International Migration from Indonesia: Stylized Facts∗ Samuel Bazzi† University of California, San Diego October 2012 In this brief note, I use new administrative and survey data to document several stylized facts on international migration from Indonesia. 1. The typical contract duration for temporary international migrants from Indonesia is 2-3 years. Two pieces of evidence support this claim. First, I can show that nearly 80 percent of international contract migrants departing through Jakarta—from where 60 percent of all legal migrants embark and nearly 100 percent of migrants destined for the Middle East depart—return within 2-3 years. The estimate is obtained by combining daily administrative data on contract migration flows through Jakarta international airport departing in 2006-2008 and returning between 2008-2011. The data are collected by immigration authorities and provided by the government’s overseas employment agency, BNP2TKI. Second, the only available nationally representative estimates based on individual-level survey data suggest that approximately 75 percent of migrants work abroad for 2-3 years (Bank Indonesia, 2009). In both cases, the median contract is 2 years. 2. There exists strong ethnic and religious sorting of Indonesian emigrants across destination countries. Using Village Potential data from 2005 on plurality destination choice by village1 and a multinomial logit specification, Table 1 shows that villages with relatively more Christians and Muslims are more likely to have migrants in Malaysia/Singpaore than in East Asian destinations. The likelihood that a plurality of migrants go to the Middle East rather than Malaysia/Singapore is sharply declining in the share of Christians in the population. I also find that the share of ethnic Arabs raises the probability that village v has a plurality of its migrants in Arab countries rather than in Malaysia/Singapore as of 2005. Meanwhile, the probability of migration to Hong Kong and Taiwan relative to Malaysia/Singapore is increasing in the share of ethnic Chinese in the population. Two other intuitive findings corroborate evidence from elsewhere and deserve mention. First, female migrant recruiters are much more likely in the case of migration to the Middle East and East Asia—primary destinations for housemaids—relative to Malaysia/Singapore. Second, the ethnic Sunda population of West Java has a well-known preference for working in the Middle East that is borne out in the data here. 3. International migration outflows are lower during months falling within the rice growing season. Table 2 demonstrates this claim using data on the the universe of international migrants returning through Jakarta between 2008m1-2011m3 (see fact 1 above) and the following specification for the log total emigrants

∗ Some of the results in this brief note are referenced in my job market paper, “Wealth Heterogeneity, Income Shocks, and International Migration: Theory and Evidence from Indonesia.” These results warranted a separate paper for two reasons: (i) to compile and provide background on a few salient quantitative results of interest to those studying international migration from Indonesia, and, more practically, (ii) to keep the job market paper’s online appendix to a manageable length. I would like to thank Sudarno Sumarto and Palmira Bachtiar for graciously sharing data. I acknowledge financial support from the Center on Emerging and Pacific Economies at UC, San Diego. Any errors are of course my own. † Dept. of Economics, UCSD, 9500 Gilman Dr. # 0508, San Diego, CA 92093-0508; [email protected] 1 Bazzi (2012) provides further details on this data source.

from district d going to destination j in month m of year y ln(emigrantsdjmy ) = α + βwet seasondm + τd + τj + τy + τd × t + εdjmy ,

(1)

where wet seasondm equals one in months during the province-specific, rice-growing wet season and zero otherwise (see Maccini and Yang, 2009); τd is a district fixed effect; τj is a destination fixed effect; τy is a year fixed effect; τd × t is a district-specific monthly time trend; and εdjmy are idiosyncratic error terms. In some specifications, I include district×destination fixed effects τdj instead of the separate τd and τj . Depending on the stringency of the fixed effects specification, I find that migration flows are 10-12 percent lower during the wet season, regardless of the time-horizon over which I restrict the sample (to avoid possible censoring). 4. The demand for Indonesian immigrants is income-elastic, particularly in Muslim-majority destination countries. Table 3 demonstrates this claim based on estimates from the following unbalanced panel data model relating GDP shocks in destination countries to demand for Indonesian immigrant labor from 1994-20092 ln(emigrantsjy ) = α +

k X

βs log(GDPj,y−s ) + τy + τj + εjy

(2)

s=0

The model is estimated using destination (τj ) and year (τy ) fixed effects. For these simple descriptive purposes, I ignore the extensive margin of migrant flows resulting in destination country-years being dropped from the sample when no emigrants are recorded in year t. Column 1 imposes βs = 0 for all s > 0. Column 2 augments this specification by allowing βs to vary across Muslim and non-Muslim destination countries. Columns 3-6 allow βk 6= 0 for k = 1, 2, 3, 4. The estimates imply that a one percent increase in destination country GDP leads to a four percent increase in the number of Indonesian immigrants with the bulk of the effect coming from Muslim-majority destinations. 5. There exists considerable spatial variation in international migration rates and destinations across Indonesia. Figure 1 plots district-level stock migration rates in 2005. Figure 2 shows the plurality destination abroad by village in Java. 6. The total number of migrants recorded in Village Potential (Podes) data correspond closely with the cumulative 2-3 year flows reported in national administrative records by BNP2TKI as well as nationally representative household survey data. Table 1 of Bazzi (2012) reports 1.1 million contract migrants abroad in Podes 2005; BNP2TKI records indicate 1.2 million Indonesians departed for overseas jobs from 2003 to 2005; I estimate 1.3 million contract migrants from 2003 to mid-2005 using sampling weights in nationally representative household survey (Susenas) data collected in June 2005.3 7. Mostly low-skilled, contract migrants typically work in construction, agriculture, manufacturing, or household services. Nearly 90 percent of ever-migrants report having worked in one of these occupations in nationally representative household survey data (Susenas) from 2005 and 2007. Nurses and medium-skill manufacturing 2 The

migration data come from BNP2TKI for 1994-2007 and the Department of Labor for 2008-9. well-suited for many microeconometric purposes, the IFLS measurement and coverage of international migration flows is problematic. Despite being representative of 80 percent of the Indonesian population, the IFLS does not survey households in a few provinces and many districts with relatively large international migration flows. A further source of underestimation lies in the long period gap between consecutive survey years (4-7 years) and the typically short-term contract (2-3 years) for Indonesian migrant workers.

3 Although

2

engineers comprise a small share of total temporary emigration flows from Indonesia. Moreover, roughly 85 percent of contract migrants returning through Jakarta international airport between 2008m1-2011m3 completed primary school or less. 8. The majority of international migrants come from agricultural households. In Susenas 2007, over 60 percent of households in rural areas with members working abroad report that a majority of household income is gained through agricultural activities. Furthermore, females—the majority gender among international migrants—account for 40-45 percent of total agricultural labor employed in rice cultivation. This estimate, based on Susenas 2004, includes employment on family and other households’ farms. Bank Indonesia (2009) reports that around 55 percent of contract migrants work in agriculture immediately prior to working abroad. 9. The majority of Indonesian households send one member abroad at a given point in time. According to Susenas 2007, 90 percent of households with any international migrants report only one household member currently working abroad, and 93 percent of households with any former international migrants report only one household member having worked abroad.

References Bank Indonesia, “Survei Nasional Pola Remitansi TKI,” Special Report, 2009. Bazzi, Samuel, “Wealth Heterogeneity, Income Shocks, and International Migration: Theory and Evidence from Indonesia,” Unpublished manuscript, 2012. Maccini, Sharon and Dean Yang, “Under the Weather: Health, Schooling, and Economic Consequences of Early-Life Rainfall,” American Economic Review, 2009, 99, 1006–1026.

3

Figures Figure 1: Variation in migration rates (2005) across districts

4

Notes: This figure shows the emigrant share by district in 2005 aggregated over all villages in Podes data.

Figure 2: Plurality Destinations Abroad by Village in Java

5

Notes: This figure shows the plurality destination choice by village in Podes 2005 data. “No Data” indicates that the village has no emigrants in 2005.

Tables Table 1: Correlates of (Plurality) Destination Choice Plurality Destination Country in . . .

ethnic Arab population share ethnic Chinese population share Muslim population share Christian population share female migrant recruiter visit pre-2008 village is urban log distance to Riau embarkation village in Java village in Bali/NTT/NTB village in Kalimantan village in Sulawesi/Maluku ethnic Javanese majority ethnic Sundanese majority ethnic Bugis majority ethnic Minang majority ethnic Madurese majority

Number of Villages by Destination Group

Middle East & North Africa

Malaysia or Singapore

East Asia

Other

4.165 (1.447)∗∗∗ -1.952 (0.528)∗∗∗ 0.079 (0.096) -0.596 (0.130)∗∗∗ 0.114 (0.021)∗∗∗ 0.004 (0.011) -0.094 (0.113) 0.257 (0.032)∗∗∗ 0.067 (0.068) 0.372 (0.118)∗∗∗ 0.096 (0.070) -0.248 (0.043)∗∗∗ 0.112 (0.050)∗∗ -0.087 (0.052)∗ -0.357 (0.094)∗∗∗ -0.180 (0.057)∗∗∗

-3.774 (1.714)∗∗ 1.925 (0.484)∗∗∗ 0.266 (0.103)∗∗∗ 0.846 (0.131)∗∗∗ -0.127 (0.021)∗∗∗ -0.031 (0.013)∗∗ -0.028 (0.116) -0.290 (0.035)∗∗∗ -0.016 (0.072) -0.357 (0.117)∗∗∗ -0.104 (0.081) 0.205 (0.043)∗∗∗ -0.152 (0.052)∗∗∗ 0.150 (0.067)∗∗ 0.289 (0.095)∗∗∗ 0.290 (0.055)∗∗∗

-0.585 (0.710) 0.063 (0.073) -0.271 (0.042)∗∗∗ -0.192 (0.040)∗∗∗ 0.021 (0.007)∗∗∗ 0.022 (0.005)∗∗∗ 0.123 (0.042)∗∗∗ 0.021 (0.017) -0.055 (0.018)∗∗∗ -0.018 (0.017) -0.005 (0.036) 0.042 (0.010)∗∗∗ 0.029 (0.022) -0.050 (0.022)∗∗ 0.067 (0.026)∗∗∗ -0.103 (0.024)∗∗∗

0.193 (0.098)∗∗ -0.036 (0.021)∗ -0.074 (0.009)∗∗∗ -0.058 (0.008)∗∗∗ -0.008 (0.003)∗∗∗ 0.005 (0.002)∗∗∗ -0.001 (0.006) 0.012 (0.002)∗∗∗ 0.004 (0.005) 0.003 (0.003) 0.013 (0.008)∗ 0.001 (0.003) 0.011 (0.004)∗∗∗ -0.012 (0.008) 0.001 -0.011 -0.007 (0.005)

8,408

21,789

2,294

673

Notes: Estimates are based on a multinomial logit specification with Malaysia and Singapore as the base category, coefficients for which are included for reference. The estimates reported in the table are marginal effects calculated at the mean of each variable. Standard errors are clustered by district. Reference categories include ethnic non-Arab/Chinese population share, village in Sumatra, and any ethnic majority other than those listed in the table. The specifications also control for whether the village is accessible by land. Riau is the city in Indonesia that is the largest hub for those emigrating to peninsular Malaysia.

6

Table 2: Migration Outflows are Lower During the Wet Season Year FE District FE Destination FE District × Destination FE

Yes Yes No No

Yes Yes Yes No

Yes No No Yes

Yes Yes No No

Yes Yes Yes No

Yes No No Yes

(1)

2005m3-2008m3 (2)

(3)

(4)

2006m3-2009m3 (5)

(6)

Departure Dates:

Month in Wet Season

Number of Observations R2 (overall)

-0.144 (0.015)∗∗∗

-0.092 (0.009)∗∗∗

-0.103 (0.007)∗∗∗

-0.100 (0.013)∗∗∗

-0.079 (0.008)∗∗∗

-0.089 (0.007)∗∗∗

7,039 0.940

36,073 0.609

36,073 0.843

7,528 0.957

40,784 0.628

40,784 0.880

(7)

2005m3-2009m3 (8)

(9)

(10)

2005m3-2009m3 (11)

(12)

Departure Dates:

Month in Wet Season

Number of Observations R2 (overall)

-0.115 (0.012)∗∗∗

-0.079 (0.008)∗∗∗

-0.088 (0.006)∗∗∗

-0.123 (0.016)∗∗∗

-0.093 (0.010)∗∗∗

-0.103 (0.009)∗∗∗

9,444 0.939

48,998 0.608

48,998 0.843

5,123 0.961

27,859 0.635

27,859 0.889

Notes: Standard errors are clustered by district in all specifications. Columns 1, 4, 7, and 10 are based on district × year × month observations. The other columns are based on district × destination × year × month observations. The data are obtained from BNP2TKI (see fact 1 in the text).

7

Table 3: On the Income Elasticity of Demand for Indonesian Immigrants, 1994-2009 βy : log GDP, y

(1)

(2)

(3)

(4)

(5)

(6)

4.213 (2.484)∗

-0.123 (2.002) 5.206 (1.900)∗∗∗

9.241 (4.514)∗∗

9.467 (4.507)∗∗

9.761 (4.660)∗∗

8.697 (4.466)∗

-5.521 (3.731)

-5.405 (4.475) -0.264 (2.465)

-1.030 (2.901) -11.954 (6.622)∗ 7.596 (5.269)

0.933 (2.988) -3.178 (2.627) -10.871 (4.768)∗∗ 8.916 (3.837)∗∗

log GDP, y × Muslim majority βy−1 : log GDP, y − 1 βy−2 : log GDP, y − 2 βy−3 : log GDP, y − 3 βy−4 : log GDP, y − 4 P4

— —

— —

3.720 [2.406]

3.798 [2.470]

4.373 [2.626]∗

4.496 [2.664]∗

Year FE Destination FE

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

229 21 0.246

229 21 0.336

228 21 0.254

227 21 0.256

225 21 0.281

223 21 0.301

s=0 βy−s [std. error]

Number of Destination × Years Number of Destinations R2

Notes: This table reports estimates of equation (2). The migration data are obtained from BNP2TKI and the Department of Labor and Transmigration and include all international contract migrants from 1994-2009. The GDP series is expressed in 2000 US dollars and is obtained from the World Development Indicators for all countries except Taiwan, for which the data are derived from the World Economic Outlook database of the IMF. The Muslim indicator variable equals one for all destination countries in which the majority of the population belongs to the Muslim faith. The destination countries include: Bahrain, Brunei, Cyprus, France, Hong Kong, Italy, Japan, South Korea, Kuwait, Malaysia, Netherlands, Oman, Qatar, Saudi Arabia, Singapore, Spain, Taiwan, United Arab Emirates, United Kingdom, and United States. The regressions include a constant that is not reported. Standard errors are clustered by destination country.

8

International Migration from Indonesia: Stylized Facts ...

According to Susenas 2007, 90 percent of households with any international migrants report only one house- hold member currently .... Table 2: Migration Outflows are Lower During the Wet Season. Year FE. Yes. Yes. Yes. Yes. Yes. Yes. District FE. Yes. Yes. No. Yes. Yes. No. Destination FE. No. Yes. No. No. Yes. No.

1MB Sizes 1 Downloads 223 Views

Recommend Documents

Refining Stylized Facts from Factor Models of Inflation
(2009) argue for rational inattention as the root of price stickiness ... Due to two essential properties of price data, simple factor models of inflation indices.

Voting in central banks: Theory versus stylized facts
Jun 3, 2016 - the CESifo conference on central bank communication, the ..... Although we call pt the interest rate, it can stand for any standard monetary.

Stylized facts of business cycles in the G7 from a real ...
data sets.5 In this paper we wish to pursue that tack. In particular, the purpose of this paper is to investigate the basic stylized facts of business cycles in the G7 ...

Stylized facts of business cycles in the G7 from a real ...
Real business cycles models cannot account for some basic stylized facts of labor dynamics .... students of business cycles and growth should draw through a time plot of this ... An illustration of this filter using the quarterly real GDP data of the

International Migration Law.pdf
Retrying... Download. Connect more apps... Try one of the apps below to open or edit this item. International Migration Law.pdf. International Migration Law.pdf.

Data Migration System in Heterogeneous Database - International ...
*(IOK-COE, Pune University, India. Email: [email protected]). ABSTRACT. With information becoming an increasingly valuable corporate asset, ...

From eurocrisis to asylum and migration crisis
immediate spill-over effect on other Member States, ability to comply with .... compliance with the most important elements of the EU acquis in the area of asylum ...

Evidence from Migration Data
sumption amenities and servicesFsometimes called the 'bright lights' hypothesis. (e.g. Henderson 1986). .... technical staff in the labour force over the last two decades. Thus, there is .... support for the view that large cities have higher shares

From eurocrisis to asylum and migration crisis
The current First Vice President of the European Commission, Frans. Timmermans, recently summarized the root cause of the unprecedented massive influx of ...

IOM International Organization for Migration OIM ... -
OIM OrganizaciónInternacionalparalasMigraciones. Regional Office for Central & East Africa. Rhapta Road, Westlands, PO Box 55040 Nairobi. Tel: +254 2 ...

Amnesty-International-Belgique-Dix-préjugés-sur-la-migration ...
Retrying... Download. Connect more apps... Try one of the apps below to open or edit this item. Amnesty-International-Belgique-Dix-préjugés-sur-la-migration-arrêtons-de-croire-n-importe-quoi.pdf. Amnesty-International-Belgique-Dix-préjugés-sur-l

Paper Facts Printing Facts
A single mature tree can release enough oxygen back into the atmosphere to support 2 human beings. • Each person in the U.S. generates approximately 2.3 ...

Data Migration from QnE Delphi to Optimum.pdf
Data Migration from QnE Delphi to Optimum.pdf. Data Migration from QnE Delphi to Optimum.pdf. Open. Extract. Open with. Sign In. Main menu. Displaying Data ...