ROUND 3 COVID-19 IMPACT MONITORING DEC 2020 Publication Date BACKGROUND UGANDA This brief presents findings from the third round of the Uganda High-Frequency Phone Survey on COVID-19 (UHFPS), which was conducted in September-October 2020. In June 2020, the Uganda Bureau of Statistics (UBOS), with the support from the World Bank, officially launched the HFPS to track the impacts of the pandemic on a monthly basis for a period of 12 months. The survey aimed to recontact the entire sample of households that had been interviewed during the Uganda National Panel Survey (UNPS) 2019/20 round and that had phone numbers for at least one household member or a refer- ence individual. The first round (baseline) of the survey was conducted from June 3 rd to June 20th, the second round was conducted between July 31 and August 21 and the third round was conducted from September 14th to October 7th, 2020. Of the 2,421 households targeted 2,227 households were interviewed in round 1, and 2,199 among them were interviewed in round 2, and finally 2,147 households were interviewed in round 3 representing a 99 percent response rate between rounds 2 and 3. BEHAVIORS RELATED TO COVID-19 Following the easing of lockdown measures, the prevalence of safe practices continued to decline in round 3. Respondents were asked about preventive COVID-19 measures that they followed during last seven days. Figure 1 demonstrates a decline in safe practices in round 3 (September/October) vis-a-vis round 2 (July/ August). The largest decline across the last survey rounds was observed in the prevalence of avoiding groups of more than ten people (16 percentage point decline) and the prev- alence of reducing the number of visits to the mar- grocery store (10 percentage point decline). Urban resi- dents demonstrated higher rates of safe practices than rural residents when it comes to avoiding groups of more than ten people and avoiding handshakes. Figure 1. Share of respondents adopting prevention measures for COVID -19 across rounds, (%) The incidence of wearing masks in public increased in third round and continued to be more prevalent among urban residents compared to rural ones (Figure 2). In round 3 (September/October), at the nation- al level, 73 percent reported wearing a mask all or most of the time while in public in the last 7 days. The comparable estimate in round 2 (July/August) was 66 percent. In round 3, the prevalence of mask wearing all or most of the time while in public was higher among urban residents and those from the richest pre-COVID consumption quintile. Figure 2. Self-reported wearing a mask while in public all or most of the time while in public during last seven day among respondents and among 10 people respondents knew, (%) There is, however, suggestive evidence of overre- ple out of ten they knew wore a mask most or all the time porting in the use of masks all or most of the time when in public in the last 7 days. On average, five people out while in public. As mentioned earlier, 73 percent of re- of ten respondents knew were reported wearing mask all or spondents reported wearing masks all the time or most of most of the time indicating 50 percent prevalence without the time while in public in the third round. However, some significant difference between urban and rural residents respondents might have provided socially desirable answers (Figure 2). This is significantly lower compared to what re- and overreported the prevalence of wearing masks. In order spondents self-reported. to check this, respondents were also asked how many peo- 1 COVID-19 IMPACT MONITORING ACCESS TO BASIC NEEDS Despite the high stated need for masks, access was not universal and was much lower among the poor- est and rural respondents as well as those residing in the Northern region. Besides the questions about the frequency of wearing masks, respondents were asked about the need and access to masks during last seven days (Figure 3). On average, in round 3 (September/October), 83 percent needed masks. The needs were higher in the Central region (100%) and among the richest quintile (87%). Among those that needed a mask, at the national level, 10 percent were unable to access a mask. Those from the poorest pre- COVID-19 consumption quintile had the highest incidence of inability to access a mask, conditional on need (33%, com- pared to 4% among households that are in the richest quin- tile and that needed a mask). Further, the respondents in Figure 3. Need and access to masks, conditional on need in round 3, across consumption per adult equivalent quintiles, (%) Northern Uganda had the largest incidence of inability to access a mask, conditional on need (20%). Economic reasons were the main behind low access the respondents from the poorest quintile (83%) compared to masks, in particular for the poorest respondents. to the richest one (66%). The reason of shops being out of Respondents were asked about the main reasons for not masks was mentioned by eight percent of respondents in being able to access masks. For more than 80 percent of re- Central region and being afraid to go out was mentioned by spondents who were unable to get masks, economic reasons 16 percent of respondents in urban areas. were to blame. Economic reasons were more important for Figure 5. Participation in any education/learning activities among children Figure 4. Households with at least one child (3-18) participating in any 3-18 age across gender and consumption quintiles in rounds 2 and 3, education/learning activities across three rounds, (%) (%) COVID-19 crisis continued to negatively affect hu- tion activities dropped more for males creating a gap in favor man capital in Uganda, and particularly among the of females (Figure 5). Worryingly, the rate of participation in poorest, by lowering participation in education/ education activities was almost halved among the children learning activities. As shown in figure 4, there was a dras- from the poorest quintile, declining from 55 percent in round tic decline in the share of households with at least one school 2 to 22 percent in round 3. The gap was particularly high at -aged child (aged 3-18) participating in education/learning pre-school and primary school levels. On a positive note, activities in round 3 (September/October). This decline fur- there was a slight increase between rounds 2 and 3 in the ther widened the rural-urban gap that reached 38 percentage share of children who did not experience any challenges in points. At individual-level, the rate of participation in educa- learning at home (from 16% in round 2 to 20% in round 3). 2 COVID-19 IMPACT MONITORING ECONOMIC SITUATION Employment and livelihood Employment rates in September/October have al- most fully returned to the pre-COVID-19 level in 2019/20. Figure 6 compares the working status of the re- spondents in June, July/August and September/October with the working status of the same measured as part of the pre- COVID-19 UNPS 2019/20. Overall, the share of respond- ents working during the last 7 days before the interview in September/October was 89 percent, which is 19 percentage points higher than in June and very close to the level report- ed in the pre-COVID-19 survey. Further, the last two phone survey rounds distinguished between those who worked in previous round and those who returned to work since pre- vious period. This allowed to see the recovery in employ- Note: This figure includes only 1669 observations that represent HHs with infor- mation for all rounds and that they did not change respondents along the way. ment across rounds. Substantial recovery in employment happened after the easing of lockdown measures. In this re- Figure 6. Respondents working status last week in UNPS 2019/20 and spect, every forth working respondent in round 2 returned across survey rounds, (%) Family businesses have been recovering as well, but there were still nine percent of households with permanently closed firms in September/October. Figure 7 shows the share of households operating a non- farm business according to the pre-COVID-19 UNPS 2019/20 and across three rounds of the phone survey. About 47 percent of households had non-farm business in 2019/20 before the COVID-19. After the COVID-19 out- break, about 17 percent of households had to temporary close their businesses. In July/August, the share of house- holds with temporarily closed businesses dropped to 10 percent and the share of households with operating busi- nesses increased to 35 percent. The share of households with non-farm businesses increased further to 40 percent Note: This figure includes 2074 observations that represent HHs with a non-farm business in the UNPS and in all UHFPS rounds. in September/October, but without substantial change in the share of households with permanently closed firms. Figure 7: Non-farm family businesses, status & fluctuation*, (% of house- holds) Sale revenues in non-farm businesses continued to increase. Respondents were asked about the sale revenues from the non-farm family business. Almost half of households with non-farm businesses in September/October reported higher sale revenues compared to July/August. This is a sub- stantial improvement compared to the situation in June when more than 90 percent of households reported either full or partial losses since the lockdown in mid-March 2020. As comparison was done to previous months after the COVID- 19, it is still not clear whether sale revenues returned to the pre-pandemic levels. Figure 8: Changes in household non-farm business sale revenues across rounds, (%) Income changes The gradual improvement in household income con- income from farming, 46 percent from non-farm business, tinued, but the situation is still quite far from the full and 36 percent from wage employment. 21 percent received recovery. There was no substantial change in the access to financial assistance from extended family members in Uganda. income sources in September/October compared to July/ The recovery in income levels continued in September/ August (Figure 9). About 77 percent of households received October (Figure 10). Most of respondents in June reported 3 COVID-19 IMPACT MONITORING either decline or no income from key income sources since households with lower or no income and the share of house- the onset of the pandemic. The reduction in income levels holds with higher income compared to the prior survey continued in July/August compared to June in many house- round declined to minimum levels for income from farming holds, but with a growing share of households reporting in- and non-farm family business. Despite these positive tenden- creases in income levels. This positive trend continued in cies, the recovery in income levels significantly lags behind September/October, and the difference between the share of the recovery in employment rates. Figure 9. Prevalence of household key income sources in the last 12 Figure 10. Changes in household income by selected sources and across months, (%) rounds, (%) Credit and Safety nets Households borrowed extensively between mid- March and August, in particular in rural areas and the Eastern and Northern regions. Figure 11 shows the share of households who borrowed money during (I) before COVID-19 crisis, (ii) between mid-March-August and (iii) between August-September 2020. The largest share of households borrowing money was recorded for the period of mid-March-August, which coincided with the strictest lockdown measures and the largest negative impacts on em- ployment and income. Every third household borrowed dur- ing that time and the share was slightly higher in rural areas and much higher in the Eastern and Northern regions. Figure 11. Shares of households taking loans across rounds residence and regions, (%) Loan purposes have changed dramatically after COVID-19 outbreak, with consumption-oriented loans dominating since the onset of the pandemic, especially among the poorest households. Households were asked about the purpose of each loan taken. Overall, among the most popular purposes for borrowing were pur- chasing inputs/capital for nonfarm enterprises (22%), buying food (21%), buying farm inputs (19%), paying for health ex- penses (15%), buying other non-food consumption (10%), paying for house construction or purchase (8%) and paying for education expenses (6%). Figure 12 shows the selected purposes in accordance with the period during which loans were taken, and by pre-COVID-19 richest and poorest quin- Figure 12. Selected purposes for loans by period, pre-Covid-19 consump- tiles and the gender of the borrower. Before the pandemic, tion per adult equivalent quintiles and gender, (%) many households borrowed to buy inputs for non-farm en- terprises, pay for education and houses. However, after the tom 20 percent compared to those from the top 20 per- pandemic started, food- and health care-related borrowing cent. Interestingly, female respondents were more likely to increased substantially. Borrowing for food and health ex- borrow to cover education expenses in comparison to male penses was also prevalent among households from the bot- respondents. 4 COVID-19 IMPACT MONITORING In-kind non-food transfers are the most common form of social assistance received by households. Since July-August, neither food nor direct cash transfers from Government or NGOs have played any significant so- cial protection role (Figure 13). The average household-level incidence of food or cash transfer receipts has not exceeded two percent since June. The household-level prevalence of in -kind non-food transfer receipts was much higher, reaching to 23 percent in September/October. The non-food assis- tance came in the form of masks, soap and mosquito nets. The absolute majority of these transfers came from the fed- eral government. Households in the Northern and Western regions were more likely to get this type of aid. Figure 13. Share of households that received different types of social assistance since the last interview across rounds, (%) Data Notes: the UGANDA COVID-19 High Frequency Phone Survey Third Round was implemented by the Uganda Bureau of Statistics (UBOS) in September- Octo- ber 2020. This survey is part of a World Bank global effort to support countries in their data collection efforts to monitor the impact of COVID-19. A World Bank team from the Development Data Group and the Poverty and Equity Global Prac- tice provided technical support. This survey is the third of a planned 12 waves of the COVID-19 High Frequency Phone Survey of households in Uganda. 2,421 suc- cessfully interview households from the 2019/20 Uganda National Panel Survey were contacted and 2,138 households were fully interviewed. These same house- holds were and will be contacted in all subsequent waves of the COVID-19 High Frequency Phone Survey. The data are representative at the regional and national level and survey weights were calculated to adjust for non-response and undercov- erage. For further details on the data, visit http://www.worldbank.org/lsms-covid19 5