1. Remittances and Temporary Migration Christian Dustmann and Josep Mestres. University College London and CReAM Royal Economic Society PhD Presentation 17th January, 2009

2. Remittances and Temporary Migration Objective: Establish How Migrant’s Intention to Return to Home Country Modifies Remittance Behaviour. Results: Changes in Return Plans Lead to Large Changes in Remittances Flows. A migrant that intends to return is 16.2 percentage points more likely to remit. In addition, the amount remitted is more than doubled. Our results imply that the yearly decline in intentions to return observed in our data will imply a large change in remittance flows : -an increase equivalent to 0.005% GDP for Host Country (Germany) - a decrease equivalent to 0.1% GDP for Home Country (Turkey)

3. Main Contributions Establish a Casual link between Intention to Return and Remittance Behaviour Show how Different Remittance Purposes are differently affected by Intention to Return Highlight the Relevance of Temporariness of Migration for Migrant Behaviour and the Implications for Migration Policies.

4. Talk Structure Motivation Data Estimation Strategy Results Policy Implications Savings and Asset Allocation Conclusions

5. Motivation

6. Remittances Migrant’s Transfers to the Home Country Remittances help economic development and poverty reduction in receiving countries. They are as well a non-negligible outflow of capital for sending countries. Little is known about how remittances interact with migrant behavior and forms of migration. Existing evidence show different motives that trigger remittances (altruistic and non-altruistic)

7. Remittances’ Motives Motivations to Remit are diverse: Altruism Exchange Insurance Family Loan Arrangements Inheritance Mixed Motives Previous Literature Lucas & Stark (1985,1988), Hoddinott(1994), Funkhouser(1995), Poirine(1997), Agarwal & Horowitz (2002), De La Briere et al (2002), Okonkwo Osili(2007), Amuedo-Dorantes & Pozo (2006), Docquier & Rapoport (2006)

8. Migrants’ Return Plans Intention and Realisations Do Not Necessarily Coincide Current Migrant Behaviour is influenced by Current Intention, not by Future Realisation. In our case, we Study How Remittance Behaviour is Affected by Migrant Intentions to Return to the Home Country. Intention to Return to Home Country in year 1984 Return between 1984 and 2003YesNoTotal Yes 28382,03% 6217,97%345 No 68365,61% 35834,39%1041 Total 96669,70% 42230,30%1386 Source: GSOEP (1984-2003) – “A Dynamic Model of Return Migration” Adda, Dustmann and Mestres (2008)

9. Data: German Socio-Economic Panel (GSOEP) 12 waves of GSOEP corresponding to years 1984-1995. Migrant Households whose head was born in Turkey, Greece, Yugoslavia, Italy and Spain. In each wave of panel, individuals are asked about –Their Intentions to Return to Home Country –Remittance Behaviour –Remittance Amount and Purpose: Family Support Savings at Home Other (ex. help neighbours and friends) Migrant Individual Characteristics Household Characteristics at Host and at Home Country

10. Summary Statistics MeanStd. Dev. Sex 0.8340.371 Age 45.21012.210 Age At Arrival 25.1788.587 Years Since Migration 19.0266.270 Number Years Education 9.5881.925 Employed (Yes=1, No=0) 0.769 Non Single (Yes=1, No=0) 0.877 Native Partner (Yes=1, No=0) 0.058 Spouse Abroad (Yes=1, No=0) 0.093 Children Abroad (Yes=1, No=0) 0.137 Rural Childhood (Yes=1, No=0) 0.418 Temporary (Yes=1, No=0) 0.517 Note: Calculations based on GSOEP data, 1984-1990,1992,1994. Individual information corresponds to the head of household.

11. Remittance Behaviour - Temporary and Family Abroad Percent Households Remitting Total Amount (in 2002 Euros) Total Amount As % HH Disposable Income Percent Households Remitting to Family Percent Households Remitting for Savings Percent Households Remitting to Others Total Migrant46.22%1,7308.26%33.63%6.61%9.99% Note: Calculations based on GSOEP data (1984-1990,1992,1994), on household level, using household weights. Information on temporary intention, spouse and children abroad corresponds to the head of household. "No Spouse Abroad" includes single heads of household. "No children abroad" includes heads of household with children in the host country and without children.

12. Remittance Behaviour - Temporary and Family Abroad Percent Households Remitting Total Amount (in 2002 Euros) Total Amount As % HH Disposable Income Percent Households Remitting to Family Percent Households Remitting for Savings Percent Households Remitting to Others Total Migrant46.22%1,7308.26%33.63%6.61%9.99% Permanent25.97%8244.04%18.84%2.76%5.80% Temporary51.09%2,0569.87%37.05%8.26%11.65% Note: Calculations based on GSOEP data (1984-1990,1992,1994), on household level, using household weights. Information on temporary intention, spouse and children abroad corresponds to the head of household. "No Spouse Abroad" includes single heads of household. "No children abroad" includes heads of household with children in the host country and without children.

13. Remittance Behaviour - Temporary and Family Abroad Percent Households Remitting Total Amount (in 2002 Euros) Total Amount As % HH Disposable Income Percent Households Remitting to Family Percent Households Remitting for Savings Percent Households Remitting to Others Total Migrant46.22%1,7308.26%33.63%6.61%9.99% Permanent25.97%8244.04%18.84%2.76%5.80% Temporary51.09%2,0569.87%37.05%8.26%11.65% No Spouse Abroad41.59%1,5016.68%29.55%6.34%9.91% Spouse Abroad66.23%2,98819.09%55%5.37%5.72% No Children Abroad41.75%1,4556.59%29.36%6.44%9.97% Children Abroad69.66%3,28118.32%57.87%7.23%9.07% Note: Calculations based on GSOEP data (1984-1990,1992,1994), on household level, using household weights. Information on temporary intention, spouse and children abroad corresponds to the head of household. "No Spouse Abroad" includes single heads of household. "No children abroad" includes heads of household with children in the host country and without children.

14. Being Temporary Migrant Changes Over Time

15. Talk Structure Motivation Data Estimation Strategy Results Policy Implications Conclusions

16. Estimation Strategy Y it : Propensity to Remit / Amount Remitted I.R. it : Intention to Return (Temporary Migration) X 1 it :Individual and Household Characteristics X 2 it : Spouse and Children Abroad. Rural Childhood Y it = β 0 + β 1 X 1 it +β 2 X 2 it + δ I.R. it + α i + u it

17. Estimation Strategy Unobserved Individual Fixed Effect α i correlated to Intention to Return I.R. it Measurement Error in Intention to Return I.R. it -> Attenuation Bias (exacerbated in FE) Reverse Causality -> Past Remittances might influence Current and Future I.R. Selection due to Return Migration –those with higher intention to Return more likely to leave –δ biased (downward if those who remain less likely to remit)

18. Estimation Strategy GMM IV Estimator (Arellano-Bond(1991)-Arellano-Bover(1995)) will give Unbiased and Consistent estimates of δ –Forward Orthogonal Deviations Transformation to eliminate fixed effect: Y o it = β 0 + β 1 X o 1it +β 2 X o 2it + δ I.R. o it + u o it –γ M.E. o it where o indicates the forward orthogonal deviation (difference between the current observation and the average of the remaining future observation) –I.V. : all past Intentions to Return I.R. it-1,I.R. it-2, …, I.R. i1 E(u o it – γ M.E. o it | I.R. it-s,s>0 )=0 to solve both Measurement Error and Reverse Causality Assuming that Real Intention I.R.* it+s uncorrelated with Measurement Error ME it (E ( ME it, I.R.* it+s )=0, s≠0 ) Future Remittances Y it+s do not affect Current Intentions I.R. it

19. Estimation Strategy – Identification of β 2 and δ Control for Unobserved Individual Fixed Effect α i -> E( u it | X 1 it, X 2 it, I.R. it ) = 0 -> Fixed Effects (or FD) estimator will give Unbiased and Consistent estimates of β 2 and δ Measurement Error in Intention to Return I.R. it -> Attenuation Bias (exacerbated in FE) -> if Real Intention I.R.* it+s uncorrelated with Measurement Error ME it (E ( ME it, I.R.* it+s )=0, s≠0 ): use as Instrumental Variable Past Values of Intentions.

20. Estimation Strategy Reverse Causality -> Past Remittances might influence Current and Future I.R. -> I.V. strategy will address it as well Selection due to Return Migration - those with higher intention to Return more likely to leave -δ biased (downward if those who remain less likely to remit) GMM IV Estimator (Arellano-Bond(1991)-Arellano-Bover(1995)) will give Unbiased and Consistent estimates of δ –Orthogonal Deviations Transformation to eliminate fixed effect –I.V.: all past Intentions to Return I.R. it = γ 0 + γ 1 X it +γ 2 Y it-s + v it, s>0

21. Propensity to Remit – LPM TotalFamily SupportSavings for LaterOther Temporary Migrant 0.134**0.096**0.101**0.057**0.029**0.030**0.033**0.043** (0.020) (0.008) (0.013) Spouse Abroad 0.097*0.1-0.026-0.022 (0.046)(0.054)(0.017)(0.024) Children Abroad 0.141**0.177**0.000-0.005 (0.031) (0.014)(0.016) Observations 8,9177,7098,9177,7098,9177,7098,9177,709 R-squared 0.1620.1720.1220.1410.0330.0340.0320.034 Note: GSOEP data (1984-1994).Household Level. Includes Time and Country Dummies. Other regressors include Age, Gender, Years since Migration and its square, Education Level, Marital Status, Household Income, Empolyment Status and Number of Adults and Children in the Host Country Household. Clustered Robust Standard Errors Reported.

22. Amount Remitted – OLS TotalFamily SupportSavings for LaterOther Temporary Migrant 1.114**0.837**0.819**0.483**0.222**0.235**0.250*0.337** (0.163)(0.162)(0.160)(0.155)(0.062)(0.064)(0.098)(0.100) Spouse Abroad 0.899*0.925*-0.219-0.204 (0.376)(0.410)(0.132)(0.170) Children Abroad 1.256**1.565**-0.008-0.051 (0.257)(0.251)(0.109)(0.119) Observations 89177709891777098917770989177709 R-squared 0.1770.1930.1350.1620.0330.0340.0330.036 Note: GSOEP data (1984-1994).Household Level. Includes Time and Country Dummies. Other regressors include Age, Gender, Years since Migration and its square, Education Level, Marital Status, Household Income, Empolyment Status and Number of Adults and Children in the Host Country Household. Clustered Robust Standard Errors Reported.

23. Propensity To Remit – FE/GMM Results Household Sent Remittances (=1 Yes, =0 No) OLSFE FE Orthogonal Deviations GMM a (1)(2)(3)(4) Temporary Migrant Coefficient0.096**0.032*0.034**0.162* (0.020)(0.015)(0.0145)(0.069) (a) AR(1) Test: z=-10.42 P-Value=0 AR(2) Test: z=-0112 P-Value=0.911 Hansen Test =31.88 P-Value=0.619 Note: GSOEP data (1984-1990,1992,1994). Household level, using household weights. All specifications include time dummies and condition on marital status and employment status of the head of the household, as well as household income, employment status of the rest of the members of the household, number of adults and children in the host country household and indicator variables for spouse and children in the home country. OLS specification includes in addition country dummies, age, years since migration (and its square), education, gender and childhood in a rural area in the home country. Standard errors are clustered by household. Instrumental variables used in GMM: lags in average intention to return (t-1, …, 1) of other household members.

24. Amount Remitted – FE/GMM Results Logarithm (Amount Remitted+1) OLSFE FE Orthogonal DeviationsGMM a Temporary0.837**0.244*0.253**1.396** (0.162)(0.121)(0.115)(0.542) (a) AR(1) Test: z=-10.42 P-Value=0 AR(2) Test: z=-0112 P-Value=0.911 Hansen Test =31.88 P-Value=0.619 Note: GSOEP data (1984-1990,1992,1994). Household level, using household weights. All specifications include time dummies and condition on marital status and employment status of the head of the household, as well as household income, employment status of the rest of the members of the household, number of adults and children in the host country household and indicator variables for spouse and children in the home country. OLS specification includes in addition country dummies, age, years since migration (and its square), education, gender and childhood in a rural area in the home country. Standard errors are clustered by household. Instrumental variables used in GMM: lags in average intention to return (t-1, …, 1) of other household members.

25. Policy Implications Migration Policies that affect Temporariness of Migration will modify Remittance Flows In our sample: Average Amount Remitted HH=1736 € (@500€ per person) Aggregate Equivalent Germany (1995):+2 billion Euros (equivalent to 0.12% German GDP) Average Yearly Decrease In Temporary Intentions (based on GMM estimates) equivalent to a yearly increase of 83 million Euros (0.005% of German GDP in 1995) in the host country (Germany) equivalent to yearly decrease of 52 million Euros (0.1% of Turkish GDP in 1995) in the home country (Turkey) Remittances(2.1% GDP), FDI(0.51% GDP), Foreign Aid(0.18% GDP).

26. Conclusions Our results show that Changes in Return Plans Lead to Large Changes in Remittances Flows. Different Remittance Purposes are Affected Differently by Return Plans. Our estimation strategy controls for Unobserved FE, Measurement Error and Reverse Causality. Policies that affect Migrant’s Intentions will have a huge impact in Remittances Flows for both Host and Home Countries. Savings and Asset Allocation are also associated with Migrant’s Return Plans.

27. Migration and Savings Allocation Savings in Home and Host Country available only in years 1992 and 1994. Migrants declare the amount of savings in both locations.

28. Savings - Home vs. Host Country OLS Amount SavingsRatio Home vs Total TotalHome CountryHost Country Temporary0.3580.229**0.2180.025*** (0.232)(0.093)(0.233)(0.008) Tobit Amount SavingsRatio Home vs Total TotalHome CountryHost Country Temporary0.421*0.164**0.2540.015** (0.255)(0.073)(0.253)(0.006) Observations1225124812251243 * significant at 10%; ** significant at 5%; *** significant at 1% Note: GSOEP data (1992 and 1994). Household level. All specifications include time and country dummies and condition on age, gender, years since migration (and its square), education, marital status, household income, employment status and number of adults and children in the host country household. Standard errors are clustered by head of household. Tobit results show unconditional marginal effects.

29. Migration and Asset Allocation Special Asset Holdings supplement in 1988 Migrants describe assets, savings and housing both in home and host countries accumulated during their stay in host country. Construct Average Temporary Migration for years 1984-1988.

30. Property Ownership – Home vs Host Country LPMOLSTobit (=1 Yes, =0 No) Logarithm of Amount Remitted Logarithm of Amount Remitted Total (Home+Host Country)0.141***1.594***1.750*** (0.053)(0.563)(0.641) Note: GSOEP data (1988). Household Level. Includes Time and Country Dummies. Other regressors include Age(divided by 10), Years since Migration (divided by 10) and its square(divided by 100), Education Level,Marital Status, Household Income, Empolyment Status and Number of Adults and Children in the Host Country Household. Clustered Robust Standard Errors Reported.

31. Property Ownership – Home vs Host Country LPMOLSTobit (=1 Yes, =0 No) Logarithm of Amount Remitted Logarithm of Amount Remitted Total (Home+Host Country)0.141***1.594***1.750*** (0.053)(0.563)(0.641) Host Country-0.078**-0.928**-0.296*** (0.034)(0.380)(0.114) Home Country0.203***2.435***2.645*** (0.049)(0.521)(0.641) Note: GSOEP data (1988). Household Level. Includes Time and Country Dummies. Other regressors include Age(divided by 10), Years since Migration (divided by 10) and its square(divided by 100), Education Level,Marital Status, Household Income, Empolyment Status and Number of Adults and Children in the Host Country Household. Clustered Robust Standard Errors Reported.

32. Property Ownership – Home vs Host Country LPMOLSTobitOLSTobit (=1 Yes, =0 No) Logarithm of Amount Remitted Logarithm of Amount Remitted Ratio of Amounts Ratio of Amounts Total (Home+Host Country)0.141***1.594***1.750*** (0.053)(0.563)(0.641) Host Country-0.078**-0.928**-0.296*** (0.034)(0.380)(0.114) Home Country0.203***2.435***2.645*** (0.049)(0.521)(0.641) Ratio Home vs Total Property0.255***0.285*** (0.063)(0.055) Note: GSOEP data (1988). Household Level. Includes Time and Country Dummies. Other regressors include Age(divided by 10), Years since Migration (divided by 10) and its square(divided by 100), Education Level,Marital Status, Household Income, Empolyment Status and Number of Adults and Children in the Host Country Household. Clustered Robust Standard Errors Reported.

33. Being Temporary Migrant Changes Over Time Change in Intention to Return in year t compared to year t-1 (in Percentage) 1985198619871988198919901991199219931994 Change21.8618.9217.0818.5216.4018.8918.9917.9217.5414.33 Temporary -> Permanent9.659.969.4112.199.1910.5110.4511.109.456.89 Permanent -> Temporary12.218.977.676.337.218.388.546.828.097.44 No Change78.1481.0882.9281.4883.6081.1181.0182.0882.4685.67 Temporary - Temporary58.0759.1958.7252.8847.8343.8440.9037.0733.7232.60 Permanent - Permanent20.0721.8824.2028.6035.7737.2740.1045.0148.7453.06 Sample Size11711115109510421012990995982952914

34. Amount Remitted – Tobit Household Sent Remittances (=1 Yes, =0 No) TotalFamily SupportSavings for LaterOther Temporary 1.271**1.055**0.833**0.568**0.267**0.297**0.320**0.377** (0.095)(0.107)(0.080)(0.088)(0.045)(0.053)(0.054)(0.062) Spouse Abroad0.986**1.365**-0.198-0.321* (0.212)(0.166)(0.108)(0.149) Children Abroad1.445**1.634**0.017-0.118 (0.146)(0.116)(0.068)(0.089) Rural Childhood0.416**0.223**0.0760.069 (0.097)(0.080)(0.045)(0.055) Observations8,9197,7098,9197,7098,9197,7098,9197,709 Pseudo R-sq0.0460.0480.0440.0540.0350.0340.0250.026 Note: GSOEP data (1984-1994).Household Level. Includes Time and Country Dummies. Other regressors include Age(divided by 10), Years since Migration (divided by 10) and its square(divided by 100), Education Level,Marital Status, Household Income, Empolyment Status and Number of Adults and Children in the Host Country Household. Clustered Robust Standard Errors Reported.

35. Probability to Remit - Logit Temporary0.1670.1360.1160.0810.0360.0410.0440.054 (0.012)**(0.014)**(0.011)**(0.012)**(0.006)**(0.007)** (0.008)** Spouse Abroad0.1130.153-0.032-0.057 (0.031)**(0.026)**(0.013)*(0.021)** Children Abroad0.1870.1900.002-0.015 (0.021)**(0.017)**(0.008)(0.012) Rural Childhood0.0540.0320.009 (0.013)**(0.011)**(0.005)(0.007) Observations 89197709891977098919770989197709 Standard errors in parenthesesMarginal Effects Reported * significant at 5%; ** significant at 1%

36. Return Plans LPM Temporary (year t-1)0.481**0.374**0.324** (0.023)(0.027)(0.031) Temporary (year t-2)0.233**0.176** (0.023)(0.030) Temporary (year t-3)0.178** (0.035) Number Observations5,6864,5913,628 R-Squared0.2990.3350.35 Note: GSOEP data (1984-1990,1992,1994).Household Level. Includes Time and Country Dummies. Other regressors include Age, Years since Migration and its square, Education Level, Marital Status, Household Income, Employment Status and Number of Adults and Children in the Host Country Household. Clustered Robust Standard Errors Reported. Return Plans in year t as a Function of Return Plans in t-1, t-2,...

37. GMM Results – Alternative IVs Dependent Variable: Household Sent Remittances (=1 Yes, =0 No) GMM (4)(5)(6) IV: Past Return Plans Head Household IV: Past Return Plans Head Household and Other HH Members IV: Past Return Plans Other HH Members Temporary0.168 a 0.163 b 0.162 c 0.034**0.031**0.069* Dependent Variable: Logarithm of Amount Remitted GMM Temporary1.288 d 1.314 e 1.396 f 0.265**0.240**0.542**

38. Probability to Remit – LPM - Returner Returner 0.0640.0550.0460.0380.0420.0390.0120.014 (0.023)**(0.023)*(0.022)*(0.022)(3.29)**(3.10)**(0.97)(1.06) Spouse Abroad 0.0860.157-0.044-0.040 (0.036)*(0.037)**(2.54)*(2.22)* Children Abroad 0.1790.2170.008-0.009 (0.023)* * (0.61)(0.68) Rural Childhood 0.0470.0300.0090.010 (0.019)*(0.018)(1.07)(0.93) Observations 8.9197.7098.9197.7098.9197.7098.9197.709 R-squared 0.120.140.090.130.03 Robust standard errors in parentheses

39. Simple Model on Remittances and Savings Max U 1 (c 1 )+γ W 1 (R) (t=1) +p[ βŨ 2 (c E2 )+ δ v(R) ] (t=2) if Return +(1 -p)[U 2 (c I2 )] (t=2) if No Return subject to w 1 =c 1 +s 1 +R; c E2 = w E2 +rs 1 ; c I2 = w I2 +s 1 If p=0 (Permanent Migration): –Savings Equalize Marginal Utility Consumption –Only Altruistic Motives to Remit If p>0 (Temporary Migration): –Both Altruistic and Non-Altruistic Motives to Remit –Additional Savings Motives due to r ≥1, β ≥ 1

40. Comparative Statics Parameters dγdγdβdβdδdδdω1dω1 dω I2 dω E2 dp ds1ds1 -+-+--+ / - dRdR+-++++