Estimating Bilateral Remittances in a Macroeconomic Framework: Evidence from top Remittance-Receiving Countries

Abstract

Based on Ratha and Shaw’s (2007) model for estimating bilateral remittances, this study attempts to develop a new method for computing bilateral and aggregate remittances for the top five remittances-receiving countries for the period 2010-16. Considering tempered altruism as a motive for sending remittances, we develop an analytical framework based on the lifecycle hypothesis of saving to compute bilateral and aggregate remittances. We compare our bilateral and aggregate remittance values with the World Bank's values based on Ratha and Shaw’s (2007) model. Our analytical framework seems to be an improvement over the Ratha and Shaw model in several ways. First, our model considers several theoretical aspects of motivations to remit like saving, investment and wealth accumulation. Second, it addresses the issues of underestimation and overestimation, i.e. inaccuracy, of bilateral and aggregate remittances in various ways (for instance, by considering GDP per capita instead of GNI per capita we control for overestimation of remittances whereas by considering every kind of migrants from all destination countries we control for underestimation) and mitigates the probability of both these issues through the proposed model. Third, it compares regional bilateral remittances between the new model and the Ratha and Shaw model, delving on the reasons behind underestimation and overestimation, i.e. inaccuracy. We conclude that our analytical model has the potential to provide a general framework for computing bilateral and aggregate remittances which can be used for most of the remittance-receiving countries.

Keywords

Altruist motive bilateral remittances lifecycle hypothesis of saving saving and investment

JEL Codes: E21, F22, F24

Introduction

Bilateral remittances are the flow of remittances between the two countries. The top five remittance-receiving countries, that is, India, China, Mexico, the Philippines and Egypt, receive the highest volume of bilateral remittances from the USA, the UK and the Middle-East countries.^{1, 2} In 2018, these top five remittance-receiving countries received around USD 241 billion of remittances out of USD 689 billion of total remittance inflows.³ Remittances have been extensively studied in the literature (Acosta et al., 2009; Aggarwal et al., 2011; Barajas et al., 2009; Catrinescu et al., 2006; Giuliano & Ruiz-Arranz, 2006 Gupta et al., 2009) either as a cause or as an effect with respect to various economic variables such as economic growth, inequality, poverty, investment and financial development, to name a few. It is also to be noted that most of these studies use aggregate remittances data, widely available with the central banks of respective countries. However, the data on bilateral remittances remains scarce to date.

An understanding of bilateral remittances is important because it determines the preference of migrant workers regarding their choice of the host country. Demand for their special skills and the corresponding higher wages are found to be one of the factors for their choice of host countries (Walmsley et al., 2005). Further, migrants’ decision to send remittances depends not only on the needs of the households at the home country but also on some host country-specific factors. For instance, some of the host countries offer more opportunities for the skilled workers, more conducive interest rates, low volatility of wages and low barriers to migration which may influence remittances inflows by migrants to the home country in a positive way.

Given the significance of bilateral remittances data, some studies have attempted to construct bilateral remittance databases from the central bank reported data for the EU countries and a few developing countries. For instance, Frankel (2010) constructs a database by utilizing data from the Inter-American Development Bank, the European Commission’s bilateral remittances data and the data constructed by Leuth & Ruiz-Arranz (2008). The latter construct a database from primary data sources, from various central banks, the International Transaction Reporting System, migration surveys and so on. Frankel’s database consists of data for 64 pairs of countries having approximately 1600 observations. Similarly, other studies (De Sousa & Duval, 2010; Docquier et al., 2012; Schiopu & Siegfried, 2006) have constructed such databases from secondary sources such as money transfer operators, central bank reports and published surveys. While Ahmed and Martinez-Zarzoso (2014) use central bank data in the case of Pakistan, this data seems to be limited in its availability in terms of the number of years. Regarding developing countries, there is a scarcity of data on bilateral remittances per se because central banks of most of the developing countries are not able to report accurate bilateral remittances data because of the usage of various informal channels such as hawala and migrant networks for informal remittance transfers (Freund & Spatafora, 2005).

Currently, a comprehensive bilateral remittance database starting from 2010 is published by the World Bank and prepared by Ratha and Shaw (2007). Their method of estimation is based on migrant stocks, income that migrants earn abroad and home and host countries’ income. The remittances sent by the migrants are estimated from the per capita incomes of the home and host countries. The resultant number, further, is multiplied by the relevant bilateral stock of migrants, to get total remittances received by that particular home country. The estimated bilateral remittances data, applying the above approach, is available with the World Bank as bilateral remittance matrices from 2010 to date. This database has been used by a few studies in various context contexts (Aggarwal, et al., 2011; Azizi, 2017; Lew & Arvin, 2012).

The bilateral remittances data of Ratha and Shaw is said to be complete in the aggregate sense but inaccurate in terms of volumes (Alvarez et al., 2015). The reason for such inaccuracy may be because of their method of estimation, which is briefly discussed as follows. First, Ratha and Shaw’s method relies on differences in per capita income alone as a basis for remittance estimation, the assumption that migrants send remittances because home country’s GNI per capita is lower than that of the host country. While per capita income difference may be one of the reasons, it may not be the only reason. For instance, in the Indian context, Nepal is a preferred migrant destination because of historical ties and cultural similarities (Subedi, 1991). The number of Indian migrants in Nepal was close to 800,000 as of 2013.⁴ Nepal’s GNI per capita has been consistently lower than that of India. Given the large Indian migrant population in Nepal, there is a rather high probability of remittances flowing from Nepal to India. But Nepal does not appear in India’s bilateral remittances matrix prepared by Ratha and Shaw. Second, the use of GNI per capita instead of GDP per capita as a measure of per capita income may overestimate remittances as GNI includes all factor income including remittances.⁵ Furthermore, some host countries have been excluded by the Ratha and Shaw model in the estimation of bilateral remittances. Such cases were observed wherein the remittances received are comparatively smaller or the number of migrants (from the respective remittance-receiving countries) are relatively fewer. For instance, remittances from countries such as Pakistan, Eritrea, Estonia and the Bahamas to India, although smaller than other countries, do occur. However, such cases are omitted or ignored in Ratha and Shaw’s database.⁶

Based on the above discussion, the present study attempts to fill the abovementioned gaps in the literature by constructing a new aggregate and bilateral remittances database. The study has mainly two objectives. First, it addresses the aspect of accuracy by considering remittances from the perspective of motives. In line with this, we propose a method of estimation with an analytical framework that is based on the lifecycle hypothesis of savings (LCHS). Second, by applying the proposed analytical model, we empirically compute bilateral and aggregate remittances for the top five remittance-receiving countries for the period 2010–2016 and compare them with the World Bank’s database (based on Ratha and Shaw’s method) on bilateral and aggregate remittances.

The rest of the article is organized as follows: The second section describes the analytical framework, the third section details empirical estimation and results and the fourth section concludes the study.

Analytical Framework

The starting point of our analytical framework is the model proposed by Ratha and Shaw (2007), described briefly as follows:

Suppose Y_i is the GNI per capita of the home country and Y_j is the GNI per capita of the host country and r_ij is bilateral remittances flowing from (host country) j to (home country) i, r_ij is estimated by the following relationship:

\begin{array}{l} r_{i j} = f (Y_{i}) when Y_{i} > Y_{j} \\ r_{i j} = f (Y_{i} + {(Y_{j} - Y_{i})}^{β}), when Y_{i} < Y_{j} \end{array}

(1)

$β$ is a parameter such that the sum of bilateral remittances equal to aggregate remittances as mentioned in equation (1).⁷

Equation (2) shows remittances at an aggregate level:

R_{i} = \sum_{i \neq j} (M_{i j} r_{i j})

(2)

where i is the home country and j is the host country. r_ij are country-level remittances where remittances flow from j to i. M_ij is the bilateral stock of migrants where migrants flow from i to j⁸.

Assumptions of Ratha and Shaw’s model: The model assumes that if the home country’s per capita income is less than the host country’s per capita income, remittances flow from the host country to the home country. In this model, remittances are assumed to be the difference between GNI per capita of the home and the host countries. The reason that migrants would always send a minimum amount of remittances is mentioned as compensation that they would send to cover for the opportunity loss of employment that the migrant creates at home.

Based on the above discussion, we conclude that Ratha and Shaw’s model has some limitations, described below. We, in the subsequent section, attempt to improve upon Ratha and Shaw’s model by modifying these assumptions.

Inaccuracy: Issues of Underestimation and Overestimation

Ratha and Shaw’s bilateral remittances estimations have been considered as inaccurate (Alvarez et al., 2015). The reasons for such inaccuracy maybe because of the model’s assumptions and the method of estimation. Their model assumes that migrants send remittances because the home country’s GNI per capita is lower than that of the host country. Furthermore, their method suffers from the following limitations that have led to inaccuracy in bilateral remittances data: omission of some host countries, use of GNI per capita rather than GDP per capita and lack of theoretical framework describing motives to remit.

While estimating bilateral remittances, Ratha and Shaw’s model omits some host countries where the migrants’ population and remittances are relatively smaller. The omission of such countries may also be because of their intention to match the aggregate of bilateral remittances values with the World Bank’s aggregate remittance data. Because of this, bilateral remittances estimated by them could be inaccurate.

Further, when Ratha and Shaw consider GNI per capita instead of GDP per capita, the difference between remittances and other factor income from abroad is not clear. Also, remittances are considered the only additional income that the household receives. In both these cases, remittances would be overestimated.

We think that such inaccuracies can also be minimized by developing an analytical framework based on an economic theory explaining the motivations to remit. Ratha and Shaw’s model lacks such a theoretical consideration. We attempt to improve upon Ratha and Shaw’s model by developing an analytical framework for computing bilateral and aggregate remittances in the following ways.

To control for possible underestimation, we assume that migrants9 from all host countries, irrespective of their migrant status and skill, send remittances because they are motivated by tempered altruism and the macroeconomic conditions at the home country.

Similarly, overestimation is controlled in two ways. First, we consider GDP per capita in place of GNI per capita as a measure of income. Unlike GNI per capita, GDP per capita excludes any income from abroad and includes income from productive activities within the domestic economy. This way, remittances are estimated separately and not as total factor income from abroad, minimizing overestimation of remittances. Second, overestimation can be controlled by considering remittances as one of the additional incomes of the household, rather than the only additional income. For instance, in developing countries, remittance-dependent households resort to earning additional income from farming, small businesses, entrepreneurship and so on. Though such households depend on remittances, it can be noted that remittances are not the only additional income which the households receive (Holden et al., 2004; Sana & Massey, 2005). Hence, we consider additional income (non-remittance additional income) in terms of value-added from businesses or farming activities in our model.

Finally, we attempt to correct for inaccuracies in the bilateral remittance values by proposing a theoretical framework. The proposed model is based on the economic theory related to motivations to remit (Lucas & Stark, 1985). In developing the analytical framework, we use the ‘tempered altruism’ motive of remittances wherein the migrants as well as the households benefit from remittances. The remittances that the household receives are not only used for their own consumption and investments but also invested for the migrants’ gain. This way, there is a mutual benefit of the migrant as well as the household. We address the consumption motive by considering the consumption function from the LCHS model and including household age dependency and unemployment rates. We address the investment motive by considering savings and wealth form the LCHS model, and with the interest rate spread as a discount rate for remittances as future income. The interest rate spread gives an idea about the stream of cash flows that the household or migrant would receive in future, from the investment of remittances.

In literature, the determinants of remittances are observed as inequality, economic growth, inflation and interest rates. We include each of these as coefficients to the variables in the model. Thus, we compute remittances according to the motives rather than just an accounting approximation.

Motives of sending remittances

Since savings and investments are observed to be the two main motives to remit to developing countries, the present study utilizes the LCHS developed by Ando & Modigliani (1963) and Jappelli & Modigliani (2003) as an analytical framework for the estimation of bilateral remittances. In this context, the LCHS has provided sufficient empirical evidence about savings rates, reasons for saving or dissaving by the elderly and the accumulation of savings into wealth for future generations (Choudhry, et al., 2010; Danziger et al., 1982–1983; Graham, 1987; Modigliani, 1993; Thøgersen, 2015). In this study, we assume that remittances are sent with the motive of improving savings as well as investments.

The LCHS delineates that consumption is a function of current and future income. If consumption is assumed to be a function of total resources, then

c_{t}^{T} = Ω_{t}^{T} (y_{t}^{T} + \sum_{τ = T + 1}^{N} \frac{y_{t}^{e T τ}}{{(1 + r_{t})}^{τ - T}} + w_{t - 1}^{T})

(3)

Where $c_{t}^{T}$ is the consumption, $Ω_{t}^{T}$ is the proportionality factor, $y_{t}^{T}$ is the current (non-property) income, $y_{t}^{e T}$ is the expected (non-property) income, $w_{t - 1}^{T}$ is the net worth in the past period, $t$ corresponds to time, $T$ corresponds to the individual’s age, N is the earning span and $r_{t}$ is the rate of return on assets. The present value of future earnings shows the earnings that would occur at age T, in the $τ^{t h}$ year. Therefore, total resources are the sum of past returns, current returns and discounted future returns. When $y_{t}^{e T}$ is constructed as annual average expected income:

y_{t}^{e T} = \frac{1}{N - T} \sum_{τ = T + 1}^{N} \frac{y_{t}^{e T τ}}{{(1 + r_{t})}^{τ - T}}

From equation (3), it would become:

c_{t}^{T} = Ω_{t}^{T} (y_{t}^{T} + (N - T) y_{t}^{e T} + w_{t - 1}^{T})

(4)

If the proportionality factor is removed, coefficients are attached and this is generalized for the entire population over all age groups, then the equation would become an aggregate version:

C_{t} = α Y_{t} + γ Y_{t}^{e} + θ W_{t - 1}

(5)

Though the LCHS takes care of the individual behavioural aspects when constructing the aggregate consumption model, it makes some drastic assumptions. The model assumes that every household has the same life span and earning span; every age group has the same earnings and average income within a particular earning span; the rate of return on assets remains constant over the entire lifetime; and that there are no bequests.

Assumptions of the new model: Our model assumes that there are bequests to give relevance to the increase of savings and, therefore, wealth creation for the younger generations, making them affluent. We further assume that the return on assets would change, making income and savings change across time. Since inflation and interest rates would differ across time, and unemployment also motivates the need for additional income, income inequality is considered to subsist, making supplemental income necessary in order to either consume equally or gradually more across time, meaning that there are differences in the allocation of money across time. We assume that all migrants, both temporary and permanent, send remittances because they are motivated by tempered altruism, despite their lengths of stay in the host countries.

With these assumptions, the equivalence of equation (4) and (5) would cease to exist, resulting in a dynamic model of LCHS.

Remittances, consumption and savings

If savings are the result of the difference between income and consumption, then:

Y_{t} - C_{t} = S_{t}

From equation (5),

Y_{t} - (α Y_{t} + γ Y_{t}^{e} + θ W_{t - 1}) = S_{t},

which becomes,

Y_{t} - α Y_{t} - γ Y_{t}^{e} - θ W_{t - 1} = S_{t}

(6)

Since $Y_{t}^{e}$ , which is the expected income, is a combination of remittance and non-remittance income, this expected income of the household would include remittances. We consider two different coefficients here, for two different variables. One for remittance income ( $r_{t})$ and another for non-remittance additional income $(Y_{t}^{e} - r_{t})$ . Accordingly, the coefficient of remittance income is $β (1 + ζ_{t})$ and the coefficient of non-remittance additional income is $η$ . Essentially, $γ$ is divided into two coefficients as $β (1 + ζ_{t})$ and $η$ .

Y_{t} - α Y_{t} - (β (1 + ζ_{t})) (r_{t}) - η (Y_{t}^{e} - r_{t}) - θ W_{t - 1} = S_{t}

(7)

The coefficients of equation (7) are selected based on the empirical literature, so that they reflect the conditions of the home country, households’ need and migrants’ motives. This way, these coefficients serve as motives for sending remittances. Their selection criteria are explained as follows:

$α$ - the coefficient of GDP per capita $(Y_{t})$ , is assumed to be income inequality: Though income inequality and consumption inequality do not have to be simultaneous or substantially equivalent in developed financial markets, they do co-occur in economies where the credit and financial markets are weak and underdeveloped. This may be the case especially with most of the low and lower-middle-income economies. Income inequality will lead to consumption inequality if credit markets do not adjust and mitigate income risk that arises out of income volatility (Krueger & Perri, 2002).

$β$ - the coefficient to remittances $(r_{t})$ , is assumed to be the age dependency ratio: Since remittances finance elderly care and educational requirements, the motive of altruism, connecting it to dependency on the migrant, is addressed through this coefficient (Adzakey, 2013; Amuedo-Dorantes & Pozo, 2009; Nangia, 2015).

$ζ$ - the additional coefficient of remittances $(r_{t})$ , is assumed to be the unemployment rate: The financial dependency in a household may appreciate due to unemployment and leading to furthering income inequality (Xue & Zhong, 2003). Therefore, we include the unemployment rate of the home country as an additional coefficient to remittances.

$η$ - the coefficient to additional income other than remittances $(Y_{t}^{e} - r_{t})$ , is assumed to be the economic growth. In the low income and lower-middle-income economies, additional income consists of business and agricultural income. Small businesses and farming are observed as two significant activities in households. The bidirectional relationship between economic development, income from businesses, agriculture, entrepreneurship and innovation are addressed through this coefficient (Easterly & Kraay, 2000; Howell et al., 2012; Wennekers & Thurik, 1999).

$θ$ - the coefficient to wealth is assumed to be inflation. We assume household wealth to be generated from past savings.

W_{t - 1} = \sum_{1}^{t - 1} S_{t}

The coefficient of wealth is inflation in our model. While redistribution of wealth may be a general consequence of inflation, there is an appreciation of wealth at the household level (Kessel, 1956; Wolff, 1979).

Since the objective is to find bilateral remittances, when (7) is rearranged, it would become:

(\frac{Y_{t} (1 - α) - S_{t} - η . Y_{t}^{e} - θ W_{t - 1})}{((β (1 + ζ_{t})) - η)}) = r_{t}

Since the motivation to remit is the differences in savings, income, inflation and economic growth between the home and host countries, we consider these differences for computing bilateral remittances. But $β$ , $α$ and $ζ$ are confined to the values of the home country. This is because $β$ and $α$ are relatively long-term features and do not change year-on-year and $ζ$ rather than $Δ ζ$ , is one of the motives of the employed migrant-sending remittances. Hence, the above equation becomes:

(\frac{Δ Y_{t} (1 - α_{t H}) - Δ S_{t} - Δ η . Δ Y_{t}^{e} - Δ θ Δ W_{t - 1})}{((β_{t H} ((1 + ζ_{t H})) - Δ η)}) = r_{t}

(8)

Where, at time $t$ , $Y_{t}$ is the GDP per capita, $α_{H}$ is the income inequality, $S_{t}$ is the cumulative savings per capita, $η$ is the economic growth, $Y_{t}^{e}$ is the additional income other than remittances and is the sum of per capita income from agriculture, industry and services; $β_{H}$ is the age dependency ratio of the home country; $ζ$ is the unemployment rate of the home country.

The result of the computation of equation (8) is the remittances per capita ( $r_{t})$ , which is part of the potential future expense of the migrant. This is discounted with the interest rate spread of the host country because we price the migrant’s future expenditure in the form of remittance outflows. We also do this to signify the least that the remittances can receive on investing since remittance-receiving countries generally have higher interest rate spreads (Agapova & McNulty, 2015). The discount factor is in the form of ${(1 + i r s)}^{n}$ . Here, $i r s$ is the interest rate spread of the host country. The exponent $n$ is an estimation based on household dependency. It is calculated based on the average number of earning members in a household, the number of dependents in the household and the average longevity (both stylized and empirical) of the home country. The estimation of the exponent term is explained in Appendix 1.

The discounted remittances per capita are multiplied with the bilateral migrants’ stock of the host country for the particular year, to get the bilateral remittances for a particular country. The sum of such bilateral remittances would be the aggregate remittances received by that country.

Empirical Analysis

Based on the analytical model and applying equation (8), we compute bilateral as well as aggregate remittances for the top five remittance-receiving countries for the period 2010–2016. The sample countries and time periods are limited by the availability of bilateral migration data.¹⁰

The Data

In equation (8), on the left-hand side, $Y_{t}$ , the GDP per capita, is defined as gross value added by resident producers, including product taxes, excluding subsidies and not considering depreciation, deductions and depletions, divided by the midyear population. GDP per capita is measured in current USD. $α_{H}$ is the GINI index, a measure of income inequality. Following Mikesell and Zinser (1973), savings per capita is calculated as gross domestic savings divided by the total population. Economic growth is defined as the growth rate of GDP at market prices. The additional income, other than remittances, is the sum of industry value-added, agricultural value-added and service value-added, all measured on a per-capita basis. This is measured in constant USD11. The age dependency ratio is defined as the ratio of the non-working population (age between 0–15 years and 64– above years) to the working population (age between 15 and 64 years). The unemployment rate is the percentage of the labour force that is not working and unemployed. Data for all the variables are sourced from the World Development Indicators (WDI) database from the World Bank.

Empirical Results

Using equation 8, we compute bilateral as well as aggregate remittances for the top five remittance-receiving countries for the period 2010–2016.12 We compare the aggregate remittances computed values based on our model with Ratha and Shaw model (reported by the World Bank) as presented in Table 1.

We examine the new aggregate remittance values and World Bank reported values in terms of the percentage difference between the values. The aggregate remittances, computed using our model, are higher than the ones estimated by Ratha and Shaw model in most of the cases. While the difference in computation method is one of the main reasons for the large variation, we find several other reasons such as a large number of migrants in some host countries (omitted by the Ratha and Shaw model), socio-cultural similarities, financial constraints, geographical proximity, historical ties and sometimes economic shocks in the home countries as other reasons for the variations. Our observations regarding reasons such as socio-cultural similarities, geographical proximity and economic shocks are similar to conclusions by earlier studies (Aguilar, 2009; Connell & Conway, 2000; Yang & Choi, 2007).

From Table 1, it is found that the aggregate remittance figures computed with the new model differ between 0.09 and 57 per cent across the top remittance-receiving countries, with the highest difference observed in the case of the Philippines in 2013. We observe three possible reasons for such differences: socio-cultural similarities, financial constraints and economic shocks.

Regarding socio-cultural similarities and historical ties in the case of the Philippines, the highest difference is observed in the case of the United Kingdom and Ukraine. While the Philippines and the UK have socio-cultural similarities, Philippines and Ukraine benefit from historical ties. As a result, there are large Filipino settlements in both these countries, so much so that there are cross-cultural marriages and multicultural households in the UK and extensive temporary labour migration to Ukraine (Charsley et al., 2012; Uy, 2016). Old settlements and frequent travel by migrants could result in a reasonably high volume of remittances that flow through unofficial channels. Further, in 2012 and 2013, the Philippines experienced large scale economic shocks driven by natural calamities. Though some percentage of remittances may have flown into the Philippines officially, a larger percentage of remittances would have flown in through informal channels. Therefore, remittance inflows could have been much more in volume than the previous years. Following this, the bilateral remittance inflows that we compute are consistently higher than the ones reported by the World Bank (based on Ratha and Shaw model).

Table 1.

Comparison of Aggregate Remittances

	India			China			Mexico			Philippines			Egypt
	WB	NM	PD (%)	WB	NM	PD (%)	WB	NM	PD (%)	WB	NM	PD (%)	WB	NM	PD (%)
2010	53.48	54.03	1.04	51.00	63.61	24.73	22.08	21.35	–3.32	21.56	29.31	35.97	12.45	15.76	26.52
2011	62.50	76.51	22.43	57.00	54.02	–5.23	23.45	22.33	–4.75	23.05	31.81	37.98	14.32	18.10	26.32
2012	68.82	92.27	34.08	57.99	55.19	–4.83	23.21	23.42	0.90	24.61	33.53	36.25	19.24	18.22	–5.28
2013	69.97	72.01	2.91	59.49	62.82	5.60	23.19	27.45	18.38	26.72	42.03	57.31	17.83	23.95	34.32
2014	70.39	70.46	0.10	61.49	60.02	–2.38	24.80	29.77	20.02	28.69	42.51	48.17	19.57	22.94	17.22
2015	68.91	65.36	–5.15	63.90	70.50	10.33	26.23	32.07	22.27	29.80	36.92	23.88	18.33	19.61	7.03
2016	62.74	63.75	1.60	61.00	67.90	11.31	28.69	31.80	10.84	31.14	38.38	23.25	18.70	23.94	28.04

Source: The world bank and authors’ own calculation.

Note: WB and NM denote the aggregate remittances (in Billion USD) values based on the World Bank and the new method respectively, proposed in the present study. PD is percentage difference.

With host countries like Mauritius, we find that financial constraints in terms of high remittance costs play a significant role in high volumes of informal remittances. For instance, the transaction cost of sending MUR 500 to India is 55 per cent, making unofficial channels a more preferred way of sending remittances. Similarly, in the case of Uganda, the fee is UGX 5900 for sending UGX 500, making it cheaper to personally hand over remittance money to the household in the home country. However, the differences seem to reduce over the years in the case of the Philippines, perhaps indicating the development of the country’s financial systems in terms of depth and liquidity which has resulted in decreasing the transaction costs. Following this, the bilateral remittances computed by us are higher than the ones reported by the World Bank (Ratha and Shaw’s model).

For a visual comparison, Figure 1 shows graphical comparison of aggregate remittances values computed from both models for each country. Further, we also present the Spearman’s rank correlation for each country in Table 2. The graph as well as correlation coefficients give an idea about the rank order change in the aggregate remittance values computed from the two different methods. While the coefficients for Mexico and the Philippines indicate significant positive relationships, the coefficients for the other three countries indicate the correlation to be insignificant but positive.

Figure 1.

Source: The world bank and authors’ own calculation.

Table 2.

Spearman’s Rank Correlation Coefficient

Country	Spearman’s Rho
India	0.28
China	0.50
Mexico	0.82**
Philippines	0.71*
Egypt	0.46

Source: The world bank and authors’ own calculation.

Note: ***, **, and * denote statistical significance at 1%, 5% and 10% level respectively.

To understand how the large migrant population in host countries is contributing to remittance inflows, we analyse regional bilateral remittances. Table 3 presents comparison of regional bilateral remittances of World Bank and the new model. We find that the differences arise from regions that contain the maximum number of migrants from the home country. For example, in the case of the Philippines, large Filipino migrant population exists in South American countries, European and Central Asia, East Asian and Pacific, North America and MENA countries. Large Filipino migrant population in the host countries, coupled with geographical proximity and low migration barriers are possible reasons for high volumes of informal remittance inflows and the resultant underestimation of bilateral remittances (Stanwix & Connell, 1995).

Another example we can be cited is of Mexico. In the case of Mexico, bilateral remittances from the North American region are higher with our model’s computation, across all the years. One of the reasons could be the highest number of migrants from this region. Other reasons are low barriers to migration and historical ties between Mexico and the North American countries. Similarly, in the case of Egypt, bilateral remittances from the European region are underestimated by Ratha and Shaw consistently over the years (Ghafar, 2017).

In such cases, the values computed by our model are higher than Ratha and Shaw’s estimates. One reason is that our method of estimation considers all types of migrants from all host countries. Therefore, in many cases, our higher computed values may also reflect the large settlements of migrants who send money to their home countries. In some cases, the values computed from our model are lower than those computed by Ratha and Shaw. We observe that the reason for this maybe because of our assumption of GDP per capita vis-a-vis Ratha and Shaw’s consideration of GNI per capita. Nonetheless, the top five remittance-receiving countries reported from our model and those the World Bank are largely the same in terms of aggregate remittances.

Table 3.

Comparison of Regional Bilateral Remittances

Country	Region	WB 2010	NM 2010	PD 2010 (%)	WB 2011	NM 2011	PD 2011 (%)	WB 2012	NM 2012	PD 2012 (%)	WB 2013	NM 2013	PD 2013 (%)
India	EAP	3.03	7.02	–79.38	3.56	9.49	–91.02%	3.91	5.82	–39.11	11.32	11.22	0.89
	ECA	5.75	8.69	–40.67	6.65	11.27	–51.47%	7.32	10.85	–38.85	5.72	10.14	–55.75
	LAC	0.03	0.02	66.43	0.04	0.04	8.67%	0.04	0.01	122.31	0.06	0.04	38.32
	MENA	25.15	24.08	4.36	29.25	35.61	–19.61%	32.20	48.24	–39.89	36.89	37.50	–1.63
	NA	12.07	13.71	–12.72	14.01	19.59	–33.20%	15.42	27.02	–54.68	13.86	12.93	6.94
	SAS	6.78	0.50	172.54	8.06	0.50	176.85%	8.88	0.32	186.22	1.53	0.12	169.97
	SUBA	0.20	0.02	171.37	0.24	0.02	163.11%	0.26	0.02	173.09	0.58	0.04	173.05
China	EAP	31.00	36.07	–15.11	34.49	25.32	30.67%	33.86	26.80	23.28	4.50	27.50	–143.74
	ECA	4.95	8.60	–53.90	13.87	9.81	34.28%	13.62	9.80	32.61	78.66	12.53	145.02
	LAC	0.34	0.04	157.30	64.81	0.03	199.80%	63.62	0.05	199.72	6.98	0.04	197.50
	MENA	0.37	0.00	196.86	0.08	0.01	170.34%	0.08	0.01	158.37	0.22	0.01	186.81
	NA	14.80	18.83	–23.95	0.08	18.81	–198.38%	0.08	18.52	–198.38	2.09	22.35	–165.79
	SAS	0.18	0.01	187.70	2.03	0.01	198.57%	2.00	0.01	198.43	0.54	0.36	40.13
	SUBA	0.12	0.06	64.80	5.87	0.04	197.30%	5.77	0.01	199.20	24.72	0.03	199.55
Mexico	EAP	0.01	0.02	–90.86	0.01	0.03	–94.94%	0.01	0.03	–110.14	0.02	0.04	–84.41
	ECA	0.17	0.25	–37.78	0.18	0.31	–51.30%	0.18	0.33	–59.59	0.18	0.67	–115.31
	LAC	0.09	0.00	185.30	0.10	0.00	188.29%	0.10	0.00	185.22	0.12	0.01	169.04
	MENA	21.80	21.07	3.39	23.29						0.00	0.00	39.49
	NA					22.00	5.70%	22.92	23.05	–0.57	22.71	26.73	–16.30
	SAS
	SUBA										0.00	0.00	177.59
Philippines	EAP	3.36	6.44	–62.86	3.66	6.81	–60.33%	3.88	7.63	–65.20	5.07	50.28	–163.38
	ECA	2.11	3.17	–40.16	2.25	3.66	–47.63%	2.39	3.61	–40.93	2.04	21.17	–164.90
	LAC	0.01	0.01	17.30	0.01	0.01	44.48%	0.01	0.01	46.13	0.05	0.11	–79.35
	MENA	4.21	3.50	18.33	4.57	4.63	–1.26%	4.85	5.02	–3.55	8.50	8.01	5.90
	NA	11.12	16.19	–37.13	11.92	16.70	–33.43%	12.64	17.26	–30.96	11.04	4.49	84.33
	SAS										0.00	0.00	187.51
	SUBA										0.01	0.00	168.83
Egypt	EAP	0.01	0.04	–105.92	0.01	0.05	-104.35%	0.02	0.05	–73.02	0.25	0.67	–92.07
	ECA	1.00	2.60	–88.79	1.13	2.93	–88.58%	1.62	2.69	–49.63	1.29	2.29	–55.77
	LAC	0.01	0.00	163.04	0.02	0.00	166.84%	0.02	0.00	176.53	0.01	0.00	93.54
	MENA	9.05	10.31	–12.96	10.26	12.87	–22.55%	14.70	12.79	13.88	14.95	18.60	–21.72
	NA	0.73	1.97	–92.45	0.82	2.04	–85.32%	1.18	1.95	–49.34	1.15	2.39	–69.74
	SAS	1.16	0.83	32.85	1.52	0.20	153.14%	2.18	0.75	97.94	0.00	0.00	164.53
	SUBA										0.18	0.01	180.84

Country	Region	WB 2014	NM 2014	PD 2014 (%)	WB 2015	NM 2015	PD 2015 (%)	WB 2016	NM 2016	PD 2016 (%)
India	EAP	11.56	11.49	0.56	7.09	10.74	–40.99	6.59	10.59	–46.63
	ECA	5.82	12.05	–69.73	6.13	12.05	–65.10	5.65	10.07	–56.30
	LAC	0.06	0.05	13.22	0.06	0.03	79.18	0.05	0.04	7.61
	MENA	36.84	36.36	1.29	13.74	23.41	–52.04	35.05	33.40	4.80
	NA	13.96	10.35	29.72	14.65	9.67	40.93	13.28	9.48	33.34
	SAS	1.57	0.11	173.05	1.21	0.06	182.26	1.15	0.05	182.79
	SUBA	0.58	0.05	170.82	0.60	0.05	172.12	0.58	0.05	168.18
China	EAP	4.68	25.39	–137.76	4.81	39.59	–156.63	4.60	39.37	–158.15
	ECA	84.68	12.21	149.60	86.82	10.38	157.29	82.88	8.11	164.33
	LAC	25.17	0.04	199.42	25.79	0.03	199.58	24.60	0.02	199.65
	MENA	0.38	0.01	193.89	0.38	0.01	194.45	0.37	0.01	194.04
	NA	2.32	22.18	–162.11	2.41	20.33	–157.61	2.25	20.28	–160.12
	SAS	0.57	0.18	105.01	0.58	0.15	119.45	0.57	0.09	143.30
	SUBA	6.01	0.03	198.28	6.17	0.02	198.86	5.88	0.02	198.69
Mexico	EAP	0.02	0.03	–69.05	0.02	0.03	–54.41	0.02	0.01	68.16
	ECA	0.19	0.72	–116.14	0.20	0.67	–106.54	0.22	0.46	–68.72
	LAC	0.13	0.01	168.27	0.13	0.01	170.31	0.15	0.00	187.84
	MENA	0.00	0.00	38.25	0.00	0.00	40.62	0.01	0.01	–29.53
	NA	24.12	29.00	–18.37	25.81	31.36	–19.42	28.27	31.32	–10.23
	SAS
	SUBA	0.00	0.00	174.97	0.00	0.00	177.97	0.00	0.00	182.90
Philippines	EAP	5.40	50.38	–161.26	5.43	43.50	–155.60	6.00	45.28	–153.18
	ECA	2.17	22.09	–164.21	2.18	20.63	–161.69	2.41	20.90	–158.70
	LAC	0.05	0.11	–76.76	0.05	0.11	–72.70	0.05	0.01	127.07
	MENA	9.04	8.09	11.15	9.06	6.50	32.88	9.87	7.42	28.39
	NA	11.73	4.36	91.63	11.74	3.09	116.62	12.79	3.16	120.71
	SAS	0.00	0.00	187.45	0.00	0.00	183.88	0.01	0.00	185.50
	SUBA	0.01	0.00	174.08	0.01	0.00	179.24	0.01	0.00	184.79
Egypt	EAP	0.27	0.59	–73.32	0.26	0.53	–69.76	0.23	0.47	–66.87
	ECA	1.42	2.28	–46.56	1.34	1.96	–38.06	1.22	1.99	–47.64
	LAC	0.01	0.00	99.60	0.01	0.00	93.67	0.01	0.00	106.53
	MENA	16.40	17.65	–7.31	15.35	14.58	5.13	13.88	18.94	–30.85
	NA	1.27	2.42	–62.24	1.19	2.53	–71.81	1.08	2.54	–80.72
	SAS	0.00	0.00	166.22	0.00	0.00	153.15	0.00	0.00	149.13
	SUBA	0.20	0.01	188.36	0.18	0.01	186.97	0.17	0.00	192.58

Source: The world bank and authors’ own calculation.

Note: Regions are delineated according to the categorization by World Bank. EAP: East Asia and Pacific; ECA: Europe and Central Asia; LAC: Latin American and Caribbean; MENA: Middle East and North Africa; NA: North America; SAS: South Asia; SUBA: Sub-Saharan Africa; PD: percentage difference.

Conclusion

This study attempts to provide a new method for computing bilateral and aggregate remittances using the Life Cycle Hypothesis of Savings (LCHS). Our method seems to be an improvement over Ratha and Shaw method. Bilateral and aggregate remittances data, based on Ratha and Shaw method, is comprehensive but said to be inaccurate. We, through our model, attempt to address the inaccuracy by the following modifications with respect to our analytical framework: considering GDP per capita instead of GNI per capita (controlling for overestimation), considering all types of migrants from all destinations (controlling for underestimation) and developing an analytical framework based on the motives to remit.

As discussed previously, bilateral remittances are an important link between demand and supply in the international labour markets. However, the lack of accurate bilateral remittance data is an obstacle for such analysis. Specific host countries demand labour with specific skills from migrants. If migrants can attain these specific skills, the respective bilateral remittance inflows can be increased, provided home country policymakers work towards the same. Such human capital formation, in turn, can accrue the benefits of economic growth and higher productivity for the home country as well (Panda, 2017).

Other than supporting consumption expenditures at the household level in the home country, remittances are also invested in education, health, business formation and so on. in many remittance-receiving countries. Therefore, increasing remittances inflows is an important policy decision for these countries (Arif et al., 2019; Kifle, 2007). Policymakers, for example, can delve on encouraging migrant remittances by providing education subsidies and formalizing their education-oriented public policies (Ambler et al., 2015). This would encourage skill development necessary to migrate to specific countries in order to increase corresponding bilateral remittance inflows. Nevertheless, one bottleneck for remittance transfers is the high costs of remittances. These need to be reduced in order to make remittances more frequent to the recipients in dire need. This depends on the level of financial development in these remittance-receiving countries because we observe that high remittance costs and huge volumes of informal remittances are features of countries with less developed financial systems.

The computation of aggregate and bilateral remittances from our model also delve on the same aspect. As long as remittances are expensive to transfer, there will be informal channels propagating for remittances. Such informal transfers would not only make a significant volume of remittances difficult to account for, but also curb any benefits of formal investment channels of remittances. Our values of bilateral remittances make a significant contribution in terms of including a probable unaccounted component of remittances. Our analytical framework, on the other hand, provides a general model that can be used to compute bilateral remittances for any remittance-receiving country. However, we are limited by the availability of regular bilateral migration data. The bilateral migration data that is reported by the World Bank is available with a lag of three years. For example, the 2016 bilateral migration data has information for the year 2013. Despite these limitations, we feel that our model can minimize the inaccuracy in the aggregate and bilateral remittances and can be used as a substitute database for computing aggregate and bilateral remittances for most of the countries.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article

Notes

Appendix 1

References

Acosta

P. A.

Baerg

N. R.

Mandelman

F. S.

(2009). Financial development, remittances, and real exchange rate appreciation. Economic Review , 94, ISSN 0732-1813, Federal Reserve Bank of Atlanta, Atlanta, GA. https://www.econstor.eu/bitstream/10419/57669/1/602140013.pdf

Adzakey

(2013). Contribution of remittances to the care of the elderly in Ablekuma south (Dissertation, University of Ghana, Legon). http://ugspace.ug.edu.gh/bitstream/handle/123456789/5306/Georgina%20Yayra%20Adzakey_Contribution%20of%20Remittances%20to%20the%20Care%20of%20the%20Elderly%20in%20Ablekuma%20South_2013.pdf?sequence=1&isAllowed=y

Agapova

McNulty

J. E.

(2015). Interest rate spreads and emerging markets: An application to the banking systems of the transition economies of central and eastern Europe . Paper presented at FMA Conference, Orlando.

Aggarwal

Demirgüç-Kunt

Peria

M. S.

(2011). Do remittances promote financial development. Journal of Development Economics, 96(2),255–264.

Aguilar

(2009). Labour migration and ties of relatedness: Diasporic houses and investments in memory in a rural Philippine village. Thesis Eleven , 98(1), 88–114. https://journals-sagepub-com-s.web.bisu.edu.cn/doi/abs/10.1177/0725513609105485

Ahmed

Martinez-Zarzoso

(2014). What drives bilateral remittances to Pakistan: gravity model approach (Discussion Paper No. 209). CEGE. https://www.econstor.eu/bitstream/10419/97486/1/786982659.pdf

Alvarez

S. P.

Briod

Ferrari

Rieder

(2015). Remittances: How reliable are the data? Migration Policy Practice , V(2), 42–46.

Ambler

Aycinena

Yang

(2015). Channeling remittances to education: A field experiment among migrants from El Salvador. American Economic Journal: Applied Economics , 7(2), 207-232. doi:10.1257/app.2014001.

Amuedo-Dorantes

Pozo

(2009). New evidence on the role of remittances on health care expenditures by Mexican households (Discussion Paper No. 4617). IZA. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1522685

10.

Ando

Modigliani

(1963). The ‘life cycle’ hypothesis of saving: Aggregate implications and tests. The American Economic Review , 53(1), 55–84.

11.

Arif

Raza

Friemann

Suleman

(2019). The role of remittances in the development of higher education: Evidence from top remittance receiving countries. Social Indicators Research , 141, 1233–1243. doi:10.1007/s11205-018-1857-.

12.

Azizi

(2017). Altruism: Primary motivation of remittances. Applied Economics Letters , 24(17), 1218–1221.

13.

Barajas

Chami

Fullenkemp

Gapen

M. M.

(2009). Do workers’ remittances promote economic growth ? (Working Paper No. WP/09/153). IMF. https://www.imf.org/external/pubs/ft/wp/2009/wp09153.pdf

14.

Catrinescu

Leon-Ledesma

Piracha

Quillin

(2006). Remittances, institutions and economic growth (Discussion Paper No. 2139). IZA. http://ftp.iza.org/dp2139.pdf

15.

Charsley

Storer-Church

Benson

Van

N. (Winter 2012).

Marriage-related migration to the UK. The International Migration Review , 46(4), 861–890. www-jstor-org.web.bisu.edu.cn/stable/41804866

16.

Chen

Cox

(2011). Six adults and one child in China. New Geography. https://www.newgeography.com/content/002474-six-adults-and-one-child-china

17.

Choudhry

I. S.

Faridi

M. Z.

Abbas

Bashir

(2010). Short run and long run saving behavior in Pakistan: An empirical investigation. Journal of Money, Investment and Banking , 16, 57–66.

18.

Connell

Conway

(2000). Migration and remittances in island microstates: A comparative perspective on the South Pacific and the Caribbean. International Journal of Urban and Regional Research , 24(1), 52–78. http://onlinelibrary.wiley.com/doi/10.1111/1468-2427.00235/abstract

19.

Danziger

Van Der Gaag

Smolensky

Taussig

M. K.

(Winter 1982–1983). The life-cycle hypothesis and the consumption behavior of the elderly. Journal of Post Keynesian Economics , 5(2), 208–227.

20.

De Sousa

Duval

(2010). Geographic distance and remittances in Romania: Out of sight, out of mind? Economie Internationale , 121(1), 81–97.

21.

Docquier

Rapoport

Salomone

(2012). Remittances, migrants’ education and immigration policy: Theory and evidence from bilateral data. Regional Science and Urban Economics , 42(5), 817–828.

22.

Easterly

Kraay

(2000). Small states, small problems? income, growth, and volatility in small states. World Development , 28(11), 2013–2027.

23.

Frankel

(2010). Are bilateral remittances countercyclical (Faculty Research Working Paper Series, RWP10-037). https://dash.harvard.edu/bitstream/handle/1/4450131/Frankel_AreBilateral.pdf?sequence=1&isAllowed=y

24.

Freund

Spatafora

(2005). Remittances: Transaction costs, determinants, and informal flows (World Bank Policy Research Working Paper No. 3704). http://documents.worldbank.org/curated/en/610101468141260179/Remittances-transaction-costs-determinants-and-informal-flows

25.

Ghafar

A. A.

(2017). Egypt’s long-term stability and the role of the European Union . Brookings.

26.

Giuliano

Ruiz-Arranz

(2006). Remittances, financial development, and growth . (IZA Discussion Papers, No. 2160). Institute for the Study of Labor (IZA).

27.

Gong

Meng

Leigh

(2012). Intergenerational income mobility in urban China. Review of Income and Wealth , 58(3), 481–503.

28.

Graham

J. W.

(1987). International differences in saving rates and the life cycle hypothesis. European Economic Review , 31(8), 1509–1529.

29.

Gupta

Pattillo

Wagh

(2009). Effect of remittances on poverty and financial development in Sub-Saharan Africa. World Development , 37(1), 104–115. https://sciencedirect.com/science/article/abs/pii/s0305750x08001903

30.

Holden

Shiferaw

Pender

(2004). Non-farm income, household welfare, and sustainable land management in a less-favoured area in the Ethiopian highlands. Food Policy , 29(4), 369–392.

31.

Howell

M. M.

Whittaker

W. L.

Gebremedhin

T. G.

Schaeffer

(2012). Examining the bidirectional relationship between entrepreneurship and economic growth: Is entrepreneurship endogenous. In Burger-Helmchen

Thierry

(Ed.), Entrepreneurship - Born, Made and Educated (pp. 283–317). Intech. https://doi.org/10.5772/35861

32.

Ikels

(2006). Economic reform and intergenerational relationships in China. Oxford Development Studies , 34(4), 387–400.

33.

Jappelli

Modigliani

(2003). The age saving profile and the life-cycle hypthesis (Working Paper No. 9). Center for Studies in Economics and Finance. https://www.researchgate.net/publication/23573835_The_Age-Saving_Profile_and_Life-Cycle_Hypothesis

34.

Kessel

R. A.

(1956). Inflation-caused wealth redistribution: A test of a hypothesis. The American Economic Review , 46(1), 128–141. www-jstor-org.web.bisu.edu.cn/stable/1816503

35.

Kifle

(2007). Do remittances encourage investment in education? Evidence from Eritrea. Journal of African Studies , 4(1). https://espace.library.uq.edu.au/view/uq:170809

36.

Krueger

Perri

(2002). Does income inequality lead to consumption inequality? Evidence and theory (Working Paper No. 9202). National Bureau of Economic Research. https://ideas.repec.org/p/nbr/nberwo/9202.html

37.

Leuth Ruiz-Arranz

(2008). Determinants of bilateral remittance flows. The B.E. Journal of Macroeconomics , 8(1), 1–23.

38.

Lew Arvin. (2012). Do happiness and foreign aide affect bilateral migrant remittacnes. Journal of Economic Studies , 39(2), 212–230.

39.

Shin

H. B.

(2013). Intergenerational housing support between retired old parents and their children in urban China. Urban Studies , 50(16), 3225–3242.

40.

Lucas

R. E.

Stark

(1985). Motivations to remit: Evidence from Botswana. Journal of Political Economy , 93(5), 901–918.

41.

Mikesell

R. F.

Zinser

J. E.

(1973). The nature of the savings function in developing countries: A survey of the theoretical and empirical literature. Journal of Economic Literature , 11(1), 1–26.

42.

Modigliani

(1993). Recent declines in saving rate: a life cycle perspective. In Baldassarri

Paganetto

Phelps

E. S.

(Eds.), World Saving, Prosperity and Growth (pp. 249–286). Macmillan.

43.

Nangia

E. N.

(2015). Energizing the elderly through remittances: opportunities for active ageing in Cameroon. International Journal of Innovative Research and Development , 4(1), 197–203.

44.

Özden

Ç.

Parsons

C. R.

Schiff

Walmsley

T. L.

(2011, June). Where on earth is everybody? The evolution of global bilateral migration 1960–2000. World Bank Economic Review , 25(1), 12–56. https://doi.org/10.1093/wber/lhr024

45.

Panda

(2017). Schooling and productivity growth: Evidence from a dual growth accounting application to U.S. states. Journal of Productivity Analysis , 48(2), 193–221. https://link-springer-com-443.web.bisu.edu.cn/article/10.1007/s11123-017-0516-4

46.

Ratha

Shaw. (2007). South-south migration and remittances (Working Paper No.102). World Bank. http://siteresources.worldbank.org/INTPROSPECTS/Resources/334934-1110315015165/SouthSouthMigrationandRemittances.pdf

47.

Sana

Massey

D. S.

(2005). Household composition, family migration and community context: migrant remittances in four countries. Social Science Quarterly , 86(2), 509–528.

48.

Schiopu

Siegfried

(2006). Determinants of workers’ remittances: Evidence from the European neighboring region . (Working Paper Series No. 688). European Central Bank. https://www.econstor.eu/bitstream/10419/153122/1/ecbwp0688.pdf

49.

Silverstein

(2011). Intergenerational exchange between parents and migrant and nonmigrant sons in rural China. Journal of Marriage and Family , 73(1), 93–104.

50.

Stanwix

Connell

(1995). To the islands: the remittances of Fijians in Sydney. Asian and Pacific Migration Journal , 4(1), 69–87. https://journals-sagepub-com-s.web.bisu.edu.cn/doi/abs/10.1177/011719689500400104

51.

Subedi

B. P.

(1991). International migration in Nepal: Towards an analytical framework. Contributions to Nepalese Studies , 18(1), 83–102.

52.

Thøgersen

(2015). Population ageing and capital accumulation: A simple OLG model with PAYGO pensions. Theoretical Economics Letters , 5, 155–162.

53.

UNESCAP. (2015). International labour migration statistics: A guide for policymakers and statistics organizations in the pacific . Author.

54.

D. M.

(2016). Philippines ranks 9th in global migrants with 5 million abroad. Business World . https://www.bworldonline.com/content.php?section=Nation&title=philippines-ranks-9th-in-global-migrants-with-5-million-abroad&id=137282

55.

Yang

Choi

(2007). Are remittances insurance? Evidence from rainfall shocks in the Philippines. The World Bank Economic Review , 21(2), 219–248. https://academic-oup-com-s.web.bisu.edu.cn/wber/article/21/2/219/1916657

56.

Walmsley

Ahmed

S. A.

Parsons

(2005). A global bilateral migration data base: Skilled labor, wages and remittances . https://www.gtap.agecon.purdue.edu/resources/res_display.asp?RecordID=1880

57.

Wennekers

Thurik. (1999). Linking entrepreneurship and economic growth. Small Business Economics , 13, 27–56.

58.

Wolff

E. N.

(1979). The distributional effects of the 1969–75 inflation on holdings of household wealth in the united states. Review of Income and Wealth , 25, 195–207. doi:10.1111/j.1475-4991.1979.tb00093..

59.

Xue

Zhong

(2003). Unemployment, poverty and income disparity in urban China. Asian Economic Journal , 17(4), 383–405. https://doi.org/10.1111/j.1467-8381.2003.00192.x