Write a short summary of the article, in which you highlight the following

1

The Effects of Remittances on Savings: A Household Level Approach

Antonio Giraldi

2

1. Introduction

The effects of remittances on different areas of the economy has been a growing literature over

the past few decades, given the raising increase in the amount of remittances sent around the

world. According to a report from the World Bank, the flow of remittances to developing

countries has more than doubled over the past ten years, going from around 178 billion in

2004, to 435 billion in 2004 (World Bank, 2014). It is because of this rapid increase in the flow

of remittances around the world, that economist are interested in determining how this source

of money affect the behavior of the economy in the receiving countries, both at the aggregate

level and the individual level.

Economists have studied the impact of remittances on different variables, such as

exchange rate, inflation, income, saving, and others, in order to provide insights on the

behavior of these variables and different policy issues. Because of the large amount of money

that goes into developing countries every year via remittances sent by migrants, it is important

to understand how this money affects both the development and the living standards of the

receiving countries. This paper examines the effects of remittances on savings in the

Dominican Republic using the 2007 Encuesta Nacional de Hogares (National Survey of

Households in its English translation), which is a household level survey conducted by the

national office of statistics of the country.

Savings play an important role in the development of a country, mainly through its high

correlation with investment. The relation between saving and investment, and its effects on

growth, was first explained in the Solow Growth model, which is one of the most important

models of growth (Romer, 2012). The high correlation between saving and investment has

been supported by several studies in the past. Baxter and Crucini find that this correlation is

higher for “more developed” countries (ϕ=0.99), than for “developing” countries (ϕ=0.85). This

fact is also supported by other studies in the Saving-Investment literature (Baxter & Crucini,

1993). Although, developing countries present a smaller correlation between saving and

investment, the correlation is still high and provides an important link between the two

variables.

In the Dominican Republic, the inflow of remittances has also been increasing rapidly

the past few decades. Since the year 2000, household remittances has more than doubled,

growing from US$1,500 million to over US$3,000 million in 2013 (see graph A1 in the appendix).

In addition, according to the World Bank statistics, in 2013 the total inflows of remittances for

the Dominican Republic accounted for 7.4 percent of GDP. This high level of remittance inflow

increases the level of consumption, investment, and savings of the country, and hence provides

a boosts in the development of the economy. In fact, Pozo, Sanchez-Fung and Santos-Paulino

argue that the inflow of remittances has played an important factor in the sustained economic

3

growth of the country, which was the fastest growing economy in Latin America from 1970 to

2003 according to the authors. They argue that migration and the inflows of remittances into

the country increases investment in physical, human, and entrepreneurial capital, which are

important factors in the development of an economy (Pozo, Sanchez-Fung and Santos-Paulino,

2010).

In addition to providing a direct boost to investment and consumption, remittances also

affect investment through their effect on savings, because of the high correlation between the

two variables, as mentioned above. Moreover, savings obtained from remittances can be used

to smooth consumption over time, and allow individual to overcome negative income shocks

easier (Amuedo-Dorante and Pozo, 2011). For these reasons, it is important to understand how

remittances affect the savings of individuals and to what extent its effect on savings affect the

development of the country. This paper investigates the effects of remittances on household’s

savings using data from the 2007 Encuesta Nacional de Hogares (ENHOGAR).

2. Data

The data was obtained from the household’s survey, Encuesta Nacional de Hogares

(ENHOGAR), which is conducted by the Oficina Nacional de Estadistica (National Office of

Statistics) in the Dominican Republic. The ENHOGAR survey is conducted every year in an

attempt to inquire into different socioeconomic issues that affect the Dominican people

(ENHOGAR, 2007).

The survey began to ask questions about remittances in the year 2007, which is the year

that is used in the analysis presented here. Although the survey collected data on remittances

since the year 2007, the survey was not conducted during the following three years, which

makes the next survey available in 2011. Data for this survey is available, but was not used in

this analysis because several issues came up because of differences in the surveys, which most

likely arose due to the three years gap between the two surveys. In sum, the year 2007

presented a more complete and clean dataset, which allows a better analysis of the data, and

therefore, only data for the year 2007 is used in this analysis.

The 2007 survey covered 12,670 households, with median monthly income of US$5,000

(RD$165,000 pesos), and median per capita income of US$857.14 (RD$28,285 pesos). Out of

the total sample, 15.3 percent (1941 households) reported receiving remittances during the

past 12 months. Table B1 in appendix presents summary statistics for the total sample. For the

individual variables, such as age and education, the head of the household was chosen as a

representative of the entire household, both in the summary and in the econometric analysis,

since the head of the household is more likely to control the inflows of remittances and the

decisions on how to spend the money.

4

As mentioned above, 1941 households reported to receive remittances during the year.

This is the sample that will be used for the analysis presented here. Households which did not

receive remittances are not considered in this analysis because there was no data reported on

savings for those observations. The nature of the survey, in terms of remittances and savings,

was as follow: first individuals were asked whether or not they received remittances during the

last 12 months. If the answer was negative (no remittance received), they moved to the next

question on the list. If the answer was positive (they received some amount of money), then a

set of questions regarding different aspects of the remittances received were asked (i.e.

country coming from, amount, frequencies, and others). One of the questions regarding

remittances was intended to inquire how the recipients spent the money: did you spend any of

this money on this activity? If yes, how much? The different categories for spending the money

were education, health, furniture, construction, home, investment, debt, savings and other (see

table B2).

In this analysis, I use the amount of saving reported by the individuals as the dependent

variable, and attempt to determine the variables that affect the level of savings coming from

remittances. Because of the nature of the survey and the data, some interesting questions are

not possible to be answered. For instance, we cannot analyze how remittance receiving

households compare with non-remittance receiving households. In addition, none of the

formal theories of savings can be tested using this dataset, since they require detailed data on

savings, and ideally over time (i.e. the life-cycle and Buffer-Stocks theories of savings).

However, some important insights can be derived from this survey, which is the focus of this

paper. There are two particular questions this paper attempts to answer; the first is whether

the level of remittances received affects the amount of savings. That is, do higher remittances

received induce a higher amount of savings or not? The second question is by how much the

level of remittances affect savings?

In order to answer these questions, I estimate Cragg’s double hurdle model (or two-tier

model) using savings as the dependent variable, and remittances and a set of covariates as

explanatory variables. The econometric methodology used in this analysis is explained in the

following section.

5

3. Results

Before presenting the results of the model, I will provide some insights of the sample being

analyzed, since the data described in section 2 was for the full sample in the survey. Table 1

below presents the summary statistics of the remittances receiving households:

Table 1: Summary statistic for Regression sample.

around 6 percent (113) of the remittance receiving households reported to save some amount

of money. The average amount saved out of the total inflow of remittances was around US$20

dollars, and the average amount saved by those who decided to save was US$340 dollars

(US$151 median). The average income of the remittance receiving household is $307 dollars

(US212 median), which is about 30 percent higher (40 percent if one uses the median Income)

than that of the total sample (US$239 dollars). This suggests that remittances receiving

households are of relatively higher income compared with the total population, which differs

from some findings in other countries (Borras & Pozo, 2007).

In addition, the average amount of remittances received was US$1,105 dollars

(US$443.5 median). Both remittances and savings are non-symmetric, presenting a right tail

(see figure A2 in the appendix), so it may be better to use the median amounts of these

variables for references, rather than the mean. In addition, because of the non-symmetry of

many of the variables in the model, a logarithmic transformation of these variables was used in

order to transform the variables so that they follow a normal distribution. I estimate both a

model with the symmetric and the non-symmetric variables. Finally, the main source of

remittances to the Dominican Republic, based on the sample from the survey, was the United

States with around 70 percent of the individuals sending money being from that country, and

accounting for more than 70 percent of total remittances received if one includes Puerto Rico.

Variable n Mean Median SD Min Max

Saving Dummy 1935 0.061 - - 0 1

Remittance 1935 1104.8 443.52 2082.182 5.913 35370

Savings in USD 1935 19.881 - 148.778 0 3405.606

Savings in USD | S>0 113 340.44 151.51 521.64 0.61 3405.61

Income in USD 1935 307.03 212.12 375.19 3.03 5454.545

Incomepp in USD 1935 102.31 60.6 152.162 2.27 2727.273

HH Size 1935 3.74 4 1.9 1 13

Age 1935 49.662 48 18.107 14 97

Edu 1935 2.327 2 1.072 0 5

Gender (1=Male) 1935 0.5064 - - 0 1

Marital Status (1=joined) 1935 0.577 - - 0 1

Zone (1=Urban) 1935 0.711 - - 0 1

6

For a full and detailed summary and information about the remittances received and the

variables used in the model, see the appendix B.

4. Conclusion

This paper investigates the effects of remittances on household’s savings using data from the

2007 Encuesta Nacional de Hogares (ENHOGAR). Both remittances and in-kind remittances are

found to increase the likelihood that a household decides to save when using the level

variables. However, only remittances affect the amount of money household save. The overall

marginal effect of an additional dollar of remittance received is estimated to be 0.0235, which

would represent around USD$80 million saved given the inflow of remittances into the country

in 2007.

Using the variables in logarithmic form, remittances still have a positive and significant

effect on the decision to save and the amount saved. Specifically, a 10 percent increase in

remittances increases expected savings by about 1.7 percent. In general, these results provide

support to some of the previous findings in the remittance literature, which suggested that

remittances not only increase consumption, but they also affect different productive sectors of

the economy, such as investment, business growth, and in this particular case savings. In

addition, by increasing the level of savings, remittances indirectly provide a boost to other

variables such as investment and future income. These indirect effects help the development

of the economy by decreasing uncertainty over time and stabilizing the level of consumption

and investment in the economy.

Further research should be directed to confirm whether these results prevail over time,

and across countries where the inflow of remittances plays an important role in the economic

conditions of the individuals and the country in general. In order to accomplish this, I reiterate

the need for more micro level surveys that allow researchers to investigate the behavior of

individuals under different economic conditions.

7

Appendix

A) Graphs

Figure A1: Inflow of Remittances to the Dominican Republic by Year

Figure A2: Distribution of Variables in Level

0

500

1000

1500

2000

2500

3000

3500

1 9

9 8

1 9

9 9

2 0

0 0

2 0

0 1

2 0

0 2

2 0

0 3

2 0

0 4

2 0

0 5

2 0

0 6

2 0

0 7

2 0

0 8

2 0

0 9

2 0

1 0

2 0

1 1

2 0

1 2

2 0

1 3

Remittances in MM

Remittances

8

Figure A3: Residual Analysis of Double-Hurdle Model (Tier 2)

Figure A4: Distribution of Variables in Logarithm Form

0

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0

U S

S a v in

g

0 10000 20000 30000 40000 Remittances

0

5 0 0 0 1 0 0 0 01

5 0 0 02

0 0 0 02

5 0 0 0

re s id

0 5000 10000 15000 20000 25000 Remittances

0

5 0 0 0 1 0 0 0 01

5 0 0 02

0 0 0 02

5 0 0 0

re s id

-30000 -20000 -10000 0 10000 Linear prediction

0

5 .0

e -0

51 .0

e -0

4 1 .5

e -0

42 .0

e -0

4

D e n s it y

0 5000 10000 15000 20000 25000 resid

9

Figure A5: Residual Analysis of Log DH Model (Tier 2)

A) Tables

Table B1: Full Sample Summary Statistics

Variable n Mean Median SD Min Max

Remittance Dummy 12670 0.1532 - - 0 1

Income in USD 12670 7910.98 5000 10274 0 190000

Incomepp in USD 12670 2671.141 857.14 4131.433 0 90000

HH Size 12670 3.77 4 1.929 1 14

Age 12670 47.61 45 16.75 12 99

Edu 12670 2.06 2 1.158 0 5

Gender (1=Male) 12670 0.6273 - - 0 1

Marital Status (1=joined) 12670 0.6427 - - 0 1

Zone (1=Urban) 12670 0.5895 - - 0 1

10

Table B2: Spending of Remittances by Categories

Table B3: Remittances Received by Country

Spendings Proportion Responding Yes Mean Value Spent in USD

Education 0.2754 $82.74

Health 0.416 $122.49

Furniture 0.0548 $25.97

Construction 0.0568 $36.55

Home 0.7721 $415.11

Investment 0.0248 $15.66

Debt 0.2299 $81.00

Saving 0.0584 $19.88

Other 0.0806 $41.58

Country N Mean Total S.D. Proportion of Total Sample Proportion of Remt Received

U.S. 1345 $1,065.884 $1,433,613.980 1,941.941 0.695 0.671

Spain 256 $1,277.266 $326,980.096 2,674.615 0.132 0.153

Italy 47 $1,339.099 $62,937.653 3,207.260 0.024 0.029

Europe 23 $1,344.747 $30,929.181 1,675.150 0.012 0.014

Puerto Rico 84 $865.760 $72,723.840 1,119.419 0.043 0.034

Latin America 36 $1,517.090 $54,615.240 2,655.908 0.019 0.026

Haiti 12 $1,274.774 $15,297.288 1,885.054 0.006 0.007

Other 132 $1,065.842 $140,691.144 2,034.701 0.068 0.066