% also means that the children less likely to attend 19.9% 20.0% school.30 15.9% 16.1% 15.0% Pre-pandemic Average: 16% The loss of a parent reduces the future 10.0% incomes of the children. Experience during the 5.0% 2004 tsunami in Indonesia shows that the loss of a mother or father resulted in worse labor 0.0% 2018 2019 2020 2021 market conditions among youth in the long Source: Estimates based on INEI - ENAHO. term and higher rates of household work among young girls, which suggests that the children are Loss of parents or primary caregivers obliged to substitute for their parents.31 A similar has a scarring effect on children study in Tanzania shows a gap in consumption among adults who were orphaned relative to Evidence from past epidemics shows that similar adults whose mothers had survived until the death of a parent or caregiver can cause at least their 15th birthday.32 serious harm to the long-term health and development of children. The health and 29. AGertler, P., Levine, D. I., & Ames, M. (2004), “Schooling and development outcomes affected by orphanhood Parental Death,” Review of Economics and Statistics 86 (1): 211–225. that are found in the literature are mostly related See the review by Ainsworth and Filmer (2005), “Inequalities in to human capital investments and psychological Children’s Schooling: AIDS, Orphanhood, Poverty, and Gender,” World Development 34 (6): 1099–1128. effects, which affect children’s long-term 30. Foster, G., and Williamson, J. (2000), “A Review of Current Literature economic prospects. on the Impact of HIV/AIDS on Children in Sub-Saharan Africa,” AIDS 14 (3): S275–S284. 31. See Gail Cas et al. 2014. Orphanhood is associated with fewer years of 32. Beegle et al. (2010), “Orphanhood and Human Capital Destruction: schooling, which worsens children’s economic Is There Persistence into Adulthood?”, Demography 47 (1): 163–80. prospects. Losing a parent increases the time a child spends in labor production inside and outside the home (as a substitute for adult labor), 27. Although the panel sample covers 30 percent of the households, the sample varies yearly. Thus, it is only possible to observe less than 2 percent of the sample during the four years of the panel. 28. Beegle et al. (2010) find that maternal orphanhood in northwestern Tanzania –an area devastated by HIV/AIDS– had a permanent adverse impact of one year of educational attainment among orphans ages 7 and 15. Ainsworth et al. (2005) also study orphanhood because of the HIV pandemic in Tanzania and find that maternal orphanhood is associated with delayed school attendance, especially among girls. 22 The long-lasting impacts of COVID-19 Orphanhood also causes children to be placed financial and food insecurity. Cumulative trauma with guardians who may not care as much increases the odds of high-risk adolescent about the education of the children; children behavior and functional impairment. Layne et al. without guardians face suboptimal care (2014) find that each additional type of trauma environments that do not facilitate adequate exposure is associated with a higher probability learning. Two recent studies in Lancet (2020) of any of the following risk behaviors: attachment examine the problem of institutionalization difficulties, skipping school, running away from worldwide. Institutional care is associated with home, substance abuse, suicide, criminality, self- lower levels of physical growth, cognition, and injury, alcohol use, and victimization through attention and a less socioemotional development. sexual exploitation. These effects are exacerbated by the length of the institutional stay. Close to 30,000 children are The loss of grandparents may reduce institutionalized in Peru, and close to 80 percent welfare in the entire household of the institutional care facilities in Lima are informal.33 Given the negative effects, reducing Grandparents play a vital role in childcare the number of children entering institutional care assistance around the world. Grandparents facilities to a minimum should be a priority. are the most common providers of informal childcare, usually because of the high costs or The psychological impact among children lack of formal childcare, negative social norms, who are orphaned by the death of parents or or the absence or separation of parents.36 Peru caregivers include an increased risk of post- is no exception. In Peru, grandparents tend to traumatic stress disorder, depression, and be involved in the lives of their descendants, suicide in adulthood. Losing a parent can have and households compose of several generations a traumatic and scarring effect on children and are common. Estimates derived from ENAHO youth. Trauma influences schooling and health 2019 show that 27 percent of Peruvians live in outcomes.34 The most widely studied long-term multigenerational households. effects of the loss of a parent are vulnerability to depression and depressive disorders. Other In Peru, grandparents take care of children so studies link parental loss with other psychiatric that the mothers can work. Grandparents may disorders, such as schizophrenia, generalized have moved closer to their children because their anxiety disorder, phobic disorder, panic disorder, grandchildren require care. A test is the female and eating disorders.35 Trauma raises the risk of labor force participation rate. If grandparents mental health issues, suicide, and abuse. care for children while the mothers work, then one would expect the female labor participation COVID-19 deaths occurred during a period rate to be higher among households with of social isolation and economic hardship, grandparents.37 If, on the other hand, the elderly leading to additional trauma. Loss of a caregiver added an additional care burden to the home, during the pandemic was aggravated by imposed then the female labor participation rate would be isolation from friends and the extended family. lower. Moreover, the losses occurred in a context of other household traumas, such as job losses and 33. Lancet (2020), https://diariocorreo.pe/peru/el-78-de-albergues-en-lima-es-informal-639569/?ref=dcr. 34. See Beegle et al. (2006); Kentor and Kaplow (2020). 35. See Agid et al. (1999); Bifulco et al. (1987); Kendler et al. (1992): Luecken (2008); Mack (2001). 36. See Chamie (2018). 37. For examples, see Du et al. (2019); Posadas and Vidal-Fernandez (2013). The long-lasting impacts of COVID-19 23 In Peru, working-age mothers who live in this may affect household dynamics, some households with elderly members exhibit potential long-term effects will be associated higher average labor force participation. with the lower availability of childcare and lower According to ENAHO 2019, the labor force female labor force participation rates. participation rate among working-age mothers with elderly members in the household is 83.9 Another likely impact of losing a grandparent percent, while the rate is only 79.3 percent among is losing an additional source of income, mothers without elderly household members. through pensions. Data of ENAHO indicate Furthermore, a means comparison of female labor that, between 2019 and 2021, the number of force participation rates among women ages beneficiaries of Pension 65, a noncontributory 15–64 is influenced by the presence of minors in pension, declined by 6 percent. In 2019, 1 the household, though the associated negative percent of the population would have been at correlation is more than offset if there are elderly risk of falling below the poverty line if the elderly household members; this trend holds if one stopped receiving pension benefits. controls for location, age-group, and educational level. This is especially true among women in the 2.4.4. Labor markets bottom 60 percent of the expenditure distribution (Appendix C). The greater negative impact of the COVID-19 crisis on labor market outcomes among youth The high incidence of COVID-19 among the may translate into lower future earnings and job elderly may have affected outcomes among quality. multigenerational households in indicators of household composition and welfare. Unemployment will be a serious problem Multigenerational households share resources, among youth after they reach adulthood such as rent, utilities, food, and responsibilities, because their greater exposure to lowering the cost of living across all members. unemployment is likely to affect their long- Resources are often shared within the household; term labor market fortunes.40 In the United this, if one individual achieves more, the benefits Kingdom, youth unemployment leaves a lasting are redistributed among all members.38 During unemployment scar: three extra months of the first two years of the pandemic, an additional unemployment before age 23 lead to two extra 150,000 elderly were among the excess deaths months of unemployment during ages 28–33.41 in Peru. This implies that as many as 150,000 households may have lost the benefits accruing Most studies agree that the impact of lost to households because of multigenerational living work experience persist through wages. For arrangements.39 In terms of childcare assistance, example, according to a study in the United estimates derived from ENAHO 2021 data show States, a six-month spell of unemployment that 9 percent of children ages under 18 were among young menage 22 results in an 8 percent living in households with grandparents, which is lower wage at age 23.42 The negative effects 2 percentage points less relative to pre-pandemic persist, and, even at ages 30–31, wages were times. Although it is too early to determine how 2 percent to 3 percent lower than they would 38. See Muennig et al. (2018). 39. This is an extreme case, first, because it assumes that all elderly who have died had been living in multigenerational households with only one elderly person, which is not likely. Second, it may also be that, as a result of COVID-19 deaths, alternative living arrangements had been adopted. 40. OECD (2010). 41. Gregg (2001). 42. Mroz and Savage (2006). 24 The long-lasting impacts of COVID-19 have been otherwise. Other authors find that the Youth unemployment may also be detrimental scarring effects of youth unemployment at age to other aspects of happiness, job satisfaction, 23 on wages persist until age 42. The authors and health. Bell and Blanchflower (2009) find estimate a penalty of 12 percent to 15 percent evidence that spells of unemployment at age 23 in the UK labor market. The penalty drops to 8 are associated with lower levels of happiness at percent to 10 percent if youth are able to avoid age 50. The negative effect increases with longer repeated bouts of unemployment.43 In particular, spells of unemployment at younger ages. unemployment immediately after graduation from college is associated with substantial and 2.5. Recommendations permanent future earnings losses.44 The immediate impacts of the pandemic A low-quality first job—such as an informal revealed the fragility of earlier social gains. one—may have negative consequences among The COVID-19 shock erased a decade of social workers for the rest of their lives. Because of progress, as seen by the sharp increase in the pandemic, the cohort that entered the labor poverty. Employment was greatly affected, and, market during the crisis obtained lower-quality although they have almost recovered, the quality jobs, and evidence shows that the quality of of jobs and the levels of income still lag. The early job matches has significant effects on the fragility of social gains and the impacts on labor future human capital accumulation and career markets show the need to take actions so that paths among workers. A study in Mexico shows jobs become more resilient to future shocks. To that young workers whose first jobs were formal achieve this, informality, which is high, at 76.8 were 10 percent more likely to have formal jobs 18 percent, and job quality, which is low, at 0.53, months later.45 Similarly, a World Bank study finds need to be addressed.48 that workers who enter the labor market during a recession show higher rates of informality Because Peru was one of the most affected even 10 to 12 years later in Brazil, Colombia, and countries, action is needed to prevent the Mexico.46 country from also experiencing the biggest long-term scarring. Long-term impacts are The potentially long-lasting effects of youth expected in human capital accumulation because unemployment depend on the general of the closure of schools, in health and nutrition conditions of the labor market in which the because of the losses in health services and food unemployment occurs. Youth unemployment security, in orphanhood because of COVID-19 leads to suboptimal short-term human deaths, and in long-term employment because accumulation, which tends to trigger a recovery of the depressed labor market. Policies in these response. However, over longer periods, full areas should serve as a roadmap. recovery may never occur. In such cases, if the labor market produces diminishing returns to additional tenure, then the negative effects may 47. Gregg and Tominey (2004). 48. According to the above estimate of the JQI. only be temporary.47 43. Gregg and Tominey (2004). 44. Oreopoulos et al. (2008); Gartell (2009). 45. Abel M., Carranza E., and Ortega M.E. (2021), “Can Temporary Wage Incentives Increase Formal Employment? Experimental Evidence from Mexico,” Working Paper. 46. Silva J., Sousa L.., Packard T.G., and Robertson R. (2021), Employment in Crisis: The Path to Better Jobs in a Post-COVID-19 Latin America. Washington, DC: World Bank. The long-lasting impacts of COVID-19 25 Promoting access to good-quality, productive three regions have completed this process. jobs requires eliminating rigidities to hiring It is therefore recommended that SUNAFIL in formal jobs. The initial and recurring costs extend its authority to both the informal sector associated with formalization should be and micro firms. In addition, lack of inspectors, reduced.49 Employers pay up to 68 percent of insufficient infrastructure, and the reactive a worker’s wages in nonwage costs, mainly nature of procedures are among the reasons associated with mandatory contributions, paid for SUNAFIL’s reduced impact on informality. It vacations, redundancy payments, and dismissal is therefore also recommended that SUNAFIL payments. Likewise, the complexity of the increase the number and competencies of labor code, which contains more than 1,400 inspectors and improve the ability of regional pages, disincentivizes formal hiring. A simplified agencies to supervise micro firms. It is likewise workers guide and the removal of unnecessary recommended that SUNAFIL should embrace requirements would therefore make the hiring a preventive approach by assisting companies process easier. Moreover, promoting better jobs during the inspection process. beyond the efforts in formality requires greater labor market flexibility. This could be achieved Training programs should be established to by promoting flexibility in firing people with support young people, women, and minorities permanent contracts, for example. though a who face greater obstacles in obtaining change in the interpretation of the Constitutional high-quality jobs. To enhance employability, Court regarding firings of indefinite term contracts training programs with a regional focus launched (contratos permanentes). The interpretation in potential high-growth sectors could help currently used is that firings without “fair cause” minorities acquire transferable socioemotional may result in the reinstatement of workers by skills. Although the first wave of youth training firms. This interpretation, applied since 2001, programs in Latin America had little effect on raises the cost of firing and significantly reduces employability and the quality of jobs, more recent the probability that a worker will be employed programs in the United States have shown through permanent contracts. This increases the promising results.51 The novelty of these programs use of fixed term contracting, which involves less is that they are less focused on technical and training, lower salaries, and less stability.50 vocational skills and more oriented to supporting beneficiaries in developing sectoral skills.52 To promote hirings in the formal sector, the above recommendations should be accompanied by measures that strengthen the 51. The outcomes of raining programs in the 1990s and early 2000s in role of SUNAFIL. The National Labor Inspection Latin America, such as ProJoven in Peru, Proyecto Joven in Argentina, Chile Joven in Chile, and Juventud y Empleo in Dominican Republic were Office (SUNAFIL) targets mainly small and large mixed in employability, earnings, and the quality of jobs. Furthermore, firms within the formal sector. It thus misses they were associated with high costs of implementation, which a large share of labor informality. Moreover, generated significant losses in cases where positive impacts were not achieved (Almeida et al. 2012, https://openknowledge.worldbank.org/ although the responsibility for supervising handle/10986/13549). micro firms was transferred temporarily from 52. Katz et al. (2020), https://www.nber.org/system/files/working_ papers/w28248/w28248.pdf. regional governments to SUNAFIL in 2018, only 49. The initial costs of hiring a worker are not as relevant as the recurring ones (Alcazar and Jaramillo 2012). 50. Legislation on long-term contracts allows workers to sue their employers in the case of “unjustified” firings. If the court ruling favors the worker, the worker can demand reinstatement. This creates negative incentives against hiring permanent workers. The problem is that neither an unfair dismissal nor the basis for a decision to require the reinstatement of the worker rather than monetary compensation is defined objectively. In addition, because legal precedents are not binding in Peru, the probability that the court will rule in favor of the worker is high in any case (Jaramillo, Almonacid, and De la Flor 2017). 26 The long-lasting impacts of COVID-19 The programs also addressed social capital deficits by acting as an intermediary during the job placement process by supporting beneficiaries in finding jobs, excelling at interviews, and keeping their jobs. The positive results of these programs are evident in employment, the quality of jobs, and income. Such programs could be replicated using the ProJoven institutional infrastructure to undertake upgraded programs. To confront long-term challenges, the government should prioritize investments in health and education. In education, efforts should be directed toward closing learning deficits. Experiences in other countries have shown success in reducing the deficits in learning associated with school closures by complementing the return to classes with a strategy to help students catch up.53 Peru could follow this example by implementing after-school or summer school remedial programs. In health, the government should work to improve the access to resilient infrastructure and better public services. In particular, the sector must raise the share of doctors and nurses to WHO levels and improve the spatial distribution of health services that are currently only concentrated on the coast. 53. Singh et al. (2022). The long-lasting impacts of COVID-19 27 Appendixes Limitations of the COVID-19 death data in estimating the socioeconomic profiles include Appendix A: COVID-19 data and strategy missing values and the restrictions imposed by the restrained set of relevant economic The Peruvian government was one of the characteristics remaining after the exclusion first to make data on deaths from COVID-19 of income and expenditures. To overcome available to the public. This effort was led by the the missing values, an imputation method was Ministry of Health, which assembled a group of applied. To overcome the lack of useful economic experts to address the number of deaths due to characteristics, complementary databases were COVID-19. The efforts came close to establishing used in the analysis. Individual characteristics the number of excess deaths, which represents of the population, including labor income and part of the strategy used to estimate deaths expenditures, are collected yearly by the INEI internationally. During the peak of the pandemic, through ENAHO. This survey is representative at the government also created platforms to follow the regional level and covers numerous individual the spread of the virus, including geographical characteristics. tools. Other civil society groups contributed to the task through the creation of the Open Covid Analysis of the distributional incidence Group, which recruited volunteer experts in of COVID-19 also requires data on the epidemiology and computer science to examine socioeconomic characteristics of the deceased official registries to identify means of collecting (for example, household expenditure). In reliable data. this way, the regressivity of the impacts of COVID-19 may be evaluated. However, SINADEF All deaths in Peru are reported in a national does not include information on the household registry, the Sistema Informático Nacional expenditures of the deceased, and ENAHO does de Defunciones (SINADEF), along with not include any information on deceased persons socioeconomic characteristics, such as date who had resided in the households surveyed. of death, sex, place of residence, civil status, Therefore, a granular approach, combining education level, and health insurance. SINADEF is SINADEF and ENAHO data, has been proposed. a joint effort of the Ministry of Health, the National Figure A.1 summarizes this analysis. Statistics and Informatics institute (INEI), and the National Registry of Identification and Civil Status The analysis follows a granular approach to register weekly deaths, including deaths from by first dividing the SINADEF data on the all causes, fetal deaths, and deaths of unknown deceased into bins by education, sex, cause. COVID19 deaths are also reported in the age, and region. Education comprises two system. The data represent a unique opportunity categories: incomplete basic education and to estimate the socioeconomic profiles of people complete basic education or higher. Sex who have died due to COVID-19 and of the also includes two categories: masculine and surviving household members. feminine. Age includes three categories: 0–39, 40–64, and 65 or more. Region of residence includes 25 categories, one for each region. 28 The long-lasting impacts of COVID-19 The combination gives 300 bins. For each of them, the mortality rate is calculated, that is, the number of deaths over the number of individuals in a bin. This procedure is carried out for SINADEF data from 2019, 2020, and 2021. Then, because SINADEF data do not include information on causes of death, an indicator of “excess deaths” is calculated to identify the share of COVID-related deaths in each bin. For each bin, a baseline death rate is calculated, that is, the death rate in a year without a pandemic, such as 2019. Then the excess deaths indicator is calculated for 2020 and 2021. The indicator represents the mortality rate in a given year, less the baseline mortality rate. Individuals included in the ENAHO are assigned to the bins on education, sex, age, and region, allowing the two datasets to be merged. Using this bin information, an excess deaths probability can be projected for each individual in the ENAHO. This represents the additional bin-specific probability of dying in 2020 and 2021. The final SINADEF-ENAHO database assembles these sources of information. The data are at the individual level and cover both household expenditure and COVID-19 deaths (proxied by the excess deaths). Using the combined SINADEF-ENAHO dataset, it is possible to calculate excess deaths by income or poverty level (for example, expenditure quintiles). This makes it possible to assess whether COVID-19 disproportionately affects poor or rich households or whether its effect was homogeneous across the income distribution. Figure A.1: A granular approach to excess deaths Indivdual-level death records, which include a limited set of socio-economic characteristics Education Total Location Gender Age Deaths Income Death rate level population Pair using variables available in both 1 M 65 1 5 999 100 5% Death rates by datasets, such as: Location, gender, income and poverty age and education 2 F 50 2 10 999999 300 3.3% levels 3 M 70 1 15 9999 400 3.75% ... ... ... ... ... ... ... ... House survey with a complete set of socio-economic characteristics The long-lasting impacts of COVID-19 29 Appendix B: Robustness exercise using Figure B.1. Excess mortality, 2020 CENSO 2017 and Poverty Map 2018 10 Excess deaths per 1,000 people To check the robustness of these results, the 8 analysis was replicated using data from the 6 2017 census. Unlike the ENAHO, the census covers information on households in all districts 4 (the smallest administrative unit in the country). 2 It includes 28.4 million households. Furthermore, 0 based on the census, INEI constructed a 0 237 417 495 573 658 758 883 1060 1354 2517 poverty map in 2018 that included expenditures Monthly per capita HH expenditure (PER Soles) projected for all households in the country. These Excess mortality Fitted values two datasets, combined with the SINADEF data, Source: Elaboration based on 2017 census and SINADEF. make it possible to estimate excess deaths at the Figure B.2. Excess mortality, 2021 district schooling-sex-age level (yielding 22,488 bins) and to compare this information across the 10 Excess deaths per 1,000 people household expenditure distribution. 8 The analysis shows that the results are 6 robust to the census database. Before 4 controlling for age, sex, education, and district 2 of residence, excess mortality is greater among richer households. Among households with 0 0 237 417 495 573 658 758 883 1060 1354 2517 monthly expenditure per capita of around S/ 27, Monthly per capita HH expenditure (PER Soles) the excess mortality was 1.5 deaths per 1,000 Excess mortality Fitted values people in 2020; whereas, among households Source: Elaboration based on 2017 census and SINADEF. with monthly expenditure per capita of around S/ 2,517, the excess mortality was 6.0 deaths per Looking at incidence disaggregated by age, 1,000 people in 2020 (Figure B.19). This increase sex, and education, the analysis yields results in income trend was also true in 2021 (Figure that are also similar to those of the ENAHO B.19). analysis. Excess mortality is greater among men, individuals ages 65 or more, and urban residents. There is no statistically significant difference in excess mortality across groups by educational attainment. Similar to the ENAHO, the main driver explaining the differences in excess deaths is the age-group of the individuals (Figures B.3 and B.4). 30 The long-lasting impacts of COVID-19 Figure B.3. Excess deaths per 1,000 population, across all expenditure levels, whereas, in by expenditure, 40–64 age-group, 2020 Apurimac, excess mortality in 2021 among the 40-64 Years Old 2021 65+ age-group was around 30 deaths per 1,000 people across all expenditure levels (Figure 5 10 15 20 25 30 35 40 45 50 Excess deaths per 1,000 people B.5). The difference was more than 20 deaths per 1,000 population between the two regions. The patterns are similar in Arequipa, Callao, Cusco, Ica, Junin, Lambayeque, Lima, Madre de Dios, Moquegua, Pasco, and San Martin. In the remaining regions, excessdeaths are increasing with household expenditure. Still, the regional 0 0 237 417 495 573 658 758 883 1060 1354 2517 differences are greater. Monthly per capita HH expenditure (PER Soles) Figure B.5. Excess mortality, selected regions, by age Excess mortality Fitted values and household expenditure Source: Elaboration based on 2017 census and SINADEF. a: Amazonas, 2021 Amazonas 2021 Figure B.4. Excess deaths per 1,000 population, , by expenditure, 65+ age-group, 2020 5 10 15 20 25 30 35 40 45 50 Excess deaths per 1,000 people 65+ Years Old 2021 5 10 15 20 25 30 35 40 45 50 Excess deaths per 1,000 people 0 0 237 417 495 573 658 758 883 1060 1354 2517 Monthly per capita HH expenditure (PER Soles) 0 0 237 417 495 573 658 758 883 1060 1354 2517 Excess mortality Fitted values Monthly per capita HH expenditure (PER Soles) Source: Elaboration based on 2017 census and SINADEF. Excess mortality Fitted values b: Apurimac, 2021 Source: Elaboration based on 2017 census and SINADEF. After controlling by region of residence, 5 10 15 20 25 30 35 40 45 50 Excess deaths per 1,000 people excess deaths are revealed to homogeneous across household expenditure in most regions. Similar to the ENAHO analysis, between- region differences are larger than within-region differences across expenditure, suggesting that regional variance is more important than socioeconomic level in explaining the deaths due 0 0 237 417 495 573 658 758 883 1060 1354 2517 to COVID-19. For example, among individuals Monthly per capita HH expenditure (PER Soles) ages 65 or more in Amazonas, excess mortality Excess mortality Fitted values in 2021 was around 5 deaths per 1,000 people The long-lasting impacts of COVID-19 31 c: Lambayeque, 2021 d: Lima, 2021 5 10 15 20 25 30 35 40 45 50 5 10 15 20 25 30 35 40 45 50 Excess deaths per 1,000 people Excess deaths per 1,000 people 0 0 0 237 417 495 573 658 758 883 1060 1354 2517 0 237 417 495 573 658 758 883 1060 1354 2517 Monthly per capita HH expenditure (PER Soles) Monthly per capita HH expenditure (PER Soles) Excess mortality Fitted values Excess mortality Fitted values 32 The long-lasting impacts of COVID-19 Appendix C: Female Labor Force Participation and Elderly Members in the Household A logit regression is performed to identify whether the presence of an elderly household member is correlated with higher female labor force participation rates among women ages 15–64. The regression is specified as follows: where is the individual, and (.) is the cumulative distribution function of the normal distribution. The results show that the presence of elderly household members negatively affects female labor force participation rates, and the presence of minors also negatively affects the rates. However, the presence of both elderly and minor household members offsets the negative effects of the presence only of additional child members. This result is significant after controlling for age, education, and location of residence of working-age women. Furthermore, the effects are more pronounced among the bottom 60 percent of the expenditure distribution. The results suggest that elderly household members may be playing a crucial role in childcare that can help working-age women enter the labor force (Table C.1). Table C.1. Marginal contributions to female labor force participation FLFP (1) (2) (3) Elderly -0.027*** -0.035*** -0.024* (0.008) (0.012) (0.010) -0.020*** -0.021*** -0.018* Minors (0.005) (0.008) (0.007) -0.031** -0.054*** 0.002 Elderly * Minors (0.016) (0.011) (0.015) Controls Age Yes Yes Yes Educational level Yes Yes Yes Area Yes Yes Yes Expenditure decile All Bottom 60 Upper 40 Source: Elaboration based on ENAHO 2019, estimated using a Probit model. The long-lasting impacts of COVID-19 33 34 The long-lasting impacts of COVID-19