Week 1 - Migration and Development (PDF)

**WEEK 1 -- MIGRATION AND DEVELOPMENT : KEY FIGURES AND FACTS** **DEFINITIONS** **INTERNATIONAL MIGRANT** = "persons who change their country of residence" (UN) **ECONOMIC MIGRANTS** = employed in a foreign country other than their country of birth/citizenship (60%) **MIGRANT (OIM) =** person mo...

**WEEK 1 -- MIGRATION AND DEVELOPMENT : KEY FIGURES AND FACTS** **DEFINITIONS** **INTERNATIONAL MIGRANT** = "persons who change their country of residence" (UN) **ECONOMIC MIGRANTS** = employed in a foreign country other than their country of birth/citizenship (60%) **MIGRANT (OIM) =** person moved across an international border or within a State away from his habitual place of residence, regardless of (1) the person's legal status; (2) the movement is voluntary or involuntary; (3) the causes; (4) the length of the stay. **FORCIBLY DISPLACED PERSON** : - **Internally displaced persons (IDPs):** forced to leave their residence because of conflicts, violence, natural disasters and not crossed borders - **Refugee**: outside country of nationality because of fear of being persecuted for religion, nationality, social or political group, and unwilling to seek protection in her own country - **Asylum seeker**: on the process of acquiring refugee status Application process 6 months in EU, necessitates housing policies and degrees of integration Disparities between labour & refugees integration (due to institutions, norms and practices) Various definitions because of *nationality* and *residence* concepts - Greencard holders = different statistics and policies - Second-generation migrants never undertook migration - Foreign-born residents **Disparities** inside migrant's categories -- World Bank - - - - - Migrant's definition uses the birth country as reference instead of citizenship, does not include the higher **internal migration** (2009, 740m people) **2 categories** - - **MATCH AND MOTIVE MATRIX** Analysis of hybrid cases **Labor economics -- match** = skills of migrants (strong/weak match) indicate the [C/B of receiving] (high skills = higher contributions than integration cost, net gains from receiving migrants) **International law - motive** = insecurity level in home country (opportunity/fear) induces a [choice or obligation] to the destination country to protect **\> Strong match** = need of policies to maximise gains in home country (facilitate knowledge, protection) and destination (ensure rights and access to labour market, attract skills) **\> Weak match** = need to share and reduce costs multilaterally (ensure work access, integration) Need of 'refugees labour pathways' implementing hybrid solutions for a stronger match (ensure resilience to shock / legal pathways) and motive (extend protection) **CRITICIZING** Migration also influenced by **complex push and pull factors** **and hybrid situations** not included in the framework development & demographic patterns at the [origins], climate change, diasporas and migration chains, structural transformation at the destination (Rybczynski Theorem) **COMPLEXITY OF COMPTABILIZING INTERNATIONAL MIGRATION** **Distinction between [stocks]** (evaluation of a phenomenon at a precise and fixed point in time) to **[flows]** (time span) Migration stock due to accumulated net migration flows Disparities of accessibility to flow data = priority to inflows recording, distinction migratory flow -- travel **GENERAL UNCERTAINTY OF DATA SOURCES** - - - - - Progress through use of **algorithms** better forecast about future changes, neutral approach **2 main datasets** on migration flows = - **UN DESA International Migration Flows data set** (compiled data from nationally available stats) - **OECD's International Migration Database** Facebook social connectedness index predicator of diaspora population and displacement trajectories from uprooting conflicts (Ukraine) **KEY INFORMATION** - - - - - - - - - - **WEEK 2 -- ARE THERE GLOBAL GAINS FROM MIGRATION ?** **BACKGROUND** **Gallup World Poll** wish to move from 40% of the poorest quartile countries stopped by **emigration constraints** : - Credit constraints and limited information at the origin - Policy barriers in destination countries US Diversity Visa Lottery -- 50k/year permanent visas ≠ 15m/2018 **Back of the envelop calculation** prevision of intense gains (23 t\$) of migration from \$ transfers & wage gap between rich and poor countries Estimates of global gains 60 - 150% of world GDP **Um & increased trade** from the movement of less than **5% of pop° from poor countries**, greater than with trade barriers necessity of eliminating trade and capital movement barriers **GAINS FROM IMMIGRATION** \- Wage disparities between countries \- **Negative relation** between workers numbers and wage levels more L = lower W \- Poor workers move from foreign to home country to receive higher W, creating ↘ W in home country & equalization, until equilibrium at W' with full migration (C) \- **Gain from immigration for home** = triangle ABC (lower W & higher W) \- **Gain for foreign** **country** = A'BC (lower L but higher W from transfers = spillover effect & development) \- **Global gains of immigration** -- increase in world GDP = **AA'B** \- Measure of the gains (area of triangle) : **½ (W-W') \* (L'-L)** **\> Marginal labour productivity** workers should earn the income created by the increased labour productivity/their contribution (=wage) if ↘ of productivity, ↘ of wages **\> Law of diminishing returns** ↗ one factor while assuming the fixity of the other causes a ↘ of productivity, necessity of balance between the 2 FP° \> **Decreasing productivity** (↗ one factor = ↘ other factor) \> **Gap** between marginal productivity of workers and wages of capitalists provided to workers **WEEK 3 -- THE EFFECTS ON DESTINATION COUNTRIES** **CASE STUDY : THE MARIEL BOATLIFT** 1980 = boatloads of 125k/↗ 7% refugees from the Cuban port of Mariel arrived in Miami Unemployment of many of the low skilled refugees = expectation of ↘ of wages for Miami low-skilled worker Denied by empirical analyses (Card, 1990) = low-skilled immigrants did not pull down the wages **Rybczynski theorem** = if the economy can absorb the increase of one FP through the modification of the P° structure, the factor prices do not change as a result of immigration (adjustment of K and L ratio)  \(a) = Miami economic resilience in low-skilled fields \(b) = real Vain high-skilled industries fell faster **CASE STUDY : ISRAEL'S RUSSIANSPEAKING COMMUNITY** 1989-96 : 670k Russian Jews, +14% pop with higher qualifications increase of skilled Israeli workers **ELASTICITY OF LABOR DEMAND AT THE ORIGIN AND DESTINATION** \- Home country decline of average and low workers' wages & decline of price of typical consumption basket about 0.5% \- Foreign country departure of emigrants raise the wages of non-emigrants **SHORT AND LONG-RUN ESTIMATES** \- **SHORT-TERM** assumption of fix K & T, mobility of labour fall in wages ↗ of marginal products of the specific factors (K and T), therefore ↗ of rentals. \- **LONG-TERM** K perfectly mobile between industries, additional labour from immigration will be absorbed entirely by the labour- intensive industry. Labor-intensive industry will also absorb additional K and L from the capital-intensive industry K--L ratio does not change, so the wage and rentals remain the same as well **factor price insensitivity**. According to the Rybczynski theorem, immigration will lead to an increase in output in the labour-intensive industry and a decrease in the output of the capital-intensive industry. **"CREDIBILITY REVOLUTION" IN EMPIRICAL ECONOMICS 2000'** more data, rigorous methodologies, strong focus on applied economics and empirical research design, evidence-based policymaking as new standard **WEEK 3 -- THE EFFECTS ON DESTINATION COUNTRIES IN THE SHORT RUN** **SPECIFIC-FACTORS MODEL** prediction of lower wage & higher rentals in home country Model's assumptions: - 2 countries: Home and Foreign - 2 goods: manufacturing and food - 3 factors of production: labour, capital and land - Perfect competition prevails - Manufacturing uses labour and capital, and Agriculture uses labour and land. - Value of **marginal product of labour** **W = P~M~ \* MPL~M~** - **Diminishing returns** in each industry increase in the amount of labour = ↘ of MP~L~ **downward sloping curve** - Home wage is determined at A where 0~M~L unites of labour are used in manufacturing and 0~A~L units of labour in agriculture - Amount of labour used in manufacturing is measured from left 0^M^ to right, amount of labour used in agriculture is measured from right 0^A^ to left **L = L~A~ + L~M~** If foreign wage W\* is inferior to home wage W (differences in technologies or endowments of R) higher wage attracts foreign workers, **increasing L~M~ by ∆L** - **Axe expansion** by ∆L, origin of agriculture from 0~A~ to 0~A~' shift of MPL~A~ of ∆L - **New equilibrium at B** = increased amount of L in manufacturing to 0~M~L' and agriculture to 0~A~'L' - Both industries have more L but fixed amounts of K and T **fall of wage** to W' due to diminishing MPL **CASE STUDY : IMMIGRATION TO THE US** Representation of the share of foreign-born workers in the US workforce categorized by educational level **false assumption of homogeneous** workforce In the middle levels of education, immigrants not numerous enough to create a significant amount of competition with US-born workers Largest % of foreign for workers w/o a high school degree & for PHDs **\>** Illegal immigrants compete primarily with the lowest-educated workers, legal immigrants compete with workers at the highest educational levels the **greatest impact on labour will be for the lowest and highest educated U.S. workers.** **\>** The negative impact on wages is modest and is offset with capital moving between industries. **RENTALS ON CAPITAL AND LAND** 2 methods to compute the rentals **MPK~M~ and MPT~A~ rise and so do their rentals**. Recurrent **support from K and T owners** to open borders by increasing of potential employees Restriction on immigration should be seen as a compromise between entrepreneurs / landowners and local unions / workers who view migrants as a potential source of competition leading to lower wages, or large immigrants groups might influence the political outcome on immigration policy **EFFECTS ON INDUSTRY OUTPUTS**  \- ↗ of labour = **production possibilities frontier shifts** outward and the output of both industries increases, from A to B \- Short run, land and capital do not move between the industries, and the extra labour in the economy is **[shared equally] between both industries** **WEEK 4 -- EFFECTS OF IMMIGRATION IN THE LONG TERM** **HECKSHER -- OHLIN MODEL** **MODEL PRINCIPLES** Total amount of K in an economy sum of K used in shoe industry K~S~ and computer industry K~C~ **K = K~S~ +K~C~** Total available L L~S~ used in shoes L~S~ and L used in computers L~C~ **L = L~S~ +L~C~** **6 ASSUMPTIONS** - - - - - - K/L ratios are measured by the slopes of 0~S~A and 0~C~A The line OsA shows the amount of L and K used in shoes and the line OcA the amount of L and K used in computers **DETERMINATION OF REAL WAGES AND RENTALS** Use of the **[marginal productivity]** of L & K determined by the K-L ratios in each industry - - - Amount of L & K used along line 0~S~A = particular K/L ratio so a particular real wage & rental Immigration = L of Home increases of **L+∆L** , the origin for the shoe industry shifts from 0~S~ to 0~S~ Extra labour will be **allocated to shoes** labour-intensive industry K can move between the industries, hence industry outputs adjust so that the K/L ratios at point B are unchanged from the initial equilibrium A all new L from immigration is allocated to the shoe industry, and K and additional L are transferred from computers to shoes, keeping **the K/L ratio in both industries unchanged & full employment of additional L** **Ratio K/L identical MPL & MPK unchanged wage and rental stay identical** **EFFECTS ON GLOBAL PRODUCTION** More L and K used in the labour-intensive industry, less L and K used in the capital-intensive industry the **output of shoes expands and that of computers contracts** Production possibilities frontier of domestic P° with initial equilibrium at point A (point of tangency between the world price line and the PPF) Increase of L at Home, the PPF shifts outward output of shoes ↗ while the output of computers ↘, **equilibrium from point A to B.** Prices identical, so the **[slopes of the PPFs] are equal**. **RYBCZYNSKI THEOREM** With 2 goods and 2 factors, an increase in the amount of a factor will increase the output of the industry using that factor intensively and decrease the output of the other industry. Effect on factor prices as the **"factor price insensitivity result"** FP prices do not need to change as the economy can absorb the extra amount of factor by increasing the output of the industry using that factor intensively and reducing the output of the other industry **CONCLUSIONS** **WEEK 5 -- THE EFFECTS ON DESTINATION COUNTRIES** *WHAT FUTURE LEVEL OF MIGRATION IS FEASIBLE ?* I. **TODARO PARADOX** Todaro paradox excessive migration flows are a structural factor that fuels both unemployment and the growth of the informal economy (underground and criminal activities) Harris & Todaro 1970 for rural to urban migration, adapted to foreign to domestic migration The extent of **feasible migration** depends on the **willingness** of politicians at potential destinations, acting as [agents for their electorates], to allow immigration: - If the median voter at the destination holds relatively little capital or skill = possible limitation of the willingness to reduce impediments to emigration from poor countries. - The wealthier, better-educated, and less-nationalist individuals in rich destination countries have more favourable attitudes toward immigration - Noneconomic attitudes such as nationalism can also play an important role **ASSUMPTIONS** When a country has a demand for immigrant workers (∆xd \> 0), the flow of immigrants to that country (∆ xs \> 0) systematically exceeds the demand **(∆ xs \> ∆ xd**) under certain conditions Conditions = occurs when the receiving country' border is porous due to geographic factors and/or constitutional and international treaty obligations **EMPLOYMENT AND INFORMAL SECTOR** The **absence of formal social assistance to unemployment in DCs** fuels under-employment & informal sector (easy for low-skilled workers, productivity, absence of labour regulations & social benefits) **Interrelation between informal & formal sectors** excessive regulations & minimum wage (\> market clearing) in the formal sector feed the informal one The higher the minimum wage, the lower the equilibrium wage in the informal sector  The increase in wage in formal sector induces a decrease in the demand for workers at this price, but the informal wage (not concerned by the minimum wage), sees an increased demand for labour. As there is no increase of wage and an increase in workers, the wage diminishes. **Structural changes** in developed countries (sectoral shares 1800-2000) Decrease of added value & employment in agriculture & manufacturing, while important increase of both in services Urbanization of the working population & concentration in developing countries II. **LEWIS MODEL (1954)** **Lewis model** conditions for the structural transformation of an economy from predominantly agricultural to predominantly manufacturing Dualistic models of economic development (traditional vs modern), to study the economic structure prevailing in LDCs **ASSUMPTIONS** **A1** Infinite availability of labour (thanks to favourable demographic development); **A2** Constraints on development will be capital and natural resources. **A3** Two sectors : - Capitalist sector where K is employed and accumulated and whose profits go to capitalists; - Subsistence sector (= agricultural sector) which does not employ reproducible K and is characterized by family-based organization. - **SUBSISTENCE SECTOR** **Low workers' productivity** (↗ of workers = ↘ of MPL & diminishing returns) driving wages to **subsistence levels** Possible fall of wages below subsistence level (B) additional workers produce less than subsistence wage (OS) As there is a correlation between wages in the capitalist sector and those in the traditional sector, capitalists have interest to maintain low wages in the traditional sector by **perpetuating the low productivity** of its workers Little motivation to disseminate knowledge of new techniques among farmers **Hidden U** In reality, some workers are employed even though their cost (wages) exceeds their productivity explained by the organisation of family enterprise with product divided amongst the members, receiving a \"wage\" equal to S, (= average product of the family farm by dividing the total product by the number of workers)  - **CAPITALIST SECTOR** Workers are employed until the marginal product are equal to the wage (OW, labour quantity N') P' As K accumulates (reproducible), there is an increase in the MPL, shifting the curve to a **new equilibrium point** (P\'\') with increased employment (N\'\' - N') The **[capital accumulation process]** continues until there is no longer a surplus of labour in the traditional sector at point S=W (above = upward pressure on W) In practice, the rate of K accumulation in the capitalist sector (savings and investment rate) determines the speed of absorption of labour and economic development - For Lewis, the **underdevelopment of LDCs is firstly linked to the insufficient accumulation** **of the capitalist sector**, which fails to absorb an adequate labour force from the traditional sector **\ ** **WEEK 6 -- RURAL-URBAN MIGRATION** *Why is there continuing rural-urban migration despite high urban unemployment?* **HARRIS & TODARO PARADOX** **Assumptions** : - with unemployment (u) at the destination country, wages should fall until u is eliminated - when w is equal between countries (no wage gaps) it will no longer be beneficial to migrate **Harris & Todaro paradox** (1970) explains rural-to-urban migration (adapted to foreign-to-domestic migration) through **[urban labour market failure]** (explains why migration continues even with u) Consequences **excess migration flow feeds the informal economy & unemployment** **VARIABLES**: Wa= current rural wage Wm=manufacturing expected urban wage M=migration P = probability of employment in urban manufacturing (urban employment rate) 1-P= urban unemployment rate **[Equilibrium urban employment rate] that stops M P = Wa/Wm** - - **OUTCOMES** - - - **POLICY DILEMMAS** **EXTENSIONS OF THE MODEL -- factors of increasing migration :** - - - - - **EFFECTS ON PUBLIC FINANCE** **What future level of emigration** is feasible **POLICY OPTIONS** **WEEK 8 -- EFFECTS AT THE ORIGIN** **Remittances** are the most important **[source of migration]** boost families' income, stabilize consumption levels, facilitate overcoming financial barriers to develop productive activities & making investments in low-income countries, enable the acquisition of long-lasting assets (homes) **MIGRATION AS RISK MITIGATION STRATEGIES** Migration perceived as **informal strategy** **for mitigating risks** tool for diversifying risk among family members, communities, economic systems & geographical regions with remittance as central factor Empirical analysis is still limited = challenges of controlling endogenous migration decisions & lack of experimental data - **Climate change & resilience** 1. 2. - **Selection at the origin : competing theories** **PARADOX OF INCOME & MIGRATION** The lack of economic opportunity at home is denounced as key reason for emigration according to migrants But greater opportunities **would not reduce migration** rising income enables others to afford migration Bias from relying on data from migrants overlooks the effect of rising incomes on poors classic selecting of the dependent variable (migration choice). **Simpson's paradox** even if income and migration are negatively related within specific groups, rising incomes can increase migration by **shifting people between groups** Education rising incomes shift people into higher education groups with higher emigration **Net aggregate effect** in DCs, increasing aggregate income is associated with higher aggregate emigration **Mattei plan** from Meloni at 2024 African Summit objective of alleviating migration pressure by promoting development in departing countries But economic growth might ease financial constraints & lead to increased migration pressure on Italy **SELECTION AT THE ORIGIN** **Less chances of moving from poorest** & least qualified Better chances of success in destination countries (positive self-selection) but won't necessarily secure appropriate placements (brain waste) Effects on sending countries ↘ in human capital (crucial for LT growth) & loss of investment (education expenses) Theory of **Pigouvian tax** (Bhagwati & Hamda, 1974) question of tax amount & benefits on home country Narrative on the **"brain drain"** is less severe than reality - - - I. **Poverty trap** **Poverty trap** critical minimum asset threshold below which households cannot accumulate sufficient assets to improve their economic situation, leading to a cycle of poverty. Analysis of **regression-based vulnerability measures** can be improved by including **asset measures** **QUASI-EXPERIMENT** **Hurricane Mitch 1998** (Central America) experiment to assess the long-term effects of an environmental shock on household assets & different recovery patterns **Two factors** highest & lowest wealth quartile, with poor & good access to market  Estimation of a **244\$ poverty threshold** below which the households are likely to experience zero growth in assets & struggle to recover Supports the existence of poverty traps **ASSET-BASED APPROACH** **El Nino in Ethiopia** recurrency of droughts since 1983-5 & Northern hotspot vulnerability 1998-2000 recovery modest direct destruction of assets, but income losses of repeated crop failures forced households to **choose between preserving assets, or selling them** **to maintain current consumption and health** - - -  **Debates about a climate-induced poverty trap** by poor countries vulnerabilities to weather variability - - II. **Empirical framework & identification strategy** **Tanzania** poor & hot country, rising temperatures with most population in rural areas & agriculture rain-fed Use of **standards empirical growth model** to assess the **convergence deceleration** due to weather shocks (JR 2002) on food consumption **𝑙𝑛 𝑌𝑖𝑡 − 𝑙𝑛 𝑌𝑖𝑡−1 = 𝛼𝑙𝑛 𝑌~𝑖𝑡~−1 +𝛽∆𝑇𝑒𝑚𝑝~𝑖𝑡~ +𝛾∆𝑃𝑟𝑒~𝑖𝑡~ + 𝛺𝑍~𝑖𝑡~ + 𝜔𝑋~𝑖𝑡~ + 𝜇~𝑖~ +𝑚~𝑖𝑡~ +𝜃~𝑟𝑡~ +𝜀~𝑖𝑡~** **Variables**: **Y** : per adult-equivalent HH food consumption **α:** average conditional convergence parameter **Temp and Pre** are temperature and precipitation "anomalies" **Detectement of a** **critical [consumption] threshold** (Hansen 2000) the effects on current consumption growth depend on whether a HHs past consumption level (*Yit-1*) is above or below a threshold **lower regime (βl)** for HH\< threshold, **upper regime (β^𝒖^​)** for HH\< it  **Baseline results Heterogeneous effects of climates** **Positive/neutral marginal effect with HH with higher previous consumption (upper regime), while lower HHs have a minimal/negative ME** **CONCLUSIONS** - - - - - **WEEK 10 -- CLIMATE CHANGE & MIGRATION : RISK EXPOSURE, VULNERABILITY AND RESILENCE** I. **Typology of climate migration** Weather (atmosphere conditions over a short period of time) vs climate change (over LT & large space) **MULTIFACETED IMPACTS ON DETERMINANTS**  The environmental effects increase the stressors on already vulnerable HHs by interconnectedness, further depleting their resilience sphere **DIFFERENT MIGRATION RESPONSES** according to the assets **Ambiguous migratory effects** of climatic shocks increase vulnerability & decreases capability to migrate **IMPORTANCE OF CONSUMPTION ON MIGRATORY DECISIONS** Existence of **'downside risk'** **of fear to face hardship** z = 100, person with yi = 120 with rain (p~RA~ = 50%), y~i~ = 60 in drought (P~DR~ = 50%), yi = 90 regardless of the weather Use of **consumption** as more accurate to define the **poverty line (z)** **Transitory poverty** individuals who may fall below the poverty line temporarily due to shocks, but above on LT **Chronic poverty** individuals consistently below the poverty line due to structural factors  Two types of **vulnerability** (risk of falling into poverty in the future) impacted by external factors - - **Resilience** Illustration of how assets influence the coping mechanisms & recovery trajectories vulnerability assessment through A & LT impacts V = vulnerability R = recovery/resilience Different recovery trajectories due to assets HHs below poverty trap have positive V, thus no R HHS with no V & good assets have positive R Assessing the global impact in both origin & destination countries (O-D), highlighting the **interconnectedness of migration, CC & economic factors** \- Slow-onset events gradual changes \- Fast-onset natural disasters Mediating channels influencing the ability to manage CC & perception of threats **KEY FACTS** Internal migration more impacted by CC, although data is scarce 10% of world population (2009, Human Development Report) & 3% of international as they receive more attention **\ ** **WEEK 11 -- CLIMATE CHANGE, ADAPTATION & MIGRATION** **CLIMATE-MIGRATION RELATIONSHIP** **Hoffman 2020 survey** on weather shocks determines a quantitative relation between env. factors & migrations - Estimation of 2000m displaced people by 2050 (Groundwell WB report, 2021) difficulty of identifying (who, mediating factors, lack of evidence of climate-induced immobility traps) due to the **intrinsic heterogeneity & context-dependency** of household responses to climate-related shocks - Need to go beyond average effects & traditional econometric tools **Debates on theorical qualification through the inexistence of "climate refugee" in IL** - **\>** Renaud, 2011 introduced the notion of '**environmentally motivated migrants'** for people who leave a constantly deteriorating environment. **\>** Increased IL recognition of the relationship of '**environmentally displaced people** (Teitiota vs New Zealand's case -- UNHCR, 2020) **CONCEPTUAL FRAMEWORK** Building CF on theorical models of poverty traps & NELM to identify granular & non-parametric evidence on - **Key channels** of CM - **Interplay** between local adaptation & household migration responses - Presence of **climate-induced immobility traps** **Empirical ML** (data driven) test longitudinal & multi-topic household data from Nigeria **Two main data resources** Nigeria General Household Surveys (2010-2019) & Standardized Precipitation-Evapotranspiration Index (SPEI, monthly data from the Global SPEI Database) Extension of asset-based poverty traps approach by **allowing multiple factors** in the definition of **welfare trajectories** Focusing on substitute management strategies = **adaptative capacity** (asset-based approach) & **household-level migration** (NELM) **Heterogeneity in HH** **migration capacity** **results in divergent trajectories** certain groups of HH are trapped in immobility, whereas for some groups [immobility is a vulnerability choice] Group 1 less likely mobile as they would benefit from the additional resources above a **"voluntary immobility threshold"** Only group 2 is mobile **EMPIRICAL TEST** **Hypothesis** **Variables** We apply **descriptive statistics** for the main variables (mean, SD, min & max weight) **Empirical method** **[Causal forests]** (Wagner & Athey, 2018) adaptation of random forests to the task of predicting heterogeneity in causal effects Estimation of **Conditional Average Treatment Effects** (CATEs) non-parametric estimates of treatment effect heterogeneity (= variation of the effect) based on the covariates (treatment effect modifiers) Ensemble of **causal trees** splitting the data to maximize treatment effect heterogeneity across leaves; Each individual tree searches for the subgroups where the treatment effects differ most final prediction is a **weighted average** over the predictions across all trees (indicating the weight on the final result) Better from traditional subgroup analysis defining a priori the subgroups here, the researcher **only chooses the set of covariates** (potential drivers of heterogeneity) defining subgroups, but not the specific subgroups or critical thresholds  **Results : treatment effect estimates** Household exposed to rainfall shocks are 6 percentage points more likely to have a family member who migrants **Result is not statistically significant & misleading** conceals the heterogeneity of mobility responses in size & direction, meaning that the relationship can have opposite effects (migration implausible for groups) **Conditional average treatment effect** Observed heterogeneity is mainly determined by **pre-shock assets** (A), **repeated climate shocks** (B) & ***in situ* adaptation score** (C) **Adaptation as substitute** for migration (H3), low assets & exposed to repeated shocks act as **immobility traps** (H1 & H2) In contrast, **relationship between consumption (D) & education (E) is almost flat** **No voluntary immobility threshold** detected (H4) Test confirming that the mechanism are relevant to the assumption that CC will increase migration **Placebo test** If we reverse the assumptions & all implied relationships are flat (no heterogeneity), then it shows that the heterogeneity is systematic & not due to noise in data  Correlation between pairs of covariates (darker shade) **Open issues : empirics** - - - **[Hypothesis]** **[Simplified Expected Findings]** ----------------------------------------------------------------------------------------------------------------- ------------------------------------------------ H1 There is a minimum level of wealth below which people cannot move due to climate impacts. H2 Households that face repeated weather shocks are more likely to get stuck in place due to limited resources. H3 Local adaptation and migration are alternative strategies to manage climate risks. H4 There is a level of wealth above which people no longer need or want to migrate. **[Treatment effect estimates]** **average** & overall impact of a treatment without accounting for differences between individuals or subgroups. **[CATES]** **conditional** because it accounts for specific factors or subgroups, revealing how treatment effects differ for different populations