03_update.Rmd

---
title: "Impact of COVID-19 on refugees and migrants, Update 3"
author: "Mixed Migration Centre, 27 May 2020"
date: ''
output: github_document
 
---


```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(tinytex)
```
\  

**This is the third update on the situation for refugees and migrants on mixed migration routes around the world in light of the COVID-19 pandemic based on data collected by the [Mixed Migration Centre](http://www.mixedmigration.org/) in Asia, Latin America, North Africa and West Africa. The objective of the global updates is to provide regular up-to-date findings on COVID-19 awareness, knowledge and risk perception, access to information, access to healthcare, assistance needs and the impact on refugees’ and migrants’ lives and migration journeys. Published once every two weeks, this series provides an aggregated overview; more detailed, thematic and response-oriented COVID-19 snapshots are also developed in each of the MMC regional offices and available [here](http://www.mixedmigration.org/resource-type/covid-19/).**


## Key Messages

•	Awareness, knowledge and risk perception: only 3 of 2,110 respondents had not heard of coronavirus. Respondents in Asia report higher awareness and taking more measures to protect themselves than in other regions. In West Africa, 20% of respondents are not taking measures to protect themselves, an increase on the previous update

•	Assistance needs: at 87% overall, the percentage of respondents who state that they are in need of extra help has remained remarkably stable since data collection began in April, with marginal differences between regions (range: 85% to 89%)

•	Access to information: while overall, governments remain the most trusted information source on COVID-19, there are wide disparities across regions: respondents in Asia more often trust (47%) and use (67%) online communities, and far less frequently cite government as an information source (17% trust and 11% use); whereas in West Africa there is greater use of other migrants and trust in friends and family

•	Access to healthcare and prevention: the same barriers to healthcare persist, with 39% citing the cost of care, 26% that they don’t know where to go. Language is seldom reported as a barrier, except in India, where 33% consider it a barrier

•	Impact on refugees’ and migrants’ lives: loss of access to work (66%) and loss of income (60%) remain major impacts of the crisis on refugees and migrants. Only in Indonesia is the figure much lower, and this is because few were earning an income beforehand. The proportion continuing to work is highest in West Africa, at 20%.

•	Remittances: The impact on sending remittances home varies widely between regions. In Latin America more than half (51%) report they are no longer able to send remittances due to loss of income, while in Asia 4% report this impact

•	Impact on migration journeys: the impact of COVID-19 on migration journeys differed between regions and countries, with generally the lowest impact among Venezuelans in Latin America and among Afghans in India and Indonesia, and the highest impact in West Africa and, to a slightly lesser extent, North Africa

## Respondents

2,110 respondents were interviewed between 6 April and 10 May 2020: 126 in Asia, 382 in Latin America, 957 in North Africa, and 645 in West Africa. In West Africa and North Africa respondents come from a variety of origin countries, while only Venezuelans are interviewed in Latin America, and only Afghans in Asia. Interviews took place in the following 9 countries:

```{r table, echo=FALSE, warning=FALSE}
table_demographics <- data.frame(Region=c("Asia", "", "Latin America", "", "North Africa", "", "West Africa", "", "", "Overall"), Country=c("India", "Indonesia","Colombia", "Peru", "Libya", "Tunisia", "Burkina Faso", "Mali", "Niger", ""), n=c(58, 68, 292, 90, 442, 515, 204, 234, 207, 2110), `Percent women` = c(43, 40, 72, 54, 29, 35, 41, 18, 28, 38), `Mean age`=c(34, 28, 34, 33, 30, 28, 28, 27, 31, 30))
library(knitr)
kable(table_demographics, align = 'c', col.names = c("Region", "Country", "Number of respondents", "Percent women", "Mean age"))


```



## Methodology

A summary of the methodology can be found [here](http://www.mixedmigration.org/4mi/4mi_faq/). The data presented is cumulative, and includes data from previous Global Updates. All figures are rounded to the nearest whole number. Figures for countries where the number  of interviews is less than or around 100 should be interpreted with caution. Unless specified, the number of observations for all analyses and visualisations corresponds to those presented in the above table. Note that for most items of the questionnaire, respondents can select several answer options, and in some analyses it was not possible to include ‘none’ answer options. 34 interviews were discarded from analyses due to questionnaire incompleteness and data quality.

## Awareness, knowledge and risk perception

Knowledge of the coronavirus is still high among respondents. Only 3 respondents, interviewed in Libya, reported that they had not heard of COVID-19. Furthermore, 92% of all respondents stated that they had seen people acting more cautiously, with a minimum of 91% in Burkina Faso, and a maximum of 97% in Indonesia.

Similarly, across regions, 90% of respondents agreed or strongly agreed that they are worried about catching coronavirus (range: from 88% in West Africa to 95% in Asia). In contrast, 19% of all respondents are not worried about transmitting coronavirus (disagreed or disagreed strongly that they were worried), with the highest percentage in West Africa (25% in the region, with a maximum of 28% in Burkina Faso) and the lowest percentage in Asia (5%).

84% of all respondents stated that they know how to protect themselves and others, but again with differences between regions, with higher proportion of respondents confident about this in Asia (91%) and Latin America (98%) than in North Africa (84%) and West Africa (75%).

Respondents know coronavirus symptoms, with dry cough (83% of all respondents), fever (81%) and difficulties breathing (80%) being cited the most frequently. Only 15% of respondents indicated that the virus can be asymptomatic, though this percentage is higher in Asia (27% in the region, and up to 40% in India). Likewise, respondents know which groups are more at risk, with older people cited more frequently (83% overall), followed by people who are already ill with another condition (58%), and health workers (34%)1. However, 22% cite babies and children under five, which are not a high-risk category (46% in Latin America, and up to 55% in Colombia).

More than 90% of all respondents take measures to protect themselves from catching COVID-19, with washing hands more regularly (76% overall) and wearing a mask (59%) being the most commonly cited. In Asia and Latin America, staying at home (88% in both regions) is much more frequent than in North Africa (55%) and especially West Africa (8%). The high percentage observed in Latin America (88%) is consistent with the fact that 86% of respondents in Latin America also stated that they had reached their final destination (13% of respondents in North Africa had reached their destination and 9% of respondents in West Africa). However, in Asia, none had reached their destination. It is likely that the severity of lockdown measures and their enforcement is influencing the degree to which respondents report staying at home, but it is interesting to explore further how many refugees and migrants have a home to stay in. Those who havereached their destination (e.g. in Latin America), or who had been immobile for some time before the pandemic hit, may be more able to stay at home (see also below, impact on migration journeys).

Respondents in Asia cited avoiding crowded spaces (65%) and keeping physical distance (60%) more frequently than respondents in any other region (range: 21% to 51% and 18% to 40%, respectively), although respondents in Latin America more frequently cited using masks (77%) and gloves (56%) than in the other regions.

Overall, a majority of respondents (53%) reported that they are able to keep the 1.5 metre distance, but again with higher percentages in Asia (87%) and Latin America (91%) than in North Africa (48%) and West Africa (31%), see Figure 1.

**_Figure 1: Do you think you are able to practice 1.5 metre distancing?_**

```{r 0, include=FALSE}
library(tidyverse)
load("rda/03_cleaned_data_20200512.rda")
```

```{r 1, echo=FALSE, fig.align='left', fig.height=7, fig.width=7, warning=FALSE}
fig1_data <- data %>%
  group_by(Region, c31) %>% 
  tally() %>% 
  mutate(Percent=round(n/sum(n)*100, digits = 1))
fig1  <- fig1_data %>% 
  mutate(c31=fct_relevel(c31, "Refused", "Don´t know", "No", "Yes")) %>% 
  ungroup() %>% 
  mutate(Region=fct_relevel(Region, "West Africa", "North Africa", "Latin America", "Asia")) %>% 
  ggplot(aes(fill=c31, y=Percent))+
  geom_col(aes(x=Region, y=Percent), width = 0.42)+
  xlab("")+
  theme_bw()+
  labs(caption = "Number of respondents
  Asia: 126
  Latin America: 382
  North Africa: 957
       West Africa: 645")+
  coord_flip()+
  scale_y_continuous(breaks = seq(0, 100, by = 10))+
  theme(legend.title = element_blank())+
  theme(legend.position = c(0.65, 0.96), legend.direction = "horizontal", legend.title = element_blank())+
  scale_fill_discrete(guide = guide_legend(reverse = TRUE))
fig1

```
\  
\


Overall, few respondents (8%) reported not doing anything to protect themselves against coronavirus. That said, when data are disaggregated by region and country, notable differences emerge. In Asia and Latin America, less than 1% of respondents reported not doing anything to protect themselves, whereas in West Africa, this figure was 20%, and up to 31% in Burkina Faso (this is an increase of almost 10% in this country since the previous update).

Out of these 166 participants (129 of them in West Africa) who aren’t taking any measures, 39% stated that this is due to a lack of protective gear (masks, gloves, sanitizer), and 38% because they do not feel it is necessary; 25% said that they simply cannot practice physical distancing due to their living situation. 15% (25 people) said they did not know what precautions to take.

Finally, less than 2% (n=36) of all respondents had been tested for coronavirus: 5 in Latin America, 8 in North Africa (with 42 refused answers, however), and 23 in West Africa (2 refused answers); no respondent was tested in Asia.

\


## Access to information

The government is overall the most-used source of information among respondents (47% reported using it), but the figure is far higher in Latin America (69%) than in Asia (only 11%), with North Africa (47%) and West Africa (42%) closer to the overall percentage. In Asia, the most-used source is the online community (67%), followed by NGOs and the UN (44%), and the government comes eighth. This may be related to language barriers: in Latin America, for example, refugees and migrants speak the same language used by authorities and local media. This is not the case in other regions, and can prove an obstacle to accessing services. Likewise, in North Africa, the online community (53%) is cited slightly more often than the government (47%) or NGOs/UN (36%). In West Africa, other migrants (58%) are cited more often than the government, especially in Niger (73%). As was previously found (see second global update here), respondents in Mali (39%) and Niger (21%) are more likely to use smugglers as a source of information than in any other region (Asia and Latin America: none; North Africa: 6%).

In most regions, the most-used sources of information are also the more trusted. For example, in Asia, the online community is cited as more trustworthy (47%, tied with health professionals) than the government (17%), see Figure 2. In North Africa, however, the government (50%) and the UN/NGOs (36%) is seen as more trustworthy than the online community (26%), although we have seen that the latter is used more in that region. Overall, the second most trustworthy source of information after the government are health professionals (40%), although these are used less than the online community and other migrants.

**_Figure 2: Who do you think is a trustworthy source of information on coronavirus?_**

```{r 2, echo=FALSE, fig.align='left', fig.height=8, fig.width=8}
fig2_data <- data %>% 
  select(193, 194, 96:112) %>% #note this includes 24 nones
  pivot_longer(cols = 3:19, names_to = "Options", values_to = "Answer") %>% 
  filter(!is.na(Answer)) %>% 
  group_by(Region, N_region) %>% 
  count(Answer) %>% 
  mutate(Percent = round(n/N_region*100, digits = 1))
fig2  <- fig2_data %>%
  ungroup() %>% 
  mutate(Region=fct_relevel(Region, "West Africa", "North Africa", "Latin America", "Asia")) %>% 
  mutate(Answer = reorder(Answer, Percent)) %>% 
  ggplot(aes(fill=Region))+
  geom_col(aes(x=Answer, y=Percent), width = 0.6, position = position_dodge2(preserve = "single", padding = 0))+
  ylab("Percent")+
  xlab("")+
  theme_bw()+
  labs(caption = "Number of respondents
  Asia: 126
  Latin America: 382
  North Africa: 951
       West Africa: 645")+
  scale_y_continuous(breaks = seq(0, 100, by = 5)) + 
  coord_flip()+
  theme(legend.title = element_blank())+
  theme(legend.position = c(0.80, 0.15), legend.direction = "vertical", legend.title = element_blank())+
  scale_fill_discrete(guide = guide_legend(reverse = TRUE))
fig2

```

\  
\

Respondents in Asia, who are often registered with UN agencies, seem to have more trust in NGOs and the UN (41%) compared to other regions (range: 15% to 36%), and West Africa respondents have more trust in friends and family (49%, counting together friends in country of departure and destination) than in other regions (range: 18% to 24%).

Most participants received information on the virus via mainstream media (69% overall), social media (62%), or in-person (47%)2, with differences between regions reflecting, again, the most-used sources of information. For example, in Asia, 86% of respondents reported receiving information via social media (whereas social media are used by 47% to 70% of respondents in all other regions). Likewise, in Latin America, 85% reported using the traditional media, which arguably reflects their higher degree of trust in the government. As previously found (see second update), respondents in West Africa used more various media, such as street advertising (37%), phone calls (49%), or simply in-person communication (60%, and up to 76% in Burkina Faso).

Out of all participants who reported using social media (n=1,293), 86% and 84% reported using Facebook and WhatsApp, respectively, which confirmed results from previous updates.

\


## Access to healthcare and prevention

Overall, only 38% of participants believe they would be able to access healthcare if they had COVID-19 symptoms, with a minimum of 31% in North Africa, and a maximum of 47% in West Africa. However, the largest differences are within regions (51% in Colombia, but 13% in Peru; 63% in Niger, but 36% in Burkina Faso). Also note that 28% of all respondents simply don’t know whether they would be able to access healthcare, with the lowest regional percentage in Latin America (18%) and highest regional percentage in North Africa (35%).

As in previous updates, the main barriers to accessing healthcare are a lack of money (39% overall), followed by not knowing where to go (26%), and discrimination against foreigners (25%). Discrimination against foreigners remains the most frequently cited barrier in North Africa (39%), Tunisia (49%), and Peru (50%), whereas it does not seem to be important in Asia (4%) and West Africa (8%), see Figure 3. MMC partners in Asia have observed that discrimination deters refugees’ and migrants’ use of public health services, and pushed them to use private services.

**_Figure 3: What are the barriers to accessing healthcare?_**

```{r 3, echo=FALSE, fig.align='left', fig.height=8, fig.width=8}
fig3_data  <- data %>% 
  select(193, 194, 115:128) %>%
  pivot_longer(cols = 3:16, names_to = "Options", values_to = "Answer") %>% 
  filter(!is.na(Answer)) %>% 
  group_by(Region, N_region) %>% 
  count(Answer) %>% 
  mutate(Percent = round(n/N_region*100, digits = 1))
fig3  <- fig3_data %>% 
  ungroup() %>% 
  mutate(Region=fct_relevel(Region, "West Africa", "North Africa", "Latin America")) %>% 
  mutate(Answer = recode(Answer, 
                          'Discrimination against foreigners limits access to services' = "Discrimination against foreigners",
                          'General insecurity and conflict prevent me from accessing healthcare' = "General insecurity",
                          'I am afraid of being reported to authorities, or arrest, or deportation' = "Afraid of being reported",
                          'I don\'t have the money to pay for health services' = "I don't have the money",
                          'I don\'t have the right or the legal documentation to access health services here' = "I don't have the right/documentation",
                          'The advice for testing and treating coronavirus is unclear here' = "The advice is unclear here",
                          'Services are overwhelmed and access is difficult for everyone' = "Services are overwhelmed",
                          'I don\'t know where to go for healthcare' = "I don't know where to go", 'Other (specify)'="Other")) %>% 
  mutate(Answer = reorder(Answer, Percent)) %>% 
  ggplot(aes(fill=Region))+
  geom_col(aes(x=Answer, y=Percent), width = 0.6, position = position_dodge2(preserve = "single", padding = 0))+
  ylab("Percent")+
  xlab("")+
  theme_bw()+
  labs(caption = "Number of respondents
  Asia: 126
  Latin America: 382
  North Africa: 951
       West Africa: 645")+
  scale_y_continuous(breaks = seq(0, 100, by = 5)) + 
  scale_x_discrete(limits=c("Refused", "None", "Other", "Don't know", "General insecurity", "There are no health services here", "Services are overwhelmed", "The advice is unclear here", "Afraid of being reported", "I don't speak the language", "Discrimination against foreigners", "I don't have the right/documentation", "I don't know where to go", "I don't have the money"))+
   coord_flip()+
  theme(legend.title = element_blank())+
  theme(legend.position = c(0.80, 0.15), legend.direction = "vertical", legend.title = element_blank())+
  scale_fill_discrete(guide = guide_legend(reverse = TRUE))
fig3
```
\  
\

A lack of legal documentation is also an important barrier in Colombia (59%, and note figure is only 10% in Peru), but less in North Africa (16%) and West Africa (9%), as discussed in the previous update. As for Asia, there are some important differences between countries, although note that the number of interviews is low. In India, the lack of legal documentation is cited by 10% of respondents, but it is not mentioned at all in Indonesia. As mentioned above, the greater likelihood of respondents in Asia being registered with a UN agency means they are more likely to have documentation to access health services. 33% of participants in India mentioned not speaking the language as a barrier, whereas this figure was only 3% in Indonesia. Language is cited more often as a barrier in North Africa (17%). General insecurity is reported as a barrier to access by 9% of respondents in Libya, and barely at all in other countries.
\



## Assistance needs

Since the first update, the percentage of respondents who state that they are in need of extra help has remained remarkably stable: 87%, with marginal differences between regions (range: 85% to 89%). Overall, 11% of respondents stated that they do not need extra help since the crisis began, with higher percentages in Mali and Libya (17% each), and Peru (19%).

For the 1,827 respondents who did report that they need extra help, the most frequently items cited were, again, cash (77% overall), basic needs such as food, water and shelter (60%, but note this is about 10% less than in the previous update), and sanitary items such as sanitizer and masks (44%). The pattern of needs between regions persists (cash is cited more often in North Africa and West Africa, but basic goods are cited more often in Latin America). In Asia, a need for access to health services was cited considerably more frequently (51%, and 68% in India) than in any other region (range: 13% to 22%). Likewise, psychological assistance was mentioned much more often in Asia (28%, and up to 36% in India) than in other regions (range: 3% to 15%).

Overall, and as in the previous update, 21% of all respondents (n=433) stated that they had received additional assistance since the coronavirus crisis began, with higher figures in Latin America (31%) than in West Africa (19%) and North Africa (17%, with a minimum of 10% in Libya, where access is particularly difficult). The regional percentage in Asia was 25%, but the highest percentage recorded across all countries was in India, with 43% of respondents reporting they received extra help (the number of interviews is quite low). It must be noted that in some countries, a proportion of respondents were referred to 4Mi by NGOs providing assistance. For the 433 respondents who received extra assistance, the most frequently cited items were food, water and shelter (67%, slightly more than what is reportedly needed overall), followed by sanitary items (29%, compared to 44% needed), cash (22%, compared to 77% needed), and information about the virus (21%, compared to 15% needed). Other gaps in assistance can be seen in Figure 4.

\
**_Figure 4: Kind of assistance received (bars) vs. assistance needed (lines)_**

```{r 4, echo=FALSE, fig.align='left', fig.height=9, fig.width=9}
fig4_received <- data %>% 
  filter(c23 == "Yes") %>%#discard those that didn't receive help
  group_by(Region, .drop = FALSE) %>% 
  mutate(N_region_part = length(Region)) %>% #get n by region
  select(193, 196, 131:143) %>%
  pivot_longer(cols = 3:15, names_to = "Options", values_to = "Answer") %>% 
  filter(!is.na(Answer)) %>% 
  group_by(Region, N_region_part) %>% 
  count(Answer) %>% 
  mutate(Percent = round(n/N_region_part*100, digits = 1))

fig4_needed  <- data %>% 
  filter(c26 == "Yes") %>% #discard those that don't need extra help
  group_by(Region, .drop = FALSE) %>% 
  mutate(N_region_part = length(Region)) %>% #get n by region
  select(193, 196, 157:169) %>%
  pivot_longer(cols = 3:15, names_to = "Options", values_to = "Answer") %>% 
  filter(!is.na(Answer)) %>% 
  group_by(Region, N_region_part) %>% 
  count(Answer) %>% 
  mutate(Percent = round(n/N_region_part*100, digits = 1)) 

fig4  <- fig4_received %>%
  ungroup() %>% 
  ggplot(aes(fill=Region))+
  geom_col(aes(x=Answer, y=Percent), width = 0.6, position = position_dodge2(preserve = "single", padding = 0))+
  ylab("Percent")+
  xlab("")+
  theme_bw()+
  labs(caption = "Number of respondents
  Asia: received = 32, needed = 110
  Latin America: received = 120, needed = 338
  North Africa: received = 158, needed = 809
       West Africa: received = 123, needed = 570")+
  scale_y_continuous(breaks = seq(0, 100, by = 10)) + 
  theme(legend.title = element_blank())+
  theme(legend.position = "none")+
  theme(axis.text.x = element_text(angle = 40, hjust = 1))+
  facet_grid(rows = vars(Region))+
  geom_point(data = fig4_needed, aes(x=Answer, y=Percent))+
  geom_line(data = fig4_needed, aes(x=Answer, y=Percent, group=Region))+
  scale_x_discrete(labels=c("Access to health services"="Health services", "Distribution of sanitary items (sanitizer/ mask/ gloves/ etc)"="Sanitizer, masks, gloves", "Cash to pay for health services"="Cash for health services", "Documentation to access health services"="Documentation", "Information about the virus: symptoms/ what to do if I have symptoms/ how to protect myself"="Information", "Other basic needs: food, water, shelter"="Food, water, shelter", "Access to work and livelihoods"="Access to work", "Psychological assistance"="Psychological support", "Other (specify)"="Other"))+
  scale_fill_manual(values=c("#C77CFF", "#00BFC4", "#7CAE00","#F8766D"))
fig4
```
\  
\

Overall, the main providers of additional assistance are still NGOs (41%, and up to 50% in West Africa, with a maximum of 72% in Niger), the local population (28%), and the government (20%), except in Libya, where family and friends were more frequently cited (54%). Assistance from the UN was far higher in Asia (69%) than in all other regions (range: 3% in Latin America to 10% in North Africa and West Africa).


## Impact on refugees’ and migrants’ lives

As found previously, the main impact of COVID-19 on refugees’ and migrants’ lives is reduced access to work (66%), followed by more stress (60%), and reduced availability of basic goods (50%).1 The inclusion of Asia data marginally increased the percentage reporting an impact on access to asylum (15% overall, and 23% in Asia). Respondents in Asia reported higher levels of worry and stress (79%) than in other regions (range: 53% to 62%), see Figure 5.


\

**_Figure 5: What impact has the crisis had on your day-to-day life?_**

```{r 5, echo=FALSE, fig.align='left', fig.height=8, fig.width=8}
fig5_data  <- data %>% 
  select(193, 194, 171:178) %>% #includes 60 none
  pivot_longer(cols = 3:10, names_to = "Options", values_to = "Answer") %>% 
  filter(!is.na(Answer)) %>% 
  group_by(Region, N_region) %>% 
  count(Answer) %>% 
  mutate(Percent = round(n/N_region*100, digits = 1))
fig5 <- fig5_data %>% 
  ungroup() %>% 
  mutate(Answer = recode(Answer, 'Other (specify)' = "Other")) %>% 
  mutate(Region=fct_relevel(Region, "West Africa", "North Africa", "Latin America", "Asia")) %>% 
  mutate(Answer = reorder(Answer, Percent)) %>% 
  ggplot(aes(fill=Region))+
  geom_col(aes(x=Answer, y=Percent), width = 0.6, position = position_dodge2(preserve = "single", padding = 0))+
  ylab("Percent")+
  xlab("")+
  theme_bw()+
  labs(caption = "Number of respondents
  Asia: 126
  Latin America: 382
  North Africa: 951
       West Africa: 645")+
  scale_y_continuous(breaks = seq(0, 100, by = 5)) + 
  coord_flip()+
  theme(legend.title = element_blank())+
  theme(legend.position = c(0.80, 0.15), legend.direction = "vertical", legend.title = element_blank())+
  scale_fill_discrete(guide = guide_legend(reverse = TRUE))
fig5

```
\  
\

60% of all respondents cited a loss of income due to coronavirus restrictions. The highest percentage was in Latin America (90%), and the lowest percentage was in Asia (42%, with a minimum of 13% in Indonesia, where it should be noted that 87% of respondents were not earning an income before the crisis). The higher percentage of people continuing to work despite COVID-19 restrictions was in Mali and Niger (20% each).

For those respondents who lost income (n=1,260), the main impact was the inability to afford basic goods (65%, which is broadly consistent with assistance needed overall, see results on needs above), increased worry and anxiety (58%), and inability to pay remittances (39%) and continue the journey (35%). There were again differences between regions and countries, however, with ability to afford basic goods most reported in Latin America (86%, compared to 32% in West Africa), as well as loss of housing (29% in Latin America), see Figure 6. The impact on the ability to send remittances home varies widely between regions. In Latin America more than half (51%) reported they are no longer able to send remittances due to loss of income, while in Asia only 4% reports this impact. This may indicate that respondents in Asia were not sending remittances before the crisis either, consistent with the large percentage who were not earning an income in Asia before the crisis.

\

**_Figure 6: What impact has the loss of income had?_**

```{r 6, echo=FALSE, fig.align='left', fig.height=8, fig.width=8}
load("rda/03_cleaned_data_20200512_c22.rda")
fig6_data <- data %>% 
  filter(c21 == "Yes") %>% #discard those that didn't lose income
  group_by(Region) %>% 
  mutate(N_region_part = length(Region)) %>% #get n by region
  select(202, 205, 181:188) %>%
  pivot_longer(cols = 3:10, names_to = "Options", values_to = "Answer") %>% 
  filter(!is.na(Answer)) %>% 
  group_by(Region, N_region_part) %>% 
  count(Answer) %>% 
  mutate(Percent = round(n/N_region_part*100, digits = 1))
fig6 <- fig6_data %>% 
  ungroup() %>% 
  mutate(Region=fct_relevel(Region, "West Africa", "North Africa", "Latin America")) %>% 
  mutate(Answer = recode(Answer, 'Other (specify)' = "Other")) %>% 
  mutate(Answer = reorder(Answer, Percent)) %>% 
  ggplot(aes(fill=Region))+
  geom_col(aes(x=Answer, y=Percent), width = 0.6, position = position_dodge2(preserve = "single", padding = 0))+
  ylab("Percent")+
  xlab("")+
  theme_bw()+
  labs(caption = "Number of respondents
  Asia: 53
  Latin America: 344
  North Africa: 547
       West Africa: 316")+
  scale_y_continuous(breaks = seq(0, 100, by = 5)) + 
  scale_x_discrete(limits=c("Refused", "None", "Other", "Loss of housing", "I am unable to continue my journey", "I am unable to pay remittances", "I am unable to afford basic goods", "Increased worry and anxiety"))+
  coord_flip()+
  theme(legend.title = element_blank())+
  theme(legend.position = c(0.80, 0.15), legend.direction = "vertical", legend.title = element_blank())+
  scale_fill_discrete(guide = guide_legend(reverse = TRUE))
fig6

```

Finally, more respondents were unable to continue their journey in West Africa (60%) and North Africa (43%) than in Asia (8%) and Latin America (4%). This is suggestive of extended or indefinite transit or long-term involuntary immobility among Afghan respondents in India and Indonesia, for whom, even though they have not reached their final destination, the Covid-19 crisis does not appear to have impacted on their migration journeys so far. Respondents in Latin America had generally reached their destination.


## Impact on migration journeys

The impact of COVID-19 on migration journeys differed between regions, see Figure 7. Overall, respondents cited increased difficulty moving around inside countries (41%) or crossing borders (38%), or simply being too afraid to either continue their journey or return home (19%). As in the previous update, a number of respondents also stated that the crisis had no impact on their journey (21%), but with a large difference between, on the one hand, Asia and Latin America (52% and 51%, respectively) and, on the other hand, North Africa and West Africa (13% and 8%, respectively). Again, this might suggest that Asia respondents are in extended transit or have been stuck for a long time, and that the restrictions put in place by the Covid-19 crisis are therefore not impacting on their journey. This is credible if we also consider that no Asia participant reported difficulties with crossing borders.
 

\

**_Figure 7: What impact has the coronavirus crisis had on your migration journey?_**

```{r 7, echo=FALSE, fig.align='left', fig.height=8, fig.width=8}
load("rda/03_cleaned_data_20200512.rda")
fig7_data <- data %>% 
  select(193, 194, 181:191) %>%
  pivot_longer(cols = 3:13, names_to = "Options", values_to = "Answer") %>%
  filter(!is.na(Answer)) %>% 
  group_by(Region, N_region) %>% 
  count(Answer) %>% 
  mutate(Percent = round(n/N_region*100, digits = 1))
fig7 <- fig7_data %>% 
  ungroup() %>% 
  mutate(Answer = recode(Answer, 'Other (specify)' = "Other", 'I\'ve been delayed because I was sick, or because I had to stop and take care of people who got sick' = "Delayed because I or other people were sick",
                         'I feel too afraid to move (to continue my journey or return)' = "I feel to afraid to move",
                         'Increased difficulty moving around inside countries'="Increased difficulty moving around",
                         'I was going to be resettled, but this is now delayed'="About to be resettled, but now delayed",
                         'Disembarked / deported back to previous country'="Deported back to previous country")) %>% 
  mutate(Region=fct_relevel(Region, "West Africa", "North Africa", "Latin America", "Asia")) %>% 
  mutate(Answer = reorder(Answer, Percent)) %>% 
  ggplot(aes(fill=Region))+
  geom_col(aes(x=Answer, y=Percent), width = 0.6, position = position_dodge2(preserve = "single", padding = 0))+
  ylab("Percent")+
  xlab("")+
  theme_bw()+
  labs(caption = "Number of respondents
  Asia: 126
  Latin America: 382
  North Africa: 951
       West Africa: 645")+
  scale_y_continuous(breaks = seq(0, 100, by = 5)) +
  scale_x_discrete(limits=c("Refused", "None", "Other", "Delayed because I or other people were sick", "Deported back to previous country", "About to be resettled, but now delayed", "Reduced access to smugglers", "Increased risk of detention and deportation", "I feel to afraid to move", "Increased difficulty moving around", "Increased difficulty crossing borders"))+
  coord_flip()+
  theme(legend.title = element_blank())+
  theme(legend.position = c(0.80, 0.375), legend.direction = "vertical", legend.title = element_blank())+
  scale_fill_discrete(guide = guide_legend(reverse = TRUE))
fig7

```

\  
\

Finally, half of all respondents (51%) stated that they had not changed their plans as a result of the coronavirus outbreak, but with much higher figures in Asia (89%) and Latin America (73%, just as in previous update), than in North Africa (43%) and West Africa (42%). Overall, few respondents changed their intended destination (2%) or route (11%). 7% decided to return home, with a maximum of 23% in Peru. The highest percentages of respondents reporting being stuck for a time were in Libya (35%) and Niger (42%), compared to regional percentages of only 2% and 5% in Asia and Latin America, respectively.

\
  
### Acknowledgement 

The owner and source of the data used for this study is the [Mixed Migration Centre](http://www.mixedmigration.org/), and this study has been published first [here](http://www.mixedmigration.org/resource-type/covid-19/) on 27 May 2020. Data analyst: Jean-Luc Jucker, PhD  
\  
\  
\  
\