J A N U A R Y, 2 0 2 2 COX’ S B AZA R PA N E L S U R V E Y: R A P I D FO L LOW- U P R O U N D 3 Status of Employment and Labor in Cox’s Bazar: PA R T I H O S T S STAT U S O F E M P LOY M E N T A N D L A B O R I N CO X ’ S B A Z A R : PA R T I H O STS 2 Timeline of COVID-19 lockdowns and CBPS data collection rounds Aug 2017: Influx 1st lockdown 2nd lockdown Jul ‘17 Jan ‘17 Jul ‘18 Jan ‘18 Jul ‘19 Jan ‘19 Jul ‘20 Jan ‘20 Jul ‘21 Mar-Aug 2019: Apr-May 2020: Nov-Dec 2020: Apr-Jun 2021: CBPS Baseline survey R1 Tracking R2 Tracking R3 Tracking This brief summarizes findings on labor market (October-December 2020) saw partial recovery impacts of COVID-19 from the third round of the in the labor market on account of the economy rapid welfare tracking surveys in Cox’s Bazar. reopening, as well as scale-up of post-lockdown These rapid phone surveys are built on the Cox’s assistance programs among hosts. Bazar Panel Survey (CBPS), which is a multi-topic survey that focused on socio-economic outcomes Round 3 of the high-frequency tracking survey was and access to services. The baseline CBPS survey, conducted between April-June 2021, amidst the implemented in March-August 2019, was designed second national lockdown which began on April to be representative of the recently displaced 5 and continued up to August 11 (lifted in camps Rohingya population (displaced between August on 19 August). Round 3 surveyed 3,652 individuals 2017 and March 2019) in Cox’s Bazar and the host from the selected adults in baseline, consisting of community. Within the host community, the survey 1,061 adults from high exposure areas, 1,034 from was further stratified into high exposure (HE, within low exposure areas and 1,399 adults from camps. 3 hours walking distance of a Rohingya camp) and low exposure (LE, more than 3 hours walking dis- Labor market outcomes from the Round 3 of the tance from a Rohingya camp) areas within the dis- tracking surveys have been organized across a trict. The overall sample size of the CBPS baseline two-part brief. Part I of the brief presents findings was 5020 households, split roughly equally across on trends observed in host communities of Cox’s Rohingya camps and host communities, and within Bazar and Part II focuses on the camp population. the latter, equally among HE and LE areas, repre- Findings are presented as cross-sections across the sentative at both the household and individual lev- four rounds: baseline, round 1, round 2, and round els. In this third tracking survey, 3,652 households 3, but are also complemented with panel analysis originally surveyed in the baseline were covered. across the rounds where feasible. Bangladesh’s local economy started experiencing impacts of the COVID-19 crisis in early to mid-March Key messages from Part II: Camps 2020, with the first case being reported on 8 March. • The camp labor market continues to face increasing A full countrywide lockdown was in place from 26 pressure with labor force participation having dou- March – 30 May 2020. The first round of the CBPS bled since the baseline (from 33 percent to 61 per- cent) and continuing to rise across rounds. high-frequency tracking surveys was conducted within the government lockdowns (between • The percentage of employed Rohingya refugees as a share of the working-age population is approaching April-May 2020) and found adverse labor market pre-pandemic levels even though more people are impacts across all communities, resulting in impli- now competing for those same jobs. cations on households’ food security and coping • Hours of engagement and nature of activities under- needs. Findings from the 2nd round, conducted taken indicate recovery in tandem with the gradual re-opening of full-scale humanitarian operations. approximately 6 months following the lockdowns STAT U S O F E M P LOY M E N T A N D L A B O R I N CO X ’ S B A Z A R : PA R T I H O STS 3 Key messages 1 Employment as a share of the labor force has fallen for hosts, coinciding with the second COVID wave and associated second nationwide lockdown. Although labor force participation has remained stable between end of 2020 (R2) and mid-2021 (R3), employment rates for hosts have deteriorated from 86 percent to 69 percent during this period, which coincides with the imposition of Bangladesh’s second nationwide COVID-19 lockdown in April 2021. High exposure areas have seen the largest decline in employment, with 68 percent of the labor force working in R3, compared to 95 percent in R2. This is in contrast with the labor market shock associated with the first set of COVID-19 lockdowns in spring 2020 (R1), which affected low exposure areas disproportionately. 2 Job losses and temporary absences increased during the second lockdown, but not as sharply as during the first set of lockdowns. Twenty percent of the working age population reported having lost their jobs or became temporarily absent1 in R3, significantly higher than the 6 percent between R1 and R2, but lower than the rate of job losses and temporary absences experienced when the first lockdown was imposed. More than a third of the working age population reported having either lost their jobs or were absent from work between the 2019 baseline and R1 when the 1st lockdown started. Job losses were similar across sectors and types of work. 3 Women’s participation in the labor force has become increasingly volatile. Women entered the labor force in response to the economic strain associated with the first set of lockdowns in 2020 and bore the brunt of the lockdowns in 2021. Women were 42 percent more likely than men with similar demographic characteristics to report losing their jobs, particularly women who were self-employed. Moreover, the rise in unemployment rates from 14 percent in R2 to 31 percent in R3 is largely driven by a sharp increase in female unemployment. Female unemployment increased from 3.1% and 20.4% in high and low exposure areas in R2, to 49 percent and 58 percent in R3. 1 Temporary absence from work is defined as individuals reporting having a job that they have been temporarily absent from in the past 7 days. STAT U S O F E M P LOY M E N T A N D L A B O R I N CO X ’ S B A Z A R : PA R T I H O STS 4 Host Communities Although labor force participation has remained stable between round 2 and round 3 at around 58 percent of working-age adults, a noticeable decline in employment and a cor- responding increase in unemployment can be observed between the two rounds. Unemployment rates increased sharply, from 5 per- increases in unemployment, with some indication cent to 32 percent between the two rounds in high of lower job retention rates for high exposure males exposure areas, and from 18 percent to 30 percent between R2 and R3, compared to low exposure in low exposure areas (Figure 1). At the same time, males. These differential impacts of the first and the share of working age hosts employed fell in R3, second set of lockdowns are in part due to the more coinciding with the reinstatement of COVID-19 lock- stringent enforcement and mobility restrictions downs. On average, 40 percent of the working age imposed in the first set of lockdowns, which par- population were employed between April and June ticularly affected the more urbanized, service-sec- 2021, compared to 50 percent at the end of 2020. tor oriented livelihoods of households in LE areas. Moreover, it appears as though HE areas experi- Contrary to what was seen during the first lock- enced somewhat of a recovery between the two down, the second lockdown seems to have affected sets of lockdowns, as evidenced by the uptick in HE hosts more than LE. High exposure areas have employment rates in R2 but were unable to main- seen a larger decline in employment, with 42 per- tain employment in the face of the second wave. cent of the working age population working in R3, The second wave in 2021 increased borrowing to compared to 63 percent in R2. In comparison, the finance private health expenditures, which may share of adults employed in low exposure areas have disproportionately affected the welfare of decreased marginally by 5 percentage points, from poorer households in sub-districts such as Teknaf 45 percent in R2 to 40 percent in R3. High exposure and Ukhia. More work needs to be undertaken to areas, which are reliant on agricultural work (50 understand the labor market impacts of the second percent of workers in HE), have faced the largest wave in HE areas. Figure 1: Labor market indicators by stratum and round (Employment and Unemployment as a share of the Labor Force) 95% 95% 96% 94% 89% 89% 89% 86% 82% 70% 69% 68% 66% 62% 58% 58% 57% 55% 53% 51% 49% 47% 42% 40% 32% 31% 30% 18% 14% 11% 11% 11% 6% 5% 5% 4% Labor Force Participation Employment Unemployment Labor Force Participation Employment Unemployment Labor Force Participation Employment Unemployment All Hosts HE LE Baseline R1 R2 R3 STAT U S O F E M P LOY M E N T A N D L A B O R I N CO X ’ S B A Z A R : PA R T I H O STS 5 Fluctuation in the shares of the labor force actively working suggest that the host com- munity labor market is susceptible to immediate impacts of lockdowns but had also seen close to full recovery of employment shares in the period between the two lockdowns. Twenty percent of the working age population Figure 2a: Increase and decrease in active stopped actively working in R3 (Figure 2a), signifi- employment (worked in past 7 days, not cantly higher than the rate of job losses and absences including temporarily absent) out of working from work between R1 and R2 (6 percent). However, age population the first lockdown period between baseline and R1 34% 35% witnessed a higher rate of respondents stopping active work (34 percent) compared to what was 20% reported during the second lockdown. Rates of tem- porary absence were also significantly higher during 10% the first lockdown in R1 (64 percent), compared to a 6% 4% much lower 22 percent in the second lockdown. Stopped working Started working between rounds between rounds Net gains in active employment (not including Baseline to R1 R1 to R2 R2 to R3 temporarily absent workers) were also signifi- cantly lower between R2 and R3 than during the non-lockdown interim recovery period between Figure 2b: Share actively employed vs R1 and R2. Only 10 percent of the working age temporarily out of working population population became new workers (not working in R2) between R2 and R3, compared to 35 percent Baseline 97% 3% of the working age population finding new work between R1 and R2. R1 36% 64% R2 88% 12% The transitions across rounds shows clear evi- dence of job losses during the first and second R3 78% 22% lockdowns (Baseline to R1, and R2 to R3), while job gains were experienced between the two lock- Actively working Temporarily absent down periods. The rise in unemployment rates in R3 Figure 3: Unemployment rate changes from shows that this increase is largely driven by R2 to R3 in HE and LE hosts by gender 58% a sharp increase in female unemployment 49% rates across all hosts. Female unemployment in both high and low expo- sure areas have increased drastically between R2 19% 20% 20% 16% and R3, from 3.1 percent and 20 percent in high and 7% low exposure areas in R2, to 49 percent and 58 per- 3% cent in R3. Even though the composition of the R2 R3 R2 R3 R2 R3 R2 R3 female labor force participants changes significantly Male Female Male Female across rounds, within women entering and exiting HE LE STAT U S O F E M P LOY M E N T A N D L A B O R I N CO X ’ S B A Z A R : PA R T I H O STS 6 the labor force, the aggregate labor force participa- Male labor force participants do not demonstrate tion rate for women has not significantly changed as much volatility across rounds with 70 percent between the two most recent rounds (37 percent in actively participating in all rounds, 10 percent hav- R2, 36 percent in R3). ing left the labor force during the major COVID-19 lockdowns in R1 and R2, and a total of 11 percent Labor market transitions across the full panel2 having joined the LF since baseline (+6% R1, +3% from baseline through R1, R2, and R3 indicate R2, +2% R3). that despite cross-sectional labor market partic- ipation for women not showing change, women Figure 4: Types of jobs held by those have been incrementally entering the labor force employed in each round of data collection since the onset of COVID. Only 10 percent of these women had been in the labor force throughout, while a combined 24 percent are (permanent) new 45% entrants from different periods (+5% R1, +8% R2, 55% 52% 54% 52% 51% 51% 59% +11% R3). In comparison, only 7 percent have left the LF permanently since baseline. A large share of the volatility in the labor market indicator is also 10% 18% 11% 18% 18% 20% driven by female participants. This churn is man- 15% 13% ifest as 1 out of 3 women (34 percent) moving in or 38% 38% out of the labor force intermittently since baseline, 26% 31% 35% 28% 31% 31% most of whom were joining or leaving temporarily in R1 and/or R2. Baseline R1 R2 R3 Baseline R1 R2 R3 Table 1: Labor market transitions based on full panel (baseline - R1 - R2 - R3) HE LE Self-employed Monthly salaried Daily/weekly wage Male Female The rates of job retention for women between R2 Always in LF 70% 10% and R3 do not vary significantly by sector, indicating that female workers were more adversely affected Permanent entry into LF 11% 24% by the second COVID-19 lockdowns than males, regardless of type of work. Further investigations Temporary entry/exit 13% 34% into female employment during the lockdown Permanent exit from LF 4% 7% periods are necessary to understand the impact of child-rearing during lockdowns on female workers, Never in LF 1% 26% as well as the impact of lockdowns on the second- ary earners within the household. 2 The CBPS rapid surveys primarily track households on each round with the secondary objective of updating labor data on 1 of the 2-3 CBPS selected adults from the baseline, meaning that there are gaps in the panel data from when subsequent rounds had interviewed two different selected adults from the same tracked household. A total of 675 adults from hosts communities (307 male, 368 female) have been interviewed in every round – baseline, R1, R2 and R3. Full panel refers to this set of individuals. STAT U S O F E M P LOY M E N T A N D L A B O R I N CO X ’ S B A Z A R : PA R T I H O STS 7 Box 1 : Women, particularly self-employed women, were more likely to face job losses during the second wave When controlling for labor and demographic variables, the main characteristic explaining job losses between R2 and R3 is gender, with females being 42 percent points more likely to have stopped working during the lockdowns than males. When estimating factors associated with job losses for females, we find that self-employed females were particularly more likely to have stopped working in Round 3, despite being active in Round 2. Even though female employment has historically been less stable than male employment across previous rounds, with a higher rate of entry and exit, Figure 5 below shows that the job losses that accrued between R2 and R3 are primarily through previously employed females, with only 35 per- cent of females working in R2 still employed in R3, compared to an average of 52% employment retention for females across previous rounds. Figure 5a: Work transitions across Figure 5b: Percentage of workers still rounds actively working in next round 14% 12% 16% 28% 84% 85% 28% 81% 13% 17% 17% 51% 42% 44% 52% 51% 22% 72% 71% 67% 35% 30% 29% 27% Baseline R1 to R2 R2 to R3 Baseline R1 to R2 R2 to R3 Baseline R1 to R2 R2 to R3 Baseline R1 to R2 R2 to R3 to R1 to R1 to R1 to R1 Male Female Male Female Stopped working between rounds Started working between rounds Worked in both rounds It is also important to note that 46 percent of women who were employed in R2 but not in R3 reported having household responsibilities (being a housewife/homemaker or engaged in domes- tic work) as a reason for stopping work, indicating that COVID-19 lockdowns and school closures could play a major role in leading women to stop working and tend to household responsibilities. STAT U S O F E M P LOY M E N T A N D L A B O R I N CO X ’ S B A Z A R : PA R T I H O STS 8 Engagement in daily wage work among Self-employed workers were more likely to have employed hosts continue to be signifi- stopped working between the two rounds than cantly higher (and increasing) compared to wage workers, which could be due to the inability of self-employed workers to conduct their business pre-pandemic levels, driven by low expo- during lockdowns. The main factor explaining the sure hosts. lower employment rates among self-employed workers are lower rates of reported temporary Daily waged employment overall has slightly absence, a modality of employment that does not increased as a share of the employed workers, align as well with own-account work as it does with driven by low exposure hosts. The shift towards wage work; around 20 percent of R2 wage workers waged work has been driven by low exposure reported being temporarily absent in R3, compared self-employed workers becoming unemployed or to only 10.5 percent of R2 self-employed workers. exiting the labor force between R2 and R3, with 40 On the other hand, self-employed workers were percent of LE R2 self-employed workers becoming more likely to be actively working in the past 7 days either unemployed or leaving the labor force in R3, than their waged worker counterparts (67 percent compared to only 29 percent of waged workers. and 59 percent respectively). Additionally, only 13 percent of self-employed R2 LE workers became waged workers in R3. The daily/weekly wage labor market among hosts mirrors the trends seen in employ- Unlike what was seen during the first lock- ment, with hours worked and weekly wage down, service sector workers were more rates reverting back to 1st lockdown levels likely to retain employment going into the after the increased rates seen during the second lockdown, but self-employment interim phase. continues to be at higher risk compared to wage work. Among actively employed workers in R3, the median hours worked per week were 30, with More than a third (34 percent) of the host popula- higher work hours recorded in low exposure (32 tion actively working in Round 2 was not working hours/week) compared to high exposure (24 hours/ during the second lockdown (R3). Regressions week) areas. On average, hourly wage deteriorated analysis on the loss of active employment from during the first (R1) and second (R3) national lock- R2 to R3 show that when controlling for labor downs, as evident from an average of 45.8 BDT per variables, respondents who worked in the service hour pay in R1, and 40 BDT in R3, compared to 78 sector in R2 were significantly more likely than BDT per hour in the pre-COVID baseline. Amidst the agricultural workers to still be employed in R3 (13 second lockdown (Round 3), both working hours percentage points) regardless of stratum. Even in and weekly wage rates have contracted to levels the more agrarian high exposure areas, 65 percent observed during the 1st lockdown. of R2 service sector workers were working in R3 (perhaps in the same or another sector), com- Daily wages in high exposure host areas may have pared to 57 percent of HE R2 agricultural workers, been hit harder in the second lockdown compared indicating that employment retention capacity/ to the first, given laborers are working more hours resilience may be higher among service sector but at lower weekly rates. Low exposure hosts workers as opposed to agricultural workers in report working the same hours but at less affected these communities. weekly rates from the first to second lockdown. STAT U S O F E M P LOY M E N T A N D L A B O R I N CO X ’ S B A Z A R : PA R T I H O STS 9 Figure 6a: Median weekly hours worked Figure 6b: Median weekly earnings by actively employed host workers during of actively employed host workers during the 2nd lockdown the 2nd lockdown 2800 2800 2800 40 40 2500 2500 36 34 32 32 31 32 2200 30 24 24 21 1400 1200 1200 1100 1000 1000 All hosts HE LE All hosts HE LE Baseline R1 R2 R3 Baseline R1 R2 R3