International Journal of Economics, Commerce and Management

United Kingdom Vol. II, Issue 9, Sep 2014

http://ijecm.co.uk/ ISSN 2348 0386

MICROECONOMIC DETERMINANTS OF MIGRANT

REMITTANCES INTO ZIMBABWE: A SURVEY OF
ZIMBABWEANS IN BOTSWANA AND SOUTH AFRICA

Mugumisi, Nathan
Department of Accounting and Finance, Faculty of Commerce
Lupane State University, Ascot Bulawayo, Zimbabwe
mugumisin@gmail.com

Abstract
This study sought to identify the micro-economic determinants of remittances flows into
Zimbabwe. The paper reviews the altruism theory, self interest, portfolio management theory,
and family implicit contracts theory. The study was based on a survey of 294 Zimbabweans,
based in South Africa and Botswana. Data for the study was collected through questionnaires
using the snow balling technique. SPSS version 16.0 was used for data analysis, using the
binary logit model. The results show that the older (age) and more educated migrants are more
likely to remit to home country. Single migrants were less likely to remit compared to those who
are divorced. We also found that married migrants are less likely to remit than divorced
migrants. The number of members from the same family and the gender of migrants were found
to be insignificant in predicting the likelihood of remitting money back to Zimbabwe.

Key words: remittance, determinants, microeconomic, logistic, Zimbabwe

INTRODUCTION
Migrant remittances have emerged as one of the major sources of foreign exchange flows into
developing countries in recent times. According to the World Bank (2012), officially recorded
remittances to developing countries totaled $381 billion in 2011, and are estimated to have
reached $406 billion in 2012, and are projected to reach $534 billion in 2015. Harnessing
migrant remittances has therefore become a major source of development financing for
remittance sending countries. Zimbabwe‟s decade long economic melt-down from 1999 to 2009,
and political crisis which surrounded the 2008 general elections, resulted in massive emigration
of both professional and non-professional Zimbabweans to neighbouring countries and abroad.

Licensed under Creative Common Page 1

Electroniccopy
Electronic copy available
available at:
at:https://ssrn.com/abstract=2671245
http://ssrn.com/abstract=2671245
 © Mugumisi

Since the turn of the century, Zimbabwe has benefited from migrant remittances from its migrant
stock estimated to have reached 4 million by 2008 (Orozco and Ferro 2008). Studies have been
done on micro and macro-economic impact of remittances, channels of remitting into Zimbabwe
(see Bracking and Sachikonye (2006), Makina (2007), Ncube and Gomez (2011), Burgsdorff
(2012), Mugumisi and Ndhlovu 2013)).
Despite the increased interest in workers‟ remittance since the turn of the century,
relatively little or work has been done to improve the understanding of determinants of
remittances flows into Zimbabwe. The question “what determines or drives remittances” is one
of the most important questions in remittance literature (Havolli 2009). There is no systematic
theory that explains remittance behaviour and almost no statistical evidence for the case of
Zimbabwe. This paper provides empirical literature of main determinants of migrant remittances
in Zimbabwe.
Remittances are decided out of the complexity of family arrangements in the migration
process. Traditionally, the economic literature on migrant remittances has distinguished two
types of theoretical models, namely the individual models and the "family" models (Bouoiyour
and Miftah2014). Individual models include the pure altruism and self interest motives, while
family models include the implicit family contracts of coinsurance and loan repayment.
According to Kelly and Solomon (2009), remittance literature is also divided into microeconomic
and macroeconomic determinants of remittance. Scholars like Russell (1986) categorised
determinants of migrant remittances into (i) socio-demographic characteristics of migrants and
their families and (ii) macroeconomic and political conditions. Social and demographic
characteristics like marital status, age, gender, education levels and household income are
categorised as micro economic determinants. Macroeconomic determinants comprise factors
like wages in host countries, inflation and exchange rates, in both home and destination
countries. It is imperative to note that aggregate remittance flows reflect the underlying
microeconomic considerations which determine the individual decisions about remittances (El –
Sakka and Mcnabb 1999). In the long run macroeconomic factors mainly affect the remittance
channel used. This paper focuses on the microeconomic determinants of migrant remittances.
This study aims to provide policy-relevant information on remittances. Providing insight
into the characteristics of remitters and to their motives for remitting is instrumental in creating a
more favorable financial environment for remittances, i.e. an environment better aligned with the
motives and needs of remitters.

Licensed under Creative Common Page 2

Electroniccopy
Electronic copy available
available at:
at:https://ssrn.com/abstract=2671245
http://ssrn.com/abstract=2671245
 International Journal of Economics, Commerce and Management, United Kingdom

LITERATURE REVIEW
The factors which affect frequency and quantity of remittance can be inferred from the
motivations of remitting. In their seminal work of 1985 Lucas and Stark identified three
explanations or motivations why migrants remit part of their incomes to their families at home,
namely for pure altruist, self-interest and to honour implicit family. The motives of remittances
largely take a microeconomic perspective, and focus mostly on the roles that families and intra-
family relationships have in decisions on migration and remittances.

Altruism
In this model the migrant derives satisfaction from the welfare of his or her family in country of
origin. In the altruistic motive transfers are motivated by the migrant‟s unselfish concern with the
welfare of family members and loved ones. Migrants‟ delight in family welfare makes
remittances a rather stable source of income for the migrants‟ families (Bouhga-Hagbe 2006:6).
This implies a utility function in which the migrant cares about the consumption of the other
members of his/her household. Altruistic migrants will remit more the poorer their family in
country of origin. The model is premised on a number of assumptions; Funkhouser (1995) in his
behavioural model of remittances suggested five testable implications under the altruistic ; (i)
the amount of remittances should increase with migrant‟s earnings(see Lucas and Stark, 1985,
Vanwey, 2004) (ii) the amount of remittances are expected to decrease as domestic family
income increases, or low income household receive more (iii) The duration of stay is expected
to have a negative impact on remittances; it is assumed that attachment to family weakens
gradually.(iv) remittances by a given migrant should decrease with the number of emigrants
from the same household. Moreso, family unifications also reduce remittances as there are less
people left behind to look after. (v) Remittances depend on the migrant‟s marital status; they
tend to increase if the migrant is married and his spouse and /or children are in the home
country. Holst and Schrooten (2000) found that remittances are positively are positively related
with income and negatively with the number of dependents in host country.

Self interest
In the pure self interest model, the migrant chooses an individual strategy of transfers. Migrants
may send remittances to native country to accumulate physical and/or financial assets back
home. A migrant with aspirations to inherit parents‟ estate may remit, especially if it is assumed
that bequests are conditioned by behaviour (SOPEMI 2006). According to Havolli (2009),
remitters are expected to have a higher chance to inherit assets, and the higher the value of
assets to be inherited the higher the remittances are likely to be. Garip (2006) found that male

Licensed under Creative Common Page 3

Electronic copy available at: https://ssrn.com/abstract=2671245
 © Mugumisi

emigrants remit significantly more compared to female emigrants possibly due to the inheritance
seeking motive. Migrants who left behind valuable assets in home country may remit in order to
make sure their assets are taken care of. Moreso, the desire to return home may promote
remittances for investment in real estate, financial assets or even public assets in order to gain
prestige and political influence in the community (SOPEMI 2006). Such remittances linked with
a desire to return are used to “buy” what Lucas and Stark (1985) called social assets i.e. the
relationship with family members and friends. The self interest model therefore highlights three
reasons of migrant‟s remittances, namely the desire to return to country of origin, investment in
community of origin, and desire to inherit assets in family of origin Lucas and Stark (1985).

Implicit family Agreements

In this model, remittances determination is placed in a family framework of decision making,
with remittances being endogenous to the migration process. Here migration is treated as
means of diversifying sources of family‟s income (Stark 1991) or a risk spreading strategy to
allocate certain members of the family as migrants and remittances as a way of redistributing
the gains (Lucas and Stark 1985). The intra-family agreements can be in form of “implicit co-
insurance agreement” or “implicit loan agreements”.
In the co-insurance model, it is assumed that there are intra-family agreements
/understandings which are safeguarded against breach. In the first phase the migrant plays the
role of insuree and the family left behind as insurer. The family will finance the initial cost of the
migration process partially or wholly. In the second phase the migrant will have settled down,
secured employment and will be remitting to improve the livelihood of family and even
undertake investment.
The loan agreement model was theorized as displaying three waves. In the first stage,
remittances are assumed to be repayments of an informal and implicit loan contracted by the
migrant for education and migration costs. In the second stage loans are made by migrants to
young relatives to finance their education, until they are ready to migrate. In the third stage,
before returning to the country of origin migrants invest accumulated capital at home therefore
the amount of remittances increases. In this model educated migrants are likely to remit more.
Ilahi and Jafarey (1999) found that the level of migrant education and amount remitted are
positively correlated.

Portfolio management decisions

Savings not needed for personal and consumption may be remitted for reasons of relative
profitability of savings in the home and host country. In this case remittances depend on interest

Licensed under Creative Common Page 4

Electronic copy available at: https://ssrn.com/abstract=2671245
 International Journal of Economics, Commerce and Management, United Kingdom

rates, exchange rates, inflation and relative rate of return on financials. Scholars however argue
that it‟s the microeconomic factors that are more significant in determining remittance flows in
the long run, while macroeconomic factors affect the remittance channel used.

METHODOLOGY
The research utilised primary data gathered through questionnaires. Participation was voluntary
after the research objectives were explained. Data from the senders (migrant workers) was
gathered from two main regional migrant destination countries of South Africa and Botswana in
November and December 2012. Zimbabweans in Johannesburg and Pretoria cities of South
Africa, and Gaborone and Lobatse in Botswana were surveyed. Postal questionnaires were also
sent and received from other parts of both South Africa and Botswana mainly using e-mail
service. To reach out to the respondents the snowballing technique was used. Probability
sampling could not be used because there is no sampling frame of Zimbabwean migrant in
Botswana and in South Africa. The snowballing technique involved the use of respondents to
identify other respondents through their own networks. The process began by identifying a few
migrants whom we had contacts with, as initial sampling points. Considerable effort was made
to ensure that the initial sampling points were of varied backgrounds in terms of age,
occupation, gender and legal status. The survey included only respondents who were willing to
participate after the research objective was explained to them. Through this snowball referral
method, we managed to obtain a sizeable sample.
The questions were closed ended questions. There could have been selection bias
resulting from the fact that news of the surveys was spread, mostly, by word of mouth, emails
and telephones. Such a „snowballing‟ effect may have resulted in a bias towards a certain
income, or age group. In Gaborone we tried to reduce bias by working with leaders of civil
society groups who linked us with Zimbabweans of various age groups and professions. In
South Africa, this challenge was mitigated by targeting a wide range of professions which
included nurses, teachers, security guards, caterers, engineers, domestic workers and other
technocrats. This enabled us to reach out to a wider Zimbabwean population with a diverse
background in terms of skills and the earnings.

Econometric model
This study employs a binary logistic regression model in order to identify the impact of socio
demographical factors on the probability of sending money abroad from South Africa and
Botswana. The binary dependent variable in the model is whether a person is remitting money
to the Zimbabwe or not, specifically 1 denotes the individual is remitting and 0 denotes

Licensed under Creative Common Page 5

Electronic copy available at: https://ssrn.com/abstract=2671245
 © Mugumisi

otherwise. Previous scholars like Holst and Schrooten (2006) and Richard(2001), used the
probability model, namely probit or logistic model to estimate the likelihood of decision to
migrate or to remit. The logistic model is used when the categorical dependent variable has two
levels i.e. “yes or no” in this case. It is used to analyse relationships between a non-metric
dependent and metric (age, length of stay in foreign country, years of schooling, number of
migrant from same family) as well as dichotomous independent variables (gender, marital
status, intention to return employment status).Logistic regression combines the independent
variables to estimate the probability that a particular event will occur, i.e. a subject will be a
member of one of the groups defined by the dichotomous dependent variable.
This study employs SPSS version 16 for data analysis. In SPSS the dichotomous
independent variables are called factors and the metric or continuous independent variables are
called “covariates”. Binary logit models analysis requires one to select a reference category, a
decision that should be informed by your research agenda. The general rule of the thumb is that
the most populated response should be the reference category; we overruled this and set “no”
as the reference category.
Logistic regression uses maximum-likelihood estimation to compute the coefficients for
the logistic regression equation. This method attempts to find coefficients that match the
breakdown of cases on the dependent variable. The Maximum-likelihood estimation is an
iterative procedure that successively tries to get closer and closer to the correct answer. When
SPSS reports the "iterations," it is telling us how may cycles it took to get the answer. The
Logistic function was specified as:

exp( z )
G( z ) 
1  exp( z )

Where G (z) is the standard logistic distribution function, and

z    1 Age   2 Length   3 Education   4 Members   5 Marital   6 Gender  

The logit model above estimates the response probability, i.e. the probability of remitting given
the independent variables above. The logistic function G (z) is a non linear function that
produces probabilities between 1 and 0.

Licensed under Creative Common Page 6

Electronic copy available at: https://ssrn.com/abstract=2671245
 International Journal of Economics, Commerce and Management, United Kingdom

Variables in the Model

Age- is the migrant‟s age in years, according theoretical literature, the age of the remitter plays
a positive role in remittance. Holding all things constant, years should be positively related to
years of experience and income. We therefore expect a positive sign for the coefficient of age.
However, beyond a certain age this tends to decline, this applies as productivity will decline with
age and the sign becomes negative.

Length- is the number of years since emigration or migration duration. It is expected that length
of stay abroad has a positive but decreasing effect on remittance. Initially early years after
migration will help the migrant settled and regularize stay in host country and this has a positive
effect on remittances. But as the years go by, the migrant will establish new social links in host
country and his/her attachment will weaken and this will tend to reduce the migrant‟s propensity
to remit (Havolli 2009).

Gender- numerous empirical studies report a significant influence of gender on the amount of
remittances. Theory suggests that male migrants would generally be expected to remit more in
order to ensure the right of inheriting assets in country of origin. While Lucas and Stark (1985)
found in their seminal work on remittances that women show a higher propensity to remit, more
recent studies have produced the opposite finding. Vanwey (2004) found that women compared
to men send more money before the wedding for escaping the social sanctions of the family.

Education – is the number of years of schooling, used to proxy the migrant‟s qualification.
Education tends to improve the migrant‟s chances to secure a decent job in host country,
allowing then to get more income and remit more. It is also easier for educated migrant works to
secure worker‟s permit and officially enter the labour market in the host country. Durand et al.
(1996), for example, found that migradollars (or remittances) of Mexican migrants increase by
4.3% with their years of education until a certain age (40s) that make older migrants less likely
to remit.

Members- is the number of migrants from the same family living and working in the diaspora.
We expect that as that as the number of migrants from the same family increases, the amount
and frequency of remitting will decrease. This is because the migrants will share the
responsibility of supporting parents and even take turns to remit or make contributions.

Licensed under Creative Common Page 7

Electronic copy available at: https://ssrn.com/abstract=2671245
 © Mugumisi

Marital- is a categorical variable to capture the migrant‟s marital status. Empirical studies have
found that married migrants are more likely to remit more. Married migrants have a
responsibility to support children and the spouse back home and therefore tend to remit more
especially if family and children are still in the country of origin. We expect a positive sign for
married, divorce (with children in home country) and widowed (with children in home country).

DATA ANALYSIS AND FINDINGS

This section presents the results and interpretation of logistic model estimated. The first step
taken was to establish the number of variables included in the model, the case processing
summary Table 1.

Table 1:Case Processing Summary

a
UnweightedCases N Percent
Selected Cases Included in Analysis 292 99.3
Missing Cases 2 .7
Total 294 100.0
Unselected Cases 0 .0
Total 294 100.0
a. If weight is in effect, see classification table for the total number of cases.

The case processing summary show the number of cases that were included in the model, the
logit model had 294 cases. Of the 294, 99.3% of the cases were included in analysis while 0.7%
was missing.

Table 2 Dependent variable

Original
Value Internal Value
No 0
Yes 1

The dependent variable for the model was remit, which is categorical. Table 2 above indicates
the coded values for the categories of the dependent variable, the response category “no” was
coded as 0 and “yes” 1. Response category “no” is the reference category, while the response
category coded 1 is the outcome we are trying is to predict.

Licensed under Creative Common Page 8

Electronic copy available at: https://ssrn.com/abstract=2671245
 International Journal of Economics, Commerce and Management, United Kingdom

Table 3 Classification Tablea, b

Predicted
do you remit funds to home
country Percentage
Observed no yes Correct
Step 0 do you remit funds to no 0 137 .0
home country
yes 0 155 100.0
Overall Percentage 53.1
a. Constant is included in the model.

The Table 3 above indicates that if we knew nothing about our variables and guessed that a
person would remit we would be correct 53.1% of the time. The benchmark that is used to
characterize a logistic regression model as useful is a 25% improvement over the rate of
accuracy achievable by chance alone. Table 4 below shows the accuracy of the populated
model, i.e. the model that includes the explanatory variables.

Table 4 Classification Tablea

Predicted
do you remit funds to home
country Percentage
Observed no yes Correct
Step 1 do you remit funds to no 87 50 63.5
home country yes 45 110 71.0
Overall Percentage 67.5
a. The cut value is .500

Table 4 above shows how the classification error rate has changed from the original 53.1%. By
adding the variables we can now predict with 67.5% accuracy. The proportional by chance
criteria is 66.375% (53.1X 1.25=66.375%). Since the accuracy rate in this case; 67.5%, is
greater than the 66.375% by chance accuracy criteria, this model is characterized as useful.

We also checked whether our intended variables would improve the model. The variable not in
the equation table above tells us whether each independent variable improves the model.

Licensed under Creative Common Page 9

Electronic copy available at: https://ssrn.com/abstract=2671245
 © Mugumisi

Table 5 Variables not in the Equation

Score Df Sig.
Step 0 Variables Gender(1) 3.985 1 .046
Age 5.915 1 .015
Marital 18.963 2 .000
Marital(1) 1.293 1 .255
Marital(2) 8.605 1 .003
Education 27.174 1 .000
Length 2.828 1 .093
Members .060 1 .807
Overall Statistics 64.184 7 .000

The variable not in the equation Table 5 above tells us whether each independent variable
improves the model. The table show that the variables, Gender, Age, Marital (2), Education and
length (at 10 level of significance) are significant and if included would add to the predictive
power of the model. If they had not been significant and able to contribute to the prediction, then
termination of the analysis would obviously occur at this point.

The populated model overall significance was tested using the Omnibus Tests of Model
Coefficients, in Table 6 below.

Table 6 Omnibus Tests of Model Coefficients

Chi-square df Sig.
Step 1 Step 74.902 7 .000
Block 74.902 7 .000
Model 74.902 7 .000

The Omnibus Test tells us whether our model is a significant improvement on the „empty model‟
(like the F-test in linear regression).In this analysis, the test of the full model versus a model with

intercept only was statistically significant,  (7, N = 292) = 74.902, p < .001. The null
2

hypothesis that there is no difference between the model with only a constant and the model
with independent variables was rejected. The existence of a relationship between the
independent variables and the dependent variable was supported.

Licensed under Creative Common Page 10

Electronic copy available at: https://ssrn.com/abstract=2671245
 International Journal of Economics, Commerce and Management, United Kingdom

Another test for the validity of model the Hosmer and Lemeshow test was also done. The model
divides subjects into 10 ordered groups of subjects and then compares the number actually in
the each group (observed) to the number predicted by the logistic regression model
(predicted).A probability (p) value is computed from the chi-square distribution with 8 degrees of
freedom to test the fit of the logistic model. If the H-L goodness-of-fit test statistic is greater than
.05, we fail reject the null hypothesis that there is no difference between observed and model-
predicted values, implying that the model‟s estimates fit the data at an acceptable level Table 7
below show the Hosmer and Lemeshow test.

Table 7 Hosmer and Lemeshow Test

Step Chi-square df Sig.

1 14.705 8 .065

In Table 7 above we fail to reject the null hypothesis, at 5% level of significance. Thenon-
significance is desirable; it indicates that the model prediction does not significantly differ from
the observed.
Although there is no close analogous statistic in logistic regression to the coefficient of
determination R2 the Model Summary Table 8 provides some approximations. Cox and Snell’s
R-Square attempts to imitate multiple R-Square based on „likelihood‟.

Table 8 Model Summary

Cox & Snell R Nagelkerke
Step -2 Log likelihood Square R Square
a
1 328.786 .226 .302
a. Estimation terminated at iteration number 5 because parameter
estimates changed by less than.001.

The Cox and Snell R square in Table 8 above indicates that 22.6% of the variation in the
dependent variable is explained by the logistic model. The Nagelkerke‟ R Square is a more
reliable measure of the relationship. Nagelkerke’s R2 will normally be higher than the Cox and
Snell measure. Nagelkerke’s R is the most-reported of the R-squared estimates. In this case it
is 0.302, indicating that 30.2% is explained by the model.

Licensed under Creative Common Page 11

Electronic copy available at: https://ssrn.com/abstract=2671245
 © Mugumisi

Although the model appears good, we need to evaluate whether each of the independent
variables included make a significant contribution to the model. Table 9 below show the
significance of the variables in the model.

Table 9 Variables in the Equation

B S.E. Wald Df Sig. Exp(B)
a
Step 1 Age .069 .021 10.969 1 .001 1.072
Length -.236 .062 14.557 1 .000 .790
Education .309 .061 25.953 1 .000 1.362
Marital 16.731 2 .000
Marital(1) -1.414 .357 15.703 1 .000 .243
Marital(2) -1.085 .419 6.694 1 .010 .338
Members -.033 .132 .063 1 .801 .967
Gender(1) -.553 .291 3.609 1 .057 .575
Constant -3.934 1.000 15.470 1 .000 .020
a. Variable(s) entered on step 1: Age, Length, Education, Marital, Members, Gender.

After testing the validity of the model, interpretation of the predictor variables was done.
Age:-The probability of the Wald statistic for the independent variable “age” (χ² (1, N = 292) =
10.969, p = .001) was less than or equal to the level of significance of .05. The null hypothesis
that the beta coefficient for age was equal to zero was rejected. The value of Exp (B) for the
variable age was 1.072 which implies an increase in the odds of 7.2% (1.072-1=.072).For each
unit increase in age, survey respondents were 7.2% more likely to remit than not to. As age
increases there is likely to be an increase in responsibility (due to marriage and increased family
size) which increases the odds of remitting. The results are consistent with the finding of Havolli
(2009).

Length:-is the number of years a person has been living and working outside the country. The
probability of the Wald statistic for the independent variable “length” (χ² (1, N = 292) = 14.557, p
= .000) was less than or equal to the level of significance of .05. The null hypothesis that the
beta coefficient for length was equal to zero was rejected. The variable had a negative
coefficient (-0.236).A negative beta coefficient results in a decrease in the likelihood of the
expected outcome (remitting). The value of Exp (B) for the variable length was 0.79 which
implies a decrease in the odds of21% (0.79-1=0.21). For each unit increase in years of living
and working outside the country, survey respondents were 21% less likely to remit to home

Licensed under Creative Common Page 12

Electronic copy available at: https://ssrn.com/abstract=2671245
 International Journal of Economics, Commerce and Management, United Kingdom

country. The results are consistent with the altruistic theory of remittances, which posits duration
of stay is expected to have a negative impact on remittances; it is assumed that attachment to
family weakens gradually. As the years of living and working outside the home country increase,
the migrant will establish social networks in host country and the migrant‟s intention to return
weakens leading to a decline in remittances.

Education: -Is the number of years of education the migrant had undergone. The probability of
the Wald statistic for the independent variable “education” (χ² (1, N = 292) = 25.953, p = .000)
was less than or equal to the level of significance of .05. The null hypothesis that the beta
coefficient for education was equal to zero was rejected. The variable has a positive coefficient
(0.309).A positive beta coefficient means an increase in education increases the likelihood of
the expected outcome (remitting). The value of Exp (B) for the variable length was 1.362 which
implies an increase in the odds of 36.2% (1.362-1=0.362). For each unit increase in years of
education, survey respondents were 36.2% more likely to remit to home country. The value of
the migrant‟s human capital is reflected in years of education. Theoretical models relying on
altruism and intra-family-investment schemes, predict that better education leads to higher
transfers. The Results are consisted to the findings of Agarwal and Horowitz (2002) and Holst
and Schrooten (2006). Educated migrants are likely to secure decent rewarding jobs and will be
able to remit more.

Marital status: -is a categorical variable, to denote the marital status of migrants living and
working outside the country. The probability of the Wald statistic for the independent categorical
variable “Married” (χ² (1, N = 292) = 15.703, p = .000) was less than or equal to the level of
significance of .05. The null hypothesis that the beta coefficient for “married” was equal to zero
was rejected. The variable had a negative coefficient (-1.414).A negative beta coefficient results
in a decrease in the likelihood of the expected outcome (remitting). This is somewhat
unexpected, since married people are expected to send back money to support families‟ bank
home. The value of Exp (B) for the variable length was 0.243 which implies a decrease in the
odds of 75.7% (0.243-1=-0.757).This means that survey respondents who are married were
75.7% less likely to remit to home country compared to those who were divorced. The
explanation may be due to fact that some married migrants having migrated with their families to
the regional destinations, leaving no close relatives in home country.
Single-migrant‟s Wald Chi square of 6.694, with a (0.000) p value less than or equal to
the level of significance of 0.05. The null hypothesis that the beta coefficient for “single” was
equal to zero was rejected. The variable had a negative coefficient (-1.085).A negative beta

Licensed under Creative Common Page 13

Electronic copy available at: https://ssrn.com/abstract=2671245
 © Mugumisi

coefficient results in a decrease in the likelihood of remitting. The value of Exp (B) for the
variable “single” was 0.338 which implies a decrease in the odds of 66.2% (.338-1=0.662). The
results mean that survey respondents who are single were 66.2% less likely to remit to home
country compared to those who were divorced. Divorced migrants may be having children back
home and are therefore expected to remit more than single migrants.

CONCLUSION
This paper provides micro-level evidence on the determinants of remittances in Zimbabwe. The
objective of the study was to provide a better understanding of transfer behavior of Zimbabwean
migrants by analyzing empirically the determinants of their remittances. The major strength of
the study was that it was based on a fairly large sample of Zimbabweans in two major regional
migrant destinations of South Africa and Botswana. The findings of the study suggest that age,
length of stay in a foreign country, marital status and education level has significant influence on
the migrant‟s chances/ probability of remitting to home country. Gender and the number of
members of same family living and working outside the country were found to be insignificant in
influencing probability of remitting to home country. The results of our analysis confirm that
remittances of Zimbabwean migrants are remarkably driven by pure altruism. This study only
focused on the determinants of remittances, there is scope for further studies to look at factors
affecting the frequency of migrant remittances into Zimbabwe.

REFERENCES
Agarwal, R. and Horowitz, A.W. (2002). “Are International Remittances Altruism or Insurance? Evidence
from Guyana Using Multiple-Migrant Households”. World Development 30 (11): 2033-2044.
Barro, R. J. (1974). “Are Government Bonds Net Wealth?”.Journal of Political Economy 82 (6): 1095-
1117.
Becker, G. S. (1974). “A Theory of Social Interactions”. Journal of Political Economy 82 (6):1063-1093.
Bernheim, B. D., Schleifer, A. and Summers, L. h. (1984). “The Strategic Bequest Motive”. Journal of
Political Economy 93 (6): 1045-1076.
Bouhga-Hagbe, J. (2006). “Altruism and Workers‟ Remittances: Evidence from Selected Countries in the
Middle East and Central Asia”. IMF Working Paper 06/130.
Bouoiyour .J. and Miftah.A. (2014).Why do migrants remit? An insightful analysis For Moroccan
case,Centre d‟AnalyseThéorique et de Traitement des donnéeséconomiques.
Burgsdorff D, (2012). An analysis of remittance flows from South Africa to Zimbabwe. PASSOP report
April 11th.
Cameron, A. C. and Trivedi, P. K. (2005). Microeconometrics: Methods and Applications. Cambridge
University Press: New York.
Durand, J; Kandel, W; Parrado, E.A and Massey, D.S, (1996).International Migration and Development in
Mexican Communities. Demography, Vol. 33 (2), 249-264.

Licensed under Creative Common Page 14

Electronic copy available at: https://ssrn.com/abstract=2671245
 International Journal of Economics, Commerce and Management, United Kingdom

El-Sakka, M.I.T. & McNabb, R. (1999) The macroeconomic determinants of emigrant remittances. World
Development, 27, pp. 1493–1502
Funkhouser, E. (1995). “Remittances for International Migration: A Comparison of El Salvador and
Nicaragua”. The Review of Economics and Statistics 77 (1): 137-146.
Garip, F. (2006): Social and economic determinants of migration and remittances: An analysis of 22 Thai
villages; Princeton University, mimeo.
Hoddinott, J. (1992) Modeling remittance flows in Kenya. Journal of African Economics, 1 (1), 233±270
Holst, E. and M. Schrooten (2006): Migration and Money – What Determines Remittances? Evidence
from Germany.DIW Discussion Paper 566.
Ilahi, N. and Jafarey, S. (1999), Guest worker Migration, Remittances, and the Extended Family:
Evidence from Pakistan, Journal of Development Economics, Vol. 58, 485-512.
Lucas, R. E. B. and Stark, O. (1985). “Motivations to Remit: Evidence from Botswana”. The Journal
ofPolitical Economy 93 (5): 901-918
Lucas, R.E.B. & Stark, O. (1985) Motivations to remit. Journal of Political Economy, 93, pp. 901–918.
Lucas, R.E.B. and O. Stark (1985): “Motivations to Remit: Evidence from Botswana,” Journal of Political
Economy, 93: 901-918.
Makina D, (2007). Survey of Profile of Migrant Zimbabweans in South Africa: A Pilot Study. University of
South Africa.
Mugumisi N and N. M. Ndhlovu (2013), In search of remittance mobilisation strategies through formal
channels in Zimbabwe; A survey of Zimbabweans living in South Africa and Botswana. International
Journal of Management and Humanity Sciences. Vol., 2 (7),605-618, 2013.
Ncube G, Gomez G.M, (2011). Local economic development and migrant remittances in rural Zimbabwe:
Building on Sand or Solid Ground. International Institute of Social Sciences. Working paper No. 523.
Ncube G, Gomez GM,( 2011). Local economic development and migrant remittances in rural Zimbabwe:
Building on Sand or Solid Ground. International Institute of Social Sciences, Working paper No. 523.
Russell, S. S. (1986), Remittances from International Migration: A Review in Perspective,
Smith Kelly, C, and Solomon, B. (2009). The Influence of Religion on Remittances Sent to Relatives and
Friends Back Home. Journal of Business & Economics Research, Vol. 7(1), pp. 91-101.
SOPEMI (2006) International Migrant Remittances and their Role in Development, ISBN 92 64-03627-X,
OECD 2006.
Orozco M, Ferro A, (2008). Worldwide Trends in International Remittances, August 2008 vol. 5 No. 3
Vanwey, L.K, (2004). Altruistic and contractual remittances between male and female migrants and
households in rural Thailand. Demography, Vol.41 (4), 739-756.
World Bank, (2012).Migration and Development Brief Migration and Remittances Unit, Development
Prospects Group. World Development, Vol. 14, pp. 677-96.

Licensed under Creative Common Page 15

Electronic copy available at: https://ssrn.com/abstract=2671245