Regional Integration and Goods and Factors Flows in the MENA Region

FEM31-07 | December 2007


« Regional Integration and Goods and Factors Flows in the MENA Region »


Sule Akkonyunlu, Bilkent University, Turkey & DIW, Germany Bilkent University, Turkey


Note :

This document has been produced with the financial assistance of the European Union within the context of the FEMISE program. The contents of this document are the sole responsibility of the authors and can under no circumstances be regarded as reflecting the position of the European Union.

Summary :

This study contributes to a understanding of several phenomena in related to migration that have been little investigated. The point of departure was an attempt at explaining the determinants of Turkish migration to Germany[1], in keeping with similar research performed by Hatton (1995) and Mayda (2007). An important financial flow that goes with migration is remittances by foreign workers. As a case study, the determinants of the Turkish remittances from Germany were studies. Moreover, we performed research into the cyclical interactions between these remittances and the output both in Turkey and Germany. We also analyzed the impact of the economic reforms implemented in 1980s and the Custom Union Agreement of 1996 on intra-industry trade in Turkey. Finally, the performance of Turkish migrants in their host country was studied using micro data (Mikrozensus) from the German Statistical Office.


1. Determinants of Turkish migration to Germany

Recruitment agreements between Germany and Turkey shaped initial migratory flows. Subsequent migratory movements, however, had their own dynamics and mechanisms, such as family unification and asylum seeking, which were independent of German policies, seeking to limit in migration. We developed a parsimonious single equation conditional error correction model using annual data from 1962 to 2004 to explain Turkish migration to Germany. This model proves to have well-defined and stable coefficients [see Akkoyunlu (2007a)]. A single cointegrating vector is found among gross migration inflows and the following explanatory variables; income ratio between Germany and Turkey, unemployment rates in Germany and Turkey, aid to GNI ratio, a trade intensity variable, share of manufacturing exports to total exports to Germany, and remittances relative to Turkish GDP. In the long run equation, relative income, unemployment rate in Turkey, trade intensity, and workers’ remittances contribute positively to migration from Turkey, while unemployment in Germany, aid, and the share of manufacturing exports to Germany contribute negatively to migration from Turkey to Germany. In the long run, relative income, trade intensity, and remittances are the most significant predictors. Thus, a 10 percent increase in relative income goes along with an increase of gross migration inflows to Germany by 49 percent. Likewise, a 10 percent increase in trade intensity is estimated to increase gross migration by 23 percent. This is a large effect, especially when compared to the findings reported in other studies, e,g. Pedersen et al. (2006) and Péridy (2006). It may be due to the fact that Germany is Turkey’s biggest trading partner. We also found that a 10 percent increase in remittances may increase gross migration inflows by 7 percent. Finally, the fact that Turkish manufacturing exports relative to total exports to Germany have a negative coefficient suggests that trade and migration are substitutes in the long run.

Deriving the final (conditional) model from the initial more general specification I took quite a few steps giving rise to the risk of type II errors. However, the diagnostic tests are satisfactory, suggesting that the conditional model satisfies the design criteria. In particular, the error correction term is highly significant and has the expected sign.

Turning to short-run effects, relative income again is the most important determinant: a 10 percent increase in the change in relative income increases the change in migration inflows by almost 20 percent. This suggests that as long as the income gap is not reduced, the pressure of migration will remain. However, compared to the unemployment rate at home, the availability of jobs in Germany matters more in the short-run. The share of manufacturing exports to total exports to Germany has a positive estimated influence once more. This may be explained with reference to the financial barriers of migration. The increase in income due to exports might help some unskilled workers afford the cost of migration. In addition, displacements and disruptions that accompany development temporarily may also trigger migration.

Aid is only effective in reducing migration in the short run, and its effect is small. Hence, financial assistance does not seem to constitute a long-run sustainable way to mitigate migration pressure. However, the majority of foreign assistance to Turkey takes the form of grants and credits, which could be made conditional on good policies, emphasizing technical assistance and training that would transfer expertise and know-how, helping human capital development, job creation, and economic growth. In this way, foreign aid would serve to reduce the income gap between Turkey and Germany, with a long-lasting indirect effect. Finally, remittances prove significant in the short run as well, supporting the hypothesis that they fuel migration. There may be several reasons for this. Remittances help to overcome liquidity constraints hampering emigration, are proof of higher income obtainable in the target country, and may well be aimed at directly supporting additional migration from a family as the optimal composition of the portfolio of the familial demographic-labor assets prompts reallocation in favor of a rewarding component. These considerations point to the possibility that an economy that receives remittances could in fact generate more rather than less migration [see Akkoyunlu, Siliverstovs, and Stark (2007)].


2. Effects of trade liberalization

The dynamic benefits of trade liberalization on migration were hypothesized to differ from its static effects. Trade liberalization and economic integration can have a positive impact on growth by favouring the reallocation of factors to serve more productive uses. Akkoyunlu, Kholodilin and Siliverstovs (2006) test this hypothesis by analyzing intra-industry trade (IIT) between Turkey and its trading partners. IIT has been shown to be an indicator of economic development in several studies, see Greenaway and Milner (2006). Akkoyunlu, Kholodilin and Siliverstovs (2006) have provided the first statistical evidence that the Turkish economic reforms of the 1980s and the Customs Union Agreement (CUA) of 1996 exerted a positive impact on Turkey`s IIT. Furthermore, they find that the reforms have a stronger impact as they affect IIT with all trading partners, whereas the impact of the CUA is only noticeable in the IIT with EU member states, as expected. And, although the CUA covers mainly industrial goods, it appears to exert a similar effect upon IIT in all categories of goods. Finally, the drastic increase in IIT after the 1980s and especially after 1996 confirms the “soft adjustment hypothesis” stating that economic integration between the EU and Turkey is accompanied by decreasing rather than increasing costs of adjustment. By accelerating economic convergence economic integration also reduces incentives to emigrate on the long run.

The balance of costs and benefits accruing to the source countries from migration is a matter of controversy in the literature. The argument usually comes down to one of remittances versus ‘brain drain’. This study (Akkoyunlu and Kholodilin, 2007) shows that remittances have been an important source of foreign exchange for Turkey. However, the evidence speaks against a brain drain is that Turkish migrants the Turkish migrants to Germany have been unskilled (Akkoyunlu and Vogel 2007).

In absolute numbers; Turkey ranks fifth in terms of remittances received, after India, Mexico, Philippines, and Egypt. Akkoyunlu and Kholodilin (2007), by analysing the cyclical interactions between the remittances of the Turkish workers in Germany, and its output effects both in Germany and Turkey find that remittances respond more strongly to changes in host country income than to home country income, measured by real GDP. Thus, Turkish migrants focus more on the economic situation in Germany than in Turkey when deciding how much to remit.

Akkoyunlu (2007b) relates Turkish remittances from Germany to economic activity in the host and home country, the real exchange rate of the Deutsche Mark (the euoro respectively after 2001), political instability in the home country and the stock of Turkish migrant workers in Germany. She estimates an OLS regression using annual observations covering the time period 1962 to 2005. The main findings are the following. First, the negative coefficient associated with Turkish income, and positive coefficients on the real exchange rate and political instability support the claim that Turkish remittances from Germany are altruistically motivated. Second, donators do take the purchasing power of their gifts into account is that real exchange of the Deutsche Mark against the Turkish Lira is positively related to remittances. Third, the coefficient of the stock of Turkish migrants does not differ from one. Thus, altruism is a permanent phenomenon is that new arrivals seem to almost make up for the decrease of remittances on the part of the longer stayers. Fourth, even though Turkish migrants remain abroad for extended periods and eventually settle there, the flow of money back home sustaining their relatives continues. While in the case of traditional emigrants remittances decrease somewhat over time and eventually – presumably when the decision is made to remain permanently in the host country – cease altogether. However, in the case of Turkey, remittances do not stop to play a consistent and important role (Akkoyunlu, 2007b).


3. Turkish workers in the German labour market

Turkish workers in Germany have been adjusting to conditions on the host country labour markets by sustaining periods of unemployment, going into self-employment, and leaving the host country. Akkoyunlu and Vogel (2007) analyzed the second alternative in greater detail, using the German Microcensus database for the 1989-2005 period. ed their presence in self-employment jobs as industrial employment diminished in Germany. Economic and sociological theories cite varies reasons for becoming self-employed, ranging from discrimination to earnings differentials. First, observed self-employment rates of all immigrants and German natives were related to by age and education, using both probit and non-parametric methods. The hypothesis that both groups have equal self-employment rates for a given on age and educational level could not be rejected. Next, the analysis was repeated for the Turks in Germany, the largest immigrant group in Germany. In this case, age and education can not explain the differences in self-employment rates between Turkish immigrants and natives. This constitutes preliminary evidence on labour market adjustment of Turks systematically from zthat of both German natives and immigrants from other countries, at least in Germany as the host country.

Next, self-employment rates were related to pull, push, and cultural determinants. The pull factor is the income difference individuals face when making the decision to become self- employed compared to being a dependent worker. However, because this differential is not exogenously given but depends on the characteristics of the individuals, separate earnings equations were estimated for self-employed and dependent immigrant workers. The push factor is proxied by local unemployment rates in Germany. Finally, cultural factors were proxied by a set of dummy [(0,1] variables indicating the type of immigrant.

The importance of pull, push and cultural components are gauged by analysis of the deviance model by making the difference of self-employment rates the dependent variable.. Push and pull factors explain most of the difference as long as the model contains only one single dummy variable identifying immigrants. However, once we distinguish different immigrant nationalities are distinguished, the cultural factor becomes dominant. While the push factor remains important, the pull component (predicted income differential) losses statistical significance. This result suggests that the self-employment decision is influenced by ethnic-specific attributes rather than sectoral earnings differentials. However, the precise nature of these cultural factors is not explored in this paper. Therefore, there are interesting questions with respect to interaction between ethnicity and self-employment that deserve attention for the future research. At the very least, immigrants apparently can not be treated as a single homogenous group, a fact that any policy encouraging self-employment would have to take into account.


4. Concluding remarks

The many issues touched on this project are just a small part of the many important and yet unexplored aspects within relating to international migration. However, the findings of this project do already provide a deeper understanding of the links between aid, trade, remittances, and migration, as well as the performance of migrants in the host country. Moreover, they identify some directions for future research.

[1] According to a private communication with Professor Nicholas P?ridy, his research team will be analysing the determinants of migration in the Middle East and North Africa (MENA) region. Therefore, we focus on other issues such as the determinants of Turkish remittances from Germany, the cyclical interactions between the remittances of the Turkish workers in Germany, and its output effects both in Turkey and in Germany, and the labour market adjustment of Turkish workers in Germany.