Policy Research Working Paper 10291 Least Protected, Most Affected Impacts of Migration Regularization Programs on Pandemic Resilience Maria José Urbina Sandra V. Rozo Andrés Moya Ana María Ibáñez Development Economics Development Research Group February 2023 Policy Research Working Paper 10291 Abstract How can regularization programs improve forced migrants’ forced migrants. The results indicate that access to the pro- resilience to shocks? This paper leverages panel data col- gram promoted better health access for eligible migrants, lected during the COVID-19 pandemic to assess whether facilitating adherence to prevention guidelines and increas- Venezuelan forced migrants who were eligible for a regu- ing detection rates. Additionally, eligible migrants had larization program in Colombia were more resilient and better housing and labor conditions, relative to non-eli- less affected by the pandemic than similar but non-eligible gible migrants. This paper is a product of the Development Research Group, Development Economics. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at sandrarozo@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Least Protected, Most Affected: Impacts of Migration Regularization Programs on Pandemic Resilience r † e Urbina⃝ Maria Jos´ Sandra V. Rozo⃝ r ‡ r § es Moya⃝ Andr´ Ana Mar´ıa Ib´ ˜ ¶ anez JEL Classification: F22, O15, R23 Keywords: Refugees, Amnesties, COVID19, Latin America. * This project was approved by the IPA’s IRB protocol 15396. Rozo acknowledges financial support from the University of Southern California and the World Bank’s Research Support Budget grant. Ib´ ˜ ac- anez knowledges financial support from the Inter-American Development Bank. The order in which the au- thors’ names appear has been randomized using the AEA Author Randomization Tool (#gGVvO2cAUzcF), denoted by ⃝ r . We would like to thank Marisol Rodr´ıguez-Chatruc for her work on the initial stages of the project. We would also like to thank IPA Colombia for their support in collecting the data for this project. † World Bank, E-mail: mj.urbina2100@gmail.com ‡ World Bank Development Research Group. Corresponding author, E-mail: sandrarozo@worldbank.org § Universidad de Los Andes. E-mail: a.moya@uniandes.edu.co ¶ Inter-American Development Bank. E-mail: anaib@iadb.org I Introduction Forced migrants, a population highly vulnerable to adverse shocks, were particularly at risk during the COVID-19 pandemic. Forced migrants often live in overcrowded houses with poor sanitary conditions that prevent many from adhering to non-pharmaceutical interventions (NPI) (Kluge et al. 2020). Their utilization rates of formal health services was infrequent due to lack of access or fear of being detected and deported when living irregularly in the destination country (Zambrano-Barrag´ an et al. 2021). Poor and unstable job conditions, lack of coverage from formal social protection systems, and weaker social networks may have led to large drops in income and reliance on informal jobs mostly in the service sector with frequent social contact. The combination of these factors con- tributed to higher infection and death rates among forced migrants in comparison to the local population (ECDC 2021), and may have led to a sharp deterioration of their eco- nomic and social conditions. Forced migrants posed an additional challenge for countries to control the pandemic. The high infection rates within migrant communities may have contributed to speed the transmission of the virus, as documented by evidence from previous spreads of infectious diseases (Montalvo and Reynal-Querol 2007, Baez 2011, Ib´ ˜ et al. 2021). The effective anez control of the pandemic required mechanisms for rapid testing, tracking of COVID-19 cases, and rolling out vaccination campaigns. Forced migrants with an irregular migra- tory status refrained, however, from using formal health services (Zambrano-Barrag´ an et al. 2021). Regularization programs might have been an effective tool to help irregular migrants bet- ter navigate the pandemic and slow the transmission of the virus.1 By providing access to healthcare, migrants might be more likely to seek medical support when contracting the virus and to vaccinate (Ib´ ˜ et al. 2021). Coverage of social protection systems, either anez through formal jobs or cash transfers, may reduce the negative income shock caused by the extended lockdowns. This paper studies the effect of a large regularization program of Venezuelan migrants in Colombia on easing the health and economic impacts of the pandemic for forced mi- grants. The results provide consequential lessons on how to support forced migrants to better handle negative shocks and to strengthen their self-reliance. II PEP: A Regularization Program for Venezuelan Forced Migrants Seven million Venezuelans had been forced to migrate from their country by 2022, as a consequence of Venezuela’s economic, political, and humanitarian crisis. Most migrated to Latin American countries, of which 2.5 million currently live in Colombia.2 A large percentage of them were considered irregular migrants as they had overstayed their visas or had illegally crossed the border. 1 Regularization programs usually provide migrants legal migratory status for a specific period of time, access to state services and job permits. Regularization programs differ from amnesties as their benefits are limited to a specific period of time. 2 https://www.r4v.info/en retrieved on December 15, 2022. 1 The Colombian government created in 2018 the Permiso Especial de Permanencia (PEP) to facilitate their integration. Prior to the PEP, the government rolled out a voluntary regis- tration process, the Registro Administrativo de Migrantes Venezolanos (RAMV), with the sole purpose of identifying the location and needs of irregular migrants. In July 2018, the gov- ernment unexpectedly announced that all migrants registered in RAMV, a little more than 440,000, were eligible for applying to the PEP program through a free and online tool. Of those, 64% registered into PEP. PEP granted legal migratory status for two years, access to education and health, eligibility to social programs, job permits, and an identification card to access private services. For a detailed description of the program see Ib´ ˜ et al. anez (2022). The design and rollout of PEP lend themselves to estimating the causal effects on the resilience of forced migrants to negative shocks. The unexpected and broad nature of the program (available to all RAMV migrants and with no eligibility criteria), allows for ruling out anticipatory effects and strategic behaviors that have affected the evaluations of other similar programs. III Data We use two waves of the Venezuelan Refugees Panel Survey (VenRePS), administered in the second semesters of 2020 and 2021. Figure A.1 in the appendix depicts the timeline for the PEP roll-out, the VenRePS waves, and the lockdowns put in place by the Colom- bian government. The baseline sample covered 2,232 forced migrant households who arrived in Colombia between January 1, 2017 and December 2018, including 1,135 who registered in RAMV and were thus eligible for PEP (from here on RAMV migrants), and 1,097 irregular forced migrants (non-RAMV migrants from here on). The sample of RAMV migrants was randomly selected from the RAMV census. The sam- ple of non-RAMV migrants was randomly selected by combining databases shared by associations of Venezuelan refugees and referrals from migrants who were being sur- veyed as part of the RAMV sample frame. The sample is representative of the three cities in Colombia that host the largest share of Venezuelan migrants (Barranquilla, Bogot´ a and Medell´ın) and a fourth “region” that aggregates smaller cities. Ib´ ˜ et al. (2022) pro- anez vides a detailed description on the survey, and shows that the initial conditions before migration to Colombia and the roll-out of PEP for the two sources of the non-RAMV samples (referrals and the list of organizations) are not systematically different. In the second wave, we re-interviewed 1,432 households (64% of the sample in the first wave). Attrition from the sample was not systematic with respect to baseline conditions, although better-off households (male headed, better educated and with both spouses) and those residing for a longer period in Colombia were less likely to drop out of the sample (see Table A.6 in the Appendix).3 3 Table A.7 also shows that the attrition rate per survey round between PEP eligible applicants and non- applicants is significant and negative, but small (5.1 p.p.). 2 IV Empirical Strategy: Intent to Treat We estimate an Intent to Treat (ITT) model by comparing RAMV and non-RAMV mi- grants and thus exploit the individual eligibility requirement for the PEP. To isolate se- lection into the RAMV, we control for a broad set of pre-migration and pre-program co- variates, including some that potentially drove selection into the RAMV. We estimate the following model: (1) Yijkt = α0 + α1 1[RAM Vijk = 1] + θ′ Xijk × ζt + ϕjk + δk + ζt + ϵijkt where Yijkt is the outcome of interest for household i located in city j of state k in year t, 1[RAM Vi jk = 1] is an indicator variable for migrants who were registered in the RAMV census, Xijk is a vector of household controls that we interact with wave fixed effects (ζt ), which captures the migrant’s economic and social characteristics before migration and the program roll-out. Appendix Table A.5 reports the descriptive statistics for the control variables. We control for the time of residence in Colombia to attenuate the assimilation bias as well as for city and state fixed effects (ϕjk , δk and wave fixed effect ζt ). We also leverage the time variation across both waves on the outcomes of interest. Even though we do not have data on pre-Covid-19 baseline conditions for the outcomes, we exploit the variation of the outcome variables through waves by interacting RAMV eligibility with the wave fixed effect. We report our main results with and without the interaction. Standard errors are clustered at the family level. The outcomes of interests focus on four dimensions that measure the resilience of forced migrants to cope with the pandemic: (i) capacity to adhere to NPIs, (ii) access to medi- cal services when required, (iii) protection and detection against COVID-19, and (iv) re- silience to the economic impacts of the pandemic (see Appendix A for the description of each index). The ITT provides the causal effect of eligibility to the PEP under the assump- tion of conditional unconfoundedness and conditional on our ability to isolate selection into the PEP by sampling RAMV migrants with and without the PEP. V Results We find that PEP strengthened the ability of eligible RAMV migrants to lessen the health and economic impacts of the pandemic shock. Results of the ITT estimates are reported in Table 1. 3 Table 1. Impacts of PEP on COVID-19 Resilience NPI Medical Protection against Economic Adherence Access Contracting Resilience Index Index Covid-19 Index Index (1) (2) (3) (4) Panel A. ITT Direct Impacts 1[RAM Vijk = 1] 0.191 0.282 0.253 0.422 (0.047) (0.013) (0.046) (0.048) R-squared 0.067 0.215 0.061 0.136 Panel B. ITT coefficient interact with wave β1 =1[RAMVijk = 1] ×1[wave] 0.080 -0.102 – 0.043 (0.094) (0.026) – (0.099) β2 =1[RAMVijk = 1] 0.144 0.333 – 0.397 (0.074) (0.020) – (0.080) Wave Diff. Effect= β1 + β2 0.224 0.232 – 0.440 (0.060) (0.017) – (0.059) R-squared 0.067 0.220 – 0.136 Observations (Both waves) 2,318 2,618 2,640 2,337 Notes: All estimations include state and sampling-city fixed effects, and control for the interaction between wave dummies and the set of individual controls listed in Table A.5. 1[wave] takes the value of one for the second wave of the panel. Standard errors clustered at the family level are reported in parentheses. Better housing conditions for eligible PEP applicants facilitated the adherence to adopt NPI. The coefficient estimate for the NPI adherence index is statistically significant and econom- ically meaningful (0.19 standard deviation –SD – higher than the mean of the control group). Less crowded homes (9.5% for more rooms per family member) with private bathrooms (4.3 percentage points –p.p.– more likely to have a private bathroom) may have helped migrants to social distance, hand washing, and to have overall better san- itary conditions to prevent infection and the transmission of the virus (see Table A.1 in the Appendix for the estimation of individual outcomes). The gap between PEP and non- PEP eligible applicants in housing conditions, and presumably the risk of contracting the virus, widened between 2020 and 2021. PEP increased access to healthcare services. The effect of PEP on the medical access index is sizeable and statistically significant. The index for PEP eligible applicants is 28% larger with respect to the mean of the non-RAMV group. Health insurance access is 49.5 p.p higher for PEP eligible applicants, providing beneficiaries more medical assistance and prescriptions when ill: 10 p.p and 5.6 p.p respectively (See Table A.2 in the Appendix). Medical access, however, seemed to deteriorate between 2020 and 2021. Despite the ex- pansion of coverage of the health system for PEP eligible applicants, the gap between eligible and non-eligible PEP applicants shrank. Higher access to healthcare services resulted in higher detection and vaccination rates of COVID- 19. Household heads who were eligible for the program reported symptoms 2.1 p.p more, relative to non-eligible households; while the probability of contracting the virus for fam- ily members of PEP eligible households was 26.8 p.p. higher relative to non-eligible household members. PEP eligible applicants also had higher vaccination rates (8.5p.p.) relative to non-applicants (Table A.3 in the Appendix). PEP eligible applicants were more resilient to the negative economic shock brought by the pan- demic. The coefficient estimate on the impact of PEP on the index of economic resilience is not only statistically significant, but economically meaningful. The effect is 0.42 SD with 4 respect to the mean of the non-RAMV group (see Table 1). The findings show that eligible applicants were much better off: their family members were 15.6 p.p less likely to skip a meal during the previous month, the number of days in which they ate proteins was 15% higher, and they were 10.5% less likely to be evicted (see Table A.4 in the Appendix). What drove the stronger resilience of PEP eligible applicants? According to Ib´ ˜ et al. anez (2022), PEP holders were 26 p.p. more likely to be employed and their labor income was 24 p.p higher. In addition, they were 22 p.p and 44 p.p. more likely to receive mone- tary transfers from the government and to have bank accounts, respectively. PEP eligible applicants seem to be more self-reliant, and better able to cope with the negative shock brought by the pandemic. VI Conclusions Regularization programs are a promising pathway to improve the lives of forced mi- grants. Beneficiaries of the PEP program were more self-reliant and thus resilient to negative health and economic shocks, relative to the ineligible migrants. Programs for forced migrants have mostly concentrated on providing humanitarian assistance, while restricting their access to labor markets in the destination country and to regular social services. The evidence of this paper offers a complementary pathway. Regularization programs, by promoting the rapid integration of migrants into destination countries and their self-reliance, may not only improve the lives of migrants but may also reduce the short-term impacts on the local populations. References Baez, Javier E (2011) “Civil wars beyond their borders: The human capital and health consequences of hosting refugees,” Journal of Development Economics, 96, 391–408. ECDC (2021) “The Economics of the COVID-19 Pandemic in Poor Countries,”Technical report, European Centre for Disease Prevention and Control - ECDC. Ib´ ˜ anez, Ana Maria, Andres Moya, Mar´ e ıa Adelaida Ortega, Sandra V. Rozo, and Maria Jos´ Urbina (2022) “Life Out of the Shadows: Impacts of Amnesties in the Lives of Mi- grants,” IZA Discussion Papers 15049, Institute of Labor Economics (IZA). Ib´ ˜ anez, Ana Mar´ıa, Sandra V. Rozo, and Mar´ ıa J. Urbina (2021) “Forced Migration and the Spread of Infectious Diseases,” Journal of Health Economics, 79, 102491. Kling, Jeffrey R, Jeffrey B Liebman, and Lawrence F Katz (2007) “Experimental Analysis of Neighborhood Effects,” Econometrica, 75, 83–119. Kluge, Hans H, Zsuzsanna Jakab, Jozef Bartovic, Veronika D’Anna, and Santino Severoni (2020) “Refugee and migrant health in the COVID-19 response,” The Lancet. Montalvo, Jose G. and Marta Reynal-Querol (2007) “Fighting against Malaria: Prevent Wars while Waiting for the ”Miraculous” Vaccine,” The Review of Economics and Statis- tics, 89 (1), 165–177. an, Patricio, Sebasti´ Zambrano-Barrag´ ırez Hern´ an Ram´ andez, Luisa Feline Freier, Marta 5 Luzes, Rita Sobczyk, Alexander Rodr´ ıguez, and Charles Beach (2021) “The impact of COVID-19 on Venezuelan migrants’ access to health: A qualitative study in Colombian and Peruvian cities,” Journal of Migration and Health, 3, 100029. 6 Appendix for Online Publication July 25, 2018 March 24, 2020 September 1, Residency permit (Permiso Especial de President Duque 2020 Permanencia – PEP) for irregular announced a nationwide refugees registered in the RAMV is nationwide lockdown announced quarantine ended December 21, 2018 October 2020 February 2021 October 2021 February 2022 March 6, Survey Survey Survey April 6, 2018 June 8, 2018 August 2, 2018 PEP Survey 2020 collection collection collection RAMV registry RAMV registry PEP program program collection First case of ends starts ends starts ends starts ends starts Coronavirus SAMPLE 2020 2021 2022 2018 7 5 MONTHS OF LOCKDOWN Wave 1 Wave 2 442,462 refugees registered in Around 281,307 3,455 Households 2,307 Households 395 municipalities in Colombia people received the (35% of the territory) PEP document COVID-19 Pandemic Figure A.1. Registry in PEP. Program Roll-out and the COVID19 Pandemic: Timeline A Dependent Variables Description We group the following individual outcomes for each dimension estimating four indices using the methodology of Kling et al. (2007): 1. Capacity to adhere to NPI: overcrowding measure (number of rooms over number of family members), whether the household had a private bathroom, and whether the household had a private kitchen. 2. Access to medical services when required: whether the household is covered by the healthcare system, received medical attention when ill, paid out-of-pocket costs, and received the required medicine. 3. Protection and detection against COVID-19: whether the household head had COVID-19 symptoms (first wave), the number of household members that had COVID-19 (second wave), and whether the household head or the partner are vacci- nated (second wave). These variables are collected only on the first wave or second wave. 4. Resilience to the economic impacts of the pandemic: whether no family member had to skip a meal the previous month, the number of days that the household ate proteins in the last week, and whether the family did not face evictions. B Disaggregated Results of PEP Impacts Table A.1. Impacts of PEP on Capacity to adhere to NPI NPI Household Household Housing Adherence had private had private overcrowding Index kitchen bathroom measure (1) (2) (3) (4) 1[RAM Vijk = 1] 0.191 0.020 0.043 0.095 (0.047) (0.015) (0.017) (0.022) R-squared 0.066 0.037 0.048 0.124 Observations 2,318 2,285 2,293 2,266 Mean values (Non-RAMV) 0.000 0.856 0.822 0.477 Notes: Dependent variables: (i) NPI Adherence Index is constructed using the outcome variables of columns (ii) to (vi); (ii) Household had private kitchen is an indicator variable [=1] if the household has kitchen for the exclusive use of the home; (iii) household had private bathroom is an indicator variable [=1] if the household has sanitary service for the exclusive use of the home; (iv) Housing overcrowding measure is the number of rooms over number of family members in the household. All columns include department (Antioquia, Atl´ antico, Bogot´ a, and Norte de Santander), wave, sampling-city fixed effects, and controls for the interaction between wave dummies and the set of individual controls listed in Table A.5. Robust standard errors are reported in parentheses. 8 Table A.2. Impacts of PEP on Medical Access Medical Covered by Received medical Paid Received the Access the healthcare attention out-of-pocket required Index system when was ill the medical costs medicine (1) (2) (3) (4) (5) 1[RAM Vijk = 1] 0.282 0.495 0.102 0.039 0.056 (0.013) (0.018) (0.022) (0.026) (0.030) R-squared 0.215 0.317 0.046 0.077 0.071 Observations 2,618 2,592 1,694 1,370 1,412 Mean values (Non-RAMV) 0.291 0.0293 0.752 0.161 0.503 Notes: Dependent variables: (i) Medical access index is the average of the variables in columns (ii)-(v); (ii) Covered by the healthcare system is an indicator variable [=1] if the person reported to have subsidized or contributory healthcare system; (iii) Received medical attention is an indicator variable [=1] if the respondent or someone in their household received medical attention the last time they needed; (iv) Paid out-of-pocket the medical costs is an indicator variable [=1] if the respondent or someone in their household had to pay for the last time they went to the hospital; (v) Received the required medicine is an indicator variable [=1] if the respondent has a family member that takes any medicine or requires treatment regularly, and had the access to buy that medicine in Colombia. All columns include department (Antioquia, Atl´ antico, Bogot´a, and Norte de Santander), wave, sampling-city fixed effects, and controls for the interaction between wave dummies and the set of individual controls listed in Table A.5. Robust standard errors are reported in parentheses. Table A.3. Impacts of PEP on COVID-19 Prevention and Detection Household head Household Household head or had Covid-19 members that the partner symptoms had Covid-19 are vaccinated (1) (2) (3) 1[RAM Vijk = 1] 0.021 0.268 0.085 (0.013) (0.100) (0.020) R-squared 0.017 0.045 0.051 Observations 2,640 1,005 2,640 Mean values (Non-RAMV) 0.085 0.605 0.241 Notes: Dependent Variables: (i) Household head had symptoms is an indicator variable [=1] if the respondent reported to have COVID-19 symptoms, this variable is available for wave 1; (ii) Household members that had COVID-19 is the total number of people in the household who reported to have COVID-19 disease; (iii) Household head or the partner are vaccinated is an indicator variable [=1] if the respondent reported that him or his partner is vaccinated with at least the first dose of the COVID-19 vaccine, this variable is available for wave 2. All columns include department (Antioquia, Atl´ a, and Norte de Santander), sampling-city fixed effects, and the antico, Bogot´ individual controls listed in Table A.5. Robust standard errors are reported in parentheses. 9 Table A.4. Impacts of PEP on Economic Resilience Economic A Household Days that ate Did not Resilience member had not to protein in the face housing Index skip a meal household eviction (1) (2) (3) (4) 1[RAM Vijk = 1] 0.422 0.156 0.640 0.105 (0.048) (0.022) (0.108) (0.024) R-squared 0.136 2,334 2,334 1,519 Observations 2,337 0.101 0.083 0.332 Mean values (Non-RAMV) 0.000 0.507 4.253 0.667 Notes: Dependent Variables: (i) Economic Resilience Index is constructed using the outcome variables of columns (i) to (iv) of Table A.3 using the methodology of Kling et al. (2007); (ii) A household member had to skip a meal is an indicator [=1] if any member of the household have to skipped a meal because there were no food in the last month; (iii) Days that ate protein in the household is the number of days that the members of the household ate protein in the last week; (iv) Housing Eviction is an indicator [=1] if the household was evicted from their home due to COVID-19 crisis. All columns include department (Antioquia, Atl´ a, and Norte de antico, Bogot´ Santander), wave, sampling-city fixed effects, and controls for the interaction between wave dummies and the set of individual controls listed in Table A.5. Robust standard errors are reported in parentheses. 10 C Descriptive Statistics Table A.5. Descriptive Statistics - Baseline Controls Observations Average STD Min Max A. Demographic variables Time in Colombia (months) 1,385 53.318 10.628 26.067 189.433 Female [=1] 1,412 0.409 0.492 0.000 1.000 Age (years) 1,412 34.287 9.142 7.000 70.000 Number of children 1,412 1.548 1.481 0.000 10.000 B. Socioeconomic variables (Info. In Venezuela before migration) Years of education before migration 1,412 13.278 2.808 0.000 19.000 Ever worked [=1] 1,412 0.977 0.151 0.000 1.000 Employed at private firm [=1] 1,412 0.605 0.489 0.000 1.000 Employed with Government [=1] 1,412 0.152 0.359 0.000 1.000 Self-employed or employer [=1] 1,412 0.181 0.385 0.000 1.000 Written contract [=1] 1,412 0.513 0.500 0.000 1.000 Had smartphone [=1] 1,412 0.581 0.493 0.000 1.000 Owner of dwelling in Venezuela [=1] 1,412 0.868 0.338 0.000 1.000 Electricity in Venezuela [=1] 1,412 0.994 0.075 0.000 1.000 Running water in Venezuela [=1] 1,412 0.860 0.347 0.000 1.000 Sewage in Venezuela [=1] 1,412 0.936 0.246 0.000 1.000 C. Migration Variables Venezuela: parents or siblings [=1] 1,412 0.443 0.497 0.000 1.000 Household Venezuela: partner/spouse [=1] 1,412 0.552 0.497 0.000 1.000 Knew of job opportunity before migrating [=1] 1,412 0.346 0.476 0.000 1.000 Gap between last job and migration (months) 1,412 1.119 4.504 0.000 30.000 Migrated for health reasons [=1] 1,412 0.101 0.302 0.000 1.000 Friends or family in Colombia [=1] 1,412 0.733 0.443 0.000 1.000 11 D Attrition Table A.6. Attrition between Waves Attrited Household [=1] (1) Household Venezuela: parents or siblings [=1] -0.033 (0.024) Household Venezuela: partner/spouse [=1] -0.068 (0.025) Household Venezuela: others [=1] 0.003 (0.034) Knew of job opportunity before migrating [=1] -0.022 (0.022) Ever worked [=1] -0.004 (0.082) Employed at private firm [=1] -0.042 (0.054) Employed with Government [=1] -0.050 (0.060) Self-employed or employer [=1] -0.048 (0.057) Written contract [=1] 0.005 (0.025) Gap between last job and migration (months) -0.002 (0.002) Years of education before migration -0.010 (0.004) Migrated for health reasons 0.038 (0.034) Friends/family in Colombia -0.037 (0.023) Had smartphone [=1] 0.007 (0.021) Owner of dwelling in Venezuela [=1] 0.003 (0.031) Electricity in Venezuela [=1] -0.077 (0.129) Running water in Venezuela [=1] 0.046 (0.032) Sewage in Venezuela [=1] -0.022 (0.045) Female [=1] -0.051 (0.023) Age (years) -0.004 (0.001) Number of children 0.000 (0.008) Time in Colombia (months) -0.002 (0.001) Observations 2,200 Notes: Robust standard errors are reported in parentheses. 12 Table A.7. Attrition per Survey Round Attrited household [=1] 1[RAMV = 1] -0.051 (0.017) Mean Values (Non-RAMV) 0.402 Observations 4,200 All Controls Municipality FE Yes Department FE Yes Wave FE Yes Baseline Controls x Wave Yes 13