Remittances and Temporary Migration

Christian Dustmann and Josep Mestres ∗

This Version: January 2008

Abstract

In this paper we study the remittance behavior of immigrants and how it relates to

temporary versus permanent migration plans. We use a unique data source that provides

unusual detail on the purpose of remittances, asset holdings, savings, and return plans,

and follows the same household over time. We commence by formalizing a simple model

that allows for the possibility of return to the home country, and where remittances are

determined together with savings and consumption. We then analyse the effect of individual

and household characteristics, the geographic location of the family, and return plans on

remittances. The panel nature of our data allows us to condition on household fixed effects.

Our results suggest that changes in return plans lead to large changes in remittance flows,

as well as to a different allocation of assets.

JEL Classification: F22,F24

Keywords: International Migration, Remittances.

∗University College London and Centre for Research and Analysis of Migration (CReAM). Email:

c.dustmann@ucl.ac.uk,j.mestres@ucl.ac.uk.

1

1 Introduction

The amount of remittances sent by immigrants back to their home countries has increased

steadily over the last three decades. As the International Monetary Fund (IMF) reports (World

Economic Outlook (2005)), remittances currently account for more than $100 billion a year,

”exceeding export revenues, foreign direct investment and other capital inflows” as a source of

foreign exchange for many developing countries. Not only has the volume of remittances in-

creased, but also its importance as a percentage of GDP: Remittances now account on average

for about 1.5% of GDP for developing countries (World Economic Outlook (2005)). Remit-

tances help economic development and are a major factor in poverty reduction1. In addition,

remittances are now one of the primary sources of foreign exchange for many receiving countries.

For immigration countries, remittances constitute a non-negligible outflow of capital. For

instance, for Germany, the volume of remittances was about 0.31% of GDP in 2003 (Bundesbank

(2006)).2 This is equivalent to 150 % of Germany’s total budget for official development aid.3

It is therefore not surprising that a large literature has developed on the subject, see Rapoport

& Docquier (2005) for an excellent survey. Key issues to understand are which migrant popu-

lations remit, for which purpose, and what determines the amount of remittances. Answers to

these questions may help to create migration schemes that affect the way remittances are chan-

neled into different purposes, thus supporting their optimal efficiency for economic development,

and raising awareness about how different policies will lead to different incentives to remit.

On the determinants of remittances, a number of papers develop models for the different

motives that may trigger remittances, and explore some of their empirical implications.4 This

research has provided us with a wealth of insight. Yet, on the empirical level we still know

relatively little about the determinants of remittances, the various forms remittances may take,

1See e.g. Adams & Page (2005), Adams (2006) and Acosta et al. (2006) for analysis.2Germany is the third largest source country of remittances payments, after United States and Saudi Arabia,

see Ratha (2003).3Official Development Assistance accounted for 0.21% of GDP in Germany in 2003, see OECD (2006).4See e.g. Lucas & Stark (1988), Lucas & Stark (1985), Hoddinott (1994), Funkhouser (1995), Poirine (1997),

Agarwal & Horowitz (2002), de la Briere et al. (2002), Faini (2006), Okonkwo Osili (2007) and Amuedo-Dorantes

& Pozo (2006).

1

and how these interact with migrant behavior and the forms of migration. One particular aspect,

which is in our view important, is the way the permanency of a migration affects the magnitude

of remittance flows, the purpose of remittances, and the allocation of asset holdings.

We address these questions in this paper. We analyse how remittance flows are related to the

permanency of migration, and to the residential location of the family. A further contribution of

our paper is to investigate asset- and property holdings of immigrants, how these are distributed

between the home and the host country, and how this is influenced by the permanency of

migration plans. We commence by formalizing a simple model that allows for the possibility of

return to the home country. Our model distinguishes between different motives for remittances.

We then give a detailed account of the various forms of remittances and asset holdings, and how

these relate to individual and household characteristics, as well as the permanency of migration.

Our empirical analysis is based on a panel data set of immigrants over the period from 1984-1994.

Our analysis distinguishes between remittances for family support, savings, and for a residual

category ”other purposes”. Due to the information our data provides us about the return plans

of immigrants, we are able to distinguish between individuals who consider their migration as

temporary, and who consider their migration as permanent. This allows us to assess of how a

permanent vs a temporary migration relates to the various remittance flows. The panel nature

of our data, and repeated information on remittances as well as return intentions, allows us to

explore and isolate the way the permanence of migration, as well as the locational distribution

of the family, affect remittance flows, conditional on observed characteristics and unobserved

fixed differences across households in their remittance propensity.

The structure of the paper is as follows: in section 2 we present our simple model; in section

3 we provide an historical background and descriptive evidence from our data; in section 4 we

show our estimation results and in section 5 we conclude.

2 A Simple Model on Remittances

A difficulty with remittances is its measurement and exact definition. If we define remittances

as all transfers from the immigration country to the immigrant’s home country (a definition

2

which we will follow below), then remittance flows consist of both savings or investments into

future consumption, and transfers for support of family or kinship in the origin country. The

motivation for both types of transfers are different. While the first can be modeled in a simple life

cycle model (see e.g. Dustmann (1997)), the second requires altruistic behavior and/or influence

through the social reference group.

Remittances for altruistic reasons, in particular to support family and kinship, can be viewed

as intra-family transfers across national borders.5 Remittances may also respond to expectations

about fulfillment of family and social commitments on the side of the social reference group in

the home country. Satisfying these expectations can be seen as a price to be paid for the option

to return back home at a later stage, or as an ”insurance” to be welcomed after return in the

home community.6

Remittance flows may further be motivated by the wish to hold assets or savings in the home

country. These may take the form of housing stock, capital investments, or simply savings.

Thus, part of remittances in this category are not different from an intertemporal allocation of

consumption, or investment into durable consumption goods across national borders.7 A positive

probability of return may affect these transactions either by inducing a preference to holding

assets and savings in the home country, or by inducing immigrants to shift more consumption

from the present to the future, or both.

To fix ideas we will commence with the development of a simple model.8 Our model allows

for temporary and permanent migrations (see Okonkwo Osili (2007) for a similar distinction).

Migrants remit for two reasons: to support their family, and to comply with expectations in

their home communities.

In addition, we allow for two savings motives, which are inherently linked to the form of

5See Lucas & Stark (1985) for an early discussion. See Cox (1987), Cox et al. (1998) for empirical analysis

of altruistic motives for private transfers. For a recent survey on the private transfer literature see Laferrere &

Wolff (2000).6Azam & Gubert (2006) stress the role of the extended family and the village in migration and remittance

decisions. Amuedo-Dorantes & Pozo (2006) investigates this motive empirically.7As Durand et al. (1996) recognize, ”sending monthly remittances (...) and returning home with savings are

interrelated behaviors that represent different ways of accomplishing the same thing: repatriating earnings”.8See papers by Funkhouser (1995), Faini (2006), Okonkwo Osili (2007) among others for models on remittances.

3

migration. Firstly, migrants may enjoy consumption in the home country more than in the host

country, and therefore delay consumption until after return. Secondly, a higher purchasing power

of accumulated savings in the home country might also motivate migrants to delay consumption

until after return. Although simple, our model provides us with some structure about the way

we should think about how the different forms of migration and remittances interact.

2.1 The Model

Suppose the lifetime of the immigrant can be divided into 2 sub-periods: period 1 is the time to

be spent in the host country, and period 2 is the time to be spent in the home country after a

possible return. Return in period 2 takes place with probability p. In the case that p = 0, the

migration is permanent. Consider the following inter-temporal utility function:

U = u1(c1) + γ W (R) + p[β ũ2(cE2) + δ v(R)] + (1− p) [u2(cI2)] (1)

In equation (1), u1 is the sub-utility in period 1 in the host country and and ũ2 and u2 are

the sub-utilities in period 2 in the home and in the host country respectively, which we assume

as being strictly concave in consumption. Further, c1, cI2 and cE2 are first and second period

consumption in immigration (index I) and emigration (index E) countries respectively. The

parameter p ∈ [0, 1] is the probability the migrant attaches to a possible return to the home

country in the second period. The utility the immigrant obtains by supporting family members

through remittances is given by γ W (R), where the amount of remittances is given by R, and

the parameter γ measures the degree of altruism. The function W (R) translates remittances R

into utility; it depends for instance on the size of the family residing in the home country.

Consumption in the second period in the case of a return may induce more utility than

consumption in the host country, due to complementarities through climate, friends etc. This

is captured by the parameter β. If β > 1, the migrant has a higher level of utility and a higher

marginal utility if he/she consumes in the home country.

Utility in the second period in the case of a return will also depend on the size of remittances

sent in the first period, through the function δ v(R), which is increasing in R. It translates the

4

level of remittances sent into ”social appreciation”. This captures the second motive for remit-

tances we have discussed above: remittances sent to fulfill expectations of the home community.

These remittances may be for family and kinship, but also for e.g. community programmes. For

zero or too small levels of remittances, v(R) may well be negative, thus reducing utility in the

second period in the home country.

The budget constraint for the first period is given by w1 = c1 +R+s. The budget constraint

for the second period is wI2 + s = cI2 in the case of a permanent migration and wE2 + r s = cE2

in the case of a return. Earnings in period 1 are denoted by w1, and in period 2 by wE2 and wI2

in home and host countries respectively. The purchasing power of the host country currency in

the home country is given by r. If r > 1, the purchasing power of the host country currency is

higher in the migrant’s home country.

The choice variables in period 1 are savings s and remittances R. Given the budget constraint,

these fix consumption in the first period (c1) and in the second period (cE2, cI2). The first order

conditions are given by:

d

d s: u11 = p β ũ

21 r + (1− p)u21 (2)

d

dR: u11 = γ W1 + pδ v1(R) (3)

where the subscript 1 denotes the first derivative.

Equations (2) and (3) determine the optimal level of remittances and savings. Savings will

be set such that the marginal cost in terms of forgone utility in period 1 is equalized to the

expected marginal return in period 2. If p = 0 (the migration is permanent), savings will

equalize the marginal utility of consumption in the two periods in the host country. If p > 0,

immigrants will save in period 1 for two reasons. Firstly, purchasing power differences (i.e.

r > 1) may induce the immigrant to shift consumption from the present to the future. Secondly,

higher (marginal) utility from consumption at home (i.e. β > 1) may induce inter-temporal

re-allocation of consumption.

Remittances will be set by equalizing the marginal cost of remitting in period 1 in terms

5

of forgone utility with the marginal return in period 2. Consider first the case p = 0, i.e. the

migration is permanent. The only motive for remittances is altruism, and the magnitude of

remittances depends on the degree of altruism γ and the degree to which remittances enhance

welfare of the family. Now consider the case of a positive probability of return. Even if γ = 0

(i.e. there is no altruism), remittances may still be positive, as they contribute to build up social

appreciation at home.

Deriving the comparative statics is relatively straightforward, and we summarize these in

table 1. Both remittances and savings increase in first period’s host country earnings. An

increase in second period’s earnings increases remittances, while it decreases savings. Also, a

higher altruistic weight increases remittances, but decreases savings.

The effect of an increase in the return probability on both savings and remittances is am-

biguous for the general case. If the only motive for remittances was altruism (δ = 0), an increase

in the return probability would decrease remittances and increase savings9. The reason is that

immigrants would rather invest into savings, as this increases their utility in the second period.

However, if δ > 0, an increase in the return probability may now increase remittances, through

increasing social appreciation, or decrease remittances, through the competing savings motive

as before. Which effect will dominate is unclear ex ante. Similarly, the ambiguity of the effect

of return on savings still persists when δ 6= 0.

2.2 Empirical Specification

We now turn to our empirical model. Our main interest is in determining how the level of remit-

tances is affected by household characteristics, and by immigrants’ return plans. We estimate

regressions of the following type:

Yit = a0 + a1Xit + a2Zit + ξRIit + �i + uit , (4)

where the indices i and t denote households and time, and Yit measures remittances or

savings. The key variable of interest is RIit, which is a measure of the temporariness of the

9Whenever r βũ21−u21 > 0, individuals prefer to reallocate resources from the first period to the second period.For example, whenever β > 1, r > 1 and wI2 = wE2.

6

migration. As we explain below in more detail, we obtain this variable from survey questions

on the migrant’s intention to return home, which we observe in every wave of the panel that we

use.

The vector X collects characteristics of the household and the head of household. We include

here the log of disposable household income, the number of adults and the number of children

(below the age of 16) living in the household, and the number of employed household members.

We also include characteristics of the head of household, like the employment status, the years

since migration and its square, the number of years of education, and whether the partner is

native born or the household head is single.

The vector Z includes other variables that affect remittance behavior. These are variables

about whether the spouse or children are living abroad, and an indicator variable whether the

head of household grew up in a rural area. Village communities are likely to exert more social

control than urban communities, and may have expectations that remittances are channeled to

the larger family and possibly the community, so that this variable picks up some of the insur-

ance motive we have discussed above. We shall concentrate below on discussing the parameter

(vectors) a2 and ξ.

Unobserved Heterogeneity

Estimation of 4 may result in biased estimates, as individuals who tend to return may have

at the same time a higher propensity to send remittances. In this case, our estimate of ξ will

be (upward) biased, as E(�i|X,Z,RI) 6= 0. Some of this bias is likely to be eliminated by

conditioning on the variables in X and Z. The repeated information we have on remittances

and return intentions allows us in addition to eliminate any remaining bias due to unobserved

household fixed effects, using within household variation for estimation. Under the assumption

that E(ũit|X̃, Z̃, R̃I) = 0 (where .̃ denotes deviations from individual specific means), a fixed

effects estimator will give us consistent estimates of ξ and a2.

One problem with this estimation strategy is that return intentions are likely to be measured

with (possibly considerable) measurement error, thus creating an attenuation bias. This problem

is greatly exacerbated when estimating the model in differences or using fixed effects (see e.g.

7

Hsiao (1986) for a detailed discussion). In order to obtain consistent estimates of ξ, we use a

GMM estimator as suggested by Arellano & Bond (1991). This estimator eliminates the fixed

effect �i and corrects for the measurement error using as instrumental variables past values of

return plans. We provide details of the econometric model in the appendix.

To apply the estimator, we first transform our data to eliminate the unobserved fixed effect,

using an orthogonal deviations transformation. This transformation is an alternative to first

difference transformation, but it is preferable if the data has gaps, as it minimizes data loss. It

subtracts from the observations t = 1, ..., T − 1 the mean of the remaining future observations

in the sample to purge the individual unobserved effect10. We then correct for the measurement

error using as instruments past values of intentions in levels. If the measurement error is ”clas-

sic”11, past values of intentions in levels are valid instruments as E(MEit, RIit−s) = 0, s > 0, and

the GMM estimator provides us with unbiased and consistent estimates of the effect of return

plans on remittance behavior.

Selection through Return Migration

A remaining problem with the interpretation of the parameters is that our sample is selected

- over the course of the panel, we observe more households who have a higher propensity to

stay permanently. This selection may be correlated with our measure for a return migration

intention: those with a higher intention to return will be less likely to be in the sample. If those

who remain in the sample have different remittance behavior (conditional on all the variables

we include in the model as well as the measure for the return intention), then this will bias our

estimate for ξ.

This bias can be signed under some assumptions: it will be downward if the residuals in the

selection equation and the remittance equation are positively correlated (indicating that those

who remain in the sample remit less than those who drop out of the sample due to return,

10Formally, the orthogonal deviations transformation of Xit is defined as (XOit =√T − t/(T − t+ 1)(Xit −

1T−t

∑(Xis)); s > t, where

√T − t/(T − t+ 1) are weights to standardize the variances. See Arellano & Bover

(1995) and Roodman (2006) for details.11That is, the real return intention RI∗it is uncorrelated with the measurement error (E[MEit, RI

∗it] = 0) and

the measurement errors are serially uncorrelated: E[MEit,MEis] = 0, t 6= s.

8

conditional on other regressors)12. We can therefore interpret the coefficients on the temporary

migration measure as a lower bound.

When conditioning on individual effects, this problem will disappear if selection is based

on ”permanent” characteristics, as in this case the selection term is constant over time and is

conditioned out.

3 Background, Data and Descriptive Evidence

3.1 Background

Between the mid 1950’s and 1973, the strong economic development in Northern Europe and the

resulting demand for labor led to a large inflow of immigrants mainly from the periphery countries

of Europe, but also from Turkey, North Africa, South America and Asia. The main receiving

countries were Belgium, France, Germany, the Netherlands, Switzerland, and the Scandinavian

countries.

The West-German economy experienced a strong upward swing after 1955, accompanied by

a sharp fall in the unemployment rate (between 1955 and 1960, the unemployment rate fell from

5.6 % to 1.3 %) and an increase in labor demand. This generated a large immigration of workers

from Southern European countries and Turkey into Germany. The percentage of foreign born

workers employed in West Germany increased from 0.6 percent in 1957 to 5.5 percent in 1965,

to 11.2 percent in 1973. Bilateral recruitment agreements were set up between Germany and

Italy, Spain, Greece, Turkey, Portugal and Yugoslavia in the 1950’s and 1960’s.

12More formally, suppose that the latent index for being selected into the sample, s∗ is linear in RI, the return

intention, with s∗i = α0 +αRIi + ei, and that an individual is in the sample if s∗i > 0. Suppose that the outcome

equation is given by yi = γ0 + γRIi + fi , and assume that ei and fi are jointly normally distributed, with

variances 1 and σ2v and correlation coefficient ρ. Then selection could be accounted for by adding the generalized

residual E(fi|s∗i > 0) = λ(ci) to the estimation equation, where λ(ci) = φ(ci)/Φ(ci), with φ and Φ being thedensity and distribution function of the standard normal, and ci = α0 +αRIi. We obtain the estimation equation

yi = γ0 +γRIi +σv ρ λ(ci) + ζi . Omission of λ(ci) results in a biased estimate for γ. The expectation of the error

term when omitting λ, conditional on RIi, is ρ σv E(λ(ci)|RIi). Since λ decreases in ci, the bias is downward forρ < 0 and α < 0.

9

Labor migration over this period was initially considered as temporary by both the immigra-

tion countries and the emigration countries. Individuals were not expected to settle permanently.

The German recruitment policy was based on the assumption that foreign workers would after

some years return to their home countries. Still, although return migration has been quite con-

siderable (see Bohning (1987)), a large fraction of foreign born workers settled permanently13.

3.2 The Data and Sample

We use for this analysis 12 waves of the German Socio-Economic panel (GSOEP 1984-1995). The

GSOEP is a household-based panel survey, similar to the US Panel Study of Income Dynamics

(PSID) or the British Household Panel Study (BHPS). Initiated in 1984, the GSOEP oversamples

the then resident immigrant population in Germany, which stems from the migration movement

we have described above. In the first wave, about 4500 households with a German born household

head were interviewed, and about 1500 households with a foreign born household head. The

data are unique in providing repeated information on a boost sample of immigrants over a long

period of time. For our analysis, we use observations for the foreign born from the over-sample,

as well as the from the standard sample.

Each individual in a household and over the age of 16 is interviewed. The household head

provides information about all other individuals in the household and below the interviewing

age. Individuals who leave households and form their own households are included in the panel.

The GSOEP data provides a rich set of survey questions on remittances and savings. It

distinguishes between remittances for family support, remittances for saving purposes in the

home country, and remittances for other motives. Information on remittances and savings is

measured on the household level. The data on remittances is both qualitative and quantitative.

Immigrants are asked whether they remit for each of the above purposes. They are further asked

to quantify the amount of money they sent back home for each of these purposes during the

previous calendar year. Information on remittances is available for the years 1984-1994, with

13The stock of foreign labor in Germany in 2004 was 3.7 million people, of which around 60 per cent originated

from the sending countries considered here (International Migration Outlook (2006)).

10

the exception of the years 1991 and 199314. All monetary variables are at the household level in

real amounts, where the reference year is 2002.

Finally, the data also contains information on asset holdings of immigrants, including home

ownership in the host and home country. This data is available in one cross section only, for the

year 1988.

A further unique feature of our data is that immigrants provide information in each wave

of the panel whether they intend to remain permanently in Germany, or whether they wish

to return home at some stage in the future. We use this information to differentiate between

temporary and permanent migrants. If economic decisions are involved, it is likely that these

are based on intentions of this sort, rather than on possible realizations at a later stage.

In addition, we have individual and household characteristics in the host country, as well as

information on family members who are living in the country of origin. There is no information

on the use of remittances by the family members in the origin country, or of other household

characteristics or income in the home country.

3.3 Descriptive Evidence

As we mention above, we measure remittances on the level of the household. When we refer

to characteristics of individuals within households, we typically refer to the head of household.

Entries in Table 2 show that the average age of household heads in our sample is 44 years,

and that migrants resided slightly less than 20 years on average in Germany. More than 85

percent of the head of households are male, and 81 percent are employed. The average net

household income is 23500 Euros (in 2002 prices). Less than 10 percent of households are single

households, and about 6 percent of household heads are married with a native partner. With

respect to members of the family living abroad, around 5 percent of heads of households report

that their partner lives abroad. The percentage of head of households that have children under

the age of 16 in another country (different from the host country) is 10 percent.

The variable ”rural childhood” measures the degree of urbanization of the area where the

14See Table A1 in the Appendix for an exact description of the variables as well as data availability.

11

immigrant has spent his childhood (up to age 15), distinguishing between city and countryside.

This variable may proxy the ”traditionality” of the immigrant’s reference group, and thus be a

measure for the degree to which social norms necessitate remittances for insurance type motives,

as we have discussed above. Almost 40 percent of all heads of households report that they grew

up in a rural area. The last variable measures the return intention of the household head. On

average, more than half of the household heads in our sample report that they would wish to

return to their home country at some point in the future.

Table 3 shows the household composition in our sample, for the period of observation 1984-

1994. The majority of households consists of a couple with children. Single-parent as well as

multiple generation households are rare.

Table 4 gives a breakdown of remittances into its different purposes, as recorded in the survey.

We report the questions that were asked in the questionnaire, and on which our numbers are

based, in the data construction section in the Appendix. The first column shows the percentage

of households that remit. About 42 percent of households report that they sent remittances

during the period over which that information has been provided (1984-1994). Our data distin-

guishes between different types of remittances, and overlap is possible. The largest fraction of

remittances are for the purpose of family support: nearly thirty percent of the households report

to remit for that reason. Around 7 percent of households transfer remittances to be saved in the

home country, while around 10 percent sent remittances for other non-specified purposes.

In the second and third columns we report the unconditional average amount of remittances

sent, and the corresponding percentage of net disposable household income. On average house-

holds remit more than 1600 Euros (in 2002 prices) per year, which corresponds to 7.5 percent

of disposable household income. More than half of that amount is remitted for family support.

In the fourth and fifth columns we display for those households that remit the total amount of

remittances and the corresponding percentage of net disposable household income. For those

households that remit, the amount sent accounts for almost 20 percent of their disposable income.

Again, family support corresponds to more than half of the amount remitted.

In Table 5 we break down remittances by purpose, as well as by household characteristics.

The numbers in the table show a stark difference between households where the head intends

12

to remain permanently, or temporarily only. While 50 percent of households with temporary

migration intentions remit, this is the case only for 30 percent of households with a permanent

migration intention. The average amount temporary households remit is around 2100 Euros,

considerably larger than the 975 Euros of permanent households. The entries in the next columns

suggest that this difference is roughly proportional across the different purposes.

The next rows draw distinction between remittances of households with different characteris-

tics. The difference between remittances for households where the spouse lives abroad as opposed

to single households or households where the spouse lives in the host country is again large, with

more than two thirds of households in the first category sending remittances as opposed to only

42 percent in the latter one. In addition, the average amount remitted for households where the

spouse lives abroad is 3440 Euros, more than two times larger than for those households whose

head is single or where the spouse lives in the host country. There are also large differences in re-

mittance probabilities and the overall amounts remitted according to whether children are living

abroad or not. Not surprisingly, the largest differences are in the category ”family remittances”,

while ”savings abroad” and ”remittances for other purposes” are more similar.

Next, we turn to assets. In 1988, a special set of questions was included in the survey asking

about asset holdings. For immigrants, questions relate to property and asset holdings, both in

Germany and in the home country.

In Table 6, we display responses to the different sets of questions regarding asset holdings. In

the first pair of columns, we report means and standard deviations for all immigrants. The next

two pairs of columns distinguish between immigrants with permanent and temporary migration

intentions. Finally, the last pair of columns reports as a reference point respective responses for

native born individuals.

The first set of questions refers to home ownership, which includes houses, apartments or any

other property. Only about 9 percent of immigrant households report owning housing property

in Germany, which contrasts with 43 percent of native born households. The value of the

property hold by immigrants is likewise considerably lower. Distinguishing between temporary

and permanent immigrants, the numbers show quite remarkable differences. While 13 percent

of immigrants with a permanent migration intention own housing property in Germany, only 5

13

percent of those who wish to return do. Likewise, the value of the housing stock is much lower

for the latter category. On the other hand, nearly one in two immigrant households with return

intentions reports owning housing stock in the home country, compared with only 24 percent

of those with permanent intentions. Again, the value of that property is much higher for those

who wish to return. Overall, home ownership of immigrant households who wish to return is

higher than of households with permanent intentions.

The next panel reports information on asset holdings. Asset holdings refer to the total

amount of assets (including cash, savings, property, etc.) net of financial obligations. For

immigrants the questions draw a distinction between assets held in Germany, and assets held in

the home country. The numbers suggest that the average amount of asset holdings of immigrants

is considerably lower than those of natives, even after taking into account that immigrants hold

assets both in the host and in the home country. Interestingly, there is hardly any difference in

overall asset holdings between temporary and permanent immigrants, but there is in the location

of where assets are held.

To summarize, while total assets held are similar for temporary and permanent immigrant

households, there are differences in property holdings, and stark differences in where assets are

held. Temporary migrants hold less property, savings and assets than permanent migrants in the

host country, but more in the home country. This suggests different wealth allocation profiles

between those migrants who want to return and those who do not. It also suggests that the

way immigrants may possibly affect the housing market in the host as well as the home country

depends largely on their re-migration plans.

4 Estimation results

4.1 Remittances

The descriptive evidence we present in the last section suggests large differences in remittance

behavior between permanent and temporary immigrant households. Some of these differences

may be due to differences in household composition of individual characteristics of household

14

members; they may also be due to differences in the family’s residential allocation. We now turn

to regression results that hold background characteristics constant.

The Propensity to Remit We commence with an analysis of the qualitative variable

whether or not the household sends remittances, and of which type. In Table 7, we report

estimation results of linear probability models.15 Specifications A include characteristics of the

head of household and household characteristics, as well as an indicator variable as to whether

the head of household considers the migration as permanent or temporary. Specifications B add

information about the whereabouts of the spouse and the children and whether the individual

grew up in a rural area. All specifications condition on time dummies and country of origin

dummies.16

The estimates in the first two columns refer to total remittances. The probability of sending

remittances increases with disposable household income, which is compatible with the predictions

of our model in section 2 and with previous empirical findings17. The magnitude of this effect

is quite considerable: an increase in household income by 1 log point increases the probability

to remit by about 8 percentage points (spec. B). Conditional on household income, remittance

probabilities increase during the first years of residence, but decrease afterwards, with the turning

point at about 15 years.

Remittances also decrease with educational attainments of the household head. Notice that

this is conditional on household income. This is in line with Faini (2006) who analyzes aggregate

data and finds that remittances are lower for the highly skilled. Faini suggests as an explanation

that skilled immigrants have longer migration periods, and a higher probability of re-uniting

with their families. However, if we condition on household location decisions and whether the

migration is temporary (spec. B), the coefficient only slightly reduces in size, and it is still

significant. One explanation is that the educational background of the head of household may

pick up features of the recipients of remittances, like neediness, and size of family, which may only

be insufficiently reflected by the variables we include in our regressions. Households where the

15Marginal effects from probit models are almost identical.16Reported standard errors are clustered by individuals.17Lucas & Stark (1985), Hoddinott (1994) and Funkhouser (1995) also report a positive association between

remittance behavior and migrant’s income.

15

head is better educated may enjoy more favorable conditions in the home country, thus reducing

the need for remittances. The better educated may also be less affected by social pressure to

remit - one of the remittance motives we discuss above.

The next set of variables measure the temporary migration intention of the head of household,

and, in specifications B, whether the spouse or children are living abroad, and whether the

immigrant head grew up in a rural area. The coefficient on the temporary migration variable

is always large and strongly significant. Unconditional on the residential location of the family,

households with temporary migration plans have a 15 percentage point higher probability to

remit (remember that only 42 percent of households remit in our sample, so that this estimate

corresponds to a 36 percent difference); conditional on family location (specification B), the

remittance probability is still 11 percentage points higher.

Therefore, the sizeable difference in remittance propensities between permanent and tempo-

rary immigrant households which we illustrate in our descriptive section remains if we condition

in addition on household characteristics, and, in particular, on the residential location of the

closest family. One reason for the sizeable coefficient estimate may be that our family location

variables insufficiently capture the location of the wider family. A further reason could be that

the ”insurance motive” we discuss above matters for remittances. The estimates may also be

driven by omitted unobserved variables that affect remittances and temporary migration plans

alike. We will address this point below.

The coefficients on the family location decisions, reported in the last rows in specifications

B, suggest a sizable effect of whether spouse or children live abroad on remittance propensities.

Households where the spouse is living abroad have a 9 percentage point higher probability to

remit. The impact of children living abroad is even larger, with households having a 16 per-

centage point higher probability to remit if children live abroad. Thus the location of the close

family seems very important for remittance behavior. Finally, having grown up in a rural neigh-

borhood (”rural childhood”) increases the probability to remit by almost 5 percentage points,

conditional on other characteristics. As we discuss above this variable may reflect differences in

social conventions in the immigrants’ home communities.

The purpose of remittances The next three pairs of columns report results distinguishing

16

between the three different purposes of remittances that are reported in our survey: remittances

to support the family, to accumulate savings in the home country, and for other purposes. As

we report in Table 4, the largest fraction of remittances is channeled towards family support.

Conditional on the other regressors, remittances for family support increase in log household

income; the effect of income is still positive, but lower in magnitude for savings in the home

country, while remittances for ”other” purposes do not respond to changes in household income.

Again, this is compatible with our theoretical model, where both remittances and savings increase

with first period income. Educational attainment is marginally significant, and with a negative

sign, for family support and savings for later. This is in line with our interpretation that one

reason for lower remittances of households with higher educated heads may be that their families

are less in need of transfers.

The variable on temporary migration is strongly related to remittances sent for family sup-

port, with a smaller impact on savings in the home country and ”other” remittances, conditional

on the location of the immediate family. One reason may be that migrants with temporary mi-

gration plans have commitments towards family members other than the spouse and children,

compared with migrants with permanent intentions. This could be either because a larger frac-

tion of their extended family is still living abroad, their extended family is larger, or because

the temporary nature of their intended migration induces a larger response to expectations and

commitments to family and kinship, in order to build social respect for the period after return.

Not unsurprisingly, remittances for family support are strongly associated with the locational

choice of the immediate family, as suggested by the coefficients on the spouse and children

variables. On the other hand, having family members abroad slightly decreases remittances for

other purposes as well as savings in the home country.

The Amount of Remittances In Table 8 we show results using the logarithm of the

reported amounts of remittances as the dependent variable18. The specifications are the same

than those in Table 7, and we report only the coefficients on the temporary measure of migration,

and the location of immediate family and rural upbringing; the full set of results is reported in

the Appendix. The first panel reports results from simple OLS regressions. As the distribution

18We use the reported amount plus one unit to avoid problems with value zero observations.

17

of remittances is truncated at zero, we also report results from tobit regressions in the second

panel of the table.

Overall, the qualitative results are similar to those we discussed in the previous section. The

magnitude of the coefficient estimates are remarkable, however: total remittances are one log

point higher when the migration is intended to be temporary, and the coefficient drops only

slightly when we condition on the location of the family. Tobit estimates are, overall, of similar

magnitude.

As before, most of the difference between temporary and permanent households is due to

family support, as columns 2 suggest. However, savings in the home country and ”other” remit-

tances are also significantly larger for households with a temporary migration intention. While

the coefficient estimate decreases when we condition on family location for family remittances,

it increases for the other two purposes - which is again similar to the qualitative results we have

obtained above.

The Allocation of Savings The strong association between the temporary character of

migration, and remittances for ”savings” for later, could be due to temporary migrants delaying

consumption in the host country for consumption in the home country after a return, which is

predicted by our model. It may also suggest that those with temporary intentions choose to save

money in the home rather than the host country19.

To investigate this further, we use data for two years of our panel (1992 and 1994) where

we have information on the amount of savings in both locations. From this information, we

construct a measure for total savings, and for the ratio of savings in the home vs the host country.

If temporary migrants have a higher propensity to save because they have a higher preference

for consumption in the home- than in the host country, or they wish to benefit form purchasing

power differentials, we should observe that they save more than permanent immigrants overall. If

(in addition) temporary immigrants have a preference for shifting savings into the home country,

then the ratio of home to host country savings should be positively related to return plans.

Results are reported in the Appendix Table A5. In the first column, we use the logarithm

of the reported total amount of savings as the dependent variable. Columns 2 and 3 distinguish

19See Bauer & Sinning (2005) for an analysis of saving behaviour of migrants in the host country.

18

between savings in home and host country. Column 4 reports the ratio of savings between the

two locations. We report OLS results in the first panel and tobit results in the second panel. The

coefficient estimates on the temporary migration variable in columns 1 are marginally significant,

suggesting an increase in overall savings if migrations are intended as temporary. When we split

up savings in home and host country, it becomes apparent that an increase in home country

savings drives this result. This is supported by findings in the last column, which shows that the

ratio of allocation of savings in home vs host country is positively related to return plans. Thus,

the estimates seem to provide modest support for the hypothesis that temporary immigrants

save more than permanent immigrants. Moreover, it points at temporary immigrants having a

preference to holding their savings in the origin country rather than in the host country.

4.2 Controlling for individual effects

As we discuss in section 2.2, our estimation results for the association between the temporary

character of migration and remittances may partly reflect that those immigrants who are intend-

ing to return home are also more inclined to remit. Thus, this omitted variable problem may

lead to a bias in our estimates of the effect of temporary migration on remittances.20

To eliminate this problem, we utilize changes in immigrants’ return intentions across the

different waves of the panel and estimate fixed effects models.

As we discuss above, the return intention variable is likely to be measured with considerable

error, leading to a downward bias in coefficient estimates, which is enhanced when we estimate

the model conditional on fixed effects. To address this, we use the GMM type estimator we

explain in section 2.2. The numbers in Table A4 illustrate indeed some variation in migrant’s

return intentions from one year to the next. The figures in the table show that the percentage

of immigrants that change intention (from permanent to temporary and v.v.) in any given year

is between 14 and 21 percent, but decreasing over time. Furthermore, migrations increasingly

lend towards changes from temporary to permanent. In addition, those migrants who do not

change intentions are increasingly allocated in the ”permanent” category.

20See also McKenzie & Sasin (2007) who emphasize the importance of controlling for unobserved characteristics

when analyzing remittances and migration decisions.

19

In Table 9 we report estimation results both for the probability to remit (Panel A) and for

the amount of remittances (Panel B). Specifications are like specifications B in Tables 7 and 8.

In the different rows of the table, we distinguish between total remittances and remittances for

different purposes. We report as a benchmark results from OLS estimation as in Tables 7 and

8 (column 1), fixed effects (FE) estimation (column 2) and GMM estimation (column 3). We

concentrate our discussion on the variable that measures ”temporary” migration.21

Comparing the first two columns in Panel A shows that conditioning on fixed effects reduces

the temporary migration coefficient considerably. This could be due to unobserved factors that

affect remittance behavior as well as temporary migration intentions, but it could also be due

to measurement error in the intention variables. In column 3 we report GMM estimates. These

are considerably larger than the FE estimates, and close to the OLS estimates. They suggest a

14.1 percentage point higher probability of sending remittances for immigrants with temporary

migration plans, as compared to 11.5 percentage points obtained from OLS results, and 3.2

percentage points in the FE specification.22 These results indicate that measurement error leads

to a considerable downward bias in our FE results as well as in our OLS results; comparing

OLS estimates with GMM estimates suggests that the possible upward bias through unobserved

heterogeneity is matched by a downward bias through measurement error.

Distinguishing between remittance purposes, and concentrating on the GMM results, shows

that return plans increase the probability to remit to support the family by 10 percentage points,

slightly higher than the OLS coefficient, and larger than the coefficients for savings and other

purposes. Remittances for family support are stronger affected by temporary migration plans

than remittances for other purposes, or for savings.

In the lower panel of the table (Panel B), we assess the magnitude of these effects, using

the logarithm of the total amount of remittances as a regressor. Concentrating on total re-

mittances (first row), the coefficient estimate on temporary migration drops in the fixed effects

specification, but is still significant, and suggests that temporary migration intentions increase

21Reported FE and GMM results condition on head of household fixed effects. Conditioning on household

fixed effects leads to very similar results.22For the GMM results, an Arellano-Bond test rejects the presence of autocorrelation AR(2) to the differenced

residuals.

20

remittances overall by 27 percent. However, GMM estimates using as instruments previous

return intentions (column 3) are again close to the OLS estimates, showing that temporary mi-

gration intentions increase total remittances by 1.12 log points. Distinguishing between different

remittance purposes shows again that remittances for family support are most sensitive to return

plans.

Overall, these results suggest that remittance behavior is strongly related to temporary mi-

gration plans, even conditional on head of household fixed effects. As the fixed effect estimator

controls for household circumstances that are fixed over time and unobserved (like for instance

the degree of altruism), the effect of temporary migration plans on remittances seems to be

driven by other considerations. One of these may be the insurance motive we have discussed

above. The appropriate level of remittances to e.g. support the extended family or community

may be considered higher when a return is envisaged. Another reason may be a differential allo-

cation of resources across the two locations - as we have discussed above, temporary immigrants

tend to shift their savings and assets to the home country. In the next section we investigate

this by more detailed analysis of asset holdings.

4.3 Asset Holdings

As we discuss above, information on asset holdings is only available for one year (1988), and

corresponds to the stock of assets accumulated up to 1988. Distinction is drawn between asset

holdings in the home and in the host country. As the stock of assets has been accumulated over

previous years, we use the average return intention for years 1984-1988 as a regressor.

We show the results in Table 10 using similar specifications as the ones used previously for

remittances. We report the coefficient estimates for the average of return intentions over the

period 1984-1988. The upper panel shows results for property ownership and the lower panel for

asset holdings. We report in the first column results of simple LPM estimation on the binary

outcome whether any of these variables is larger than 0; in columns 2 and 3, we report results for

the total amounts from OLS and Tobit estimations; in the last two columns, we report results

for the ratio of home country vs. total asset or property value.

21

Concentrating first on property ownership, the results show that temporary immigrants are

more likely to hold property, and the value of property they hold is higher than that of those with

permanent intentions. Further, return plans have an opposite association with owning housing

property in the home and and the host country. Immigrants who want to return to the home

country are 8 percentage points less likely to own a house in the host country. On the other

hand, they are 20 percentage points more likely to own a house in the home country.

Although immigrant households with temporary intentions overall hold more property, the

total value of assets held does not differ between the groups, as estimates in the second panel

suggest. As before, however, households with temporary intentions are 13 percentage points

more likely to hold assets in the home country, after controlling for household income and

composition; as the results in the last two columns show, the ratio of home country held assets

to total assets is positively related to return plans.

5 Discussion and Conclusion

To obtain an idea of the magnitude of the relationship between remittance flows and permanent

versus temporary migration intentions, we provide some simple estimates based on the GMM

results in Table 9 (Panel B). Over the period we consider, the average yearly flow of remittances

sent home by the immigrants in our sample amounts to 1736 Euros per household, or 504 Euros

per individual23. This corresponds to an aggregate of more than 2 billion Euros (equivalent to

0.12% of the German GDP) in 1995, for the population of immigrants that are represented in

our sample24.

Now consider an increase in permanent intentions of migrations of 30 percentage points25.

This change is equivalent to a drop in remittances sent of 40.1 percent of the total amount

23We obtain this number by dividing the average remittances per household by the average household size for

our sample during the years 1984-1994. This amount is in line with official aggregate statistics. Total remittance

flows in 1995 were 4120 million Euros (in 2002 prices) according to Bundesbank (2006), which corresponds to

574 Euros per immigrant, based on the total immigrant population.24Immigrants from Turkey, Ex-Yugoslavia, Greece, Italy and Spain accounted for 60 percent of the total

immigrant population in Germany in 1995 (OECD (2006)).25This corresponds roughly to the change in our sample over the observation period

22

remitted. This drop corresponds to around 813 million Euros, or around 0.048% of the German

GDP in 1994.

The drop in remittances is even more important for receiving countries. To put this number

into perspective, consider Turkey. Remittances are an important source of income for Turkey. In

1994, remittance flows corresponded to 2.1% of the Turkish GDP, much higher than foreign direct

investment (0.51%) or aid (0.18%) 26. An increase in permanent intentions to stay in Germany

of Turkish immigrants by 30 percentage points corresponds to a decrease in remittance flows of

371 million Euros, using our GMM estimates in Table 9 (Panel B). This is equivalent to around

0.76% of Turkish GDP in 1994, an amount larger than the foreign direct investment and foreign

aid received by Turkey in 1994 combined. Although these are rough calculations, they highlight

the magnitude of the effects of temporary vs permanent migration on remittance behaviour.

The particular form of migration has also implications in terms of home ownership and asset

accumulation. Using the estimates based on the results in column 3 in Table 10, an increase in

ten percentage points in the permanency of migration is equivalent to a 3 percent higher value

of property wealth held in the host country for each migrant household. More importantly, it

corresponds to a reduction in home ownership in the home country by 30 percent. It further

decreases the overall financial wealth held in the home country by around 21 percent.

Our results have important implications for migration policies, and emphasize the importance

of the particular form of migration for immigrant behavior. Our findings suggest that migration

policies that encourage temporary migration are likely to lead to considerably higher remittance

flows than migration policies that encourage permanent settlement. In addition, these migration

policies are also likely to affect the location of migrants’ savings and asset holdings.

Thus, our analysis suggests that remittances need to be discussed in conjunction with the

particular form of migration. Policies that are likely to have an impact on the permanency of

migrations will also affect remittance flows, and possibly dramatically so.

26OECD (2006), WorldBank (2006)

23

References

Acosta, P., Calderon, C. A., Fajnzylber, P. R. & Lopez, H. (2006), ‘Remittances and development

in latin america’, The World Economy 29(7), 957–987.

Adams, R. H. J. (2006), ‘International remittances and the household: Analysis and review of

global evidence’, J Afr Econ 15(2), 396–425.

Adams, R. H. J. & Page, J. (2005), ‘Do international migration and remittances reduce poverty

in developing countries?’, World Development 33(10), 1645–1669.

Agarwal, R. & Horowitz, A. W. (2002), ‘Are international remittances altruism or insurance?

evidence from guyana using multiple-migrant households’, World Development 30(11), 2033–

2044.

Amuedo-Dorantes, C. & Pozo, S. (2006), ‘Remittances as insurance: evidence from mexican

immigrants’, Journal of Population Economics 19(2), 227–254.

Arellano, M. & Bond, S. (1991), ‘Some tests of specification for panel data: Monte carlo evidence

and an application to employment equations’, The Review of Economic Studies 58(2), 277–

297.

Arellano, M. & Bover, O. (1995), ‘Another look at the instrumental variable estimation of error-

components models’, Journal of Econometrics 68(1), 29–51.

Azam, J.-P. & Gubert, F. (2006), ‘Migrants’ Remittances and the Household in Africa: A Review

of Evidence’, J Afr Econ 15(2), 426–462.

Bauer, T. K. & Sinning, M. (2005), The savings behavior of temporary and permanent migrants

in germany, IZA Discussion Papers 1632, Institute for the Study of Labor (IZA).

Bohning, W. (1987), Studies in International Migration, St. Martin’s Press, New York.

Bundesbank, D. (2006), ‘Time series eu3212:transfers by foreign workers to non-residents’.

24

Cox, D. (1987), ‘Motives for private income transfers’, Journal of Political Economy 95(3), 508–

46.

Cox, D., Eser, Z. & Jimenez, E. (1998), ‘Motives for private transfers over the life cycle: An

analytical framework and evidence for peru’, Journal of Development Economics 55(1), 57–80.

de la Briere, B., Sadoulet, E., de Janvry, A. & Lambert, S. (2002), ‘The roles of destination,

gender, and household composition in explaining remittances: an analysis for the dominican

sierra’, Journal of Development Economics 68(2), 309–328.

Durand, J., Kandel, W., Parrado, E. A. & Massey, D. S. (1996), ‘International migration and

development in mexican communities’, Demography 33(2), 249–264.

Dustmann, C. (1997), ‘Return migration, uncertainty and precautionary savings’, Journal of

Development Economics 52(2), 295–316.

Faini, R. (2006), ‘Remittances and the brain drain’, IZA Discussion Paper (2155).

Funkhouser, E. (1995), ‘Remittances from international migration: A comparison of el salvador

and nicaragua’, The Review of Economics and Statistics 77(1), 137–46.

Hoddinott, J. (1994), ‘A model of migration and remittances applied to western kenya’, Oxford

Economic Papers 46(3), 459–76.

Hsiao, C. (1986), Analysis of panel data, Cambridge University Press.

International Migration Outlook (2006).

Laferrere, A. & Wolff, F.-C. (2000), Microeconomic models of family transfers, Vol. Handbook

on the Economics on Giving, Reciprocity and Altruism, MacMillan and St Martin Press.

Lucas, R. E. B. & Stark, O. (1985), ‘Motivations to remit: Evidence from botswana’, Journal

of Political Economy 93(5), 901–18.

Lucas, R. E. B. & Stark, O. (1988), ‘Migration, remittances, and the family’, Economic Devel-

opment and Cultural Change 36(3), 465–81.

25

McKenzie, D. & Sasin, M. (2007), ‘Migration, remittances, poverty and human capital. concep-

tual and empirical challenges’, World Bank Policy Research Working Paper 4272.

OECD (2006), ‘International migration data’.

Okonkwo Osili, U. (2007), ‘Remittances and savings from international migration: Theory and

evidence using a matched sample’, Journal of Development Economics 83(2), 446–465.

Poirine, B. (1997), ‘A theory of remittances as an implicit family loan arrangement’, World

Development 25(4), 589–611.

Rapoport, H. & Docquier, F. (2005), ‘The economics of migrants remittances’.

Ratha, D. (2003), Workers remittances: An important and stable source of external development

finance, Global development finance, The World Bank.

Roodman, D. (2006), ‘An introduction to difference and system gmm in stata’, Center for Global

Development Working Paper Series .

WorldBank (2006), ‘World development indicators’.

World Economic Outlook (2005), Technical report, International Monatery Fund (IMF).

26

Appendix

Data Construction Appendix

We use 12 waves of the German Socio-Economic Panel data corresponding to the years 1984-1995.

Our sample consists of migrant households whose head was born in Turkey, Greece, Yugoslavia,

Italy or Spain. All our income variables are reported in real terms (deflated to Euros in year

2002) and at household level. Household income corresponds to the net monthly income of the

household transformed to year level. The exact wording of the question is ”If everything is taken

together: how high is the total monthly income of all the household members at present? Please

give the monthly net amount, the amount after the deduction of tax and national insurance con-

tributions. Regular payments such as rent subsidy, child benefit, government grants, subsistence

allowances, etc., should be included. If not known exactly, please estimate the monthly amount.”

Individuals declared the amount of remittances sent the previous year during 1984-1995 (ex-

cept for the years 1992 and 1994). The wording of the question is ”(Last year) did you personally

send or take money to your homeland?”. In case of an affirmative answer, individuals are asked

for the overall amount and the purpose: ”And how is this amount distributed between support

for your family, savings for later and other”. ”Savings” correspond to the amount of savings in

the home country. ”Other” corresponds mainly to transfers to people other than family (friends,

neighbors, etc.). We aggregate these amounts to the household level and lag them for one year

to match them time-wise with the rest of observed variables.

Information on household savings in the host country is available from 1992 and corresponds

to the net monthly savings of the household transformed to year level. The question explicitly

asks the head of household ”Do you usually have an amount of money left over each month for

major purchases, emergencies, or savings? If yes, how much?”. This implies that information

on the amount of savings in both the home and the host country is available for two years (1992

and 1994).

Asset information is drawn from questions in special survey in year 1988. Asset holdings

27

refer to the total amount of asset holdings (including cash, savings, property, etc.) but net of

financial obligations. The wording of the question is ”If you could add up all the wealth of this

household (including cash, goods and property you own but without furniture), what will be the

approximate total value of it? Please make sure to subtract all the mortgages, loans and credits

that you could have on them”.

Property includes the houses, apartments or any other property at market prices. For each

type of property, the wording of the question is ”Are you the owner of (specific type of property)?

If yes, how much do you estimate its commercial value is, that is, how much money will you get

if you sold it now?”.

Investment in the country of origin include the purchase of land, property, furniture, ma-

chinery, investment in family or own company and other during their migration in Germany

until 1988. The wording of the question regarding investment in the country of origin is ”Did

you make one of the following major investments since you have been living in Germany?”. All

entries correspond to the aggregated household amounts declared in the year 1988, in Euros,

deflated to the base year 2002.

See the appendix table A1 for a summary of data availability.

28

Econometric Specification Appendix

Return plans are likely to be assessed with measurement error. To see how this affects our

estimate of ξ, define the observed return plan as RIit = RI∗it + MEit, where RI

∗it is the true

measure of return intention and MEit is the measurement error. Specification 4 can be rewritten

as:

Yit = a0 + a1Xit + a2Zit + ξRIit + �i + ηit − ξMEit (5)

where vit = �i + ηit − ξMEitThe OLS estimator of ξ as N →∞ converges to:

plimN→∞ ξ̂OLS = ξ +Cov(RIit, �i)

V ar(RIit)− ξV ar(MEit)

V ar(RIit)(6)

This estimator is inconsistent because it does not control for the fact that those with a higher

unobserved propensity to remit also want to return back home; it does also not control for

possible measurement error in the observed intention. Using a fixed effects transformation the

unobserved �i can be eliminated27:

Yit − Ȳi = a1(Xit − X̄i) + a2(Zit − Z̄i) + ξ(RIit − R̄I i) + (ηit − η̄i)− ξ(MEit − M̄Ei) (7)

However, measurement error makes the FE estimate of ξ inconsistent:

plimN→∞ ξ̂FE = ξ[1−T − 1T

V ar(MEit)

V ar(RIit − R̄I i)] < ξ (8)

In order to get consistent estimates of ξ, we need to find a estimation technique that first

eliminates the fixed effect �i and then corrects for the measurement error.

Our strategy will be the use of an entire set of instruments in a GMM procedure as suggested

by Arellano & Bond (1991). This procedure is designed for ”small T, large N” panels with

individual fixed effects that are heteroskedastic and correlated within individuals. It instruments

the endogenous variable using past values of it in levels. We will discuss below in further detail

the choice of instruments.

We eliminate the fixed effect �i by transforming our data using a orthogonal deviations

transformation. This transformation is an alternative to first differences transformation, but it

27See Hsiao (1986) (p.63).

29

is preferable if the data has several gaps in order to minimize data loss28.It subtracts to each

observation t = 1, ..., T −1 the mean of the remaining future observations in the sample to purge

the individual unobserved effect.

If we define XOit as the orthogonal deviation of Xit29, specification 5 is transformed to

Y Oit = a1XOit + a2Z

Oit + ξ(RI)

Oit + v

Oit (9)

where vOit = ηOit − ξMEOit , thus eliminating the fixed effect �i. We use as instrumental variables

past values of intentions in levels. If the real return intention RI∗it is uncorrelated with the mea-

surement error MEit, and the measurement errors are serially uncorrelated(E[MEit,MEis] =

0, t 6= s), past values of return intentions RIit−1, ..., RIi1 are appropiate instrumental variables

as E(MEit, RIit−s) = 0, s > 0.

More precisely, the matrix of instruments used is

Zi =

Zi1 0 ... 0

0 Zi2 ... 0

...

0 0 ... ZiT−1

where Zit = (RIit−1, ...,RIi1). Note that for each time period t, the number of instruments used

is different.

The GMM estimator ˆξGMM will provide us with unbiased and consistent estimates of ξ. In

order to test for the exogeneity of the instruments used, we compute the Arellano-Bond test for

autocorrelation, which is an AR(2) test applied to the differenced residuals.

28See Roodman (2006).29Formally, the orthogonal deviations transformation of Xit is defined as XOit =

√T − t/(T − t+ 1)(Xit −

1T−t

∑(Xis)); s > t where

√T − t/(T − t+ 1) are weights to standarize the variances. See Arellano & Bover

(1995) for a detailed explanation.

30

Parameters dp d γ dβ d δ dω1 dωI2 dωE2

d s1 + / - - + - + - -

d R + / - + - + + + +

Sex Age Age At Arrival Years Since Migration Number Years Education Household IncomeNumber Children in HouseholdNumber Adults in HouseholdNumber Employed Individuals in HouseholdEmployed (Yes=1, No=0) Non Single (Yes=1, No=0)Native Partner (Yes=1, No=0)Spouse Abroad (Yes=1, No=0) Children Abroad (Yes=1, No=0) Rural Childhood (Yes=1, No=0) Temporary (Yes=1, No=0)

Couple With Children Couple Without Children 1-Pers.-HH Single Parent Multiple Generation-HH Other Combination

Total

Household Composition

0.817

0.33311.3018.9826.068

0.9120.0620.0540.109

4.943.921.61

0.3920.568

Note: Calculations based on GSOEP data, 1984-1990, 1992, 1994. Individual information corresponds to the head of household. Household Income in 2002 Euros.

Per cent

61.9814.87

Table 3: Household Types

43.99324.56719.3709.419

Table 1 : Comptarative Statistics

Mean0.873

Std. Dev.

Table 2 : Summary Statistics - 1984-1994

23,496.931.0202.4101.570

1.95413,006.51

1.1791.0790.918

Proportion

95.790.5969.97

Male Head

Note: Calculations based on GSOEP data, 1984-1990, 1992, 1994. The first column shows the percentage of the particular household type, the second column the percentage of male heads of household in that household type.

100

17.4490.6574.07

12.67

Total Remittances 41.37% 1,644.74 7.50% 3,965.48 17.98%

Family Remittances 28.43% 878.03 4.28% 2,116.35 10.25%Savings Remittances 7.45% 371.59 1.59% 895.82 3.82%Other Remittances 10.49% 395.12 1.63% 953.31 3.91%

Family Savings Other

Percentage Amount Percentage Percentage PercentagePermanent 29.38% 974.68 19.83% 4.03% 7.38%Temporary 49.71% 2,109.75 34.40% 9.84% 12.66%

No Spouse Abroad 42.69% 1,657.04 28.61% 7.90% 11.37%

Spouse Abroad 67.45% 3,440.60 60% 6.60% 5.11%

No Children Abroad 41.25% 1,571.04 27.21% 7.88% 11.23Children Abroad 68.69% 3,271.84 57.53% 9.17% 9.67%

All 41.37% 1,644.74 28.43% 7.45% 10.49%

Note: Calculations based on GSOEP data (1984-1990,1992,1994), on household level. Average amount over the total sample (in 2002 Euros). Information on temporary, spouse and children abroad corresponds to the head of household. "No Spouse Abroad" includes single heads of household. "No children abroad" includes heads of household with children in the host country and without children.

Table 4: Purpose of Remittances

Table 5: Remittances by Household Characteristics

Total

Amount if Household Remits

Total As Percentage of HH Disposable Income

Note: Calculations based on GSOEP data (1984-1990,1992,1994), on household level. Columns 2 and 3 report the average amount for all households, and columns 4 and 5 for those households that remit a positive amount (in 2002 Euros).

Amount

Percentage Total As Percentage of HH Disposable Income

Households Remitting

Permanent Immigrants Mean Std.Dev. Mean Std.Dev. Mean Std.Dev. Mean Std.Dev.

Proportion 8.6% 13.2% 5.5% 43.8%Average Value Property 13,353 53,854 22,677 69,813 6,824 37,687 67,193 107,669

Proportion 38.5% 24.5% 48.3%Average Value Property 25,561 55,316 12,621 33,744 34,622 64,880

Proportion 73.1% 71.8% 74.1% 79.8%Average Value 31,912 93,987 47,114 117,511 20,613 69,720 104,636 161,861

Proportion 69.2% 55.8% 76.0%Average Value 46,549 73,317 31,584 50,719 54,102 81,418

In Host Country

In Home Country

Asset Holdings

Note: Calculations based on GSOEP data, 1988, on household level. Average amount over the total sample (in 2002 Euros). Property Ownership includes house, apartment or any other property. Asset holdings refer to the total amount of asset holdings net of financial obligations, including cash, savings, property, etc. Investment includes the purchase of land, property, furniture, machinery, investment in family or own company and other in the country of origin during migration history.

In Host Country

In Home Country

Table 6: Home Ownership and Assets

Total Immigrants Natives

Home Ownership

Temporary Immigrants

Specification A Specification B Specification A Specification B Specification A Specification B Specification A Specification BAge/10 0.028 0.005 0.026 0.006 0.005 0.006 -0.005 -0.006

(0.010)** (0.012) (0.009)** (0.011) (0.005) (0.005) (0.005) (0.006)Years Since Migration/100 0.118 0.143 0.084 0.077 0.016 0.008 -0.005 0.023

(0.054)* (0.064)* (0.050) (0.063) (0.024) (0.030) (0.029) (0.035)YSM-Squared/100 -0.043 -0.040 -0.031 -0.024 -0.007 -0.003 -0.001 -0.005

(0.014)** (0.016)** (0.013)* (0.015) (0.006) (0.007) (0.007) (0.009)Log HH Income 0.087 0.078 0.061 0.052 0.025 0.031 0.019 0.018

(0.018)** (0.019)** (0.013)** (0.014)** (0.007)** (0.009)** (0.010) (0.012)Number Adults HH Host -0.036 -0.025 -0.041 -0.024 -0.009 -0.011 0.015 0.013

(0.011)** (0.011)* (0.010)** (0.010)* (0.005) (0.006) (0.007)* (0.007)Number Children HH Host -0.014 -0.012 -0.017 -0.008 -0.002 -0.003 0.002 -0.000

(0.007) (0.008) (0.007)* (0.008) (0.004) (0.004) (0.004) (0.005)Employment Head HH 0.170 0.168 0.146 0.138 0.019 0.025 0.003 0.005

(0.022)** (0.024)** (0.021)** (0.023)** (0.009)* (0.010)* (0.012) (0.014)Number Employed HH 0.022 0.023 0.004 0.008 0.019 0.016 0.024 0.025

(0.012) (0.012) (0.011) (0.011) (0.006)** (0.006)** (0.008)** (0.008)**Number Years Education -0.011 -0.010 -0.008 -0.006 -0.003 -0.004 0.002 0.001

(0.004)* (0.005)* (0.004) (0.005) (0.002) (0.002) (0.002) (0.003)Non Single Head HH 0.078 0.036 0.069 0.012 0.024 0.025 0.014 0.016

(0.031)* (0.037) (0.029)* (0.033) (0.010)* (0.013)* (0.013) (0.016)Native Partner -0.160 -0.154 -0.091 -0.085 -0.053 -0.053 -0.061 -0.056

(0.029)** (0.034)** (0.024)** (0.029)** (0.009)** (0.010)** (0.015)** (0.018)**Temporary 0.148 0.115 0.109 0.070 0.039 0.042 0.046 0.054

(0.015)** (0.016)** (0.014)** (0.015)** (0.007)** (0.007)** (0.008)** (0.009)**Spouse Abroad 0.092 0.161 -0.039 -0.038

(0.035)** (0.037)** (0.017)* (0.018)*Children Abroad 0.165 0.208 0.003 -0.016

(0.023)** (0.023)** (0.013) (0.013)Rural Childhood 0.046 0.030 0.009 0.010

(0.019)* (0.018) (0.009) (0.011)Observations 8919 7709 8919 7709 8919 7709 8919 7709

R-squared 0.14 0.15 0.10 0.14 0.03 0.03 0.03 0.03* significant at 5%; ** significant at 1%

Note: Dependent variable: Household sent remittances (=1 Yes, =0 No). GSOEP data (1984-1990,1992,1994). All specifications include time and country dummies. Household level. Standard errors are clustered by head of household.

Table 7: Probability to Remit

Linear Probability ModelTotal Family Support Savings for Later Other

Specification A Specification B Specification A Specification B Specification A Specification B Specification A Specification B

Temporary 1.242 0.995 0.879 0.575 0.308 0.336 0.355 0.423(0.119)** (0.125)** (0.111)** (0.113)** (0.055)** (0.057)** (0.067)** (0.070)**

Spouse Abroad 0.881 1.446 -0.320 -0.309(0.299)** (0.302)** (0.141)* (0.140)*

Children Abroad 1.461 1.806 0.009 -0.129(0.190)** (0.189)** (0.104) (0.102)

Rural Childhood 0.421 0.270 0.076 0.088(0.153)** (0.143) (0.072) (0.086)

Observations 8 919 7 709 8 919 7 709 8919 7709 8919 7709R-squared 0.15 0.17 0.11 0.15 0.03 0.03 0.03 0.04

Specification A Specification B Specification A Specification B Specification A Specification B Specification A Specification B

Temporary 1.298** 1.072** 0.852** 0.583** 0.273** 0.299** 0.326** 0.379**(0.135) (0.149) (0.116) (0.125) (0.052) (0.059) (0.064) (0.070)

Spouse Abroad 0.847** 1.229** -0.209 -0.335(0.277) (0.223) (0.135) (0.186)

Children Abroad 1.470** 1.659** 0.019 -0.116(0.179) (0.142) (0.076) (0.100)

Rural Childhood 0.433** 0.238 0.078 0.072(0.164) (0.139) (0.057) (0.074)

Observations 8919 7709 8919 7709 8919 7709 8919 7709Pseudo R-sq 0.045 0.047 0.044 0.053 0.035 0.034 0.025 0.026

* significant at 5%; ** significant at 1%

Note: Dependent variable: Logarithm of amount remitted. GSOEP data (1984-1990,1992,1994). Household level. All specifications include time and country dummies and condition on age, years since migration (and its square), education, marital status, household income, employment status and number of adults and children in the host country household. Standard errors are clustered by head of household. Tobit results show unconditional marginal effects.

TobitTotal Family Support Savings for Later Other

Table 8: Amount Remitted

Total Family Support Savings for Later OtherOLS

Panel A:Dependent Variable: Probability to Remit OLS FE GMM

Household Sent Remittances (=1 Yes, =0 No) 0.115 0.032 0.141 1

(0.016)** (0.015)* (0.028)**

Family Support (=1 Yes, =0 No) 0.070 0.020 0.100 2

(0.015)** (0.014) (0.023)**

Savings for Later (=1 Yes, =0 No) 0.042 -0.001 0.035 3

(0.007)** (0.009) (0.015)*

Other (=1 Yes, =0 No) 0.054 0.028 0.056 4

(0.009)** (0.011)** (0.018)**

Panel B:Dependent Variable: Logarithm of Amount Remitted

Total Amount Remittances 0.995 0.244 1.124 5

(0.125)** (0.121)* (0.221)**

Amount Remittances Family Support 0.575 0.150 0.742 6

(0.113)** (0.107) (0.172)**

Amount Remittances Savings for Later 0.336 -0.022 0.27 7

(0.057)** (0.075) (0.124)*

Amount Remittances Other Purposes 0.423 0.203 0.422 8

(0.070)** (0.086)* (0.142)**

Observations 7 709 7 984 5239Number of Never Changing Person ID 1 411 1152

* significant at 5%; ** significant at 1%

Arellano Bond AR(2) Test :(1) z=0.325 P-Value=0.746 (2) z=0.555 P-Value=0.579 (3) z=-2.48 P-Value=0.013 (4) z=-1.351 P-Value=0.177(5) z=0.364 P-Value=0.716 (6) z=0.108 P-Value=0.914 (7) z=-2.711 P-Value=0.007 (8) z=-1.521 P-Value=0.128

Note: GSOEP data (1984-1990,1992,1994). Household level. All specifications include time and country dummies and condition on age, years since migration (and its square), education, marital status, household income, employment status and number of adults and children in the host country household. Standard errors are clustered by head of household. Instrumental variables used: Lags in intentions to return (t-1, …, 1). Reported coefficents correspond to the coefficient on the temporary intention variable.

Table 9: Probability to Remit and Amount Remitted - Fixed Effects and GMM

(=1 Yes, =0 No)

Dependent Variable:Property OwnershipTotal (Home+Host Country)

Host Country

Home Country

Ratio Home vs Total Property

Dependent Variable:Asset Holdings Total (Home+Host Country)

Host Country

Home Country

Ratio Home vs Total Assets

* significant at 5%; ** significant at 1%

(0.061)0.196***

0.255***

OLSLPM OLS Tobit

1.750***1.594***0.141***

(0.795)

0.09(0.627)-0.134

(0.641)

(0.063)

-0.296***(0.380) (0.114)2.435***

(0.623)1.601** 1.913**

2.645***(0.521) (0.641)

0.11

(0.638)

(0.039)0.016

(0.054)0.129**(0.061)

-0.202(0.594)

(0.529)

Tobit

-0.01

(0.053)-0.078**(0.034)0.203***

(0.563)-0.928**

(0.049)

(0.055)

Note: GSOEP data (1988). Household level. All specifications include time and country dummies and condition on age, years since migration (and its square), education, marital status, household income, employment status and number of adults and children in the host country household. Property ownership includes the purchase of house, apartment or any other property, in the host and in the home country. Asset holdings refer to the total amount of asset holdings net of financial obligations, including cash, savings, property, etc., in the host and in the home country. Investment in the country of origin includes the purchase of land, property, furniture, machinery, investment in family or own company and other during migration to Germany. Standard errors are clustered by head of household. Tobit results show unconditional marginal effects. Reported coefficents correspond to the average intention to return up to 1988 (1984-1988).

Table 10: Asset Holdings

0.223***(0.064)

0.285***

Logarithm of Amount Remitted

Logarithm of Amount Remitted

Ratio of Amounts

Ratio of Amounts

Return Intention Intention to Return to the Home Country

Asset Holdings Home Net Amount of Assets in the Home CountryAsset Holdings Host Net Amount of Assets in the Host CountryProperty Home Property Ownership in the Home CountryProperty Host Property Ownership in the Host Country

Table A1: GSOEP Data AvailabilityVariable Name Description Availability

Total Remittances Total Amount sent to Home Country 1984-1990,1992,1994Family Remittances Amount sent to Support the Family 1984-1990,1992,1994Savings at Home Country Amount Saved in the Home Country 1984-1990,1992,1994Other Remittances Amount sent to Others 1984-1990,1992,1994Savings at Host Country Amount Saved in the Host Country 1992-2003

1984-2003Spouse Abroad Spouse in the Home Country 1984-1997Children Abroad Under Aged Children in the Home Country 1984-1997Number of Adults Number Adults Host Country Household 1984-2003Number of Children Number Children Host Country Household 1984-2003Household Income Annual Net Household Income 1984-2003

1988

Note: German Socio Economic Panel (GSOEP) data.

198819881988

Specification A Specification B Specification A Specification B Specification A Specification B Specification A Specification B

Age/10 0.250 0.075 0.234 0.080 0.039 0.045 -0.038 -0.042(0.080)** (0.093) (0.072)** (0.085) (0.041) (0.045) (0.039) (0.049)

Years Since Migration/10 0.910 1087 0.680 0.604 0.116 0.046 -0.059 0.147(0.427)* (0.510)* (0.381) (0.477) (0.196) (0.247) (0.224) (0.267)

YSM-Squared/100 -0.341 -0.312 -0.256 -0.193 -0.053 -0.018 -0.003 -0.029(0.108)** (0.123)* (0.098)** (0.117) (0.051) (0.061) (0.056) (0.066)

Log HH Income 0.790 0.746 0.537 0.477 0.223 0.274 0.166 0.157(0.157)** (0.172)** (0.110)** (0.116)** (0.062)** (0.074)** (0.086) (0.104)

Number Adults HH Host -0.356 -0.249 -0.387 -0.231 -0.074 -0.093 0.110 0.096(0.087)** (0.092)** (0.079)** (0.080)** (0.041) (0.047)* (0.053)* (0.060)

Number Children HH Host -0.148 -0.125 -0.179 -0.098 -0.018 -0.031 0.028 0.007(0.060)* (0.066) (0.058)** (0.061) (0.030) (0.035) (0.034) (0.040)

Employment Head HH 1 378 1 362 1 156 1 092 0.152 0.195 0.032 0.051(0.170)** (0.187)** (0.157)** (0.171)** (0.075)* (0.082)* (0.094) (0.105)

Number Employed HH 0.213 0.210 0.041 0.061 0.147 0.126 0.192 0.195(0.097)* (0.101)* (0.084) (0.086) (0.047)** (0.050)* (0.061)** (0.066)**

Number Years Education -0.084 -0.068 -0.061 -0.036 -0.020 -0.030 0.016 0.009(0.036)* (0.038) (0.035) (0.036) (0.016)