Information about Middle East Oriente Medio

3. The Macroeconomic Impact of Migration

Raphael Espinoza, Ghada Fayad, and Ananthakrishnan Prasad
Published Date:
November 2013
  • ShareShare
Information about Middle East Oriente Medio
Show Summary Details

3.1 Introduction

The dominance of foreigners in the labor force of the GCC countries is a very peculiar characteristic of these economies and one that has many macroeconomic consequences. Migration in the region has been steadily increasing, reaching extreme levels recently with non-nationals constituting over 90 percent of the labor force in Qatar and the UAE in 2011, 50 percent of the labor force in Saudi Arabia in 2009, and 77 percent of total employment in Bahrain in 2010. Recent efforts to “nationalize” the labor force aimed at limiting the supply of foreign workers and increasing the demand for national labor in the private sector. Nevertheless, by 2010 the GCC was the third region of immigration in the world after North America and the EU, this remaining true despite a shortage of statistics about, and hence likely underestimation of, migrants in this region.1

This chapter, which will focus on the macroeconomic consequences of migration, argues that the massive influxes of foreign labor have helped the region avoid one the main elements of the resource curse: the Dutch disease effect of large oil windfalls, which involves an overappreciated real exchange rate and thereby a loss of competitiveness of the export sector. The GCC countries have indeed avoided strong appreciations of their exchange rates, although this moderation has not translated into a particularly positive performance in non-oil exports. In addition, measured as the evolution of the share of non-oil exports in total exports, diversification experiences have been far from homogeneous, with relatively more success in the UAE, Oman, and Bahrain. This chapter investigates the fundamental determinants of real exchange rates in the region, using estimation techniques that do not assume homogenous relationships.

Immigration has substantially relaxed supply shortages in the non-traded goods markets while demand has been contained via extremely restrictive immigration policies (such as various restrictions on property ownership, the absence of any naturalization process, and restricted family reunion). Theoretically, the adjustment process following an oil boom in a labor-importing country will depend on the potential effect of remittance outflows in alleviating the resulting upward pressure on the real exchange rate. Since imported workers employed in the non-tradable sector spend only part of their expenditures on non-traded goods (van Wijnbergen 1984) and remit a major part of it (especially in the GCC), further upward pressures on the real exchange rate are alleviated.

The reaction of remittances to commodity booms may therefore be of economic significance, though this link is largely unexplored in the empirical Dutch disease literature. In this chapter, we present novel evidence on the mitigating effect of international migration and remittances on the Dutch disease in the context of oil-rich and labor-poor countries.

We estimate an econometric model for the fundamental determinants of real exchange rates using a number of panel estimation techniques. We find, across our estimation methods, that the vast pool of foreign labor in the MENA net labor importers and the resulting large remittance outflows they generate act as a stabilizing factor by easing down the otherwise upward pressure on the real exchange rates caused by large oil windfalls. More specifically, we find no evidence of Dutch disease generated by large oil windfalls: the coefficient on oil export revenues is mostly negative and statistically insignificant. At the same time, the coefficient on remittance outflows is consistently negative and statistically significant across all estimation methods.

The rest of this chapter is structured as follows. Section 3.2 highlights some important facts about the GCC labor markets, including the extent of immigration and issues concerning migrants’ rights, labor market segmentation, labor force growth, and unemployment. Section 3.3 discusses diversification in the GCC. Section 3.4 presents a simple theoretical model of Dutch disease in an oil-rich labor-importing economy. Section 3.5 discusses the estimator of choice and econometric model for our panel, including a treatment of cross-sectional error dependence. Section 3.6 presents our estimation results on the real exchange rate effect of remittance outflows and oil export revenues in the GCC countries. Section 3.7 concludes.

3.2 Background on GCC Labor Markets

There are two dominant features about GCC labor markets that distinguish them from the rest of MENA: (i) their high reliance on foreign (skilled and unskilled) workers, and the pervasive segmentation of labor markets between foreign workers in private jobs and nationals in public-sector jobs,2 and (ii) the existence of large wage disparities between foreigners and nationals in the private sector, as well as between private- and public-sector wages for nationals. While these features have been a common characteristic in all GCC countries, the specific labor market issues they entailed have differed across GCC countries: in countries with relatively large indigenous populations such as Oman and Saudi Arabia, the limited scope for continued expansion in public-sector employment coupled with growing labor forces have put youth unemployment as a main concern. In these countries, governments initiated ambitious private-sector nationalization plans. However, in countries with small populations like Qatar, unemployment levels in general have been at record lows. Another common feature is very low female labor force participation rates.

3.2.1 Immigration and Remittances

In the GCC, “the employment of large numbers of foreigners has been a structural imperative […], as the oil-related development depends upon the importation of foreign technologies and requires knowledge and skills alien to the local Arab population” (Kapiszewski 2006: 2). Migrant workers have indeed become a structural feature of GCC economies, given the long-term infrastructure development plans.

Immigrants’ Rights

Although most migrants benefit from the work opportunities given in the GCC, international organizations have noted some issues with the treatment of migrants and made recommendations to improve their treatment. The International Labor Organization (ILO) has called for a reform of the sponsorship system, where workers’ visas depend on the employer—giving employers control over expatriate workers. This system ties migrant workers to employers and fosters abuses (IOM 2012). Many workers also arrive in the GCC with large debts due to “recruitment commissions” charged by intermediaries. In addition, working conditions and salaries are often unspecified (Plant 2008). The combination of employer control, large debts, and unspecified income, can lead to situations tantamount to forced labor.

Migrants, the destination countries, and the countries of origin in general benefit from these jobs and from the savings and remittances generated by the work opportunities. Survey data shows that the majority of migrant workers benefit from significantly improved economic conditions compared to what they earn in their country of origin (Plant 2008). Thus, a comprehensive approach towards ensuring migrants’ rights and their fair treatment in legal systems and policies would benefit all stakeholders. Progress on this front has however been slow, despite improvements made in some of the GCC countries recently (IOM 2012). In addition, to facilitate transparency in the setting of wage conditions, the ILO has recommended that the GCC countries introduce a fair minimum wage, in line with international labor principles.

Composition of Foreign Workforce

The composition of the foreign labor force in the GCC countries (mostly Arab and South and Southeast Asian workers) has evolved throughout the years. While Arabs from neighboring countries dominated the Gulf foreign workforce at the beginning of the oil era owing to cultural, religious, and linguistic factors, the balance is now leaning in favor of Asian workers, in what is referred to in the literature as “de-arabization” of the Gulf labor market. The reasons for this new stream of migration relate both to favorable characteristics of the Asian workforce and increased detachment towards non-GCC Arabs.3Kapiszewski (2006) estimates that while Arabs constituted 72 percent of the immigrant populations across all GCC countries in 1975, this share declined to only 32 percent by 2004. Between 1975 and 2004, the share of Arabs in the foreign population declined from 22 to 15 percent in Bahrain, from 80 to 30 percent in Kuwait, from 16 to 6 percent in Oman, from 33 to 19 percent in Qatar, from 91 to 33 percent in Saudi Arabia, and from 26 to 13 percent in the UAE.

The skills composition of the foreign workforce has been diverse. The GCC countries, known for hosting a large number of low-skilled construction workers, have also successfully attracted a large number of highly skilled foreign laborers (Table 3.1). A recent UN survey indicated that all GCC countries are aiming at maintaining high-skilled immigration, despite plans to lower immigration levels in general and immigration of temporary workers in particular (Table 3.2). However, controlling unskilled immigration in countries with large, labor-intensive, infrastructure projects is challenging when the local population is small, well-paid, and unwilling to accept unskilled jobs.

Table 3.1.Share of high-skilled migrants by region in 2000
Table 3.2.Government views and policies on immigration, 2009
BahrainKuwaitOmanQatarSaudi ArabiaUAE
ViewSatisfactoryToo highSatisfactoryToo highToo highToo high
Permanent settlementLowerNo InterventionNo InterventionNo InterventionLower
Temporary workersLowerLowerLowerLowerLowerMaintain
Highly skilled workersMaintainMaintainMaintainMaintainMaintain
Family reunificationMaintainMaintainRaiseMaintainMaintainLower
Integration of non-citizensYesNoYesYes
Source: UN World Population Policies (2009)
Source: UN World Population Policies (2009)


Reflecting the pattern of migrants and their high propensity to remit, remittances from the GCC in 2009 amounted to about 61 billion US dollars, 20 percent of world remittance outflows, and have constituted an important source of income for many countries in the region (see Chapter 9). Over the last decade, the GCC countries ranked among top source countries in the world, in gross terms as well as in percent of GDP (Figure 3.1). In the GCC, immigrants’ incentives to remit are intensified by the countries’ strict immigration rules: migrants are considered as temporary workers (even if they are essentially permanent migrants) with no potential for any naturalization process, and are thus not allowed to fully integrate in the local economy because of various restrictions on family reunion, property ownership, and rights.4 The marginal propensity to remit in the Gulf is high for both skilled and unskilled migrants. Unskilled migrants are normally associated with higher propensity to remit as they are financially constrained, travel often alone, and leave their families back home. Docquier et al. (2011) analyze the relationship between migrants’ education, the restrictiveness of immigration policies in migration destinations, and remittances. They find that skilled migrants remit more from the GCC compared to other advanced host countries. More specifically, they find that immigration policies conducted in the Gulf make the skills ratio more effective at increasing remittance outflows, while policies conducted in European countries act to reduce the amount of remittances sent home by skilled migrants.

Figure 3.1.Remittance outflows in percent of GDP, average 2000–10

Source: World Development Indicators

3.2.2 Labor Market Segmentation and Wage Disparities

Labor markets in the GCC are characterized by a high degree of segmentation between the private and the public sectors and between nationals and expatriates. Foreign workers are overwhelmingly employed in the private sector and this has been the case for several decades. For instance in 2010, the share of foreigners in private-sector employment was 90 percent in Saudi Arabia and 80 percent in Bahrain. Nationals, by contrast, mainly work in the public sector, with different proportions across countries and over time (Table 3.3). In Kuwait and Qatar, the public sector accounts for close to 90 percent of total employment of nationals. In those countries, nationals also represent the smallest share of the total population. In Bahrain, Oman, and Saudi Arabia, the share of nationals employed in the public sector is lower and unemployment is also more of a concern. These employment patterns reflect differences in oil wealth and varying degrees of saturation in public-sector employment: given their small national population and their surpluses in fiscal accounts, Qatar and Kuwait can easily offer all nationals public-sector jobs.

Table 3.3.Public-sector employment as a share of total employment of nationals in the GCC
Saudi Arabia70827372

Significant wage disparities exist between foreigners and nationals in the private sector, and between public- and private-sector wages for nationals. Detailed wage data for Saudi Arabia show that nationals are paid significantly more than foreign workers in the private sector, and that these wage differentials decrease as the skill level of nationals and non-nationals increases. This pattern in wage gaps is even more pronounced when skills level is defined by occupation: among agricultural workers Saudi wages are nine times higher than those of expatriates, while among managers and directors there is essentially no wage gap (Figure 3.2). Wage disparities for nationals between public-and private-sector jobs are also a function of education. Differences in job security and work hours notwithstanding, for high-skilled employees, the public sector is not necessarily more lucrative than the private sector. A recent graduate with a bachelor’s degree would typically be hired at a low-rank public-sector job and earn about 6,500 Saudi Riyals (SR) a month, less than the SR7,700 average wage for similarly educated Saudis in the private sector. For Saudis with secondary or lower education, however, the lowest-paying public-sector job pays about 30 percent more than a private-sector job.

Figure 3.2.Saudi to non-Saudi monthly wages in the private sector by education, 2009 (ratio)

Source: Saudi Arabia Ministry of Labor Statistics of 2009

Similar patterns exist in Bahrain where nationals are paid two to three times more than foreigners in the private sector, and increasingly so over the last decade. On average, over the last decade, public-sector jobs paid Bahrainis 1.7 times more than private-sector jobs. However, foreigners are paid about as much as Bahrainis in the public sector. The ratio of public to private-sector wages for foreigners has been consistently high over recent years, and increasingly so, reaching six in 2010 (Figure 3.3).

Figure 3.3.Wage ratios by sector and nationality in Bahrain, 2002–10

Source: Bahrain Labor Markets Regulatory Authority

Wage disparities are also a function of gender, with the wage gaps between Saudis and expatriates consistently higher for males than for females. While Saudi males on average earned three times as much as expatriates in 2010, for females the average ratio was only 1.5 and in several job categories the ratio was below one.

3.2.3 Unemployment

GCC total unemployment rates (for both nationals and expatriates) have been contained, below 5 percent for most countries according to the International Labor Organization (ILO) (Figure 3.4). To a large extent, this reflects low unemployment rates for expatriates. Available data from national sources show that in 2011 the unemployment rate for Qataris, while still low at 3.9 percent, is eight times larger than the rate for both Qataris and expatriates, and in Saudi Arabia the unemployment rate of nationals is twice as large at 10.5 percent. High levels of unemployment do exist however and are concentrated among two categories of jobseekers: youth and women. Qatar is the only country in the region where youth unemployment is recorded at below 10 percent. In Saudi Arabia, for instance, high youth unemployment, which has averaged around 25 percent over the last decade, is the result of a combination of growing populations and thus labor forces (with the Saudi labor force growing 1.5 times over 1999–2009), saturation of public-sector jobs, and private-sector job creation that has mainly benefitted foreigners. Youth unemployment is also near 30 percent in Bahrain.

Figure 3.4.Youth and total unemployment rates (in percent, latest available date)

Source: International Labor Organization KLIM database

Women represent the other main group of unemployed. Figures from Qatar and Saudi Arabia show nationals’ unemployment rates for women about four times as high as for men in recent years, and this difference dates back at least a decade. Along with high unemployment, female labor force participation rates have been very low in some countries in the region, for instance at 12 percent for Saudi females and 34 percent for Qatari females on average over the last five years.5 The more available country-wide rates including both national and expatriate females show significantly higher participation rates, as high as 52 percent in Qatar in 2011, 43 percent in Kuwait and 39 percent in Bahrain in 2010.

3.2.4 Nationalization

While all GCC countries’ development plans target higher employment of nationals in the private sector through ambitious nationalization plans, these plans have not yet been successful in “nationalizing” the private sector.

Indeed, much of the difficulty in achieving the GCC’s nationalization objectives relates to a low degree of substitutability between national and foreign workers (Fasano and Goyal 2004). Reasons for the low substitutability include a shortage of highly skilled nationals, an almost perfectly elastic supply of cheap foreign labor, a general aversion among nationals to taking menial jobs, and the role of the public sector as the dominant employer of nationals (see also Chapter 4). These factors have all contributed to generating large differences in pay between nationals and foreigners and to cementing the segmentation between the two categories of workers (Abdalla et al. 2010). They have also allowed for the coexistence of high youth unemployment alongside rapid job growth in the private sector.

Up until recently, nationalization policies in the GCC have been basically immigration policies aiming at affecting the quantity and cost of hiring foreign workers through a combination of mandatory and administrative measures. Examples are quotas on the proportion of nationals employed by private companies in specific professions or sectors and time-specific cash benefits for employing nationals. These measures however have proven difficult to monitor and implement, especially given the unlimited access of private companies to foreign labor at internationally competitive wages. Other measures have aimed at increasing the relative cost of hiring expatriates, such as regulating the supply of work permits for foreigners and imposing fees for use of expatriate labor.

More recently, immigration policies have become part of a broader effort to increase employment of nationals in some GCC countries, and it is now recognized that successful nationalization of the private workforce will also require new job creation through diversification, such as the promotion of small and medium-sized enterprises (SMEs).6 For instance in Saudi Arabia, several labor market initiatives were co-launched in 2011, including the “Nitaqat” program, which is the updated Saudization policy. Nitaqat differs from past Saudization schemes in that (i) the percentage of Saudis required to be employed by each company is based on its area of activity and the size of its workforce, (ii) many SMEs are included in Nitaqat, as all Saudi companies with at least ten employees are required to participate, and (iii) Nitaqat involves harmonization between several government bodies to improve compliance.7

3.3 Diversification in the GCC

Before we provide evidence on whether the GCC economies have been harmed by a Dutch disease effect, we review the performance and composition of their non-oil sectors. As a share of total exports, non-oil exports in the GCC have remained quite low, mostly below 20 percent, though there was some improvement in the last two decades in Oman and in the UAE (Figure 3.5). The performance of non-oil exports looks more promising when expressed in percent of non-oil GDP: the share increased in Kuwait, Oman, Saudi Arabia, and the UAE (though it remains below 20 percent), and even exceeded 100 percent in Oman.

Figure 3.5.Share of non-oil exports in total exports (in percent)

Source: IMF and authors’ calculations

What are GCC countries exporting other than hydrocarbons? With the exception of Oman and the UAE, around 50 to 60 percent of non-oil exports in the GCC are petrochemicals (Figure 3.6). It is important to single out petrochemical industries because this sector is heavily subsidized by the hydrocarbon sector and thus its performance is artificially boosted by the subsidy policies (see Chapter 4 for a discussion on subsidies). Other non-oil exports include agricultural (live animals, animal products, vegetables products, prepared foods, and beverages), mineral products, base metals, electrical machinery in most GCC countries, in addition to free-zone exports in Dubai. Qatar and the UAE are also examples of countries that have managed to have diversified services export sectors (transportation, including airlines, tourism related to international sports events, etc.).

Figure 3.6.Share of petrochemical exports in total non-oil exports (in percent)

Source: IMF and authors’ calculations

3.4 Simple Theoretical Model

In this section, we develop a simple theoretical model to illustrate the mechanism through which immigration and remittances help alleviate potential real exchange rate appreciation in the oil-exporting, labor-poor, GCC states. We augment the Dixit and Norman (1980) dual trade general equilibrium model allowing for migration of labor and remittances. We consider a resource-rich economy with the following equilibrium conditions (where subscripts denote partial derivatives):

Equation (1) is the income–expenditure equilibrium. The expenditure side is represented by the concave expenditure function e(q, u) expressed in terms of the relative price of non-traded with respect to traded goods q, and the aggregate welfare level u. The implicit assumption here is that the price of tradables is normalized to one so that everything is expressed in terms of the (non-resource) tradable good. The production side consists of a convex non-resource production function represented by g(q, L) and an inflow of foreign exchange resource rents represented by N. We are assuming here that oil production does not employ any resources, so the increase in oil revenue simply shows up as an increase in transfers received from abroad. The labor force L consists of the indigenous population I and the immigrant workers M. Changes in the labor force are driven only by changes in the inflow of foreign labor so that dL=dM (since dl = 0), which is in turn driven by resource rents, i.e. dM=MNdN where MN>0. The latter effect reflects the long-standing GCC policy response to oil price shocks of increased demand for and import of foreign workers, as well as a supply-side “Alberta effect” where a booming oil sector attracts immigrants seeking to share the rents (Corden 1984).

Equation (2) reflects equilibrium in the non-traded goods market. As the first derivative of the expenditure function with respect to q represents demand for non-traded goods, and similarly the first derivative of the production function with respect to q represents the supply of non-traded goods,8equation (2) simply says that the non-traded goods market clears. With the economy’s excess demand for the tradable good always met by the rest of the world, excess demand for all goods becomes the excess demand for non-tradables. The real exchange rate (the relative price of the non-tradable good) adjusts to clear the non-tradable goods market. The final equation represents equilibrium real product wage determination under perfect competition and assumes that the marginal product wage of migrants is lower than that of the locals. This seems to be a stylized fact in GCC countries with (almost perfectly) segmented labor markets where nationals are mainly employed in the highly paid public sector, and migrants are distributed across construction and other sectors based on their skill levels. However, we assume away any heterogeneity in the migrants’ wages. The non-traded sector in our setup is labor-intensive, hence gqM>0 by Rybczynski’s theorem (in a two-sector model, an increase in the endowment M leads to a (more than proportional) increase in the supply of the good that was M-intensive). Finally, we assume for now that there are no remittance transfers by the GCC immigrants to their families in their home countries, and allow for this possibility in the next section.

Total differentiation of equations (1) and (2) and simple manipulations produce the following expression for the change in real exchange rate dq in terms of the change in the resource windfall (dN):9

where (φqq=qgqq/gq>0) is the elasticity of supply of non-traded goods with respect to the real exchange rate; (εqq=qeqq/eq<0) is the corresponding non-traded goods’ demand elasticity with respect to the real exchange rate; (η=eequ/eqeu>0) is the income elasticity of demand for non-traded goods; (λ=qeq/e>0) is the share of non-tradables in total expenditure; and (εqM=MgqM/gq>0) is the elasticity of supply of non-traded goods with respect to inflows of immigrant workers.

Equation (4) suggests that a resource windfall dN generates two opposite effects on the real exchange rate. First, there is a positive demand-side effect (first term in brackets) which is increasing in the income elasticity of demand for non-tradables η and in the share of non-tradables in total expenditure λ. This is the standard spending effect of a resource windfall where additional (public and private) spending partly falls on non-traded goods. In our model, such demand effect is exacerbated by the inflow of foreign workers who also consume non-traded goods, and is increasing in migrants’ wages gM and in the immigration response to increased windfalls MN. Second, a negative supply-side effect is at play since the oil boom is accompanied by an influx of foreign labor (second term in brackets). This depreciative supply-side effect is higher the more elastic is non-tradable supply with respect to increases in immigrant workers εqM and the greater their productivity gq/M. In the GCC context, this is equivalent to relaxing supply shortages or bottlenecks by importing almost everything: physical capital (or rather, unskilled construction workers) and human capital in the form of skilled workers (we do not now distinguish between both types of inputs). Finally, equation (4a) which defines the denominator of equation (4) gives the standard result in such models that the higher the real exchange rate elasticities of supply and demand for non-tradables, the lower the real appreciation caused by any factor included in the right-hand side of equation (4).10

The upshot is that a real depreciation is possible if the supply-side effect of immigrants is large enough and their contribution to domestic demand small enough. This seems to be unlikely in this setup where migrants are assumed to spend all their incomes in the GCC host countries. We now relax this restrictive assumption and allow immigrants to save and remit a proportion of their incomes back to their home countries. In fact, restrictive immigration policies in the GCC (such as various restrictions on property ownership, the absence of any naturalization process and hence a sense of “permanent temporary residence”, and restricted family reunion) encourage migrants to remit even larger proportions of their incomes back home.

More specifically, we model immigrants’ remittances as an exogenous fraction θ<1 of their wages. Equation (1) becomes:

And the rest of the model equations remain the same. Total differentiation of equations (5), (2), and (3) generates the following:

where εMM=MgMM/gM<0 is the elasticity of migrants’ wages to the inflow of foreign workers, and all other terms are as defined above.11

Equation (6) for the model with remittance outflows differs from its counterpart (equation (4)) for the model without remittance transfers in that now a real depreciation is more likely. This can be seen from the first term illustrating the demand-side effect which is smaller the higher the fraction of migrants’ wages repatriated back home. The term −θεMM, which is overall positive, suggests that the faster migrants’ wages adjust downwards to additional inflows of foreign workers, the lower are overall remittance outflows and hence the lower is their depreciative effect. The supply-side effect is unchanged as expected. It is thus more likely here that the supply side is big enough to offset any positive demand-side effects so that a real depreciation is observed following resource windfalls. This suggests that not only will the usual Dutch disease trade-off between non-traded and non-resource traded be absent (labor only moves internationally and not intersectorally), but the non-resource tradable sector can actually expand, and pro-industrialization (as opposed to de-industrialization) occurs. This may well have been the case for the GCC countries, as they have witnessed an expansion of their non-oil sectors over the last three decades.

3.5 Estimation

Our theoretical model shows that the GCC countries can avoid the structural adjustment of a resource windfall to a new long-run structure, namely a larger non-traded goods sector, by jumping instantaneously to this new structure. This has been made possible since “all sorts of capital—skills, capital equipment, and infrastructure—can be redeployed or bought and sold on world markets, so that bottlenecks are not encountered and relative prices need not change” (van der Ploeg and Venables 2010). With remittance outflows and oil export revenues as our main right-hand-side determinants of real exchange rates in equation (6), we may not find any significant real exchange rate appreciation driven by oil export revenues. We do however expect a significant depreciative effect of remittance outflows. With no available time series on the stock of immigrants in GCC countries and given evidence of high correlation between the size of remittance transfers and the stock of immigrants, we posit that the coefficient on remittances captures both the supply-side and demand-side effects.

The macroeconomic (supply- and demand-side) shocks that fundamentally determine the equilibrium long-run real exchange rate, and hence constitute our set of control variables, have been well identified in the literature. In addition to the leading variables of interest—international transfers in the form of public windfalls (oil export revenues) and private flows (workers’ remittances)—the equilibrium long-run real exchange rate is generally determined by international financial conditions, government spending, terms of trade, commercial policy, and productivity growth (Montiel 1999).

For our sample, however, we restrict regressors to remittance outflows (rem), oil export revenues (oil), government spending (gov), and net foreign assets (nfa). We do not include trade openness, the proxy for Montiel’s commercial policy, as a control variable. The most common measure of trade openness, sum of exports and imports to GDP, is highly collinear with oil export revenues in the GCC. We choose to drop the terms-of-trade variable from our estimation for the same reason, as oil prices drive terms of trade for major oil exporters.

Given the dynamic nature of real exchange rates, we investigate the alleviating effect of immigration and the resulting remittance outflows by estimating a dynamic Error Correction Model (ECM) of the long-run relationship between the real effective exchange rate (reer)—our proxy for the relative price of non-traded goods—and its above-mentioned determinants, all measured in percent of non-oil GDP.

The prior expectation would be that oil export revenues have an appreciating effect on the real exchange rate, but that remittances have a depreciating effect. The effect of government consumption and net foreign assets is more ambiguous. For the former, the effect depends on whether government spending falls more heavily on tradable or non-tradable goods and for the latter, on whether the country is a net creditor or debtor.

Remittance outflows are potentially endogenous due to reverse causality: endogeneity of the demand for foreign labor may result in endogeneity of the resulting remittance outflows. More specifically, real exchange rate fluctuations following a boom or contraction in the oil sector (due to a change in world oil prices) generate different labor demand structures than the existing ones across sectors, namely the oil sector, the tradable non-oil sector, and the non-tradable sector. Corden (1984: 366–7) makes an indirect case for remittances’ endogeneity in oil-exporting countries when discussing the “Alberta effect”, where booming oil sector revenues accrue to the government, which redistributes them to the public through lower taxes and better public facilities. This policy attracts immigrants seeking to share the rents. In the Gulf countries, this adjustment is accompanied by import of foreign labor, to deal with labor supply shortages, not only in the expanding non-tradable sector but in other sectors as well (van Wijnbergen 1984). It should be noted, however, that reverse causality running from real exchange rate misalignments (following oil booms) to immigration and remittances is expected to be positive, i.e., working against our hypothesis of a negative effect of remittances on the real exchange rate. This suggests that correcting for such reverse causality can only strengthen our results.

Our basic model is estimated with the following auto-regressive distributed lag model, which we present, for illustrative purposes, with one lag on both the dependent and explanatory variables:

where µt and σt are country and year dummies that control, respectively, for country-specific time-invariant unobserved heterogeneity and global shocks or common factors affecting all countries in the sample (such as oil price shocks). εit are error terms that are assumed to be identically and independently distributed across i and t. Our time dimension is annual data from 1980–2009 and the countries we include in the sample are labor-importing oil exporters, namely: GCC countries, Australia, Libya, Netherlands, Norway, Russia, and the United Kingdom. All variables (except net foreign assets) are expressed in logarithmic values.

Equation (7) can then be expressed in error correction form. Manipulating and rearranging terms, we separate the short-run adjustments from the long-run equilibrium relationship and capture the speed of adjustment:

Where φ=(1λ1);δ=δ0+δ11λ1;α=α0+α11λ1;etc.;ηi=μi1λ1;τt=σt1λ1

  • (reeri,t1ηiδoili,tαremi,tγgovi,tθnfai,t) is the long-run relationship between the REER and its fundamental determinants. More specifically, it is the deviation of reeri, t-1 from its predicted value given by (ηi+δoili,t+αremi,t+γgovi,t+θnfai,t). These are the long-run coefficients that we report below.
  • δ0Δoili,t+α0Δremi,t+γ0Δgovi,t+θΔnfai,t are the short-run adjustments which are assumed to be homogeneous for the pooled ECM and allowed to vary across countries in the PMG model (see below).
  • φ is the error correction term or speed of adjustment. It must be negative and less than one (in absolute value), for a stable equilibrium to exist. The larger is φ, the faster is the speed of adjustment back to the long run.

We estimate two variants of the model. We first use a pooled ECM model to estimate a homogeneous cross-country response of the real exchange rate to its determinants. Unit root and cointegration tests indicate that all model variables, except for oil revenues, are nonstationary and cointegrated.12 We therefore estimate the pooled ECM without oil revenues, keeping in mind that government consumption is strongly correlated with oil revenues. More specifically, since in oil-exporting countries oil revenues are the main source of budget financing, including government spending, which is the main mechanism through which oil revenues are injected to the economy, would be a good proxy for spending booms out of oil windfalls (see also Chapter 5).

For robustness, and since we are including countries at different stages of development, we also estimate a separate version of the ECM that accommodates potential heterogeneity in individual countries’ responses. For this, we use the pooled mean group (PMG) estimator (Pesaran, Shin, and Smith 1999), which allows the short-run coefficients to vary across countries, while imposing homogeneous long-run responses. This is particularly relevant when dealing with real exchange rates: while it is expected that short-run real exchange rate movements are affected by country-specific factors, long-run real exchange rate changes are driven by the same fundamentals.

The PMG estimator does not require pre-testing for the presence of unit roots in the panel variables.13 Since the PMG estimates the model for each country separately, it does not allow us to include year fixed effects. As a result, an important issue which arises in this heterogeneous setting is potential error cross-sectional dependence, i.e., the potential for errors to be contemporaneously correlated across panel members due to unobserved (global) common factors. These major oil producers are indeed all exposed, potentially in different ways, to common global factors of changing world oil prices. In order to ensure that regressions residuals are cross-sectionally independent across countries, we use a recently developed augmented version of the PMG estimator suggested by Binder and Offermanns (2007) in our panel. The ECM is augmented by cross-sectional averages of all the variables of the model, which are taken as proxies of the common factors. Just like all the variables in the model, the countries’ responses to these global factors are thus allowed to be heterogeneous in the short run. The literature also suggests assuming a common country response to global factors and hence correcting for cross-section dependence by simple cross-sectional demeaning of model’s variables prior to estimation. We report results from both approaches below.14

3.6 Results

Our findings in Table 3.4 show the important stabilizing effect of immigration and remittance outflows on the real exchange rates of GCC countries.

Table 3.4.Effect of remittance outflows on the REER, 1980–2009
Dependent variable: REERPooled ECMPooled Mean Group
Long-run coefficients(1)(2)(3)
Oil export revenues−0.049−0.017
Remittance outflows−0.254*−0.274***−0.133***
Error correction coefficient−0.110***−0.133*−0.311**
Country fixed effectsYesYesYes
Year fixed effectsYesCSDCSA
Note: ***, **, and * indicate significance respectively at the 1, 5, and 10 percent levels. In regressions (1) and (3), we use two lags. In regression (2), we use the Akaike Bayesian optimal lag selection criterion. CSA means that the model was augmented with cross-sectional averages of variables. CSD means that all model variables were cross-sectionally demeaned prior to estimation. Controls, not reported here, include government spending and net foreign assets (which are positive and significant for regressions (2) and (3)).Source: Authors’ calculations
Note: ***, **, and * indicate significance respectively at the 1, 5, and 10 percent levels. In regressions (1) and (3), we use two lags. In regression (2), we use the Akaike Bayesian optimal lag selection criterion. CSA means that the model was augmented with cross-sectional averages of variables. CSD means that all model variables were cross-sectionally demeaned prior to estimation. Controls, not reported here, include government spending and net foreign assets (which are positive and significant for regressions (2) and (3)).Source: Authors’ calculations

Across all models, we find that remittance outflows significantly depreciate the reer in this group of countries. The correlation of remittance outflows with the stock of immigrants suggests that importing foreign workers does indeed mitigate Dutch disease, as expected. Oil export revenues, in contrast, do not seem to exert the expected appreciative effect on the reer (Table 3.4). For the major labor importers in the sample, namely the GCC and Libya, if oil booms are always accompanied by imports of foreign labor, it may be that the usual Dutch disease-type bottlenecks are simply not present. In addition, saving a large share of oil revenues in foreign currencies may also mitigate Dutch disease. Norway is often seen as an example of a country that has successfully avoided Dutch disease through sound management of resource windfalls.15 Consequently, in this set of countries, there is little to suggest that Dutch disease is a serious problem.

The coefficient that captures the speed of adjustment is significant at the 1 percent level, suggesting a strong long-run relationship and feedback effects between the real exchange rate and its fundamental determinants.

Our results can be interpreted in light of the diversification efforts of the GCC countries. First, our analysis suggests that while the Gulf countries’ open immigration policies and consequent access to a perfectly elastic supply of foreign labor have played a significant role in alleviating the bottlenecks and relative price pressures that often tend to crowd out non-oil exports in oil-exporting economies, the current level of diversification in most GCC countries has not yet reflected such considerable advantage.

Second, GCC countries adopting active policies to boost the employment of nationals in the private sector, which has been historically dominated by foreigners, might face a trade-off between creating new jobs in the private sector and shifting existing jobs from foreigners to nationals. Job creation initiatives that promote economic diversification, such as the development of SMEs, are intended to add to labor demand in the non-oil economy, and therefore require improvements in competitiveness. Nationalization policies aimed at increasing the share of nationals in the private sector may on the other hand reduce the number of foreign workers and reduce competitiveness. In their pursuit of jobs for nationals, GCC countries will therefore need to maintain competitiveness, as raising the share of nationals in private-sector employment is not a simple matter of substituting foreign for national workers.

The experience of other labor-importing countries shows that it is possible to combine a high reliance on foreign workers with strong jobs growth for nationals. Singapore, for example, has seen consistent increases in employment of nationals over the past decade, with employment of both nationals and expatriates going up in boom years (whilst the brunt of job losses during downturns was borne by foreign workers). Underpinning this, Singapore’s immigration policies have simultaneously attracted highly skilled foreign workers and restricted the inflow of low-skilled workers (Ruppert 1999).

3.7 Conclusion

This chapter has attempted to show the stabilizing effect of immigration and the resulting remittance outflows on the real exchange rates of the GCC countries. This is particularly important for these major oil exporters who face the risk of Dutch disease and the ensuing undermining of competitiveness in non-oil export sectors. Using a number of estimation techniques, our findings not only suggest a significant negative depreciative effect of remittance outflows but also show the absence of any Dutch disease threat of oil windfalls in a panel of resource-rich, labor-poor economies. This result suggests that the GCC countries have managed to avoid the structural adjustment of a resource windfall (creating a spending boom and a relative increase the price of non-traded goods) that typically dictates an expansion of the non-traded sector at the expense of the non-oil traded sector. As the GCC countries continue in their plans to diversify from oil and increase the employment of nationals in the private sector, it will be important to keep in mind what lies behind this success.

Technical Appendix

Derivation of equation (4):

Keeping in mind (as explained above) that dL=dM=MNdN (since dI=0, total differentiation of equation (1) gives:

Total differentiation of equation (2) gives:

Combining equation (A1) with equations (2):

Replacing the value of du in (A3) in (A2), we get:

Multiplying and dividing the left side of equation (A4) by qeq=qgq allows us to express the left side of (A4) as:

And similarly multiplying and dividing the left side of equation (A4) by qeq=qgq and additionally multiplying and dividing the first term by e and the second term by M, we get:

Equating (A5) and (A6) and replacing the defined elasticities in both gives equation (4).


    AbdallaI. M.Al-WaqfiM. A.HarbN.HijaziR. H. and ZoubeidiT. (2010). “Labour policy and determinants of employment and wages in a developing economy with labour shortageLabour24: 16377.

    • Search Google Scholar
    • Export Citation

    Baldwin-EdwardsM. (2011). “Labour immigration and labour markets in the GCC countries: National patterns and trends.Research paper Kuwait Programme on Development Governance and Globalisation in the Gulf States.

    • Search Google Scholar
    • Export Citation

    BinderM. and OffermannsC. J. (2007). “International investment positions and exchange rate dynamics: A dynamic panel analysis.CFS Working Chapter No. 2007/23.

    • Search Google Scholar
    • Export Citation

    CordenW. M. (1984). “Booming sector and Dutch disease economics: Survey and consolidationOxford Economic Papers36: 35980.

    DixitA. and NormanV. (1980). Theory of International Trade. Cambridge: Cambridge University Press.

    DocquierF.MarfoukA.ÖzdenC. and ParsonsC. (2010). “Geographic, gender and skill structure of international migration.Report written for the Economic Research Forum.

    • Search Google Scholar
    • Export Citation

    DocquierF.RapoportH.ÖzdenC. and SalomoneS. (2011). “Remittances, migrants’ education and immigration policy: Theory and evidence from bilateral data.CReAM Discussion Paper No. 19/11. London: Centre for Research and Analysis of Migration.

    • Search Google Scholar
    • Export Citation

    FasanoU. and GoyalR. (2004). “Emerging strains in GCC labor markets.IMF Working Paper No. 04/71. Washington DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation

    GylfasonT. (2006). “The Dutch disease: Lessons from Norway.Paper prepared for Trinidad and Tobago Chamber of Commerce and Industry.

    • Search Google Scholar
    • Export Citation

    ILO (2013). “Domestic workers across the world: Global and regional statistics and the extent of legal protection.Geneva: International Labor Organization.

    • Search Google Scholar
    • Export Citation

    ImK.PesaranM. H. and ShinY. (2003). “Testing for unit roots in heterogeneous panelsJournal of Econometrics115: 5374.

    • Search Google Scholar
    • Export Citation

    IOM (2012). “The Middle East and North Africa: Annual report 2011.Geneva: International Organization for Migration.

    KaoC. (1999). “Spurious regression and residual-based tests for cointegration in panel dataJournal of Econometrics90: 144.

    • Search Google Scholar
    • Export Citation

    KapiszewskiA. (2006). “Arab vs. Asian migrant workers in the GCC countries.Paper presented at the United Nations Expert Group Meeting on International Migration and Development in the Arab RegionBeirutMay15–172006.

    • Search Google Scholar
    • Export Citation

    LucasR. E. B. (2008). “International migration and economic development: Lessons from low-income countries.Stockholm: Expert Group on Development Issues, Swedish Ministry for Foreign Affairs.

    • Search Google Scholar
    • Export Citation

    MontielP. J. (1999). “Determinants of the long-run equilibrium exchange rate: An analytical model,” in L. E.Hinkle and P. J.Montiel (eds) Exchange Rate Misalignment: Concepts and Measurement for Developing Countries. New York: Oxford University Press26490.

    • Search Google Scholar
    • Export Citation

    PesaranM. H. (1997). “The role of economic theory in modelling the long runThe Economic Journal107: 17891.

    PesaranM. H.2007. “A simple panel unit root test in the presence of cross-section dependenceJournal of Applied Econometrics22: 265312.

    • Search Google Scholar
    • Export Citation

    PesaranM. H. and ShinY. (1999). “An autoregressive distributed lag modelling approach to cointegration analysis,” in S.Strom (ed.) Econometrics and Economic Theory in the 20th Century: The Ragnar-Frisch Centennial Symposium. Cambridge: Cambridge University Press371413.

    • Search Google Scholar
    • Export Citation

    PesaranM. H.ShinY. and SmithR. (1999). “Pooled mean group estimation of dynamic heterogeneous panelsJournal of the American Statistical Association94: 62134.

    • Search Google Scholar
    • Export Citation

    PlantR. (2008). “Temporary contract labour in the Gulf states: Perspectives from two countries of origin.Paper for the Gulf Forum on Temporary Contractual Labor International Labor OrganizationGeneva.

    • Search Google Scholar
    • Export Citation

    PloegF.vander and VenablesA. J. (2010). “Absorbing a windfall of foreign exchange: Dutch disease dynamics.OxCarre Working Papers No. 052.

    • Search Google Scholar
    • Export Citation

    RuhsM. (2009). “Migrant rights, immigration policy and human development.UNDP Human Development Reports Research Paper No. 2009/23.

    • Search Google Scholar
    • Export Citation

    RuppertE. (1999). “Managing foreign labor in Singapore and Malaysia: Are there lessons for GCC countries?World Bank Policy Research Paper No. 2053. Washington: World Bank.

    • Search Google Scholar
    • Export Citation

    United Nations (2009). “World Population Policies 2009.New York: United Nations Department of Economic and Social Affairs, Population Division.

    • Search Google Scholar
    • Export Citation

    WijnbergenS. van (1984). “Inflation, employment, and the Dutch disease in oil-exporting countries: A short-run disequilibrium analysisQuarterly Journal of Economics99: 23350.

    • Search Google Scholar
    • Export Citation

    World Bank (2011). “World Bank prospects: Migration and remittances dataset.” Available online at <> accessed May282013.

    • Search Google Scholar
    • Export Citation
1World Bank bilateral matrix of immigrant stocks 2010: see World Bank (2011).
2Indeed, the prevalence of nationals, and not foreigners, in the public sector is not only a feature common to the GCC but also worldwide; however, the striking feature is that the employment of nationals in the public sector constitutes the largest share of their total employment.
3Lucas (2008: 61) also mentions the higher cost of labor from neighboring Arab countries, as well as the “greater ease of denying family accompaniment to the contract workers from Asia,” as reasons for the shift of balance from Arab to Asian workers.
4It remains unclear whether strict immigration rules are driven by concerns in the GCC countries about maintaining their national identity and security, especially with citizens constituting a small minority of the population, or are solely based on a clear economic plan of mobilizing a cheap workforce to the provision of goods and services without adding to demand pressures (Ruhs 2009).
5Qatar’s third National Human Development Report, as part of its efforts to support the achievement of Qatar National Vision 2030 and the program of National Development Strategy 2011–16, puts the spotlight on Qatari youth and identifies expanding education opportunities for young people and enhancing their participation in the labor force as one of the key challenges ahead.
6While a large part of job creation takes place in small and medium-sized enterprises, SMEs are generally more financially constrained than large firms and less likely to have access to formal finance. In the GCC and the MENA region in general, SME financing through banks is very limited. Data shows that SME loans represented only 2 percent of total loans over the last few years in the GCC. However, there has been an increased focus in the GCC on SME development. For instance, Saudi Arabia has substantially scaled up programs to alleviate financing obstacles for SMEs, which are viewed as major job creation vehicles. The Saudi Industrial Development Fund supports bank lending to SMEs, and several specialized credit institutions provide additional lending. In Qatar with record low unemployment rates, SME development is solely for the purpose of diversification, infrastructure development, and privatization.
7Under Nitaqat, companies are classified according to a basic color scheme: red and yellow for noncompliant companies subject to sanctions, and green or premium for companies that fulfill Saudization requirements and accordingly receive benefits. Sanctions and benefits are defined in terms of companies’ ability to apply for or modify work permits for expatriates. For instance, red companies are not allowed to apply for or renew work permits, and existing foreign employees will be able to take jobs with companies categorized as green or premium.
8The production function can also be expressed as the inner product of the price q and the optimal supply function x(q, L) g(q, L) = q*x(q, L). With production decisions assumed to maximize total profit, the optimally chosen supply function is thus obtained by differentiating the production function with respect to their prices. Similarly, minimizing expenditure to attain a target utility u at given prices q gives us the Hicksian compensated demand function, which is the first derivative of the expenditure function with respect to q.
9For details of the derivation, please see the technical appendix at the end of the chapter.
10It can also be shown that a resource windfall followed by the import of foreign workers is strictly welfare-enhancing by solving for du, and that the resulting change in domestic wages is ambiguous by solving for dw. The change in domestic wages includes the negative effect of the import of foreign labor and the resulting increase in labor force, as well as the positive effect reflecting the adjustment process through which Dutch disease happens: real exchange rate appreciation driven by an excess demand for non-traded good post-resource windfall raises demand for labor and domestic wages for locals and immigrants proportional to their shares in the total labor force, and hence increases overall wages. This is by definition the process through which labor is drawn out of the tradable sector to the booming non-tradable sector and Dutch disease occurs.
11Derivation of equation (6) is very similar to derivation of equation (4) shown in the technical appendix, and is left to the interested reader.
12Based on a number of unit root tests such as Im, Pesaran, and Shin (2003) and Pesaran (2007). Panel cointegration tests included Kao’s (1999) homogeneous residual test and the Fisher test.
13Pesaran et al. (1999) show the consistency of the PMG estimator in the case of I(0) and regressors. In the PMG model, the parameters of interest (long-run coefficients and speed of adjustment) are obtained by maximizing a concentrated log-likelihood function of the panel data model (defined as the product of likelihoods of each group). Starting with an initial estimate of the long-run homogenous parameters (such as static fixed effects), estimates of error-correction coefficients and the other short-run coefficients (including country-specific intercepts and error variances) can be computed (also using maximum likelihood) as the averages of the estimated coefficients for each cross-section. These average estimates can then be used to obtain an updated estimate of the long-run parameters. The same process is repeated until convergence is achieved. The long-run parameters are consequently nonlinear functions of the short-run parameters.
14Several practical points on the PMG estimation are worth noting. First, the time dimension of the data has to be long enough to allow estimation of the model for each of the cross-sections separately. Second, the lag order has to be long enough to ensure that the residuals of the error correction model are serially uncorrelated but not too long to cause a serious loss of degrees of freedom. In this respect, there is a trade-off between the loss of degrees of freedom when including too many lags (relative to the time series dimension) and the loss of consistency when including too few lags. Augmenting the model with lags addresses the potential endogeneity of remittances. In this respect, Pesaran (1997) and Pesaran and Shin (1999) show that for inference on the long-run parameters, sufficient augmentation of the order of the auto-regressive distributive lag model can simultaneously correct for the problem of residual serial correlation and endogenous regressors. The optimal number of lags is best chosen according to an information criterion such as Akaike Information Criterion (AIC) or the Schwarz Bayesian Criterion (SBC).

    Other Resources Citing This Publication