OCTOBER - DECEMBER, 2020 COX’S BAZAR PANEL SURVEY: RAPID FOLLOW-UP ROUND 2 IMPACTS OF COVID-19 ON WORK AND WAGES IN COX’S BAZAR: PART 1 – HOST COMMUNITIES This is the second in a series of briefs to disseminate findings from high- frequency rapid follow-ups on the CBPS sample, by the Poverty and Equity GP of the World Bank. CBPS COVID-19 BRIEF - 2, POVERTY AND EQUITY GP 2 H OST COMMUNI TY COVI D-19 LAB OR MA RK E T This brief summarizes findings from rapid welfare tracking surveys in Cox’s Bazar. Two rounds of tracking surveys were implemented via phone interviews in 2020 to monitor the impacts of the COVID-19 crisis on labor markets, wages, and household coping strategies. The first round was conducted during the COVID-related lockdowns in April-May 2020. A second round was conducted from October-December 2020 (roughly 6 months after the government-imposed lockdowns). These rapid phone surveys are built on the Cox’s Bazar Panel Survey (CBPS), which is a multi-topic survey that focused on socio-economic outcomes and access to services. The baseline CBPS survey, implemented in March-August 2019, was designed to be representative of the recently displaced Rohingya population (displaced after August 2017) and the entire Bangladeshi host community in Cox’s Bazar. Within the host community, the survey includes two strata: high exposure (HE, within 3 hours walking distance of a Rohingya camp) and low exposure (LE, more than 3 hours walking distance from a Rohingya camp) areas within the district. The overall sample size of the CBPS baseline was 5020 households (and two adults per household), split roughly equally across Rohingya camps and host communities, and within the latter, equally among HE and LE areas. Key modules of the baseline survey, including detailed labor market modules were administered to two randomly selected adults in each household. The first tracking survey re-interviewed 3,012 adults originally interviewed in the baseline, while the second survey interviewed 3,438 adults baseline adult respondents (958 in HE, and 927 in LE areas among hosts). This brief (Part 1) focuses on key findings among the host community, with findings for Rohingya in camps discussed in an accompanying brief (Part 2). KEY MESSAGES: 1. Host community areas have experienced a sharp increase in seeking behavior induced by the economic downturn is re- unemployment, particularly in low exposure areas, driven flected in a surge in labor market participation. This is driven by a combination of COVID-induced job and earning losses, by a fivefold increase in the number of unemployed persons and unsuccessful job seeking in the face of falling house- looking for work in low exposure areas. Increased job search hold incomes. The relatively rural, agrarian economy in high has been so substantial that despite the number of employed exposure areas appears to have recovered to pre-COVID levels persons returning to pre-COVID levels, employment as a share across headline labor market indicators. However, in more of the labor force has declined. urbanized, service sector-dependent low exposure areas, job 2. 3. Increased job search may be driven by the In addition, there is evidence of continued strain on the local economy incomplete recovery in employment, with through (i) a shift to daily wage work and longer work hours and (ii) an incom- lockdown-induced temporary absences plete recovery in earning levels. These patterns are pronounced in low expo- turning into permanent job losses, par- sure areas. The host economy has increased its reliance on daily wage work as a ticularly in low exposure areas. In these source of employment. These daily wage workers are also working longer hours areas, more than a third of those report- in a week to compensate for lower wage rates compared to pre-COVID levels. ing temporary absence in round 1 are still Two-thirds of the self-employed (own-account workers (and micro-enterprise absent, currently unemployed, or have owners or workers) continue to report sub-par earnings, citing lack of consumer dropped out of the labor force. demand as the main reason. CBPS COVID-19 BRIEF - 2, POVERTY AND EQUITY GP 3 ABOUT THE COX’S BAZAR PANEL SURVEY AND HIGH FREQUENCY ROUNDS Bangladesh’s local economy started experienc- TIMELINES ing impacts of the COVID-19 crisis in early to mid- March 2020, with the first case being reported Baseline Mar-Aug 2019 on 7 March. A full countrywide lockdown was in place from 26 March-28 May 2020. The first Round 1 Apr-May 2020 round of the CBPS high-frequency tracking sur- veys was conducted within the government lock- Round 2 Oct-Dec 2020 downs (between April-May 2020) and focused on capturing key trends in the labor market related bor market in host communities in Cox’s Bazar. to the economic contraction. Round 2 surveyed 958 adults in high exposure upazilas (primarily Ukhia-Teknaf) and 927 adults Findings from the 2nd round, conducted approx- in low exposure upazilas (Cox’s Bazar Sadar, imately 6 months following the lockdowns (Oc- Ramu, Pekua, Chakaria). Findings are presented tober-December 2020) are summarized in these as cross-sections across the three rounds: base- 2-part briefs, on host communities and camps line, round 1 and round 2, but are also comple- respectively. This part focuses on recovery pat- mented with panel analysis across the rounds terns and potential sustained impacts on the la- where feasible1. 1 In this second tracking survey, 3,438 adults out of the 9,045 adults originally surveyed in the baseline were covered. Of these, 1,710 adults were surveyed in the first round of tracking as well. CBPS COVID-19 BRIEF - 2, POVERTY AND EQUITY GP 4 FINDINGS FROM THE ROHINGYA CAMPS IN COX’S BAZAR UNDER THE SAME TRACKING SURVEYS ARE PRESENTED IN A COMPANION PART 2 BRIEF. KEY MESSAGES FOR CAMPS: 1. Among the Rohingya, labor force participation and unemployment rates indicate similar patterns of economic stress as with hosts. Compared with baseline, employment rates within the labor force have declined from 64 to 24 percent, representing roughly 35,000 lost jobs. This contraction in economic activity has been accompanied by a surge in labor force participation, to almost 60 percent of the working age population, compared with 33 percent at baseline. This increase in labor force participation is entirely driven by unsuccessful job seekers, with the number of unemployed increasing by 280 percent (roughly 170,000 individuals). 2. The bulk of the work available in camps are manual daily-wage labor, mostly taken up by men. In this context, men both disproportionately lost jobs when humanitarian operations were curtailed and were also quicker to recover access to the same upon resumption of activities. Employment composition in camps have changed since the lockdowns with men recovering employment at faster rates, while women remaining unemployed i.e., left looking for work. This pattern is driven by the almost complete dependence on humanitarian site development and maintenance activities for work, work which is socio-culturally dominated by men. 3. While employment remains contracted, weekly earnings for those who are employed have recovered as workfare programs have returned to standard operations. Weekly earnings levels, which had dropped during the lockdowns, have recovered, driven by recovery in the number of hours worked. The latter has increased in tandem with the gradual restoration of humanitarian operational activity to pre-COVID levels. Average hourly wage rates remained constant at government-mandated fixed rates throughout 2020. REC OVERY ACROSS HOSTS HAS BEEN UNEVEN HIGH EXPOSURE AREA HOSTS HAVE RECOVERED TO PRE-COVID EMPLOYMENT RATES WHEREAS LOW EXPOSURE AREA HOSTS NOW FACE THE HIGHEST UNEMPLOYMENT RATES SINCE 2019. Overall, the number of labor force participants has increased by 36 percent relative to baseline, driv- en by a threefold increase in the number of individuals actively and unsuccessfully looking for work. This trend is driven by a fivefold increase in unemployment among low exposure hosts. This segment of the population, living in the more urbanized, industry and service-sector dependent upazilas of Cox’s Bazar, also faced the brunt of the economic downturn related to the lockdown. On the other hand, HE hosts appeared to have recovered in terms of reverting to baseline levels of employment and unemployment rates, albeit with an increase in the number of labor force participants as a share of the working age population. This pattern of more urbanized regions facing slower recovery appears consistent with other parts of the country. Even within the two major urban centers of Dhaka and Chittagong, a similar study2 finds that Chittagong residents reported higher employment recovery than Dhaka, the relatively more ur- banized capital city. 2 “COVID-19 Monitoring Survey in Poor and Slum Areas of Dhaka and Chittagong: Bangladesh Labor Market Situation as of Round 2”. Poverty and Equity Global Practice, World Bank Group. Study conducted between September 2 - October 11, 2020. CBPS COVID-19 BRIEF - 2, POVERTY AND EQUITY GP 5 Figure 1: Labor force indicators for hosts (2019-2020) 95,3% 94% 95 % 96% 88,7% % 89% 89% 86 82% 66% 58% 53% 55 % 51% 49%47% 42% 40% 14% 18% 11,3% 11% 11% 4,7% 6% 5% 4% Labor force Employment Un- Labor force Employment Un- Labor force Employment Un- participation employment participation employment participation employment ALL HOSTS HIGH EXPOSURE LOW EXPOSURE Baseline Round 1 Round 2 JOB SEARCH AND UNEMPLOYMENT HAS INCREASED, DRIVEN BY WOMEN AND SECOND- ARY HOUSEHOLD EARNING MEMBERS, SIGNALLING CONTINUING ECONOMIC STRAINS. More than twice as many individuals have joined over the last year in the host community were the labor force (22 percent) in the last year than women. Population counts confirms this trend. those who left (9 percent), explaining the net in- While the number of women in the labor market crease in participants compared to pre-COVID. has almost doubled in the last year, the increase Within these, the share of women entering the of the number of men during the same period labor force, at 27 percent, is almost double that was only 17 percent. This increase in labor force of men, which may be explained in part by the is characterized by more people looking for jobs. lower base of labor force participation among Indeed, while the number of unemployed wom- women. Population counts confirm that the la- en and men have increased by 725 and 225 per- bor force volume has indeed expanded by about cent respectively, these have not translated into 37 percent, characterized by two factors: (i) a 22 jobs, and employment changes have remained percent increase in the number of employed; modest (74 percent in women and 5 percent in (ii) a 300 percent increase in the number of un- men). In addition, two-thirds of new entrants employed. LE hosts have entirely driven this are the spouse or child of the household head, increase in the number of unemployed (+500 indicating entrance of secondary household percent) with HE hosts reporting clear trends of members into the labor force. The patterns point recovery: increased number of employed (+36 towards a community which is coping with the percent) and decreased number of unemployed economic downturn by sending more members persons (-2 percent). out in search of jobs. The phenomenon is likely to strain the local labor market further should While the labor force is still predominantly com- the number of jobs available not expand at the posed of men, 7 out of 10 new labor force entrants same pace as the labor market. CBPS COVID-19 BRIEF - 2, POVERTY AND EQUITY GP 6 RATES OF TEMPORARY ABSENCE FROM WORK, WHICH ACCOUNTED FOR TWO- THIRDS OF EMPLOYMENT DURING THE LOCKDOWNS, HAVE GONE BACK TO NEAR PRE-COVID LEVELS. One of the key characteristics of the labor market Figure 2: Current status of participants who were during the lockdowns were high reported rates temporarily absent during lockdowns of temporary absence from work (64 percent on average) within the employed population. The High Exposure 70% 15% 4% 11% impacts of the lockdowns on host communities were primarily manifested through this report- Low ed absenteeism rather than as unemployment, Exposure 60% 4% 13% 23% spurred by people expecting to go back to their jobs post-lockdown. Current rates of absentee- Currently working now Still absent ism stand at 12 percent on average, still higher Unemployed now Left the LF than pre-COVID levels (3 percent). HOWEVER, NOT ALL JOBS WERE RECOVERED. Panel findings between round 1 (during lockdowns) and round 2 (6 months after lockdowns) show that about two-thirds of those who reported being temporarily absent during the lockdowns have regained active employment. Return to work rates in round 2 were 10 percentage points higher in HE than in LE. However, many absences have persisted. 15 percent of temporary absences reported in round 1 in high exposure areas and 36 percent of absences in low exposure areas have converted into unemployment or quitting the labor force. EMPLOYMENT HAS RECOVERED ACROSS ALL JOB TYPES AND SECTORS TO PRE-COVID LEVELS, BUT PANEL TRANSITIONS HIGHLIGHT DETERIORATION IN JOB QUALITY. Figure 3: Changes in shares of employment types (2019-2020) 27% Irregular/ 21% daily/ weekly 31% 19% Monthly 23% 16% 54% Non-wage 57% 53% Baseline Round 1 Round 2 The share of irregular/daily wage laborers in comparison (Figure 3). Employment recov- among employed hosts have marginally in- ery across sectors has also been imbalanced. creased (31 percent) from 2019 baseline lev- Increased shares of active employment un- els (27 percent). Both monthly salaried work der agriculture during lockdowns and the and non-wage employment have decreased fall back to pre-COVID levels afterwards in- CBPS COVID-19 BRIEF - 2, POVERTY AND EQUITY GP 7 dicates that agriculture was protected from riculture, or the increased use of such work the lockdowns, thus retaining the highest as a safety net and coping mechanism in a share of active employment during this pe- time of economic crisis. riod. Service sector jobs portray similar out- comes, but the industrial sector, mainly con- Changes in relative terms show that even sisting of construction labor in this context, though the shares of daily and monthly evidently faced significant contraction but wage, and non-wage male workers have had has recovered, nonetheless. a more stable trend in the last year, employ- ment quality has deteriorated among wom- Panel transitions however confirm an under- en. The share of women working with daily lying shift in job quality: 2 out of 3 individu- contracts has more than doubled in the last als on average continue to work in the same year (11 percent in baseline vs 25 in round 2). type of occupation as before, consistent Changes in population counts confirm these with findings from Dhaka and Chittagong trends: While the number of males engaged labor markets as well3. However, close to a in daily wage labor has increased a little (driv- fifth of this population (17 percent) shifted to en by HE areas), the number of women en- doing daily-wage labor from monthly sala- gaged in this kind of contract has increased ried workers and self-employment, indicat- 225 percent (with HE areas reporting larger ing that the scale of fall in job quality may increases). The majority of these women be- be more widespread than employment num- long to female headed households (58 per- bers suggest4 . cent), and/or households whose heads have lower education (89 percent of the heads Changes in absolute terms confirm this having reported no schooling or below pri- pattern and suggest an even larger influx mary education levels in the baseline), indi- into irregular daily wage labor than sug- cating that these women belonged to vul- gested by proportional changes, particular- nerable households at baseline. ly for HE hosts. The number of individuals engaged in daily wage labor among hosts has increased by 29 percent on average, Table 1: Changes in absolute population of types of the number of monthly salaried jobhold- workers from baseline to R2 ers, despite seeming most protected, have decreased (-5 percent) while the non-wage Type of work All hosts HE LE population has increased by 10 percent. Irregular/daily HE hosts represent higher volatility in job +29% +69% +14% labor modality shifts compared to LE hosts. Giv- en the higher reliance of HE hosts on ag- Monthly -5% -20% -1% riculture, the larger shifts among this host community could represent a combination Non-wage +10% +15% +7% of a seasonal shift into irregular work in ag- 3 About 63 percent of the workers returning to work in Dhaka and Chittagong were in the same type of occupation as before the crisis. “COVID-19 Monitoring Survey in Poor and Slum Areas of Dhaka and Chittagong: Bangladesh Labor Market Situation as of Round 2”. Poverty and Equity Global Practice, World Bank Group. Study conducted between September 2 - October 11, 2020. 4 Sector level trends are not reported as they may be affected by seasonality in agricultural work that cannot be accounted for due to different survey months. CBPS COVID-19 BRIEF - 2, POVERTY AND EQUITY GP 8 DAY LABORERS ARE WORKING MORE HOURS PER WEEK, POTENTIALLY TO COMPEN- SATE FOR LOWER WAGE RATES AND PRIOR ECONOMIC STRAINS FACED DURING LOCKDOWNS. Irregular/daily wage laborers in host communities reported working 56 hours a week compared with roughly 40 hours/week in 2019, or in other words, have moved from working 5 days a week at baseline to 7 days a week currently. This may be related to the fact that earnings for these laborers have not recovered to pre-COVID levels: weekly earnings are 17 percent lower than baseline levels with daily wage rates having dropped by 41 percent on average. HE hosts report an 8-percentage point higher decrease in daily wage rates than LE hosts (48 percent vs. 40 percent). Along with the higher propen- sity to shift into irregular and daily wage work in HE areas, the increase in hours worked suggests that the road to recovery may be a long one. Monthly salaried workers have evidently been most protected throughout the period, particularly among LE hosts. Not only did they face the lowest impacts during the lockdowns but have also recov- ered to pre-COVID earning levels overall. Figure 4: Trends in hours worked/week for daily Figure 5: Trends in weekly earning for daily wage wage laborer (2019-2020) laborers (2019-2020) 3 011 2 994 2 942 2 525 2 476 2 377 56 56 56 Weekly earnings (bdt) Hours worked/week 1 510 40 40 1 491 1 461 36 32 24 21 All High Low All High Low hosts Exposure Exposure hosts Exposure Exposure Baseline Round 1 Round 2 THE MAJORITY OF NON-WAGE WORKERS CONTINUE TO REPORT REDUCED EARNING LEVELS, WITH UKHIA-TEKNAF HOSTS DEMONSTRATING SLOWER RECOVERY. During the lockdowns, 82 percent of non-wage round 1 and 2, indicating faster recovery. workers reported lower than usual monthly earn- ings with similar rates among HE and LE. Current- Among the self-employed, own-account non-wage ly, 6 months later, a relatively smaller share of workers or single person enterprises are more like- non-wage workers, 63 percent, report decreased ly to report continued earning losses than those earnings indicating some level of recovery. running enterprises with along with other part- ners or employees. These own-account workers However, similar to what is observed in wage-em- are primarily engaged in trade and services such ployment modalities, HE hosts are experiencing a as rickshaw pullers, tom-tom drivers, grocery slower recovery with respect to LE hosts. 73 per- shop owners, tailors, street vendors., are also cur- cent of HE hosts are still reporting lower earnings rently reporting reduced earnings. However, the than usual, compared to 58 percent of LE. More- types of difficulties they reported facing currently over, the share of non-wage HE hosts reporting were similar to those reported at baseline in 2019, similar to usual earnings has remained constant which suggest their chronic vulnerability. (16 percent) from round 1 to 2. In LE areas, this has increased from 16 to 33 percent between