25
Working Paper in Economics and Business Volume V No.5/2015 Internal Migration in Indonesia: Duration and Factors Hera Susanti Arie Damayanti October 2015 Department of Economics, Faculty of Economics and Business University of Indonesia

Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Working Paper in Economics and BusinessVolume V No.5/2015

Internal Migration in Indonesia: Duration and Factors

Hera SusantiArie Damayanti

October 2015

Department of Economics, Faculty of Economics and BusinessUniversity of Indonesia

Page 2: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Working Paper in Economics and BusinessChief Editor: Hera SusantiEditors: Muhammad Halley Yudhistira, Rus’an NasrudinSetting: Rini Budiastuti, Moslem Afrizal

Copyright c©2015, Department of EconomicsDepartment of Economics Building 2nd FloorDepokWest Java, Indonesia 16424Telp. 021-78886252Email: [email protected]: http://econ.feb.ui.ac.id/katagori/working-paper/

Page 3: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Contents

Contents 3

List of Tables 4

List of Figures 4

1 Introduction 1

2 Literature Review 2

3 Model Framework 4

4 Comparative Statics 64.1 Change in the Consumption Preference . . . . . . . . . . . . . . . . . . . . . . . . 64.2 Change in Purchasing Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64.3 The Change of Migration Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74.4 Change in Human Capital . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

5 Methodology 8

6 Migrants Profile and the Pattern of Migration 9

7 Factors Affecting the Duration of Migration 12

8 Factors Affecting The Duration of Migration: Total and Based on Mi-grants’Original Region 16

9 Conclusion 19

10 References 19

Page 4: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

List of Tables

1 Total Profile of Respondents and Situation on 1993, 1997, 2000 and 20079 . . . . . 102 Percentage of Migrants Based on Characteristic, 1993-2007 . . . . . . . . . . . . . 113 Reason for Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Details of Migration with the Reason of Working and Family Matters Year 1997,

2000 and 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Number of Migrants Based on the Year of Leaving and Return, 1993-2007 . . . . . 126 Survivor Function of Some Variables . . . . . . . . . . . . . . . . . . . . . . . . . . 137 Factors Affecting The Duration of Internal Migration 1993-2007 . . . . . . . . . . . 18

List of Figures

1 Survivor Function for Education Variable . . . . . . . . . . . . . . . . . . . . . . . 152 Survivor Function for Burden Variable . . . . . . . . . . . . . . . . . . . . . . . . . 153 Survivor Function for Status of Original Region Variable . . . . . . . . . . . . . . 154 Survivor Function for Real Income Variable . . . . . . . . . . . . . . . . . . . . . . 16

Page 5: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Internal Migration in Indonesia: Duration and Factors

Hera Susanti1 , Arie Damayanti2

Abstract

We analyse the behavior of internal migration in Indonesia and estimate factors influencedthe migrants’s decision to return. We adopt the international migration model to estimate theduration periods of the Indonesian internal migration. The characteristic variables are developedfrom the Indonesian Family Life Survey (IFLS) data within period of 1993-2007, while the controlvariables are using various regional data fron the National Statistical Bureau of Statistics. Themain conclusion indicates that the return decision was mainly influenced by the opportunity toincrease migrant welfares. Hence, migrants’s characteristic and education level proved to affectthe duration. The migrants’s engagement to their family and community was remain strong, andeven stronger if the status of the home region was rural area. The duration also tends to be longerafter the implementation of regional autonomy.

JEL Classifications: C41, J61, O15

Keywords: Migration Duration, Internal Migration, Return Migration, Survival Analysis, IFLS

1. Introduction

The relationship between economic develop-ment and labor mobility has become an inter-esting discussion on regional development. Theeconomic development of hosting region canencourage migrants; hence, the migrants canalso influence the development of host region’seconomy. This situation raises some issues re-garding human resource disparity, which hasbecome a major problem in regional develop-ment.

In a less developed region, high quality hu-man resource within productive age group hastendency to migrate, partly due to the lowlevel of living in the origin region, and atthe same time because of the expectation to

1Faculty of Economics and Business, University ofIndonesia

2Graduate Program of Economics, Faculty of Eco-nomics and Business, University of Indonesia

have a higher salary and higher quality of life.This condition drove a higher level of dispar-ity, mainly because the migrant dominated byhigh quality and high motivated labor in pro-ductive age compared to those who stayed intheir original region. Hence, a lot of evidenceindicates that there are not only economic fac-tors that are encouraging the migration, butalso demographic and human security. There-fore, the discussion of migration should includethe social, cultural and law issues.

Recent policy about deployment and mobil-ity in Indonesia was developed using indirectapproach. This approach regulates centers ofeconomic activity which in turn will affect mo-bility. However, the development of economicactivity needs not only physical, but also hu-man capital. Migrants who back to their regionbecame potential worker. Their experience aswell as their skill developed during their mi-gration became potential asset for their orig-

1

Page 6: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 2

inal region. Even if their skill not as good aslabor who are still migrated, if the number ofthis ex migrants is significant, then their ex-perience and skill will enhance output of theirregion.

To take advantage of the returned migrants,it is essential to know their duration of migra-tion, the characteristic of migrants and whatfactors drove them to return to their region.By knowing the probability of migration dura-tion, for example, the local government is ableto prepare the appropriate incentive to attractpotential migrants to advance their homecom-ing.

This study analyses internal migration in In-donesia, whether there was a tendency for mi-grants to return to their home region, how longmigrant tends to stay in the host region beforedeciding to returns and factors influencing thedecision.

2. Literature Review

Unlike other kind of production factors, themigration of labor naturally followed by the mi-gration of the owner of this particular factor ofproduction. Therefore the decision to migrateincluding return migration and repeat migra-tion will be affected not only by economic fac-tor, but also non-economic factors embeddedto the owner of the factor of production. Thesenon-economic factors might include social, po-litical and regulation aspect.

Empirical evidence shows that in many coun-tries, there are some other factors that mightaffect the decision of return migration, such asdemographic3, economic, and human security4.Some non-economic factors also consider to af-fect the decision to migrate, such as social net-work (the migration decision was motivated bywillingness to follow migrants’ friend or fam-ily), looking for new experience, been encour-aged by their country of origin (to reduce thenumber of unemployment, to enhance poten-tial migrants’ education level and improve their

skill, enhance national income through the re-mittances, and due to the influence of global-ization (Wickramasekera, 2002; World Bank,2009)). On the other hand, due to the highcost of migration (including physical and psy-chological cost), people might become demoti-vated whether to go for migration or return totheir original region which relatively underde-veloped.

Technology development promotes not onlygoods and services mobility, but also the hu-man mobility as well. Nowadays the risk of mi-gration has become lower due to the ease ofobtaining information, ease of communication,ease of transportation, and due to bilateral andregional cooperation. This situation has pro-moted higher level of labor mobility. The in-crease of labor mobility has changed the discus-sion focus of the migration. Previous discussionon retun migration focused on permanent mi-gration, and senior or pension resident who re-turned to their origin after they migrated whenthey was young. Nowadays, the discussions fo-cus on non-permanent migration, and migra-tion of younger people who migrated to followtheir parents (Conway & Potter, 2009). Thismigration pattern was predicted to develop fur-ther, some researchers even consider that nowa-days, the majority pattern go migration is tem-porary migration (see the work of Dustmannand Weiss, 2007; Dustman et.al. 2010; Magrisand Russo, 2003; and Kirdar, 2010).

There are tendencies that migrated laborwill return to their origin (return migra-tion), or continue to migrate to other regionswhich caused repeat/onward migration (Bor-jas, 2008). There are various factors which mo-tivated migrated labor to return to their origin.

3Including ageing of population structure in devel-oped countries, higher level of education, the gap be-tween the needs and availability of labor force, etc.

4Human security is not only related to war, but alsonatural disaster, political situation, and other factorswhich forced people to migrate (for example see WorldBank, 2009; Conway and Potter, 2007; Wickramasekera,2008)

Page 7: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 3

For first generation migrants, return migra-tion commonly caused by pension, the deathof the spouse, or caused by what it is calledby failed migration5. For the next migrants’generation, their motive factors became morediverse. Some wanted to continue their careerin their origin, whether in social, culture, oreconomics6, which is possible due to the widerknowledge and more developed skill comparedto other labor who did not have school or work-ing experience in other places (Mulder & VanHam, 2005). In the same study, Mulder andVan Ham concluded that decision to returnfrom migration also drove by willingness to dovarious things that has not been done beforethey migrate (corrective moves), which usu-ally were non-economic motive driven activitiessuch as taking care of their parents, taking careof their children, etc. Women also have moreopportunity to take full time job when they re-turn. This fact supports research by Newbold(1996) which covered that women’s economiccondition are better when they return for mi-gration. On the other hand, men’s commonlyhave a better economic condition if they do on-ward migration.

It is interesting that some studies revealedthat migrant’s decision to return does not al-ways relate to wage differences between theirorigin and their destination. Stark (1992) pro-posed the theory of relative deprivation whichstated that migrants will return to their origi-nal region which relatively in a worse economiccondition to minimize risk. Meanwhile Mes-nanrd (2000) stated that return migration isa way to cover migrants’ problem with cap-ital constraint. Most researchers agreed thatlocation-specific preferences is another fac-tors which drive return migration, despite thehigher purchasing power in the country of ori-gin using currency from their migration desti-nation country. .

Some behavior models have been developedby proposing the use of some new theory.Those theories are conducted by considering

several factors such as expected future returnsif migrants return and risk factor outside mi-grants’ origin (Stark, 1992), better purchas-ing power in destination country/region utiliz-ing migrants’ earning from destination coun-try/region and ability to do more activities incountry of origin (Dustmann, 1995, 1997), thebound with country of origin7, etc. Assumingthat there is tendency for migrants to consumemore in their country of origin, then the ad-vantage of doing migration could be diminishedbased on migration cycle, meanwhile the costof migration tend to be positive and increase.

In line with the development of research re-lated to human capital, Dustmann and Weiss(2007) argued that decision to migrate is a re-spond to the most efficient utilization of humancapital which in turn will generate high return.This conclusion was generated based on hisstudy which was developed from Roy model.This model assume that worker have two dif-ferent basic skills, which gained by ”learning bydoing”. Therefore, migrated workers not onlyrespond to the level of income, but also chanceto improve their skills. For migrants who returnto their country, their abroad working experi-ence will increase their level of income com-pared to the situation where they did not goabroad.

5This term is related to back migration which causedby difficulties to adapt, for instance fail to reach earn-ing target, dropout, failed to adapt with new situation,or having bad experience such as racial discrimination-either open or hidden- and due to rejection (anti mi-grants) in their destination country

6Some books have discussed the motivation ofback migration. Those books are written by Conwayand Potter (2009), David Ley and Audrey Kobayashi(Hongkong experience), John Connell (Polynesia), Con-way et. al, (Trinidad) and Graeme Hugo (Australia)

7Many researchers included this factor in many in-tepretation, for instance Hill (1987), Djajic & Mil-bourne (1988), Dustmann (2003, 2007) dan Carrion-Flores (2006)

Page 8: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 4

3. Model Framework

This model is adopting international migra-tion model by Dustmann and Weiss (2007).This model was customized by including mi-gration cost, and reinterpretation toward pref-erence and assumption so that it will matchwith internal migration condition in Indonesia.This international migration model was appliedby considering several matters, which are:

1. Due to the globalization, the access to in-formation regarding the situation of des-tination country became more open. Soas the communication transportation fa-cilities and access to the country of ori-gin. This situation caused similar charac-ter of mobility between country, and mo-bility between regions.

2. Indonesia’s geographical condition whichconsist of big and small islands and widevariation of ethnic groups, culture, andlanguages, which became natural barrierso that regional migration in Indonesia isnot as easy as regional migration in othercountry which only consist of mainlandand relatively homogeny population.

The basic model of migration duration as-sumed that migrants have preference to con-sume either in the destination region (cA) or inthe original region (cB). Migrants will also de-termine their location of residence. Every year,migrants have to decide whether they will stayin their current place, or back to their origin. Ifthe migrants decided to go back to their orig-inal region, then there will be a return migra-tion. Period between early times they migrateuntil they return is an optimal duration of themigrants.

Some assumptions were developed in thisstudy. Migrants are working until the end oftheir life, but they did not concern about theuncertainty of their wage in the following years.The migrants have preference to consume ei-ther in their original region or in their desti-nation region, however they prefer to consume

in their original region. When there are returnmigration, the duration of living in the regionof destination was considered as optimum du-ration of migration for the individual. At thebeginning of migration, potential wage in coun-try of destination is higher than in the coun-try of origin. However, migrant’s human capi-tal will increase due to working experience inthe country of destination. Therefore, after mi-grant went back to their country of origin, theirpotential wage became higher, even higher thenwhat they could get in their migration desti-nation, and lifetime income will increase alongwith the duration of migration.

In order to simplify, discount rate = 0, there-fore wage rate in both region assumed to beconstant at all time. This analysis only consid-ers interior solution, that the focus of analysiswill be only to migrants who decided to returnto their original region. Assume that migrantsstarted to migrate on t = 0 and passed away ont = T . Their duration of migration was statedas h, where h ε (0, T ). If h = 0, means mi-gration did not happen, otherwise if h = 1, itmeans migrants decided to stay in their regionof destination.

The utility function of lifetime migrant canbe stated as follows:

U =

h∑t=1

v(cA)

+ ξ

T∑t=h+1

v(cB)

(1)

Since interest rate was assumed to be zero,then the wage received by migrants either inregion of destination or in region of origin wasassumed to be constant at all time, then theutility functionwillbe as follows:

U = hv(cA)

+ (T − h)ξv(cB)

(2)

Where

• v(c) = c1−α

1−α , is the utility migrant func-tion, α < 1.

• ξ = ξ(age, zi) is a parameter representsmigrant’s preference regarding the loca-tion of consumption, ξ > 1.

Page 9: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 5

Parameter α was assumed < 1, it shows thatmarginal utility level decrease when consump-tion level increase. Parameter ξ was assumed> 1, it shows that for the same level of con-sumption, migrant’s marginal utility is higherif the consumption happened in the original re-gion. The value of this parameter can be differ,depends on migrant’s age at the beginning oftheir migration (age) and character of the mi-grant (zi), such as gender, education level, andtheir employment status.

Migrant will maximize (2) towards cA, cB

and h, considering lifetime budged constraint

h(cA + δ) + (T − h)pcB ≡ Y (h) (3)

Where

• Y (h) = Lifetime income

• p = pB/pA, is the consumption price ra-tion in the original region relative to des-tination region

• δ = cost occurred for living outside origi-nal region = δ(family, distance, access)

Since p is a consumption price in the orig-inal region relative to the destination region,then 1/p is the purchasing power in originalregion relative to destination region. If p < 1,means consumption price in destination regionis higher then in the original region. Converselyif p > 1, means consumption price in originalregion is higher then in the destination region.

During migration, there are some costs oc-curred due to the situation where migrant wasnot living in his original region (δ). Thesescosts include cost for communication with fam-ily members, cost occurred when migrants wentback to their original region for holiday or othernecessities. These costs will be higher if mi-grant left his core family (family), and if thereis long distances between region of destinationand region of origin (distance), and if there islimited access to the country of origin (access).The farther the location of destination originfrom original region, or the harder the effort

needed to contact the original region, then thehigher the value of δ. Based on the situationmentioned above, then factors driving returnmigration could be identified as follows. Thefirst factor is due to local consumption prefer-ence is higher (ξ > 1). The second factor is dueto the cost of living outside migrant’s originalregion became higher (δ < 1).

By utilising Lagrange equation, optimumconsumption in both regions can be generated.Consumption value in both regions will be in-crease along with the increase of lifetime in-come after deducting the cost of migration.However, consumption preference in originalregion is still higher compared to consumptionpreference in destination region.

Human capital is an aggregation that de-scribe production capacity owned by an in-dividual in the form of education and skills.These skills can be developed by conductingthe concept of ”learning by doing”. Working inthe well developed destination region will en-hance the skills of migrants which in turn willincrease their market value. Although the wagerate is assumed to be constant, the experienceof working in other region was considered asfactors which enhanced migrant’s capacity thustheir capacity considered as higher than thosewho never work outside their original region.Therefore, the migrant’s wage rate after theyreturned to their original region will be higherin line with their duration of migration. Thusit can be concluded that migrant’s wage ratein their original region is affected by his dura-tion of migration. The value of lifetime incomeis affected by the summation of the wage re-ceived during migrant’s working time, either indestination region (A) or in the original region(B).

hwA + (T − h)wB(h) ≡ Y (h) (4)

Where wA represents wage in destination re-gion and wB(h) is wage in original region aftermigrant returned from migration.

Page 10: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 6

According to the assumption than when h =0, that is when migrant started to migrate tothe destination region, the wage rate in thedestination region is higher than in the orig-inal region wA > wB(0). After return to theiroriginal region, migrants’s wage rate increasedin line with the duration of migration. Thusit is possible that the wage rate in the origi-nal region is higher than in the destination re-gion, if migrants improved their relevant skillsneeded in the original region during the mi-gration. Therefore, this situation reflects thatthere is intention to accumulate human capitalduring the migration.

Lifetime income Y (h) also increase in linewith h.

Yh(h) = wA−wB(h)+(T−h)dwB(h)

dh> 0 (5)

It will increase if migrant’s wage rate in des-tination region (A) is higher than in the orig-inal region (B), and if the experienced gainedduring migration improve migrant’s income ca-

pacity(dwB(h)dh > 0

). However, since living in

the destination region will increase migrants’potential income in their original region, thentheir lifetime wealth could be decreased if theyare living too long outside their original region,since wage rate in original region is higher thanin the destination region, where wB(h) > wA.

The duration of migration is also affected byadditional utilities gained due to living outsideoriginal region. It affects migrant’s lifetime util-ity in two ways:

• Delay in migrant’s return will be detri-mental since migrant’s will lose opportu-nity to get higher utility, such as if themigrant return to their original region.

• On the other hand, delay in migrant’s re-turn also has a potential advantages, sincethe longer the duration of migration thehigher the lifetime income.

Thus the lifetime income will diminishmarginal utility derived from additional health

gained in destination region (A) and increasecost due to delay in return of migration. Thiscost occurred in the form of losing opportunityto consume in the original region. This matteris important in the discussion of regional de-velopment policy. The government needs to de-velop various programs that could enhance mi-grants’ lifetime income so that migrants couldshorten their duration of migration and workin their original region.

4. Comparative Statics

Based on the model above, comparativestatic analysis can be conducted to knowingdeterminant factor of the duration of internalmigration.

4.1. Change in the Consumption Preference

In order to determine its utility, migrant wasassumed to have consumption preference (cA,cB), where the parameter for consumption is(ξ). The value of ξ can be differ based on char-acter of the migrant such as age, education,field of work, and the employment status.

∂Γ

∂ξ= −αv

(cB)< 0 (6)

As stated before, it was assumed that mi-grants prefer to consume in their original re-gion, thus ξ > 1. If there is increase in mi-grant’s local preference, then the marginal util-ity to stay in the destination region will be di-minished.

4.2. Change in Purchasing Power

p represents consumption price in the origi-nal region, relative toward destination region,pB/pA, thenp shows purchasing power in thedestination region, while 1/p shows purchas-ing power in the original region relative to thedestination region. If p < 1, it means thatconsumption in the destination region is moreexpensive then in the original region. There-fore purchasing power in the original region ishigher then in the destination region. On the

Page 11: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 7

contrary, if p > 1, it means consumption in theoriginal region is more expensive then in thedestination region. Therefore purchasing powerin destination region is higher. In the case ofinternal migration, both situations could hap-pen. However since this study assumed that themovie of doing migration is to achieve higherlevel of well being, then it was assumed thatdestination region is more well developed thenthe original region, therefore p < 1.

∂Γ

∂p= α

[vA(cA) dcAdp

(1 − pk)

]+ vAA

(cA) dcAdp

[Yh(h) − δ(h)] > 0

(7)

By assuming that migrant’s income only af-fected by wage, then the increase of purchasingpower in the original region, or the decrease inp, will diminish marginal utility of staying inthe destination region. Thus migrant tends tohaste his return.

• The first part explained that increase ofpurchasing power in the original region(1/p increase, p decrease) will increasemarginal utility of total consumption inboth regions. Migrant tends to shortenthe duration of migration, considering thatmarginal utility in original region is higherthan marginal utility in destination region.Thus delaying their return will cause util-ity loss, because it will reduce their life-time consumption.

• The second part explained that an increaseof purchasing power in the original regionwill reduce the marginal utility.

4.3. The Change of Migration Cost

Cost occurred during migration not onlycame from daily consumption, but also addi-tional costs occurred due to living outside orig-

inal region (δ). Thesecostsinclude communica-tion cost, the cost of returning to theirhome-towneither in the holiday or in otherdays. Thiscost of migration willincrease if the migranthas dependant in theirhometown (corefamily,or other relatives), or if migrant has strongbond withtheirhometown.

∂Γ

∂p= α

[vA(cA) dcAdδ

− ξvB(cB) dcBdδ

]+ vAA

(cA) dcAdδ

(Yh − δ) − vA(cA)Q 0

(8)

Equation (8) shows that there are ambiguouseffects of increase in migration cost towards theduration of migration.

• On the first part, increase in migrationcost will reduce consumption in both re-gions (due to the loss of income. Therefore,marginal utility of the duration of migra-tion is more affected by other factors suchas preference or purchasing power.

• On the second part, the impact of increasein migration cost towards the duration ofmigration depends on the level of increasein marginal utility to stay in original re-gion compared to the loss of benefits toconsume in destination region (A) due tothe cost of migration.

4.4. Change in Human Capital

In order to make comparison, this study as-sumed than preference to consume and pur-chasing power are the same in destination re-gion and original region. Therefore the cost ofmigration will be zero and consumption eitherin destination or original region became iden-tical. In other words, ξ = 1, p = 1, δ = 0.

The accumulation of human capital reflectedfrom the wage. If the human capital gainedfrom destination region is needed in the origi-nal region, then migrant will be valued higher

Page 12: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 8

in his original region. This situation will drivethe migrant to return to his hometown. Basedon this assumption, then the only motivationto migrate is in order to increase the wealth inthe form of human capital.

Change of optimum duration of migration asa result of the change of wage in the destinationregion can be generated by using total differ-ential, thus

∂h

∂wA=

1dcA

dh

> 0 (9)

Where: 0 < dcA

dh < 1And,

∂h

∂wB=

1 + ∂2wB

dh2

dcA

dh

Q 0 (10)

∂Γ

∂ (wA − wB)= vA

(cA)> 0 (11)

Based on equations above, it can be seen thatif there is increase of wage in A, then the dura-tion of living in the destination region will alsoincrease. However if there is increase of wagein the original region, the duration of migra-tion to A is not always shortened. By assum-ing that migrant always get benefits by stayingin the destination region, then migrant will re-turn only if he achieved the accumulation ofhuman capital. Otherwise, he will continue tostay in the destination region. The implicationsis that if there is a huge impact from the ac-cumulation of human capital towards wage inthe original region, wB(h) > wA, then migrantwill always return to their hometown after mi-grated to other provinces. However, an increasein the difference of wage between destinationand original region will increase the durationof migration.

5. Methodology

This research is using survival analysis or du-ration analysis. Survival analysis was consid-ered as the most adequate analytical methodconsidering data that was censored at the end

of period, and its ability to describe, quan-tifying, and analysing development of a sit-uation and predict the emerge of an occa-sion. In order to explain the determinants ofduration migration, Cox proportional hazardwhich commonly abbreviated as Cox modelwas implemented. Moreover, since during thisstudy, there is portability that migration hap-pen more than once, then recurrent event sur-vival analysis was utilized.

In order to develop empirical estimationmodel, Indonesian Family Life Survey (IFLS)was utilized. The advantage of using IFLS isthat this data has longitudinal character in in-dividual and household level. This data is alsoconsidered as rich data in explaining the char-acter of a situation. Given that IFLS has donequite a long time, since 1993, it is expected tobe sufficient to obtain one or more events ofthe duration of migration. Unfortunately, thissurvey is not conducted every year, so thereare some missing information. Also, this sur-vey that covered the whole region of Indonesiaonly conducted until 2007, meanwhile the nextsurvey is estimated to be published on 2016.

Migration data is available for 4 years, whichare 1993, 1997/1998, 2000, and 2007. Hav-ing regard on the situation of respondent, thisstudy employed the last 3 surveys, namelyIFLS2, IFLS3, and IFLS4. However the datacovers the years of 1993-2007. IFLS1 was usedas the basis for determining the basic respon-dent, as well as to capture the behavior andcharacteristics of respondents in 1993.

This study defines migrant as someone whomoves away for more than 6 months, and agedover 12 years. Besides due the availability ofdata, the determination of the age over 12 wasbased on consideration that the individual atthis age has had strong bond with its originalregion, and for some people, especially in ru-ral area, they has begun to work to help theirparents. Since this research assumed that mi-grant who return to their original region canparticipate to the regional development by giv-

Page 13: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 9

ing their contribution, then all migrants couldbe set as sample. Migrant might not intendedto work when they return, however since thismigrant gained experience by doing work in thedestination region, they have improved theirknowledge, insight, and skill - learning by do-ing process -, so that their human capital wasimproved as well.

As a control variable, regional data collectedby Central Bureau of Statistics (Badan PusatStatistic, BPS) was used. These data areRegional Income at the level of City/County1993-2007, Population Census 1990, 2000,and 2010, National Labor Survey (SurveiTenaga Kerja Nasional, Sakernas) 1992-2007, and Consumer Price Index (CPI).Meanwhile, distance data was estimatedusing www.distancecalculator.globefeed.

com/Indonesia_distance_calculator_asp

and some data regarding distance betweencity/county was collected from province gov-ernment’s website. Distance calculation in thisstudy utilised straight line/distance/flying/airdistance calculation. Even though the resultgenerated will be different with overlanddistance, which is feasible to do by respon-dents, therefore this approach was consideras adequate to describe the distance betweendestination and original region.

Use of the period 1993-2007 in this studyraises the need to do some adjustments to thedata, such as the adjustment of area code, ru-ral/urban area and the detail questions in thequestionnaire. This adjustment was needed be-cause during this period there are some re-gional development and integration, and alsoadjustment in the variables, due to the emer-gence of more detail questions (such as ques-tion about wealth) or there was addition ofnew variables (such as welfare). Data sustain-ability is also important in the duration anal-ysis. Considering that IFLS is a longitudinaldata, some respondents has been interviewedin the previous survey. For those who had beenrespondent, information about their migration

only covers migration since their last interview.Therefore, the use of data from particular yearmight cause lost or discontinue data of old ornew respondents. Hence, for the old respon-dent, the integration of migration data withinSakerti years was needed, so that their datacould be compared with new respondents’ data.On the other hand, for new respondents, thereis no limitation in the use of historical informa-tion, therefore there was also many data loss orinaccurate.

6. Migrants Profile and the Pattern ofMigration

The profile of the migrants and its migrationpattern can be seen in Table 1. Average dura-tion of migration is over 3.36 years, and variedbetween 1 year until 14 years8. However basedon year on year data, there is a tendency ofincrease in the duration of migration. The ma-jority age of migrants are in the level of youngadult, which is around 24,5 years old at thetime they migrated, and 28 years old at thetime they returned. There is a tendency thatfrom years to years the age of migrants be-comes younger at the time they migrated. If in1993 the average age of migrants who startedto migrate is 30.9 years old, in 1997 the aver-age age of migrants who started to migrate is23.27 years old.

Migrants’ average education level at the timethey migrated are high school with the dura-tion 9.9 years and tend to get higher from yearto year. However if we see from the distribu-tion of data, there are about 25 percent mi-grants who get education less than 6 years (el-ementary school and below), 20 percent juniorhigh school and 40 percent senior high school.

8This average value of duration was underestimated,seeing that at the end period of this research there aremigrants who have not returned yet to their original re-gion, either because they delayed their return, they didanother migration, or they decided to become perma-nent migrant.

Page 14: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 10

Table 1: Total Profile of Respondents and Situation on 1993, 1997, 2000 and 20079

VariableTotal 1993 1997 2000 2007

MeanStd.

MeanStd.

MeanStd.

MeanStd.

MeanStd.

Dev. Dev. Dev. Dev. Dev.

Duration 3,36 2,42 3,81 2,72 3,12 2,28 3,24 2,24 3,88 2,69Age (migrate) 24,52 10,17 30,91 13,24 27,11 11,64 24,74 10,31 23,37 9,02Age (return) 27,91 10,50 34,72 13,52 30,25 12,05 28,02 10,75 27,28 9,56Distance 249,03 350,23 201,06 312,01 235,97 358,19 243,82 348,56 244,84 330,82Education (year)- migrate 9,88 3,52 8,63 4,13 9,42 3,71 9,89 3,47 10,10 3,43- return 10,22 3,60 9,05 4,19 9,75 3,78 10,25 3,57 10,49 3,54Number ofObservations 5179 1103 2530 4130 3290

Source: IFLS 1993, 1997/98, 2000 and 2007.

This situation shows that migration in Indone-sia is dominated by migrants with low educa-tion level. However, there were slight increasesin migrants’ education level at the time theyreturned.

Migrants also tend to go to areas that arenot too far away, with an average of less than250 km from the area of origin. High standarddeviation figures also show large distributionof the data. Based on the data, the majority ofmigrants moved to a place with a distance lessthan 150 km from their original region.

Table 2 shows that percentage numbers ofmale and female migrants are relatively equal.However, it can be seen that the number ofmale migrants become higher. Majority of mi-grants’ status are single, and if they are mar-ried mostly they were not bringing along theirfamily. There is tendency that migrants moveto a region with similar status of one levelhigher. Despite that there are some migrantswho decided to move to big city or vice versa,usually migrants who come from rural areawill move to another rural area or small city,

9The profil of respondent for every year describesmigrants’ situation during the year, therefore it mightoccur that the same respondents profile emerge in dif-ferent year as long as migrant has not returned to theiroriginal region.

while migrants from small city will move to an-other small city or to the big city. Data showedthat majority of the migrants came from ru-ral area. Although there are some rural areaswhich changed its status, but mostly status oforiginal regions is still the same when migrantsreturned. Majority of the migrants migratedto area with the status of small city. However,there was a tendency that migrants might cometo the big city.

The reason to migrate as showed in Table 3and 4 also interesting to see. The major reasonsof the migration upon leaving are more diverse,such as working matters (49.24%) followed byfamily matters (24.72%), and to pursue educa-tion (17.7%). However most reason to returnare due to family matters (59.54%) followed byworking matters (23.19%). As for working mat-ters, major reason to migrate is due to there arework to do or to find a job, meanwhile there isvery view migrants who migrate for the reasonof they have a pension. Therefore, there aremany migrants who returned to their originalregion for the reason of working matters, notonly because they have their pension, but alsobecause there is another job for them in theoriginal region.

Table 5 shows interesting migration patternin the period of 1993-2007. In general, it can be

Page 15: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 11

Table 2: Percentage of Migrants Based on Characteristic, 1993-2007

Variable Total 1993 1997 2000 2007

GenderMale 52,06 47,5 49,92 51,12 52,96Female 47,94 52,5 50,08 48,88 47,04

Marital Status During MigrationSingle 55,68 39,4 50,47 58,11 59,34Married 41,16 56,05 45,54 38,79 38,29Divorced 3,17 4,55 3,99 3,1 2,37

Original Region Status when Migrants are 12 Year OldRural Area 65,70 60,79 62,84 64,93 65,88Small City 24,19 25,50 25,41 24,87 24,81Big City 10,10 13,71 11,75 10,20 9,31

Original Region Status when Migrants ReturnedRural Area 55,65 52,20 54,20 53,09 53,58Small City 26,75 30,01 28,93 27,28 26,59Big City 17,60 17,78 16,87 19,62 19,83

Destination Region Status at The Beginning of MigrationRural Area 33,47 48,37 39,08 33,03 31,27Small City 31,50 30,54 30,84 31,70 31,92Big City 35,03 21,09 30,08 35,28 36,81

Source: processed from Sakerti data 1993, 1997/98, 2000 and 2007.

Table 3: Reason for Migration

No Reason Leave % Return %

1 Work 1956 49,24 706 23,192 Education 682 17,17 144 4,733 Family 982 24,72 1813 59,544 Willingness to be Independent 260 6,55 207 6,805 Others 92 2,32 175 5,75

3972 3045

Source: Proceed from IFLS 1993, 1997, 2000 and 2007

Table 4: Details of Migration with the Reason of Working and Family Matters Year 1997, 2000and 2007

No Purpose of Migration 1997 2000 2007

1 Working Matters* 100,0 100,0 100,0a. Working/related to work 73,2 70,3 74,8b. Find a job 21,7 23,2 17,2c. Pension 0,1 0,2 0,4

3 Family Matters 100,0 100,0 100,0a. Married 34,6 33,1 37,2

∗Total of migration, did not distinguished between

leave for migration/return

Source: Proceed from IFLS 1993, 1997, 2000 and 2007

Page 16: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 12

Table 5: Number of Migrants Based on the Year of Leaving and Return, 1993-2007

Year Leave % Return %

1993 389 7,51 . .1994 311 6,01 58 1,121995 334 6,45 106 2,051996 335 6,47 134 2,591997 554 10,70 343 6,621998 320 6,18 441 8,521999 307 5,93 249 4,812000 1010 19,50 321 6,202001 267 5,16 836 16,142002 272 5,25 216 4,172003 264 5,10 203 3,922004 269 5,19 224 4,332005 278 5,37 209 4,042006 269 5,19 263 5,082007 . . 1576* 30,43

Source: Sakerti Data, 1993, 1997, 2000, 2007,

processed∗including censored data

seen that from year to year there is decrease onmigration, meanwhile there are increase in re-turn migration. In 1997 and 1998, there wasslight surge on return migration, which eachbecame 6.6 percent and 8.5 percent comparedto the previous year which around 2 percentrespectively. This might be a reaction of eco-nomic crisis which affected majority of big andsmall city in Indonesia, especially in Java. How-ever, on 2000 there was surge of leaving mi-gration, which about 19.5 percent respectivelycompared to the previous year which only 5.9percent.

The data shows interesting fact, which is30.3 percent of migrant who leave on 2000, re-turned on 2001, and other 11 percent returnedon 2002. This situation might cause anothersurge of return of migration on 2001. This sit-uation might occur due to failed of migration,therefore migrants who leave on 2000 decidedto return to their original region on 2001. How-ever, it might be worth to consider that thissurge of return of migration might cause by theimplementation of regional autonomy on 2000.This is especially seen in the increase in the

number of migrants who return from year toyear, compared with the condition before 1997.

If migrants decided to return since they havepositive expectation towards the implementa-tion of regional autonomy, then some migrantswho returned in that year might be motivatedto develop their career in their original region.This situation might have positive impact fororiginal region, since the return of some mi-grants will enhance human resource capacityin that region. This discussion shows that de-cision to return from migration seems stronglyaffected by the occurrence and economic situ-ation on the particular year, or previous year.

7. Factors Affecting the Duration of Mi-gration

Table 6 shows survivor function of some se-lected variables along with its rank test logvalue. Survivor function shows several possi-bilities that migrants still stay in each year. Itcan be seen that despite gender variable, othervariables shows difference in survivor patternwithin the group of variable. Such as in the

Page 17: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 13

Table 6: Survivor Function of Some Variables

No Variable 1 2 3 4 7 10 13 Pr>chi2

1 Gender1 M 0,821 0,672 0,556 0,456 0,184 0,096 0,034 0,3733 F 0,835 0,693 0,565 0,470 0,178 0,111 0,048

1 Single 0,832 0,679 0,555 0,460 0,156 0,101 0,0552 Marital Status 2 Married 0,834 0,697 0,596 0,518 0,217 0,111 0,080 0,044

3 Divorced 0,835 0,684 0,572 0,522 0,168 . .

1Not Passing

0,836 0,721 0,566 0,494 0,258 0,258.

Elementary2 Elementary 0,849 0,735 0,630 0,534 0,141 0,141 .

Education Level When 3 Junior High School 0,830 0,710 0,607 0,493 0,103 . .3 Migrate 4 Senior High School 0,825 0,699 0,619 0,553 0,167 0,125 0,125 0,015

5 Academy 0,896 0,788 0,646 0,530 0,321 0,161 .6 Bachelor 0,875 0,780 0,780 0,690 0,375 . .7 Others 0,907 0,761 0,435 . . . .

1Not Passing

0,814 0,688 0,504 0,455 0,241 0,241.

Elementary3 Junior Highs School 0,791 0,642 0,514 0,407 0,080 . .

4 Education Level When Return 4 Senior High School 0,835 0,714 0,633 0,560 0,165 0,165 0,165 0,0005 Academy 0,909 0,825 0,702 0,564 0,243 0,081 .6 Bachelor 0,727 0,529 0,529 0,529 0,000 . .7 Others 0,967 0,904 0,833 0,698 0,249 . .

5 Distance

1 <150 km 0,850 0,721 0,591 0,490 0,184 0,120 0,0552 150-500 km 0,804 0,654 0,554 0,464 0,201 0,096 0,0223 500-1000 km 0,779 0,626 0,502 0,419 0,165 0,081 0,030 0,0004 >1000km 0,770 0,590 0,493 0,396 0,171 0,038

6 Age When Migrate

1 12-18th 0,833 0,690 0,535 0,436 0,155 0,088 0,0282 18-25 th 0,809 0,661 0,549 0,438 0,166 0,085 0,0353 25-35 th 0,825 0,667 0,556 0,471 0,225 0,149 0,060 0,0004 35-55 th 0,888 0,765 0,669 0,594 0,244 0,108 0,0725 >55 th 0,835 0,718 0,631 0,576 0,174 . .

7 Status of Original Region1 Rural Area 0,837 0,705 0,578 0,484 0,201 0,122 0,044

When Migrate3 Small City 0,822 0,680 0,561 0,468 0,206 0,116 0,046 0,0305 Big City 0,812 0,645 0,515 0,423 0,202 0,104 0,041

8 Status of Original Region1 Rural Area 0,774 0,586 0,442 0,348 0,117 0,055 0,016

When Return3 Small City 0,805 0,669 0,544 0,428 0,182 0,109 0,050 0,0005 Big City 0,873 0,747 0,647 0,555 0,305 0,236 0,110

Source: IFLS 1993, 1997, 2000 and 2007, proceded

Page 18: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 14

variable of marital status, it can be seen thatthere is different behavior between single, mar-ried or used to be married migrant. For mi-grants who used to be married, they have rel-atively low survivor function. Even there is no”survive” in the 102th year. In general, the sur-vivor function reaches 50 percent respectivelyat the 3rd and 4th year, except for migrantsfrom rural areas which reached 58 percent inyear 2. What was found to be interesting isthat migrants with bachelor degree who leavewhen they have graduated, then until the 7th

year the probability to stay in the destinationregion is 37.5 percent respectively. However,when they returned, in the 2nd year the proba-bility to stay in the destination region is about53 percent, this value last until the 4th year,and there are few who stay in the 2th year, andthere is no survival in the 10th year. The emer-gence of same survival value from year to yearmight due to missing data, so that at the endof research period, there are respondents whostill stay in the destination region. It might be-cause the delay their returned or because thedecided not to return to their original region.

In different format, by using the KaplanMeier Analysis, we can predict the return onmigration based on migrants’ characteristics.Based on education background (Figure 1.) itcan be seen that migrants with high educationlevel tend to stay longer than other 2 groups.Migrant’s survivor function for migrant withelementary and high school degree tend to bealike. However, after the fifth year, it can beseen that many of migrants with high schooldegree decided to return to their original re-gion, while about 13 percent decided to stay.Meanwhile, migrants with elementary degreetend to stay longer in the destination region,and about 21 percent decided to stay in thedestination region beyond the time when thisresearch was conducted.

Group 1 represents migrants with high levelof education. This group has wider opportunityto work compared to other groups. Many mi-

grants from this group decided to return onthe year 4, 6, and 7 due to various factors.Beside due to the situation where they havefailed or due to family matters (such as returnto get married), another possibility is due toeconomic crisis on 1997/98 which impact hitthis group. Meanwhile, decision to return after2000 might cause by regional autonomy, so thatmigrants decided to return and have career intheir original region.

For migrants with low level of education,they decision to return in the first and secondyear might be caused by their failure in theirwork, trying to find new job opportunity, ordue to family matters. This situation is sup-ported by the fact that these migrants werein the young age. However, if migrants havefound suitable job, they will tend to maintainthe job. The explanation of this situation isthat for those who have low education level,they have very limited job opportunity andthey have lower opportunity to get better jobin rural area/small city compared to those whohas higher education level.

Other factors that also affect the resilienceof a migrant to stay in the destination regionare the cost of migration. In order to calculatethe cost of migration, this study used severalvariables as an approach. These variables areburden (beban), status of the region (deskot)and distance (jarak). Beban=1 if migrant issingle or married and bring along his family.Beban=0 if migrants is married and left hisfamily in his original region. It can be seen inFigure 2 that migrants who has married or usedto married and left their family in their originalregion have higher survivor function comparedto single migrants or migrants who bring alongtheir family. This situation might occur sincemigrants who left their family in their originalregion were actually do not have other optionsthan working in the destination region to pro-vide a decent life for their family, or becauseskills they have can only be used in the desti-nation region, or due to other factors related

Page 19: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 15

Figure 1: Survivor Function for Education Variable

Figure 2: Survivor Function for Burden Variable

Figure 3: Survivor Function for Status of Original Region Variable

Page 20: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 16

Figure 4: Survivor Function for Real Income Variable

to their original region.

Migrants who came from rural area tend toreturned sooner than those who came fromsmall and big city (Figure 3). Meanwhile mi-grants who came from big city tend to havehigher survivor function compared to two othergroups. This situation might occur due to mi-grants’ bonding with their original region ordue to migrants’ communication skill. Migrantswho came from big city relatively do not haveproblem with their communication ability aswell as their ability to adapt, considering thatthey had adequate access and other support-ing infrastructures. However, for those whocame from rural area, it takes longer timeto adapt. In this matter, communication withfamily/relatives in their original region or peo-ple who came from the same region becamean important matter in the adaptation process.This is a factor that determines migrants’ in-capability to survive at the early years of theirmigration. However, in the 4th-5th years, thecurve became sloping. The same thing hap-pened to migrants who came from small city,but their survival level is higher.

Meanwhile, in order to see whether wealthand prosperity affect to migrants’ tendency toreturn to their original region, we randomly setvalue between Rp. 1000,000,- /month as the de-terminant of real income. Figure 4 shows thatthere is tendency for migrants with wage levelhigher than Rp. 1000,000,-/month has lowersurvivor function compared to migrants with

lower wage level. If this wage level representsmigrants’ competence level, then it can be con-cluded that migrants’ with lower competencylevel has higher endure compared to migrants’with higher competency level. This phenom-ena supports negative selectivity theory, whichstated that someone with low education leveltend to move to other region which more pros-per and offered opportunities to get higher in-come.

8. Factors Affecting The Duration ofMigration: Total and Based on Mi-grants’Original Region

The result of regression estimation in Table7 showed that migrants are very rational. De-cision making process regarding their return ishighly affected by their condition at the par-ticular time, where migrants will always com-pare the condition of their original region withthe destination region before they decided toreturn. This situation caused insignificance ofvariables taken at the time they started to mi-grate in the regression. On the other hand, theimplementation of regional autonomy also af-fected migrant’s behavior. More than 60 per-cent of migrants who stay in the destinationregion at the time when regional autonomy wasimplemented tend to keep staying in the desti-nation region. This situation happened for thewhole region which was estimated. Migrantsalso compared economic development between

Page 21: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 17

original region and destination region in orderto make decision about their return. For in-stance they concerned about increase in realincome in both region as well as local economicdevelopment.

Migrant’s preference is a factor that stronglyaffect the duration of migration. Short dura-tion of migration tend to occur when migrantis male, his first migration was at his youngage and married. However, if observation wasdone for each island, gender factor only signif-icant in Java, not significant in Sumatra andother regions. The regression result also showsthat the opportunity to pursue higher educa-tion level will extend the duration of migration.There are several causes related to this finding.First, pursuing education will make migrantstay longer to finish their education program.Second, opportunity to improve their educa-tion will cause migrant to practice their skillin the destination region before they made de-cision whether they will return or stay. Bothcases support point of view stated that the aimof migration is not only to improve migrant’scapacity but also to show that not all returnmigrants are those who failed in the destina-tion region. It is expected that the first causewill support regional development, and also itis expected that if migrant decided to work indestination region, they will return so that theoriginal region will have advantages from theirimprovement in knowledge and working skill.However, even if migrant decided to stay inthe destination region it does not mean thatthe original region will lose their best humanresource. In order to elaborate this phenomenadata and further research regarding this groupof migrants is needed.

In accordance with the expectation that wel-fare level is the main consideration for migrantto choose whether they will migrate longeror shorter. The duration tends to be fasterif the assets are likely to increase. However,increase in migrant’s family consumption willextend the duration of migration. This situa-

tion occurred since family consumption coversnot only food consumption, but also educationand other non-food consumption. This couldbe considered as the responsibility of migrantstoward their family.

Judging from the cost factor of migration,there is no proof that the distance affect theduration of migration. But specifically for mi-grants in Java, the higher the increase in dis-tance between original and destination region,the harder migrants decide to return. Thesematters also related to the cost of return ofmigration, either in financial or psychologicalaspect. it is interesting to observe that for mi-grants who came from Java and other regionsoutside Sumatra, if the original region is acity, then the duration of migration tends tobe longer. In other words, migrants who camefrom rural area tend to return to their originalregion sooner. This situation shows 2 things.First, migrant’s bound with original region ishigher for migrant who came from rural areacompared to migrant who came from city. Sec-ond, non-material cost of migration for migrantwho came from city in no as high as migrantwho came from rural area. The second condi-tion might be due to technology and accessi-bility factor, which make migrant feel close totheir family and original region even thoughthey are physically separated. Better accessi-bility also caused easiness for migrant from bigcity to return when necessary.

The regression results also show that if therate of growth of wages in the destination re-gion increases, migrant will delay his return for69.2 percent. However the growth of wages inthe original region is not significant in affectingthe duration of migration. This situation sup-ports the hypothesis stated that migrants havetendency to extend their duration of migrationif there is increase in wage difference betweendestination and original region, since this dif-ference in wage will increase their lifetime in-come. However migrants not always decided toreturn when there is increase of wage rate in

Page 22: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 18

Table 7: Factors Affecting The Duration of Internal Migration 1993-2007

VariableIndonesia Java Sumatra Others

Coef HR Coef HR Coef HR Coef HR

Age when leave-0.0315*** 0.969*** -.0251*** 0.975*** -.037*** 0.964*** -.0384*** 0.962***(0.00429) (0.00416) (0.00466) (0.00454) (0.00773) (0.00745) (0.0120) (0.0115)

Ln(wealth)0.0600*** 1.062*** 0.0397 1.040 0.0541 1.056 0.141*** 1.151***(0.0190) (0.0202) (0.0273) (0.0284) (0.0394) (0.0415) (0.0444) (0.0512)

Real consumption ratio -0.0364* 0.964* -0.307*** 0.736***return/leave (rkr) (0.0189) (0.0182) (0.101) (0.0746)

Consumption square ratio0.000187** 1.000** 0.0131*** 1.013***(9.42e-05) (9.42e-5) (0.00425) (0.00431)

Education improvement-0.187*** 0.830*** -.212*** 0.809*** -.155*** 0.856*** -0.222*** 0.801***(0.0244) (0.0202) (0.0372) (0.0301) (0.0384) (0.0329) (0.0539) (0.0432)

Ratio of wage growth in -1.026*** 0.358*** -.062*** 0.346*** -.013*** 0.363*** -1.095*** 0.334***destination/original region (0.112) (0.0400) (0.136) (0.0469) (0.219) (0.0795) (0.222) (0.0743)Status of original region -0.324*** 0.723*** -.411*** 0.663*** -0.396** 0.673**

(0.0667) (0.0482) (0.0860) (0.0570) (0.169) (0.113)GDP growth rate in -1.192*** 0.304*** -3.296** 0.0370** -3.669*** 0.0255***destination region (0.384) (0.117) (1.597) (0.0591) (0.929) (0.0237)

Growth rate square0.349*** 1.418*** 1.064*** 2.898***(0.107) (0.151) (0.254) (0.735)

Regional Autonomy-0.832*** 0.435*** -.797*** 0.451*** -0.769** 0.463** -1.274*** 0.280***(0.140) (0.0608) (0.205) (0.0924) (0.300) (0.139) (0.248) (0.0692)

Gender0.218*** 1.243*** 0.242*** 1.273*** 0.157 1.170(0.0621) (0.0772) (0.0812) (0.103) (0.111) (0.129)

ln (distance)-0.0417** 0.959** -0693*** 0.933***(0.0201) (0.0192) (0.0219) (0.0204)

ln (real consumption)-0.104** 0.901**(0.0490) (0.0441)

Marital Status-0.148* 0.863* -0.278* 0.758*(0.0787) (0.0679) (0.151) (0.114)

GDP growth ratio in -0.0354 0.965destination/original region (0.0255) (0.0246)

Observations 1,257 713 362 228No of Failures 800 433 233 136Wald chi2 509,98 346,02 110,22 126,18

Robust standard errors in parentheses***p<0.01,**p<0.05,*p<0.1

Page 23: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 19

the original region.

Besides wage rate, regional variables areGDP growth rate and regional autonomy.Overall GDP growth rate in the destination re-gion extend the duration of migration for about60 percent. When analyzed by region, increasein GDP growth rate will extend the durationof migration more than 90 percent in Java andother regions. In Sumatra, if GDP growth ratein destination region is higher than in the orig-inal region, then duration of migration will alsoextend although the value is not significant, i.e.about 3.5 percent.It also interesting to observemigrants’ behavior in the destination region af-ter year of 2000 (the year when regional au-tonomy was implemented). They tend to havepossibility to stay longer for about 56 percentcompared to those who have returned before,and this result is significant in all regions.

Especially for migrants who came fromSumatra, they have different behavior. Thereare only two factors indicated that shorteningthe duration of migration. Those factors are in-crease in family asset, and gender. The effect ofboth factors are not significant. This situationmight due to culture factor or other factor thatneed to be studied further.

9. Conclusion

From the regression it can be seen that theregression result supports the hypotheses. Itcan be seen that migrants are very rational,that is before they decided to return or stay,they will compare the economic developmentbetween original and destination region in or-der to make decision.

(a) Migrants’s preference factors strongly af-fected the duration of migration. Dura-tion of migration tend to be significantlyshorter if the migrants are male, theyare young and not married at the firsttime they migrate. However, opportunityto pursue higher education level will ex-tend duration of migration.

(b) Welfare level is the main consideration formigrant to decide whether they will ex-tend or shorten their duration of migra-tion. The duration of migration tend to belonger after the implementation of regionalautonomy, and if there is an improvementof economic condition in the destinationregion, either in the side of regional eco-nomic or regional wage growth. However,economic and wage growth in the originalregion not always affecting the durationof migration. This situation was indicatedfrom the regression result that was highlyinsignificant so that it has to be excludedfrom the equation.

(c) As a migration cost factor, there is no ev-idence that distance variable affected theduration of migration. However, the longerthe distance between destination and re-gional region, then it became harder formigrants to decide to return. This mightdue to the cost of return migration, eitherin material or psychological aspect.

10. References

[1] Armstrong, Harvey dan Jim Taylor (2000), Re-gional Economics and Policy, 3rd edition, Black-well Publisher.

[2] Arntz, Melanie (2006), What attracts human cap-ital? Understanding the skill composition of inter-nal migration Flows in Germany,

[3] Aydemir, A., & Robinson, C. (2008). Global labourmarkets, return, and onward migration.CanadianJournal of Economics, 41(4), 1285-1285.

[4] Bauer, Armin dan Myo Thant (2010), ed, Povertyand Sustainable Development in Asia: Impact andResponses to the Global Economic Crisis, AsianDevelopment Bank.

[5] Beine, M., Docquier, F., & Rapoport, H. (2001).Brain drain and economic growth: Theory and ev-idence. Journal of Development Economics, 64(1),275-289

[6] Bell, M., Australia. Bureau of, I., & Population,R. (1996). Understanding internal migration. Can-berra: A.G.P.S.

[7] Bell, M., Blake, M., Boyle, P., Duke Williams, O.,Rees, P., Stillwell, J., & Hugo, G. (2002). Cross-national comparison of internal migration: issues

Page 24: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 20

and measures. Journal of the Royal Statistical So-ciety: Series A (Statistics in Society), 165(3), 435-464.

[8] Blossfeld, Hans-Peter dan G?tz Rohwer (2002),Techniques of Event History Modelling: New Ap-proachs to Causal Analysis, 2nd ed. LawrenceEribaum Associates, Inc, London.

[9] Borjas, G.J. (1987). Self-Selection and the Earn-ings of Immigrantsss. American Economic ReviewVol. 77, 1667-1717.

[10] Borjas, G.J., Bronars, S.G. and S.J. Trejo (1992).Self-Selection and Internal Migration in the UnitedStates.Journal of Urban Economics Vol. 32, 159-185.

[11] Carrion-Flores, Carmen E (2006), What MakesYou Go Back Home? Determinants of The Dura-tion of Migration of Mexican Immigrantsss in theUnited States, Mimeo, The University of Arizona.

[12] Chau, Nancy & Oded stark (1999), Migration un-der Asymmetric Information and Human Capi-tal Formation, Review of International Economics,7(3), 455-483

[13] Collyer, Michael (2004), The Development Impactof Temporary International Labour Migration onSouthern Mediterranean Sending Countries: Con-trasting Examples of Morocco and Egypt, Work-ing Paper T6, Development Research Centre onMigration, Globalisation and Poverty

[14] Conway, D., & Potter, R. B. (2009). Return migra-tion of the next generations: 21st century transna-tional mobility. Farnham, England: Ashgate

[15] Dmurger, Sylvie dan HuiXu (2010), Return Mi-grantsss: The Rise of New Entrepreneurs in RuralChina, GATE Working Paper No 1008.

[16] Dmurger, Sylvie dan HuiXu (2010), The Durationof China’s Internal Migration - What makes RuralMigrantsss Stay Away Longer, Preliminary draft,December 2010.

[17] Devillanova, Carlo (2004), Interregional Migrationand Labor Market Imbalances, Journal of Popula-tion Economics, 17:229-247

[18] Dustmann, Christian (2003), Retun Migration,Wage Differentials, and The Optimal MigrationDuration, European Economic Review, 47.

[19] Dustmann, Christian, Itzhak Fadlon and YoramWeiss (2010), Return Migration, Human CapitalAccumulation and the Brain Drain, CReAM Dis-cussion Paper No 13/10, Centre for Research andAnalysis of Migration, Department of Economics,University College London

[20] Dustmann, Christian and JosepMestres (2008),Remittances and Temporary Migration, CReAMDiscussion Paper No 09/09, Centre for Researchand Analysis of Migration, Department of Eco-nomics, University College London

[21] Dustmann, Christian and Yoram Weiss (2007), Re-

turn Migration: Theory and Empirical Evidence,CReAM Discussion Paper No 02/07, Centre forResearch and Analysis of Migration, Departmentof Economics, University College London

[22] Fan, C Simon dan Oded Stark (2007), The BrainDrain, ’Educated Unemployment’, Human CapitalFormation, and Economic Betterment, Economicsof Transition Volume 15(4), 629-660

[23] Fasbender, Karl (1989), Rural Migration and Re-gional Development: The Example of Indonesia,INTERECONOMICS, July/August 1989

[24] Fox, John (2002), Cox Proportional-Hazards Re-gression for Survival Data. Appendix to an R andS-Plus Companion to Applied Regression. Hosmer,David W. Jr. dan Stanley Lemeshow (1999), Ap-plied Survival Analysis: Regression Modelling ofTime to Event Data. John-Willey & Sons, Inc.

[25] Hugo, Graeme J.,Terence H. Hull, Valerie J Hulland Gavin W. Jones (1987), The Demographic Di-mension in Indonesian Development, Oxford Uni-versity Press.

[26] King, Russel (1986), Return Migration and Re-gional Economic Problems, Mackays of ChathamLtd, Kent.

[27] Kirdar, Murat G. (2010), Source Country Char-acteristics and Immigrantsss’ Migration Durationand Saving Decisions, Discussion Paper No. 4899,Forschungsinstitut zur Zukunft der Arbeit Insti-tute for the Study of Labor

[28] Kleinbaum, David dan Mitchel Klein (2005),Survival Analysis: A Self-Learning ext, 2nd ed.Springer.E-book.

[29] Kong, Sherry Tao danTadjuddin NE (2011), Occu-pational Choice and Mobility Among Migrantsssto Four Cities, dalam Manning, Chris danSu-darnoSumarto, Employment, Living Standardsand Poverty in Contemporary Indonesia, ISEAS.

[30] Lehmer, F., & Ludsteck, J. (2011). The returns tojob mobility and interregional migration: Evidencefrom Germany. Papers in Regional Science, 90(3),549-571.

[31] Liu, Xian (2012), Survival Analysis: Model andApplications, Higher Education Press, John Wiley& Sons Ltd. United Kingdom.

[32] Magris, Francesco and Giuseppe Russo (2003),Frontiers Openness and the Optimal MigrationDuration, CELPE, Universit degli studi di Salerno,via Pon te don Melillo.

[33] Martin, Philip (2005), Migrantsss in the global la-bor market, A paper prepared for the Policy Anal-ysis and Research Programme of the Global Com-mission on International Migration.

[34] Massey, Douglas S, et.al. (1994), An Evaluation ofInternational Migration Theory: The North Amer-ican Case, Population and Development Review,Vol. 20, No. 4 (Dec., 1994), Population Council

Page 25: Internal Migration in Indonesia: Duration and Factorsecon.feb.ui.ac.id/wp-content/uploads/2015/10/201505.pdfInternal Migration in Indonesia: Duration and Factors Hera Susanti1, Arie

Damayanti, A. and Susanti, H. /Internal Migration in Indonesia: ... 21

[35] Mueller, Charles F (1982), The Economics of La-bor Migration: a Behavioral Analysis. AcademicPress Inc. London.

[36] Muhidin, Salahudin (2002), The Population of In-donesia: Regional Demographic Scenarios Using aMultiregional Method and Multiple Data Sources,Population Studies, Rozenberg Publishers, Ams-terdam.

[37] Mulder, Clara H dan Maarten van Ham (2005), Mi-gration Histories and Occupational Achievement,Population, Space and Place 11, 173-186.

[38] Newbold, K. B., & Bell, M. (2001). Return andonwards migration in Canada and Australia: Ev-idence from fixed interval data. International Mi-gration Review, 35(4), 1157-1184.

[39] Newbold, K. B., & Liaw, K. L. (1994). Return andOnward Interprovincial Migration through Eco-nomic Boom and Bust in Canada, from 1976-81to 1981-86. GEOGRAPHICAL ANALYSIS, 26(3),228-245.

[40] Stark, O. (1995). Return and dynamics: the path oflabor migration when workers differ in their skillsand information is asymmetric. The Scandinavianjournal of economics, 97(1), 55-71

[41] Wescott, C. dan J. Brinkerhoff (2006), ed, Convert-ing Migration Drains into Gains: Harnessing theResources of Overseas Professionals, Asian Devel-opment Bank.

[42] White, M. J., Lindstrom, D. P. (2005). CH 11.Internal Migration (pp. 311-346). Boston, MA:Springer US.

[43] Wickramasekera Piyasiri (2002), AsianLabour Mi-gration: Issues and Challenges in an Era of Glob-alization, International Migration Paper 57, ILOGeneva

[44] Strauss, John, Firman Witoelar, Bondan Sikokidan Anna Marie Wattie (2009), The Fourth Waveof The Indonesia Family Life Survey: Overview andField Report, Volume 1. the RAND Labor andPopulation working paper series. April 2009. WR-675/1-NIA/NICHD

[45] Tirtosudarmo, R. (1999). The Indonesian State’sResponse to Migration. Journal of Social Issues inSoutheast Asia, 14(1), 212-228.

[46] Titus, M.J. (1978) Interregional Migration In In-donesia As A Reflection Of Social And RegionalInequalities. Tijdschrift voor Economie en SocialGeografie 69 (1978) Nr. 4

[47] Tjiptoherijanto, Prijono (1996), Sumber DayaManusia Dalam Pembangunan Nasional, LembagaPenerbit FEUI.

[48] Van Lottum, J., & Marks, D. (2012). The de-terminants of internal migration in a developingcountry: quantitative evidence for Indonesia, 1930-2000. Applied Economics, 44(34), 4485-4494.