GOVERNANCE, GENDER, DEVELOPMENT ASSISTANCE AND

MIGRATION

by

Dane Rowlands

The Norman Paterson School of International Affairs

and

Ann Weston

The North-South Institute

Report for Citizenship and Immigration Canada

MAY 12, 1998

Contents

	Page
Executive Summary	3
1. Introduction	6
2.A statistical analysis of migration	7
(a) The effects of basic welfare measures	14
(b) The effects of demographic features	15
(c) Governance variables	16
(d) Gender-differentiated measures of development	18
(e) Conclusions from the statistical analysis	18
3.Migration, governance, arid development assistance	20
(a) Governance and migration: the indirect effect	21
(b) Governance and migration: the direct effect	24
(c)Targeting governance with ODA and the implications for migration management	26
(d) Conclusions	30
4. Migration, gender, and development assistance	32
(a) Experiences	33
(b) Underlying factors	38
(c) Policy responses	42
(d) Conclusions	46
5. Aid allocations and sources of immigrants to Canada.	47
6. Conclusions and Recommendations	49
Appendix 1: Data definitions and sources	52
References	56
Table 1. Anova Analysis Results	11
Table 2: Canadian ODA and Immigration Rankings (1997)	48

EXECUTIVE SUMMARY.

In recent years, donor countries have increased their support for improving governance and gender equity in developing countries. This report, building on Rowlands and Weston (1996), considers whether such aid will also achieve specific migration management objectives. Certainly both gender and governance are critical elements of a nation' 5 social structure, and as such are linked to migration. While the heightened interest in other countries' intemal affairs may challenge state sovereignty, this paper emphasizes the common policy interests that may promote cooperation rather than confrontation.

In Section 2 we analyze a cross-section of international migration data for the year 1990 to identify significant macro-level factors. The latter include wellbeing, demographic, governance and women's status. Data limitations meant that sample sizes ranged from 128 countries (for many of the welfare measures) to 46 countries (the two governance measures). An ANOVA (analysis of variance) test revealed that as welfare improves, emigration pressures rise, although these results must be treated with caution as several variables were closely correlated amongst themselves). Rising migration was also associated with high population density and, more surprisingly, low population growth. The governance variables perform extremely poorly, whereas gender development is positively related to emigration.

Regression analysis was used to explore more subtle linkages, after taking into account possible correlations (e.g. between welfare and gender measures). In general, it performed reasonably well for this type of cross sectional data set, with equations estimating the migration-to-population ratio explaining as much as 30% of the variation in migration rates. There is some statistical support for a migration 'hump' -- high emigration rates are associated with a per capita income band of $1900 to $4600, peaking at $3300. Higher GNP growth rates were associated with lower emigration rates. Both population size and population density were clearly an important determinant of emigration. Curiously, ethnic fractionalization and high population growth were associated with reduced emigration. Also disappointing, for this study, was the inconsistent effect of governance on emigration rates, perhaps reflecting the limited and poor quality of the data. Finally, a higher measure for the UNDP's gender empowerment index, indicating more empowerment for women, was associated with higher levels of migration.

In Section 3, we note that the theoretical grounds for claiming poor governance affects migration are stronger; good governance may affect migration indirectly as well as directly, whether defined narrowly in terms of economic policy and administration, or broadly to include political, social and human rights. Through its positive impact on development and, with it, social and economic upheaval as well as improved information and transportation, good governance should have an indirect effect on migration. If factors such as personal security, the adequacy of public services, and political inclusion affect migration decisions, governance issues should emerge as key determinants of migration levels.

Governance targets associated with migration management appear to be easily integrated with standard ODA pro grams, which recognize the importance of political and administrative reform. While still needing refinement, migration theory provides some guidance for structuring ODA programs so as to enhance our capacity to manage migration flows. If good governance reduces incentives to leave by the more effective provision of public goods and services (as opposed to private ones which may be used to finance migration), and enhances rather than destroys community ties, it appears to be well suited to the goal of migration management.

In Section 4, we note that the proportion of women in international migrants has increased; they accounted for 53% of immigrants to Canada from developing countries from 1993 to 1995. Increased migration by women has resulted from economic restructuring in both sending and recipient countries. In the former, women have assumed the responsibility for family survival following adjustments in national labour markets. Sometimes the relatively large surplus of unskilled female labour has led women to seek employment in other countries. Governments have also actively encouraged them to do so, to generate foreign exchange remittances. On the demand side, the transition to service economies has increased opportunities for immigrant assembly workers and service workers. Migrant women face difficult work conditions, sometimes the result of their uncertain legal status, and often because of segmented labour markets; many lost their jobs with the downtum in East Asian manufacturing.

Supply-side initiatives to reduce the need for women to migrate range from more labour-intensive economic policies, such as policies to promote community development, to micro-credit and skills development -- measures popular with many donors. But more macro evidence suggests that such projects may lead to increased out migration in the short-term, at least amongst lower income groups or where the gender development index is low. An important question is whether some gendered aspects of governance might influence migration flows -- either in general or specifically of women. There may be some public good considerations; improving women's status may facilitate their individual migration, but it may reduce the pressures for families to move. In order to test these various propositions, more comprehensive collection of gender-differentiated data is urgently needed.

In Section 5, we compare the flow of immigrants to Canada with our major aid recipients, to determine whether any might be suitable candidates for ODA programs that target migration management issues. The three largest source countries, China, Pakistan, and the Philippines, may appear suitable targets for governance and gender based ODA, but their economic and population sizes suggest Canadian and other ODA may he too small to influence their policy let alone have much effect on their migration levels. Of the six countries supplying 1,000 to 10,000 migrants to Canada, in only the two relatively small economies (Haiti and Ghana) and with whom Canada has a long aid relationship, might Canadian aid have any impact.

A final section presents some preliminary recommendations and suggests directions for future refinements of the analysis:

· Additional aggregate statistical research should be encouraged due to the promising results of this preliminary investigation.

· Governments should coordinate their activities in terms of defining, collecting, and publishing detailed data on immigration that is disaggregated by sex, by a common definition of migrant and refugee, and detailed in terms of country of origin.

· The performance of governance-based ODA and governance-based conditionality needs to be monitored and governments should encourage the sponsors and reviewers of such pro grams to conduct and publish evaluations of their efforts.

· Similarly, there is also a need for specific evaluation of projects targetting women in migrant source communities or regions. Further research is needed to determine whether these projects will be able to offset the broader macroeconomic tendencies that underlie women's migration as much as men's.

· It may be worthwhile to examine in more detail the potential for integrating migration management initiatives into countries such as Ghana and Haiti with whom Canada has a longstanding aid relationship; for other source countries, initiatives might be considered in conjunction with other ODA donors.

1. Introduction

This report was commissioned by Citizenship and Immigration Canada as part of an effort to develop a more comprehensive understanding of the forces underlying international migration, and the policy instruments that may be used in an effort to manage these flows. This report follows up some of the themes identified in Rowlands and Weston (1996), and focuses specifically on whether directed official development assistance (ODA, or aid) can be used to achieve specific migration management objectives. 1

The specific question addressed is: Will aid directed to improved governance or gender issues in the source countries affect quantitatively or qualitatively the subsequent migration flows? The two issue areas of governance and gender were chosen primarily because they are topics of current debate in the development field. Recent publications by the Canadian International Development Agency (CIDA) highlight these issues as crucial components of Canada's new ODA strategy. 2 The consequent increase in policy relevance is, unfortunately, somewhat offset by the paucity of relevant previous research. Thus many of the conclusions must remain tentative at this stage.

Gender and governance are linked in more fundamental ways, however. Both are critical elements of a nation's social structure, and reflect important aspects of the distribution and use of power. The link between these social elements of a nation and migration also draws attention to the ways in which social structures in one country can affect other states. And while the heightened interest in the internal affairs of foreign countries has often been seen as a challenge to state sovereignty and a source of possible conflict, this paper emphasizes the common policy interests that may promote cooperation rather than confrontation.

Because of the absence of both substantial theory and evidence on either governance or gender in the context of migration, the research was expanded to include a substantial empirical investigation that could provide some empirical foundations for the discussion. The empirical analysis is used both to investigate the migration phenomenon generally, but more importantly it is used specifically to investigate the role of governance and gender as determinants of migration.

As a consequence, the paper has four substantive sections. Section 2 presents the preliminary results of an econometric analysis of international migration. The purpose of this section is to use the available data to identify those factors that can be linked in a statistically significant way with the magnitude of migration flows, and those factors that appear to have no such link. The subsequent section reviews the literature on governance and migration and governance-based aid. Section 4 reviews the linkages between gender issues and migration and considers the role of development assistance with a gender component. Section 5 briefly examines the connections between Canada' s current ODA allocations and the major sources of migrants moving to Canada. A final section presents some preliminary conclusions and suggests directions for future refinements of the analysis.

2. A statistical analysis of migration

The phenomenon of international migration in the aggregate is the consequence of numerous individual, household, family and community decisions. The analysis of these micro-level decisions is clearly impossible in the absence of micro-data that characterizes both migrants and non-migrants. Macro-level analysis of migration data instead attempts to identify those factors that, on a systematic basis, tend to be associated with higher or lower levels of migration. The analysis presented here attempts to explain the number of emigrants from a particular nation as a function of macro-level indicators from source countries.

It should be noted that this analysis is compromised by the absence of good data, and consequently the results of this section need to be interpreted with some caution. The data set has been compiled as a cross section for the year 1990. To determine the number of emigrants from a specific country, we added up the number of immigrants reporting that country as their country of origin (or birth) as collected by officials in the major recipient countries of Canada, the United States, the developed countries in Europe, Australia, and New Zealand. The limitations of the data are clear. First of collected by officials in the major recipient countries of Canada, The United States, the developed countries in Europe, Australia, and New Zealand. The limitations of the data are clear. First of ah the quality of the data are variable and the definitions of immigrants used by each destination country are not entirely consistent. Some developed country recipients may not be captured at ah in the data. 3 Secondly, the focus is clearly on 'South-North' migration; 'South-South' migration is not reflected in the data. Illegal migration is obviously excluded from the data set. In addition, many of the explanatory variables were available for only a sub set of countries, thus reducing the sample size. Details of the data set are provided in the appendix to this report.

These imperfections do not invalidate the data, however, as the measures are sufficiently consistent and the data sufficiently complete to be a fair reflection of actual migration pattern. Furthermore, the recipient countries included in the sample represent both a large proportion of the group of recipient countries (including the major countries of immigration, Australia, Canada, the United States, and the major European recipients) as well as key members of the Development Assistance Committee (DAC) of the Organization for Economic Cooperation and Development (OECD). 4 This latter group is clearly the most relevant given our focus on development assistance as a tool of migration management. To our knowledge this is the first statistical analysis using a cross-sectional data set developed in this manner. Given the basic absence of any macro-level formal statistical research of this issue, we believe that the analysis is worth pursuing despite these data limitations.

The explanatory variables that were collected measured source country characteristics in four broad areas. The first set of measures examines the average material well-being in the source country. The measures used are: gross domestic product (GDP) per capita, GDP growth, the United Nations Development Program's Human Development Index (HDI), levels of education, access to water, and public expenditures on health. The second set of measures is demographic: population, population density, rate of population growth, and the level of ethno-linguistic fractionalization. The third set of variables reflects governance issues: political stability and administrative capacity. The ethnolinguistic fractionalization measure, and the political stability and administration quality measures, are clearly somewhat subjective in nature, and there are competing variants designed to reflect essentially the same thing. The measures used in this paper are from Mauro (1995) and, while somewhat dated, they have proven to be one of the most useful and commonly used data sets of this kind. The final three variables measure the status of women: female education levels, and the UNDP's gender empowerment measure (GEM) and gender development index (GDI).

Since the availability of the explanatory variables was inconsistent, and the inclusion of some would alter the sample size for regression analysis, the first step in the statistical investigation was to perform analysis of variance (ANOVA) tests on emigration levels. The emigration and emigration as a portion of population ratios were arranged according to the ascending order of each explanatory variable. The resulting series was then divided in two, and an ANOVA process used to identify statistically significant differences between the two halves of the dependent variable series. For example, the GDP per capita series was arranged in ascending order with the corresponding emigration rate variable. The emigration rate series was then divided in two approximately equal parts. The first half-series corresponded to the emigration rates of the low GDP per capita countries in the sample, while the second half-series corresponded to the high GDP per capita countries. The se two series were then compared for statistically significant differences. Sample sizes ranged from 128 countries (many of the welfare measures) to 46 countries (the two governance measures).

The ANOVA analysis provides a simple indicator of whether there is any relationship (causal or otherwise) between the emigration measures and each explanatory variable. Table 1 summarizes the results of the ANOVA tests for the emigration-per-population dependent variable, which theoretically should be more relevant as a measure of emigration pressure in a country than just total emigration. 5

The first two columns of Table 1 are self explanatory. Column three provides the average emigration per population rate (multiplied by 1 million for scaling purposes) for the sample half with the low values of the explanatory variables from column one. Column four provides the same information for the sample half with the high values of the explanatory variable. The final column presents the level of error for which there is a statistically significant difference between the two sample halves. A statistically significant difference implies that there appears to be a significant relationship between the explanatory variable and the emigration rate. In general, levels of significance below 5% are deemed to indicate a relationship, while any level of significance above 10% would suggest no relationship. Between the 5% and 10% levels caution should be used in assuming any difference between the sample halves.

The ANOVA results are indicative of the effects perceived in the subsequent regression analysis, and several basic relationships can be identified. The key welfare measures (the first seven explanatory variables in Table 1) all suggest that as welfare improves, emigration pressures rise. As the sample includes some of the poorest LDCs, we may wish to interpret this result as indicating the presence of minimum threshold level of development for which welfare levels are sufficient to provide the resources and information required for emigration.

However there may be some regional effects here as well. Poor countries are concentrated in Sub Saharan Africa, which is geographically more isolated from the major recipient countries than Latin America, the Caribbean, Southeast Asia, or North Africa. 6 Furthermore, it is important not to infer that each of the explanatory variables has its own relationship -- causal or otherwise -- with emigration rates. Several of these variables are closely correlated amongst themselves (as discussed below) and thus may reflect the same underlying features of a country.

TABLE 1: ANOVA ANALYSIS RESULTS

Variable	Sample size	Average (low)	Average (high)	Level of significance
HDI	127	435	2629	0.43%
Life expect.	128	273	2754	0.0043%
GDP per cap.	128	939	2084	6.47%
education	127	373	5256	0.018%
% w/o water	96	2738	670	1.02%
Health budget	104	963	1302	35%
GNP growth	102	2180	925	7.87%
Pop. density	96	385	2005	0.58%
Pop. Growth	117	2399	673	0.94%
Ethnic division	87	2009	884	9.5%
Pol. stability	46	1172	1957	31%
administration	46	1459	1669	78.7%
Female.educ.	127	466	2565	0.078%
GDI	96	520	2621	0.26%
GEM	68	547	3109	0.80%

The demographic variables also provide some interesting results. High emigration rates are associated with high population density and, more surprisingly, low population growth. 7 Higher levels of ethnic division are associated with lower levels of emigration, though the connection does not appear to be statistically significant.

The governance variables both perform extremely poorly, indicating that political instability and administrative capacity are not related to emigration rates, at least in a simple fashion. On the other hand, gender development is strongly related to emigration. All three measures indicate that greater gender development is associated with higher rates of emigration. These statistically strong results are possibly due to a positive correlation between welfare and development in general, and the improvement of conditions for women specifically.

The ANOVA results are useful for identifying variables that have potentially strong + perhaps even causal -- relationships with emigration. The difficulty is that the method captures only very simple relationship pattern, and does not take into account the potential for more subtle linkages. Furthermore the analysis cannot identify the linkages that result from complex interdependencies between the explanatory variables. To investigate the possible presence of these other linkages, regression analysis was used.

One of the potential difficulties with using regression analysis to analyze migration flows is the possibility that different variables that are hypothesized to be affect migration may also be correlated amongst themselves. This problem -- known as multicollinearity -- compromises the regression analysis. Essentially the regression analysis determines how much variation in the dependent variable (migration in this case) is associated with variations in the explanatory variables. If two explanatory variables exhibit similar variation patterns (that is, they are correlated) then the regression process cannot easily distinguish the effects of each one. As a consequence, minor changes in the regression equation can lead to substantial changes in the coefficient estimates for the correlated variables.

As suspected, many of the welfare and gender measures were found to be closely correlated. For example, the human development index (HDI) strongly reflected per capita GDP and life expectancy. Thus, including both HDI and either of the other two variables in a single regression equation would lead to estimation problems. A prior statistical test was used to identify groups of variables that exhibited similar variance patterns, and the regression analysis relied on these correlation tests to prevent multicollinearity and the consequent compromise of the statistical estimates.

Regressions were run using four different dependent variables: total emigration, the emigration per population rate, the log of emigration, and the difference in the log of emigration and population (i.e. the log of the emigration per population rate). As expected, the regressions on total emigration were very poor and the only statistically significant explanatory variable was population, which had the expected positive effect, and the square of the population, which had a negative estimated coefficient. The focus of the discussion will instead be on the regressions using the other three dependent variables. For the most part the results of these other regressions were complimentary. A variety of regression equations were used in order to investigate the effects of each variable without introducing multicollinearity and without jeopardizing the analysis by having too few observations. 8 We examine the effects of the explanatory variables in the following subsections.

(a) The effects of basic welfare measures

The first observation to be made is that while welfare measures such as GDP per capita and education both appeared to be statistically significant in many oft he regression equations, the overall explanatory power of the regressions was improved when HDI was used as a compo site indicator in their place. Life expectancy was highly correlated with the other welfare measures, and was not used in the regressions. Access to water and health budgets were occasionally significant statistically, but the results were far less robust relative to the other we~ are measures. GNP growth, on the other hand, generally had a strong negative effect on emigration when it was included in the equations for estimation. The inclusion of GNP growth rates, however, often overwhelmed the effects of other welfare variables such as GDP per capita, reducing the statistical impact of most other explanatory variables to statistical insignificance. This instability in the regression equations is problematic, and may reflect the need for better equation specification or the sensitivity of the analysis to the sample available.

Nonetheless, some inferences do seem reasonable and may provide some preliminary insights into the emigration process. Though weak, there is in fact some statistical support for the idea of a migration 'hump'. This migration hump is a band of income per capita within which emigration rates tend to be higher. Income below or above the band limits is thus associated with lower levels of emigration. Using the regressions with emigration-to-population rates as the dependent variable, we can interpret the band numerically. The significant estimates of the coefficient for GDP per capita concentrated around the level of 2.0, while the coefficient estimates for the square of GDP per capita were in the neighbourhood of -0.0003. A regression was run using only GDP per capita, its square, and a constant. The resulting parameter estimates were 2.21 for the linear coefficient and-0.000336 for the squared term coefficient. The intercept was 1032. The solution of the resulting quadratic equation indicates that the band of income associated with high emigration rates is $1934 to $4638. 9 The associated maximum level of migration occurs at per capita income levels of $3288.

Thus, countries with per capita income levels of just over $3000 U.S. in 1990 (adjusted), such as Panama, Thatland, and Tunisia, should have elevated migration rates in comparison to other countries.

Higher rates of GNP growth were associated with lower rates of emigration. This result was strong and robust. In terms of the magnitude of the effect, a 1% increase in GNP (ah other factors held constant) would reduce the emigration rate by approximately 0.0004. To put this number in perspective, the average emigration rate in the sample was approximately 0.002 for the regression equation in question (the full sample average is approximately 0.0005). Thus an increase in the growth rate by 1% would reduce emigration rates by approximately 20% for the sample of countries in the analysis. While some caution is again needed in interpreting these results, it would nonetheless appear that rapid economic growth reduces migration pressure significantly.

(b) The effects of demographic features

The demographic variables also provided some useful results. While population size was clearly an important determinant of emigration, so too was population density. Higher levels of density were associated with higher rates of emigration. Additional tests suggested that the impact declined as density increased, as indicated by the fact that the parameter on a quadratic term had a negative estimated coefficient. The se effects were quite significant statistically in many oft he equations. The parameter estimates suggest that the effects of density on emigration diminishes after it reaches levels of 4255, which is a relatively high level. 10 Again, however, the sample sensitivity suggests that caution is needed in interpreting this result.

Curiously, ethnic fractionalization and high population growth were associated with reduced emigration. These resu1ts were again often significant statistically, and were relatively robust in the face of different equation specifications. There is no clear interpretation of these results from a theoretical perspective, though a more detailed examination of the sample would perhaps yield some insights. While some sociological explanations may be required, these results would appear to be counterintuitive. Even if they are correct, policy interpretations should be cautious. For example, rapid population growth may not be a causal factor for emigration, merely a correlated tendency. To suggest that emigration rates could be lowered by encouraging population growth or ethnic division would appear to be suspect even if such policies were morally or politically acceptable. Indeed, it may well be that case that the population growth is affecting the equation through the emigration rate's denominator: faster population growth may imply a larger non-migrant population, thereby reducing the rate without affecting the volume of migration. in addition, higher emigration rates will necessarily reduce population growth.

(c) Governance variables

The governance variables did not appear to have any consistent or robust effect on emigration rates. Clearly this result is the most disappointing in terms of this particular study. In only a few equations did political stability have a statistically significant effect. In the se cases high political stability was associated with reduced emigration. The administrative capacity variable was never significant. The political stability variable's occasional significance and relatively stable coefficient are suggestive that improved equations may yet yield some significant results.

A likely cause of the poor results, however, is the unsatisfactory quality of the data. First of all, many countries were omitted from the samples with the governance variables, thereby affecting the statistical results. The absence of data may not be independent of the quality of governance or the degree of political stability. As a consequence, the samples used in the analysis may be systematically biased in terms of the governance variables.

A more immediate concern regarding the data is that they are outdated: the most readily available data were from 1980, a full decade before the other variables. These data, discussed in some detail in Mauro (1995), are based on surveys by Business International, an economic research firm since incorporated into the Economist Intelligence Unit. The numbers reported by Mauro and used here represent the average scores for the country for the 1980-1983 period. While there is likely to be some correlation between past and current performance in these regards, it is likely that stronger inferences would be possible by using more current information. Such data, however, are not widely available yet. If the political stability numbers are in fact indicative of a causal relationship, however, then governance-based development assistance may prove to be a valuable instrument for both development and migration management.

Finally, it is important to recall that the migration being explained here is for South-North migration, and that in some instances refugees are not included in the definition of migrant in a consistent manner. Illegal migrants are also excluded from the data, due to the obvious problems of availability. So some of the migration associated with poor governance will not be reflected in the data here either because of reporting issues, or because political refugees must often flee across land borders to neighbouring countries that often have similar economic characteristics and thus would not appear as a South-to-North migration phenomenon.

Despite the relatively poor performance of the governance variables, the analysis here has highlighted some potential linkages and some possible approaches to addressing the statistical deficiencies that were encountered. Additional research would clearly seem to be warranted.

(d) Gender-differentiated measures of development

This research project also examines in detail the question of gender and migration and, to a lesser extent, development assistance that targets women. Unfortunately the measures of female education levels and the UNDP's gender development index (GDI) were too closely correlated with other key explanatory variables to include in the regressions. Therefore the measure used in the regressions was the UNDP's gender empowerment index (GEM). As with political stability, the estimated coefficient for GEM was only occasionally significant. Again, the coefficient size was fairly consistent and always positive. Thus, a higher measure for GEM (indicating more empowerment for women) was associated with higher levels of migration. This result could be reflecting basic development effects (there was some evidence that multicollinearity may have been a problem when using the GBM measure), or it could be a more subtle effect in its own right. For example, it is conceivable that more empowered women may either have the capacity to migrate, or may be willing to tolerate more migration by family members. Such an interpretation, however, is clearly open to question, and more analysis is required to refine any inferences.

These results for the gender-based measures are not surprising. Prior theory and analysis is very limited, and provides little guidance in the formation of statistically testable hypotheses. In addition, the effect of differentiated gender development may affect the gender composition of migration rather than the overall levels. Migration data disaggregated by gender, however, are not available for the current study, and thus any remarks on this point would be speculative. The collection of such data, however, is strongly recommended.

(e) Conclusions from the statistical analysis

In general, the equations performed reasonably well for a cross sectional data set of this nature. The equations estimating the migration-to-population ratio explained as much as 30% of the variation in migration rates. Over 40% of the variation in the log of the emigration to population rate could be explained in the equations, suggesting that the functional forms for the equations of estimation need to be investigated and tested more rigorously. The equations using the log of total migration had 65% of the variance explained in some cases, but much of the explanatory power in these equations seemed to be derived from the effect of population as an explanatory variable.

The statistical analysis appears to support the following preliminary conclusions:

1. There does appear to be some evidence of a migration band that peaks at income per capita levels of around $3250 (U.S., adjusted by the World Bank Atlas method 11 ).

2. Faster economic growth does appear to discourage migration.

3. There is limited evidence that demographic characteristics such as population density and ethnic composition affect migration rates, but further investigation is required.

4. Governance characteristics do not appear to influence migration, bat statistical problems compromise the results and further research is recommended.

5. Gender-related variables do not appear to have a strong effect on migration, but again statistical and theoretical problems are significant. Sex-differentiated data should be collected, and theoretical advances should be encouraged.

6. On the whole there is insufficient development of migration theory at the aggregate level, a deficiency that adversely affects the construction and interpretation of statistical analysis. Despite the se problems, the use of aggregate-level statistical approaches appears to be a very promising means of examining important migration and migration policy questions.

The last of these conclusions warrants some elaboration and emphasis. While the current literature on migration has identified many factors that may potentially affect migration levels and patterns, there is a need to formalize more precisely the exact nature of the hypothesized relationships. Such a process of refinement will require both theoretical innovation and empirical testing. Some of the posited relationships may be too simplistic, and the lack of precision may lead empirical investigators to reject certain relationships simply because they have not identified the appropriate functional form. In the absence of better theory, the empirical task will be much more difficult. Further research, including the accumulation of improved data on migration, is clearly worthwhile given the results of this study.

3. Migration, governance, and development assistance

There has been little research Into the linkages between governance and migration. The statistical analysis conducted for this study does not identify any strong statistically significant relationship between these two issues either, though future refinements may find evidence of a stronger connection. Nonetheless it is valuable to examine the potential linkages and place them in a suitable analytical framework.

Governance can be thought of as affecting migration in two ways. The first effect is indirect, relating governance quality to social and economic development in general. The second effect is direct, with the quality of governance directly changing propensities to migrate Irrespective of the level of development.

To begin with it is useful to define good governance. 12 Good governance in the narrow sense refers to the quality of public administration. Consequently, the elements of good governance include issues such as public sector management, accountability, the legal framework, degree of transparency, and levels of corruption. The International Monetary Fund's interpretation of governance is restricted to economic issues (IMF, 1997). The IMF has demonstrated a willingness to use a broader interpretation that includes certain political factors, but only in so far as these affect economic efficiency in an immediate sense. The World Bank's definition acknowledges the importance of political dimensions more explicitly. In their 1993 Governance Report (World Bank 1993) the Bank includes the form of political regime as a key aspect of governance. However, despite the recognition of the potential importance of political structures on the development process, the Bank, like the Fund, claims that explicit interference in this area is contrary to its mandate.

Wider interpretations of the governance agenda have been identified by the regional development banks and the community of ODA donors. For some of the regional development banks there has been a willingness to pursue explicit political reform goals in their lending programs. The Inter-American Development Bank (IDB) supplements the standard list of good governance elements with social equity, gender equity, and the presence of participatory structures (Peace Initiatives, 1997). The community of ODA donors and their coordinating body, the OECD's DAC, have pushed the governance definition even further to include issues of human rights, democracy, and levels of military spending.

While there is clearly a fairly broad spectrum of potential definitions for governance, it is often presumed that all of these forms of governance will tend to have the same qualitative linkage with the development process. In the following discussion, therefore, we will refer to governance as a generic term in most instances, referring to specific definitions as warranted.

(a) Governance and migration: the indirect effect

Governance issues have attracted the attention of development practitioners, theorists, and institutions primarily since the mid-1980s. One of the key publications signaling the change in thinking is the influential 1989 World Bank report on Sub-Saharan Africa. The Bank concluded that a key factor behind the failure of Bank projects in that region was an absence of adequate public sector administrative capacity (World Bank, 1989). Subsequent studies have provided both theoretical and empirical evidence to support an emerging consensus that the quality of public administration -- governance in the narrow sense --is indeed an important and perhaps even necessary prerequisite for successful development. 13 Indeed, it may be possible to go even further to say that efficient public administration is an element of successful economic, social and political development.

The political and human rights dimensions of a broader interpretation of governance do not appear to have as clear a linkage to economic development as the more administrative aspects. while there is currently a resurgence in the view that democracy and basic human rights are essential for development (socially and economically), neither the theoretical nor empirical evidence is sufficiently strong to prove the case definitively. 14

So there are grounds to believe that governance in the narrow sense at least, and possibly even in the broader sense, Will have a positive impact on economic and social development. As a consequence there should be an indirect effect on migration. Rowlands and Weston (1996), Appleyard (1992), Russell and Teitlbaum (1992), Bohning (1991) and others have reviewed the linkages between development and migration already. The available evidence suggests that at the early stages of development, emigration will tend to increase. A key reason for this effect is the ability of potential migrants to afford the initial costs associated with emigration. Furthermore, the development process implies structural economic and social transformations that break down traditional geographic and social relationships. The consequent dislocation makes emigration far more likely. Finally, improvements in economic and social development are often accompanied by greater access to information and communications, both of which serve to advertise international disparities in living conditions and identify opportunities for relocation.

As an economy moves through the development process, however, the pressures for emigration appear to diminish. Growth becomes more stable, higher quality employment opportunities become more available, and the living standards improve sufficiently to diminish the incentives for migration.

So to the extent that good governance contributes to economic development, it will affect the migration process by helping to determine the time at which a country reaches the beginning of the development-migration cycle, and possibly the speed at which it moves through that cycle. In contemplating the promotion of the good governance agenda, therefore, policy makers may wish to be cognizant that their efforts may hasten the onset of emigration pressures in countries that are close to the beginning of the cycle. In a world that was inordinately concerned with migration management, in which the starting point of the migration-development cycle was known, and in which policy makers could actually manipulate the development process with precision, it is possible to conceive of an attempt to sequence the order and timing of countries entering the development cycle in some optimal manner. None of these conditions, however, exists. The subordination of economic development to migration concerns appears unrealistic even if its morally objectionable implications did not rule it out. Furthermore there is no precise identification of when a country is likely to enter a migration-4evelopment cycle, a concept that has not yet been confirmed statistically in any meaningful way. Finally, there is no possibility of actually controlling either the governance or development processes sufficiently to conceive of such policy fine tuning.

With respect to the speed at which a country moves through and beyond the development stages in which migration pressures are highest, a faster transition could lead to reduced levels of migration. An important qualification, however, is that rapid development may also be accompanied by more severe disruptions of economic and social structures. If the effects of the speed of transition dominated those of the disruption, and a reduction in migration is the goal of both the source and recipient country, then assistance to promote good governance (especially in the narrow sense) seems to be a justifiable policy option. In evaluating its potential as an instrument, however, governance targeted development assistance would have to be compared against other uses for those resources in terms of achieving development goals. In other words, since the effect under consideration -here is indirect and possibly in the wrong direction, the link between governance and migration must be viewed through the effects on the intermediating factor of development in general.

(b) Governance and migration: the direct effect

The direct effect of governance on migration refers to the reaction of potential migrants to the quality of public administration and political capacity in their home country. The connection is most observable in the extreme: 'failed states' in which civil order has largely disappeared generally exhibit large-scale movements of population. Haiti and Rwanda are recent examples of large refugee flows induced by political turmoil. It is crucial to stress that these are refugee movements, which only in rare circumstances lead to South-North migration. Most of these flows are between less developed countries. As the focus of this paper is on South-North migration of a more economic sort, we will not deal explicitly with these cases.

Nonetheless, the lessons are clear: people do respond to political circumstances in their country, and may choose to emigrate when these circumstances deteriorate sufficiently. Laroque (1987) discusses this linkage in general, while Adepoju (1995) identifies this cause of migration in the context of Sub-Saharan Africa. Weintraub and Diaz-Briquets (1992) describe a direct linkage between political disturbances, such as civil war, and the major outflows of migrants from the Central American region. What is not clear is the responsiveness of migration to political and administrative deficiencies that are not extreme. The anecdotal evidence relates primarily to cases in which a 'crisis' has clearly occurred. There is no strong anecdotal or systematic empirical evidence to link governance issues with migration under more normal circumstances.

In the literature, the linkages between governance and migration are not always transparent. For example, Weintraub and Diaz-Briquets (1992) indicate that emigration rates from Costa Rica have traditionally been relatively low in comparison with its neighbors. It is not clear to what extent the political stability and enduring democracy have contributed to this result directly, or whether the effect has been primarily the result of better economic performance in general. The superior economic performance over the 1980s may, of course, also be the result of the political stability of the country. Adepoju (1995) clearly focuses on the indirect effect in which political instability provokes emigration through the intermediating effect of inhibited development.

The theoretical pedigree of the notion that governance issues should affect migration levels, however, is impeccable. One of the frequent complaints about the models of migration that are based on narrow economic and neoclassical foundations is the absence of the social and political context in which migration decisions are made. If factors such as status, 15 personal security, the adequacy of public services, fairness, and political inclusion affect migration decisions, as they surely must at least at the margin, then governance issues should emerge as key determinants of migration levels independent/y of the indirect effects through the development process. It seems reasonable to expect, therefore, that if there are two countries with identical levels of economic development, we should expect the one with better measures of governance to have lower levels of emigration. While the preliminary statistical work does not reveal this relationship, it would appear worthwhile to examine its potential explanatory power in much more detail in future work.

There is a crucial theoretical insight that may help to differentiate between the direct and indirect effects of governance on migration, and therefore the potential value of governance-targeted ODA. One of the reasons why emigration is presumed to increase with the level of development is that the latter process increases income for the population in general. This additional income can then be used to offset the private costs of migration that might otherwise reduce mobility. Better public administration and political sophistication, however, are public goods. Therefore the provision of good governance reduces the incentive to emigrate without directly increasing the financial ability to relocate. Furthermore good governance is more likely to lead to improved community ties, for example through participatory political processes, than to the kind of social dislocation more commonly associated with economic development.

To conclude, there is no strong empirical evidence upon which to claim that poor governance affects migration except in extreme conditions of civil turmoil. The theoretical grounds for expecting poor governance to encourage migration are, however, quite reasonable. There is also no strong empirical evidence to contradict this hypothesized relationship. Until better empirical data are available, it may be worth examining how ODA may be used to affect migration through improvements in governance.

(c) Targeting governance with ODA and the implications for migration management

Overseas development assistance can affect migration either directly, primarily in the context of refugees, or indirectly, by altering the factors that 'push' migrants to leave their home countries. Aid can also influence 'pull' factors, for instance when it is spent in Canada on assistance to foreign students or used for refugee settlement. Aid flows may also generate personal connections that may contribute to the attraction of immigrants to the donor country. These effects and the associated literature are reviewed in Rowlands and Weston (1996). The extent to which ODA can be used to manage migration, however, remains unclear. It is important to recognize that ODA levels are fairly small, and by itself ODA cannot be expected to dramatically alter the conditions that lead to migration. Furthermore, migration is a complex phenomenon, and it is difficult to know how ODA can be deployed so as to contribute to the management of migration flows. As noted earlier, basic economic development may lead to increased migration pressure in the initial period by raising incomes and making it possible for people to afford the initial costs of migration, and by leading to economic and social changes that disrupt traditional ties to community. It may only be after much del ay that the benefits of development are sufficient to reduce the pressure for migration. If ODA is effective in promoting economic development, then the potential for this initial effect must be taken into account. The most efficient use of ODA as a migration management instrument, therefore, may be to promote forms of development that minimize these initial pressures while still facilitating development. Using ODA to promote good governance practices may meet these criteria.

Before embracing this policy option, several questions need to be answered. First of ah a more conclusive link between good governance and migration needs to be established. While it makes sense theoretically, and there is some prima facie evidence to indicate that there is a connection between bad governance and increased migration, the strength and precise nature of the connection remains unclear. ODA linked to good governance may reduce migration, but it may not be the best use of aid resources even for the purposes of managing of migration, to say nothing of meeting effectively the wider objectives of donors and recipients.

In order to compare governance-based ODA with other ODA targets as a means of managing migration it is necessary to determine how effective such programs have been in the past. As a relatively new target, however, the literature on ODA effectiveness in this area is limited. The preliminary answer must be that it is too soon to tell. However there are some results that may be useful as early indicators of effectiveness.

Governance-based ODA typically comes in two forms. In the first case the resources are used to provide technical assistance for improving government administration. There is a long tradition of this type of assistance both by bilateral and multilateral donors. Many of these programs are in the form of personnel training either through study programs or through exchanges of bureaucrats. Other forms of assistance in this category would include the financial and technical support for administrative reform initiatives. Ranis (1992) cites decentralization as a potentially useful process for the Philippines. Other frequently identified reforms include administrative changes to the judicial sector, policing, tax collection, and financial supervision. Assessments of these programs are not particularly useful for our purposes, as these evaluations focus primarily on the extent to which training and administrative changes meet organizational targets, not on whether the pro gram had an impact on subsequent development. Expecting such assessments hardly seems reasonable, however, given the very limited nature of these pro grams, the fact that any effects would tend to occur only after considerable delay, and the fact that other development factors will overwhelm any effects of the program. Nonetheless, the direct targeting of administrative personnel and administrative reform deserves attention in the future both in its own right and as a supplement to the other types of governance assistance.

This second form by which ODA may be used to promote governance reform is through conditionality. Conditionality refers to the tying of resource provision by donors and lenders to the implementation of identified policies. While the IMF and World Bank have always used conditionality as a component of their operations, bilateral donors have only recently (since the late 1 980s) become interested in linking resource provision with reform efforts. As indicated earlier, the extension of this conditionality into the area of governance has been a somewhat more recent phenomenon.

The purpose of conditionality is to provide incentives for the recipient to engage in specific activities that the donor deems desirable. These incentives are frequently deemed necessary because many governments must overcome significant internal barriers to introduce governance reforms. As the definition of governance widens, the need for conditionality increases. In many cases donors seek to pressure governments into reforms that the ruling group fundamentally opposes. For example, a totalitarian regime may not regard democratization or improved human rights as being in its interest. Outside pressure must therefore be brought to bear to encourage these changes. The desirability of this pressure, and its effectiveness, is still a matter of some contention. The Canadian government, however, has clearly indicated a willingness to emphasize governance and democratic reform as part of its ODA strategy; good governance ranks high on the Canadian government's agenda (Brem and Rawkins, 1992). Canada and other ODA members, however, tend not to use direct and explicit conditionality. Instead there is a preference for linking bilateral assistance to World Bank and IMF programs (which are conditional) or to withdraw assistance from countries that fail to meet certain expectations of behaviour, expectations that may be identified and coordinated explicitly within the DAC.

Assuming that the promotion of good governance through conditional lending is a good idea in terms of its effect on migration specifically and development in general, there are three key problems that need to be addressed. First of ah, current evidence clearly indicates that such reforms are best carried out when the recipient is involved in the identification of reform goals and schedules. This 'ownership' of the reform process is a key determinant of its efficacy. Secondly, reform efforts must be coordinated across bilateral donors and multilateral agencies to ensure consistency. The coordination mechanisms currently in existence may need some strengthening in this regard. The need for enhanced coordination in terms of the migration effects of ODA is made more difficult by the fact that donor interest in the migration effects of its programs is weak when it comes to countries from which it receives few immigrants(Weintraub and Diaz-Briquets, 1992). Finally, the use of conditionality implies a willingness to suspend ODA in cases of non-compliance. Donors must be willing to live with the consequences of this requirement, which may inflict further hardship on an already disadvantaged or oppressed population.

The biggest advantage of using conditionality is the capacity to lever additional domestic resources into the reform efforts. In addition, it has been alleged that the conditional programs of the IMF and World Bank act as a catalyst for other resource flows, though this view is now somewhat controversial (Bird and Rowlands, forthcoming). Ghosh (1992) stresses the need to use limited ODA resources to lever additional resources, as does Oualalou (1992).

How effective have these past aid programs been? Macroeconomic studies of ODA have shown the effects on development to be quite weak, with the possible exception of Asia (e.g. Mosley, Hudson and Horreil, 1987). Diaz-Briquets and Weintraub (1991: 24-25) among many others identify instances in which aid has hampered development efforts, and raise doubts about the ability to integrate migration-motivated ODA goals with more traditional objectives in the context of a single program. In general, ODA effectiveness generally remains a debated issue. With respect to migration specifically, there are instances in which ODA has been identified as an important factor in increasing emigration (De Wind and Kiniey 1988, Ranis 1992).

Despite these pessimistic findings, others conclude that aid can be of some assistance in managing migration (Molle, de Koning, and Zandvliet, 1993). While the governance-specific programs require more data for evaluation, closely related policies have been examined. The evidence on these is mixed, though the specific objectives of the programs have perhaps been more controversial. For example, the effects of conditional lending programs by the IMF and World Bank are very controversial. Bilateral ODA programs based on conditionality have also had problems, though typically these now occur in conjunction with IMF and Wor1d Bank programs and so the effects cannot be separated. The allegations of ineffectiveness are linked to both perceived deficiencies in the program objectives as well as compliance. On the other hand, while Ranis (1992) gives evidence of negative ODA effects on migration in the Philippines (support for labour-displacing mechanization in agriculture), he also points out the need to support the structural adjustment programs of the IMF and World Bank by providing assistance to mitigate their disruptive effects. However Ranis also suggests that misguided aid programs have allowed governments to postpone needed economic reforms and possibly, we might add, political ones as well.

(d) Conclusions

This section of the report has focused on the effects of governance on the migration process. Perhaps not unexpectedly, there is not a lot of compelling research upon which to base any strong conclusions. There are reasonable grounds for presuming a significant connection to exist between governance and migration, but there is no strong evidence to indicate that such a relationship exists in anything but extreme cases of political and administrative dysfunction. Furthermore, our experience with governance-targeted ODA, both direct and through conditionality, is limited and generally not of sufficient duration to offer firm lessons on effectiveness.

Despite the se shortcomings, a few simple lessons are worth emphasizing.

1. There is clearly an increased willingness to use ODA in an effort to improve the governance capacity of recipient states. For more developed countries with whom the ODA relationship is fairly weak, this effort may take the form of greater cooperation and encouragement rather than financial resources per se. Conditionality to encourage governance reform can also take place within the context of other relationships, including regional political and economic associations such as the Organization of American States (OAS), or the Free Trade Agreement of the Americas (FTAA).

2. There is considerably more interest on the part of governments to use ODA instruments to help manage migration flows.

3. There are no grounds to suppose that governance-targeted ODA should have a direct negative effect on migration. The theoretical and limited empirical evidence suggests the opposite.

4. Governance targets associated with migration management appear to be easily integrated with standard ODA programs, which themselves are becoming more and more sensitive to the importance of political and administrative reform.

5. while still in need of considerable refinement, migration theory does provide some guidance for how to structure ODA programs so as to enhance our capacity to manage migration flows. By reducing incentives to leave by the more effective provision of public goods and services (as opposed to private ones which may be used to finance migration), and by enhancing rather than destroying community ties, the provision of good governance appears to be well suited to the goal of migration management. For example, a more efficient public health system or enhanced democratic participation in government policy making will provide benefits to individuals without necessarily providing them with financial resources personally that can be used to pay the costs of migrating. Thus ODA of this nature may be preferable to programs which generate formal employment opportunities that allow a few persons to acquire the resources which they can use to facilitate emigration. Nonetheless, care must be taken to examine governance-related ODA carefully to ensure that it is informed by migration theory and aid experience.

6. While governance-based ODA is being used in the context of conditionality, insufficient time has elapsed since their inception to provide useful lessons for potential migration policy initiatives. These programs should be monitored, however, and a future study of the effectiveness of governance-based conditionality is probably warranted.

4. Migration, gender, and development assistance

In Rowlands and Weston (1996) we noted that the proportion of women in international migrants has been increasing, especially within Asia. For instance in the case of the Philippines, women outnumber men emigrants by a factor of 12 to 1, in Indonesia they accouflt for two-thirds of all official migrants and in Sri Lanka over 80% (Lim and Oishi 1996). Their predominance means they have been particularly affected by the financial crisis in Asia in 1997/98, of which one of the consequences has been the exodus of thousands of migrant workers (The Globe and Mail, January 12, 1998, p. B7).

Here, we review some of that evidence in greater detail. In addition we consider the importance of a gendered analysis of migration. 16 As noted by Status of Women (1996). The goal of the gender-based analysis process is to integrate a gender perspective into policy analysis and development (p. 11). In the case of migration, this requires differentiating between women’s and men’s experiences of migration, the underlying factors, and the impact on women and men of various policies or programs intended to influence migration flows, whether directly or indirectly. It is in the latter section that we address questions relating to the earlier discussion of governance -- namely, are there particular gendered aspects of governance that might influence migration flows. For instance, if; as suggested earlier, gender equity policies are part of good governance, how will this affect the migration of women? More specifically, if donors wish to promote gender equity whether through specific projects, at an institutional level or through broader, macroeconomic policy initiatives, what effect might this have on migration, at a general level and particularly of women?

(a) Experiences

We begin here by reviewing briefly the experiences of women who mi grate from developing to developed countries, and then review the situation amongst developing-countries. More women than men entered Canada, from 1993 to 1995-- with women accounting for 52.8% of the total, and in the case of immigrants from developing countries, women share was even higher, at 53.2% (NSI 1996, p. 128). This characteristic was evident also in the 1980s (Boyd 1996, p. 194) Most women in recent years have entered in the family class and as independents (especially as domestic workers); in the somewhat smaller business category men predominated (NSI 1996, p. 131).

Despite the large numbers entering under family reunification schemes or as accompanying spouses, immigrant women have virtually as high a labour force participation as non-immigrants (62% compared to 63% in 1991, Citizenship and Immigration Canada or CIC 1996, p. 6) and a higher percentage work full-time (50% compared to 45%). They tend to work in low-skill jobs (clothing, other manufacturing, personal services). In Canada female immigrants are disproportionately employed in clothing (whereas in the U.S., they are over represented in clothing, electronics and domestic/personal services) (Boyd 1996, p. 200). 13% of foreign-born women are in machining and production occupations compared to 5% of Canadian-born women. Whereas foreign-born women accounted for 18% of the female labour force in Canada in 1986, they accounted for a larger share of women workers in the textile sector-- primary textiles (21%), textile products (37%) and clothing (49%)-- as well as electrical products (30%), accommodation (18%) and personal services (21%). In total these industries employed almost half (45%) of women from developing countries; one in eight are in clothing. In the U.S., as many as 1 in 9 women from Central and South America is in clothing (Boyd 1996, PP. 201-206).

A more recent series of profiles of immigrants in Canada confirm that immigrant women’s

occupational concentration has continued. In 1991, for manufacturing as a whole, 10% of immigrant women worked in manufacturing compared to 4% for women born in Canada (CIC 1996, p. 6.) The concentration was even higher for particular groups: 32% of employed women from Viet Nam, 22% from China and 15 per cent from India worked in manufacturing (CIC 1996 a, b, and c).

Zlotnik (1995) challenges the view that female migration has increased markedly, or that international migration flows, at least from South to North, have been highly feminized. Drawing on the varying experiences of the US, Germany, Belgium and the UK, for which some sex disaggregated data is available, she agrees that the data show other participation of women in international migration directed to developed countries has been far from trivial (". 232). But the numbers vary considerably, depending on whether one is looking at women as a proportion of the foreign population, permanent immigrants or legalized aliens (especially undocumented farm workers); migrant workers or family members; gross or net migration.

The experience also vanes from country to country. In Germany and Belgium she found that when the labour migration programs had ended, the impact of incentives to return to countries of origin, coupled with the continuation of immigration under family reunification criteria, led to an increase in the new inflows of women relative to men (".246). In the UK it seems to be the process of family formation and reunification, especially among South Asians, that explains women’s higher share in net migration from developing countries ("p. 250-251). But in Europe net migration has been relatively low; and in terms of total/gross immigration from developing countries, men still predominate, though women’s share is still higher than immigration from other developed countries. Women’s share is also higher for particular regions (e.g. Latin America, East/Southeast Asia compared to South/West Asia or North Africa). The author concludes with an appeal for more visibility being given to this issue, by ensuring the collection of sex-differentiated statistics.

In Asia, the picture is somewhat clearer, though a lack of comprehensive gender-differentiated data, coupled with unrecorded (illegal) flows, makes it difficult to be conclusive about the overall numbers. Women’s share of temporary out migration from Asian countries (including within Asia) rose from less than 15% in 1976 to 27% in 1987 and a decade later was certain to be even higher, given what was known about the gender composition of flows from major exporting countries -- the Philippines, Indonesia, Sri Lanka and Thatland (Lin and Oishi 1996, p. 2)-- though regional totals are not available. In Indonesia, women accounted for two-thirds of migrants from 1984 to 1994 (Amjad 1996, p. 346).

In Asia, too, the types of industries which predominantly employ foreign workers include textiles and electrical machinery industries as well as domestic services (in Taiwan -- Lee and Wang 1996, p. 288), textiles and shoes (in Korea -- Kang 1996, p. 267) and it is likely that this is where most migrant women are employed though the data is not available to prove it.

Another result of industrial expansion in some countries has been the stimulus given to rural -urban flows of young women workers; certainly this has been the case in Thatland and parts of China where certain regional flows are female-dominant (Skeldon 1997, p. 17).

There has been a lot of concern expressed about the status of migrant women workers, and the implications for their own welfare and the overall impact of their migration on their families and home countries. This derives from the nature of their work, as well as the recruitment process, and their legal status in the host country. Problems have been experienced in both developed and developing countries. To deal with legal employment first, there have been allegations of gender discrimination e.g. with respect to wages, and more serious infringements of human and labour rights facing workers in manufacturing in Korea (N.B. without specific references to how women are treated) such as wages often below minimum and certainly below expected levels, physical abuse, industrial accidents (as a result of inadequate training) and long hours without overtime pay (Kang 1996, PP. 270-275). These problems are greater in smaller workshops and in the informal sector where rules are even harder to enforce. In Korea, as in Taiwan and Japan, industrial workers, despite being known as technical trainees, rarely receive the training nor even the salary they are led to expect by the broker (ibid. p267). In many countries migrant workers are not meant to leave the employer for whom they were recruited; or they may not leave for a certain period (two years for domestic workers in some Asian countries, Amjad 1996, p. 358). But often they do so to improve their employment conditions (e.g. higher wages, less abuse).

In some cases expectations may not be met as there is no labour contract. Even where there are standard employment agreements, however, as in the case of domestic workers in Canada, there may be misunderstandings as the provisions oft he agreement are not enforceable; rather the final contract is negotiated between employer and employee, and even this is difficult for the employee to have enforced. For instance the requirement for two years of satisfactory live-in domestic service to apply for landed immigrant status can act as a disincentive to complaints about employers or to leaving (Grandea 1996, p. 23). Sometimes women working as domestics or in entertainment are not covered by labour rights and social security provisions (Lin and Oishi 1996, p. 6).

Women are particularly vulnerable where prostitution is involved, even if it is legal. In other sectors, where migrant women are employed illegally, they usually put themselves at risk by depriving themselves of certain rights (e.g. in Taiwan they are not covered by health insurance; Lee and Wang 1996, p. 294). They are especially vulnerable to abuse by brokers or employers. According to Lin and Oishi (1996), the majority of female migrants from Indonesia and Sri Lanka seek work illegally and are routinely misled by brokers who also overcharge them for their services.

Another challenge for migrant workers, and especially for female migrant workers, is that they tend to face a highly segmented labour market, thus diminishing the scope for them to move to other jobs to improve their conditions. Their concentration in manufacturing makes them vulnerable 40 stop-go migration policies in response to changing export market conditions, as seen recently in the East Asian financial crisis (though lay-offs of men may well be larger in the infrastructure sector, as a result of many contracts being cancelled by governments).

On the issue of remittances, the greater dependence of women on brokers suggests that they may well have to pay a higher share of their wages in brokerage fees than men do. There is no evidence in the literature surveyed here of women’s remittances to their families, and whether they are able to offset the higher fees plus lower wages with a higher propensity to save. Nor is it possible to generalize about the contribution of women’s remittances to the reduction of poverty or income inequalities in their countries of origin. One study suggests that Filipina migrant domestic workers are more highly educated than the general female population in the Philippines, taking into account differences in age, and they tend to come from the richer regions and higher income groups rather than the poorer regions and lowest income groups. But poorer regions have been a source of internal migration; also in recent years they have experienced a faster rate of growth in overseas migration (Amjad 1996, p. 353).

Finally, few of the se studies discusses the impact of women’s migration on their own lives, for instance whether their work abroad contributes to their personal development or improved economic status on their return to their country of origin. Some work has shown that weaknesses in the financial sector have discouraged their families from saving some of their remittances (Lin and Oishi 1996, p. 7). Returning migrant workers may have saved too little themselves to become self-employed, while their other employment prospects may not be improved by their work abroad -- in fact some of this work, notably domestic service, may well have involved de-skilling (Mehmet 1997, p. 25).

(b) Underlying factors

In Rowlands and Weston (1996) we set out a number of theories about migration -- some of which have been used to attempt to predict migration patterns, for instance to assess the likely consequences of changes in economic circumstances or policies. Most of these theories have been developed without any reference to gender, making their relevance to the discussion of women migrants somewhat questionable. For instance it would be interesting to know whether the 'Migration Band' of$350-1 500 per capita is also a useful predictor in the case of women~- or is the starting point of the band higher for women than in the case of men? Do women respond to a given wage or income gap in the same way as men? Do women experience the same 'migration hump'? Are migration networks as important for women as they are for men?

To a large extent the movement of women migrant workers is the result of deliberate policy choices on the part of the governments in both host and home countries. Even without government intervention, however, it is likely that certain economic fundamentals, especially in the context of accelerating outward-oriented regional integration, would lead to migration of women, with both supply-push and demand-pull factors playing their part. We examine these in turn.

On the supply-side there are many reasons for the growth of women migrants. Government promotion of women migrant workers forms part of a more general strategy of using the export of labour-to deal with macroeconomic problems of unemployment (or underemployment) and foreign exchange shortages. Whilst recognizing that there may be economic and social costs, the significant volume of remittances has led many governments to adopt migration as a deliberate development tool and within this women migrant workers are a key element. Such policies have become institutionalized through the creation of vested interests in the public sector responsible for managing migration flows and the private sector recruitment agencies.

Tyner (1996) is interested in those supply-side institutional factors which explain the gendered aspects of migration from the Philippines. Here, the share of women in annual out migration of some 500,000 people has risen to about 40% (p. 406). They are disproportionately employed abroad in domestic services and service industries like entertainment. A factor driving the promotion of Filipina women workers is the intense competition with other labour-surplus countries for labour contracts; governments and/or recruiters increase the marketability of migrant workers by ascribing desirable traits onto their worker pool (13. 410) and by depicting women as domestic workers, nurses and entertainers rather than engineers or construction workers. Patterns of labour migration are socially constructed through the activities of government and private recruitment institutions that control the availability and distribution of labour contracts (p.414). How far labour export agencies are directly responsible for these sexist recruitment strategies, rather than merely responding to stereotypes of labour importers, is less clear.

At other times, in response lo concerns about the conditions facing migrant women, some source governments have sought to influence women’s out migration e.g. by instituting a minimum age and/or skills lo minimize the number of women moving to work as domestics or entertainers, and even by restricting women’s migration to certain countries.

In some cases migration has resulted from the restructuring associated with the adoption of more open economic policies, with women assuming the responsibility for family survival in the wake of the associated adjustments in national labour markets. Their willingness lo do so partly reflects relative freedom from social constraints and already high levels of labour force participation (Lin and Oishi, 1996, p. 3) as shown by the differences in levels of female migration from Pakistan, for example, and the Philippines.

The relatively large surplus of unskilled female labour has led women (e.g. from Indonesia) lo seek employment in other countries (Amjad 1996, p. 356). Another push factor is relative (as opposed to absolute) poverty; as already noted, many overseas domestic workers do not come from the poorest areas or backgrounds. Rather, migrants from poorer areas tend lo mi grate within countries, whereas those who migrate outside are from richer areas (Amjad 1996, p. 353).

The existence of informal social networks has also sustained female migration, perhaps more importantly than for men. Women, especially young women, are more likely than men lo move as part of chain migration, following their sisters or other relatives who are already working as overseas contract workers. They also rely more than men on informal social networks. (Lin and Oishi 1996, p. 4). One study found that 32% of Sri Lankan female migrants used informal channels compared with 16.5% of male workers.

Fertility is one important factor in determining long-term migration pressures. Differences in fertility rates have created differences in demographic structures. For instance the age structure in the Philippines is very different from Japan; and this has contributed on the one hand lo the labour surplus and supply of potential migrants in the Philippines and the labour shortages and demand for immigrants in Japan (Miller and Martin 1996, p. 193). Martin (1997, p. 6) argues that Mexican-US migration has peaked and emigration pressure will decline. One reason is the steady reduction in the birthrate from 7 children per woman in 1970, as a result of which the number of new entrants lo the labour market will fall from 970,000 in 1997 lo 500,000 lo 550,000 by the year 2010. Al the same time he predicts rising rates of economic growth in Mexico will be sufficient not only to absorb those who are currently unemployed and underemployed, but also non-working women who rejoin the labour force.

On the demand-side, within Asia, rapid economic and income growth in some countries until 1997 created employment opportunities often associated with women. For instance, throughout the newly industrialized economies, increases in living standards led to the demand for several hundred thousand domestic workers and entertainers (Stahl 1997, pp. 2-3). In some countries, immigration regulations were relaxed for these categories, thus triggering the inflow of certain types of women workers.

As countries have moved into being service economies, the demand for immigrant assembly workers has also grown. Active immigration policies form part of the official and private sectors response to increased competition in world markets for goods. Kang (1996, p. 277) talks of production cost-cutting methods which transcend national boundaries. Industrial policy may either encourage investment lo flow lo (female) labour surplus countries or allow immigration. Of course some countries have chosen both strategies, with immigration to satisfy the needs of small businesses, many of whom may be too small to relocate (Kang 1996, p. 267). According lo Kang (1996, p. 278), The government policy on foreign workers in Korea has been part of the state’s overall industrial policy, since the import of unskilled foreign labor is often closely related to the process of industrial restructuring. This is typical of other countries -- both richer developing and developed countries like Canada, where for instance the clothing industry has relied on immigrant workers to remain competitive. Others argue, however, that dependence on cheap unskilled labour slows down the restructuring process as it removes the incentive to invest in higher value-added products that require skilled labour.

Gendered characteristics of the local labour market can also be critical in determining the level and nature of women’s migration. For instance, low female participation will create opportunities for migrant women workers. In Malaysia, female labour participation only reached 50% in 1990. This was an increase from the 37% recorded in 1970 (Lin 1996, p. 330), and was associated with a shift of Malaysian women from low-skilled self-employment and agriculture to more skilled and higher-waged employment in the urban areas. The shift created space for Indonesian women to work in domestic services, as well as some agriculture. Evenso, the relatively low participation rate of Malaysian women, compared to other countries (e.g. 65% in the Philippines, Amjad 1996, p. 357) coupled with the manufacturing sector’s need for low-wage assembly workers, led to a large number of Indonesian women working in manufacturing in Malaysia. Malaysian men, for their part, preferred to mi grate to neighboring Singapore (Lim 1996, p. 328). It is possible that if more developed countries were to tap into their hidden or underemployed labour supply (notably amongst women) this would diminish the need for immigrants, provided that some of the jobs they do presently were upgraded.

(c) Policy responses

Sending and receiving countries have introduced various policies or programs which affect women migrants. Some of these are intended directly to influence the numbers of women migrants, others to respond to their particular needs, e.g. to attempt to improve their welfare and indirectly to enhance the net impact of their migration on their home countries or communities. Finally, there is a range of policies which seek to affect migration flows more generally that have specific effects on women. Donor countries, like Canada, may play a role in all three, whether through CIDA development assistance projects, through aid conditionality (as described above), or through multilateral organizations like the World Bank, the UN and even the World Trade Organization.

As mentioned above, some measures have directly sought to limit the migration of women -- for instance through limits on employment permits, though often with little success. Others have focused on reducing both demand and supply factors. Some receiving countries have adopted 'pronatalist' policies, to increase fertility rates and thus to lower immigration needs over the medium-term (Lin 1996, p. 334). Others have considered measures to increase labour participation rates of women as well as men and thus to reduce the need for immigrant workers. This may be through training or through the provision of publicly provided childcare. Another strategy would be to deliberately pursue higher skill goods and service industries and phase-out low-skilled labour-intensive industrial production.

This would accelerate the relocation of some industries to labour-surplus countries and thus reduce the need for emigration in search of work. Other supply-side measures might include more labour-intensive economic policies, such as policies to promote community development, and slower-paced adjustment policies to minimize the social and economic disruption from economic restructuring. Certainly, this is an area where donors, like Canada, and especially international organizations like the UNDP and the World Bank have sought to have some influence.

Box: CIDA and DAC Policies on Women in Development and Gender Equity

CIDAÓs goal is the full participation of women as equal partners in the sustainable development of their societies. To achieve this, CIDA supports initiatives which aim to increase women’s participation in decision-making; to improve their income levels and economic conditions, their access to health and family planning services, educational levels and skills; and to protect and promote their human rights. In addition CIDA is committed to working for the elimination of discriminatory barriers against women CIDA (1 996a). In 1996/97, 4% of CIDA expenditure was on projects directly promoting Women in Development. Bui this likely underreports the value of all Canadian assistance in this area, as the tracking system used by CIDA does not capture WID spending in projects categorized under other headings (e.g. Good Governance) (CIDA 1998).

The OECDÓs Development Assistance Committee in 1997 approved a set of guidelines on gender equality and women’s empowerment which included particular references to poverty, economic development, democratic processes and human rights, education, health, environmental sustainability, and conflict resolution (DAC 1997).

Al a more micro4evel, some labour recipient countries have sought to reduce immigration through development projects -- as cited in Rowlands and Weston (1996). It is important to note that, as in the case of good governance measures, initiatives to promote gender equity may have both direct and indirect effects on migration, and we should differentiate between the impacts on women's migration and migration in general. Investments in reproductive health services might lead to lower population growth and lower overall emigration, but most likely only in the medium-term. 17 In the case of initiatives to reduce women’s poverty (such as microcredit, skills development), which have become popular with donors in recent years (see Box), an additional benefit may be assumed to be diminished push-factors. Bui the empirical evidence discussed in section 1 (albeit relating migration of men and women to more macro indicators) suggests that, instead of reducing the incentive to migrate, such projects may lead to increased out migration in the short-term (though it is not clear whether this is migration of women or their male family members). Certainly this appears to be the case for lower income group according to the migration hump theory, or where the GDI is low, though exactly when the peak will be reached is not clear. Increasing women’s skills (provided these are transferable and recognized) would give them the option of working abroad in other, more secure, higher paying jobs. As a general comment it is worth noting one analyst’s conclusion, namely that migration-responsive policies are more likely to be effective than those that are migration-directive (Skeldon 1997). In other words, it may be easier to address the consequences of women's migration -- e.g. to improve specific aspects of migrant women's lives -- than it is to implement policies which will influence the direction lei alone the number or skill-type of women who migrate.

Some donors have worked with non-governmental organizations to address the particular vulnerabilities of migrant women, given both the nature of the work in which they are employed (entertainment, domestic) and their social isolation. This type of work might involve legal clinics; it could lead to increased migration flows, by reducing the risks and increasing the net returns to women migrant workers. Another approach is through the ratification and enforcement of ILO conventions -- notably Nos. 97 (migration for employment 1949, which confers equal treatment with nationals) and 143 (migrant workers 1975, basic human rights) -- and the more comprehensive UN International Convention of the Rights of All Migrant Workers and Members of Their Families, though none of these address female migrant workers specifically. The UN Convention would require signatories to treat migrant workers no less favorably than nationals with respect to working conditions and wages, and to allow them to join unions. It was adopted in 1990 but will not come into force until ratified by 20 countries.

Another way of regularizing these labour flows, would be through the labour services aspects of the General Agreement on Trade in Services (GATS). As Stahl (1997, p. 4) underlines, the GATS has focused on highly skilled professional services, rather than low and medium skilled services which involve the bulk of migrants, at least in the Asia-Pacific region, Óto the detriment of ordinary international migrant worker’s let alone women migrant workers. Were the GATS or even APEC to consider coverage of these temporary workers, it might help to ensure their protection, or at least to increase their rights.

Another important initiative in the near-term in Asia, would be specific programs or policies specifically designed to help the large number of women returnees to reintegrate in their countries of origin. If adequate measures are not put in place, there could be a surge in the application for immigration to more advanced and stable countries like Canada, or just illegal immigration. More generally, even in non-crisis countries, measures to train women in business management might help them to transform their savings while overseas (however little) into working capital for small businesses; this might reduce the incentive for them to become migrant workers again.

(d) Conclusions

To conclude:

1. The flow of women migrants is increasingly raising questions about the appropriate responses of governments (whether in source, recipient or donor countries) or even in international organizations -- and particularly whether gender-specific targetted initiatives are needed.

2. Many of the arguments about governance-related conditionalities or projects, and the implications for migration, may be relevant in the case of gender, and in many respects the promotion of gender equity forms part of the good governance agenda.

3. Once national and international agencies begin to collect gender-differentiated migration data, it would be useful to analyze The relationship between flows of migrant women and the various economic, social and political indicators considered in section 1.

4. In addition the relationship between women' s migration and specific projects promoting women' s status needs further analysis.

5. For the lime being, initiatives to promote women's income generation activities within their home countries -- both through skills development, facilitating access to credit, and more labour-intensive policies -- should be complemented by measures to assist women returnees.

6. Regularizing The flow of women (and men) migrant workers, e.g. through The GATS, should also be considered.

5. Aid allocations and sources of immigrants to Canada

In this section we briefly examine the pattern of Canada's ODA allocation and the source countries of Canadian immigrants. Data were taken for the top twenty-four recipients of Canadian ODA. 18 These were compared against the list of major source countries of Canadian immigrants in order to identify any similarities. Table 2 provides the comparative figures for ODA and immigration, as ranked by ODA receipts.

Several countries appear to be good candidates for ODA targeting based on migration management needs. As an initial step, however, the data in Table 2 was divided into three groups for analysis. The first group was the countries that were the source of over 10,000 immigrants to Canada in 1997. These rankings have been reasonably stable over that past few years. There were three countries in this group in the top 24 recipients of Canadian ODA: China, Pakistan, and the Philippines. The se countries ranked 2^nd, 6^th and 10^th in terms of percentage of Canadian ODA received. All of these countries, however, are quite large in terms of population and GNP and thus may not be suitable candidates for ODA pro grams that target migration management issues. The impact of Canadian and other ODA on the se countries may well be too small to notice much effect on migration levels, and may not provide sufficient influence over policy. Ah of these countries, however, do make suitable targets for governance and gender based ODA, and if future analysis indicates that these types of programs do affect migration patterns, then a revision of the ODA structure may be possible and warranted. Clearly any such revisions would have to balance migration management with other ODA objectives. However the governance and gender related ODA efforts appear to complement traditional ODA objectives as well, and thus there may not be a substantial amount of conflict in identifying program structures that serve both purposes.

Table 2: Canadian ODA and Immigration Rankings (1997)

Country	Oda ($ million)	Immigration Rank	N° Inmigrants
1. Bangladesh	67.7	15	3262
2. China	46.3	1	24608
3.Rwanda	37	97	153
4. Haiti	36.4	30	1647
5. Indonesia	31.8	85	226
6. Pakistan	29	5	12146
7. Ghana	25.7	34	1262
8. Egypt	23.3	22	2039
9. Senegal	22.7	111	106
10. Philippines	22.6	6	11384
11. Peru	21.3	45	685
12. Vietnam	17.1	23	1995
13. Cameroon	16.8	112	103
14. Mali	15.1	142	29
15. Thailand	14.9	91	190
16. Guinea	14.3	137	40
17. Mozambique	13	160	16
18. South Africa	12.4	27	1749
19. Malawi	12.2	166	11
20. Ethiopia	11.4	41	808
21. Zambia	11.3	106	117
22. Tanzania	11.1	79	269
23. Kenya	10.8	68	373
24. Bolivia	10.8	134	45

The second group of countries are those that sent over 1,000 migrants to Canada in 1997. There were six of these in the top twenty-four Canadian ODA recipients as well: Bangladesh, Haiti, Ghana, Egypt, Vietnam and South Africa. Of these it may be preferable to focus migration-related ODA on Haiti and Ghana, as the se countries have relatively small economies. Thus, ODA may have more effect. These two countries have a long ODA relationship with Canada, much of it focused on administrative reform and democratization. As relatively major contributors of immigrants to Canada, the effect of ODA on migration may be more easily analyzed.

The third set of countries are those that contributed fewer than 1,000 immigrants to Canada in 1997. While these may represent fairly substantial emigration outflow rates for some of the countries in question, the levels may simply be too low for Canada to have much interest in analyzing emigration-ODA linkages. However it is important to consider opportunities for cooperation with other donor countries that may have a migration interest in these countries. Therefore, Canada may wish to examine the total emigration numbers to identify major source countries for which coordinated ODA may have a migration effect.

6. Conclusions and Recommendations

This report has covered a substantial amount of analytical territory. While preliminary in nature, several of the resulting conclusions have important policy implications. In particular it seems appropriate to highlight the following points.

1. The empirical analysis yielded several useful observations regarding statistical linkages between source country characteristics and emigration rates. Many of these linkages are likely to be causal. The results in the report are perhaps unique in terms of identifying these linkages using formal statistical approaches. While the analysis requires refinement and improved data, it nonetheless provides fairly strong evidence upon which to base future analysis of emigration. It is strongly recommended that additional aggregate statistical research be encouraged due to the promising results of this preliminary investigation.

2. The empirical analysis identifies several factors that appear to affect or accompany emigration. Both demographic variables and those that measure material welfare levels appear to have particular significance in determining emigration rates. The two areas of specific interest to this report --governance and gender -- appear to have somewhat more ambiguous effects. The governance measures appear to have the least influence, though the primary data for these variables is also the least reliable. The variables that measured the progress of women in the development process were difficult to interpret given their high correlation with other general welfare measures. Nonetheless, there was weak evidence to suggest that improving women' s conditions did affect (i.e. raise) overall emigration rates. The absence of robust results for the governance and gender variables suggest that more research is required before either becomes a specific target for major migration management policy initiatives. Unfortunately there are currently some fairly significant statistical problems. While improvements to governance and gender-based measures are ongoing, there are clearly difficulties with the measures of migration. We recommend that governments coordinate their activities in terms of defining, collecting, and publishing detailed data on immigration that is disaggregated by sex, disaggregated by a common definition of migrant and refugee, and detailed in terms of country of origin.

3. The theoretical underpinnings of the governance-migration linkage appear to be sound. Improving the quality of public life in source countries should reduce migration pressures. Improved governance capacity may also speed up the country's progress though the migration 'band' during which emigration rates accelerate. Unfortunately it would appear to be too early to identify specific governance-based ODA initiatives that have proved to be effective. While the discussion of governance issues in the development process has a reasonably lengthy pedigree, ODA donors have not integrated these issues into their programming in a substantial manner until relatively recently. This issue, however, will deserve reexamination before too long. The performance of governance-based ODA and governance-based conditionality needs to be monitored and governments should encourage the sponsors and reviewers of such pro grams to conduct and publish evaluations of their efforts.

4. With respect to gender it is clear that more gender differentiation is needed both in data collection, analysis and policy responses. In particular we recommend that the UN and the OECD spearhead the collection of migration data by gender. The types of migration, particularly with respect to sectors of employment and conditions of work, in which women are involved differ in several respects from those involving men, and the se should be taken into account in any attempt to stabilize or moderate these flows. While it is possible to design development assistance projects targeting certain groups of women which might reduce their need to migrate, and others which would help groups to resettle, further research is needed to determine whether these projects will be able to offset the broader macroeconomic tendencies that underlie women's migration as much as men' s. In the meantime it may be more effective for development agencies to focus their efforts on working with governments of source countries to pursue economic policies which enhance women's employment prospects at home rather than abroad.

5. Finally, there are some suitable candidate countries that are significant sources of Canadian immigrants and with whom we have a long standing and significant ODA relationship. On these grounds, Haiti and Ghana appear to stand out as candidates for integrating migration management initiatives into Canadian ODA pro grams. It may also be worthwhile to examine the potential for integrating such initiatives into other country ODA programs in conjunction with the efforts of other donors.

6. The area of migration and migration management has not been analyzed extensively using systematic empirical evidence. This report indicates that such analysis may be worth pursuing. In addition to identifying the relationship between macro-level indicators and country emigration rates, such a framework also allows for the analysis of factors with a specific policy dimension. It should be stressed that the results in this report can be improved considerably in terms of reliability and in terms of identifying the degree and sources of sensitivity. Nonetheless they provide a reasonable starting point for subsequent empirical analysis. Future analysis should include an explicit examination of the geographic dimension of emigration, and should also attempt to capture the effects of hysteresis (as suggested by the 'network' and 'relative deprivation' theories of migration -see Massey et al. (1993), Stark and Taylor (1991) and Russell and Teitlbaum (1992)). The issue of network-induced migration may also be of considerable importance in Haiti, one of the countries identified in section 5 as being of potential policy interest for Canada.

The governance dimension of migration management policies will also become more interesting as the evaluations of governance-based ODA projects grow in number. From the perspective of migration policy, however, it may be useful to delay such a study until a stronger empirical link is identified between governance conditions and migration.

There is also a need for specific evaluation of projects, institutions or broader policies targeting women in migrant-source communities or regions.

It should be noted that the report has not addressed the difficult political and ethical problems associated with managing migration. By focusing on the reduction of migration pressure as opposed to the use of administratively imposed limitations on mobility, however, many of these difficult issues are avoided. By fostering improved welfare in source countries we may eventually reduce the need for people to use migration as a means of personal development, thereby avoiding the potentially substantial personal and social cost imposed on both the migrants and their home communities.

APPENDIX 1: DATA DEFINITIONS AND SOURCES

Migration Statistics

U.S.: Source: Statistical Abstract of the United States (1996): "Immigrants, by Country of Birth: 1990" from Immigration and Naturalization Service, Statistical Yearbook, annual; and releases.

Definition:

Immigration statistics are prepared from entry visas and changes of immigration status forms. Immigrants are aliens admitted for legal permanent residence in the United States. The category, immigrant, includes persons who may have entered the United States as non-immigrants or refugees, but who subsequently changed their status to that of a permanent resident. Refugees are considered non-immigrants when initially admitted into the United States.

Canada: Source: Citizenship and Immigration Canada (1998), unpublished data. Data does not include refugees.

Australia: Source: "Settler Arrivals by Region/Country of Birth 1989-90" from Bureau of

Immigration, Multicultural and Population Research Statistical Focus: Historical Series, Oct. 1945- June 1995. June 1996.

Definition of Settler:

Settlers comprise of persons arriving in Australia who hold permanent visas, regardless of intended period of stay, New Zealand citizens who indicate an intention to settle, and those who are otherwise eligible to settle (e.g. overseas-born children of Australian citizens).

New Zealand: Source: Statistics New Zealand.

Europe: Source: "Recent Trends in stocks and flows of immigrants in the member countries of the Council of Europe: A Statistical Approach" in Political and Demographic Asects of Migration Flows to Europe. Council of Europe, 1991.

The major sources for immigration are taken from the National Institute of Statistics and Eurostat in Luxembourg. Illegal immigration is not taken into account. The criteria seems to be the "intention to reside for a minimum duration other than one year" taken from a United Nations definition.

We1fare Measures: Source: United Nations Development Programme (UNDP), Human Development Report: New York: Oxford University Press, 1992.

1. Human Development index (HDI). Definition: The HDI includes three key components: life expectancy, educational attainment and income. Educational achievement is measured by two educational stock variables: adult literacy and mean years of schooling. The measure of educational achievement is adjusted by assigning a weight of two-thirds to literacy and one-third to mean years of schooling. A high HDI indicates a high standard of living.

2. Life Expectancy. Definition: Life expectancy at birth, years (1990). The number of years a newborn infant would live if prevailing patterns of mortality at the time of its birth were to stay the same throughout its life.

3. GDP per capita (adjusted). Definition: GDP per capita adjusted to facilitate international comparisons.

4. Education. Definition: Educational Attainment (1990). The measure of educational achievement is adjusted by assigning a weight of two-thirds to literacy and one-third to mean years of schooling. A high number indicates a high level of educational achievement.

5. Percentage without water. Definition: Population without access to safe water (millions) (1990).

6. Health Budget. Definition: Public expenditure on health as % of GDP, 1990. Health expenditures: expenditures on hospitals, health centers and clinics, health insurance schemes and family planning.

7. GNP growth. Definition: GNP per capita annual growth rate %, 1980-89. The higher the percentage, the higher the rate of annual growth from 1980-89.

Demographic Variables Source (unless stated otherwise): United Nations Development Programmed (UNDP), Human Development Report: New York: Oxford University Press, 1992.

8. Population. Definition: Estimated population, 1990. Some: figures taken from The World Bank World Development Report (1991).

9. Population Density. Definition: Population Density (per l000ha) (1990). The total number of inhabitants divided by the surface area. A high number indicates a high population density.

10. Population growth. Definition: Annual Population Growth % 1960-1990. p.170. A high number indicates a high annual population growth rate from 1960-90.

11. Ethnic division. Source: Mauro, Paolo. '1Corruption and Growth" in The Quarterly Journal of Economics Vol. CX Issue 3. Cambridge: MIT Press, Aug. 1995. Definition: The higher the number, the greater the amount of fractionalization. The index of ethnolinguistic fractionalization from 1960 is from Taylor and Hudson (4972).

Governance Variables Source: Mauro, Paolo. "Corruption and Growth" in The Quarterly Journal of Economics Vol. CX Issue 3. Cambridge: MIT Press, Aug. 1995.

12. Political stability. Definition: The higher the number, the higher amount of political stability. The political stability index is the average six indices: institutional change, social change, opposition takeover, stability of labor, neighboring countries and terrorism.

13. Administration. Definition: A high value of bureaucratic efficiency means the country is more efficient. The bureaucratic efficiency index is the average of three indices: judiciary, red tape and corruption.

Gender Variables Source: United Nations Development Programmed (UNDP), Human Development Report: New York: Oxford University Press, 1992.

14. Female education. Definition: Average number of years of schooling received per person age 25 and over. Ah figures are expressed in relation to the male average, which is indexed to equal 100. The smaller the figure the bigger the gap, the closer the figure to 100 the smaller the gap, and a figure above 100 indicates that the female average is higher than the male.

15. Gender Development index (GDI). Definition: The UDI value attempts to capture achievement in the same set of basic capabilities included in the HDI: life expectancy, educational attainment and income but adjusts the HDI for gender inequality. The lower the number the more deprivation.

16. Gender empowerment measure (GEM). Definition: (p.39 and pp.123-24) The GEM measures gender inequality in key areas of economic and political participation and decision making. The first two variables used are women's and men's percentage shares of administrative and managerial positions and their percentage and share of technical jobs, reflecting their economic participation and decision making power. The third variable is women's and men's percentage shares of parliamentary seats, reflecting their political participation and decision making power.

References

Adepoju, A. (1995): Emigration Dynamics in Sub-Saharan Africa, International Migration 33(3/4): 315-390.

Ámjad, Rashid (1996): "Philippines and Indonesia: On the Way to a Migration Transition," Asian and Pacific Migration Journal. 5:2-3.

Appleyard, R. (1992): "Migration and Development: A Global Agenda for the Future, International Migration 30 (Special issue): 17-31.

Bardhan, P. (1997): The Role of Governance in Economic Development. (Paris: OECD Development Centre).

Belnye, F., D.Q. Dang, D. Ivascanu, 1 Ketcheson, M. Shulman, and R. Slavens (1998): The Role of Political Conditionality in the International Financial Institutions. Working Paper, The Norman Paterson School of International Affairs, Carleton University.

Bohning, W. and M. Schloeter-Paredes, eds. (1994): Aid in Place of Migration?: Selected Contributions to an ILO-UNHCR Meeting (Geneva: ILO).

Bohning, W. (1991): Integration and Immigration Pressures in Western Europe, International Labour Review 130(4): 445-458.

Boyd, Monica (1996): "Female Migrant Labor in North America: Trends and Issues for the

1 990s," in Simmons, Alan (editor). International Migration, Refugee Flows and Human Rights in North America: The Impact of Trade and Restructuring. (New York: Centre for Migration Studies).

Brem, M. and P. Rawkins (1992): Development Assistance, Migration and the Canadian Agenda: Linking Policy, Programmes and Actions ILO Working Paper 92-69 (Geneva: ILO).

Canadian International Development Agency (1998): How CIDA Spends ODA Pro gram Funds -The 1996/97 Priority Coding Report. mimeo.

Canadian International Development Agency (1997): Performance Report (Ottawa: CIDA).

Canadian International Development Agency (1996): Government of Canada Policy for CIDA on Human Rights, Democratization, and Good Governance (Ottawa: CIDA).

Canadian International Development Agency (1996a): "CIDAÓs Policy on Women in Development and Gender Equity." Ottawa.

Canadian International Development Agency (1993): Women in Development: A Policy Statement (Ottawa: CIDA).

Citizenship and Immigration Canada (1996): "A Profile of Immigrants in Canada," Ottawa from http ://www.ci. gc.ca/english/ref/profile/9608imne.html

Citizenship and Immigration Canada (1996a). "A Profile of Immigrants from Viet Nam in Canada," Ottawa from http :/Iwww.ci.gc.ca/english/ref/profilel9608vtne.html

Citizenship and Immigration Canada (1996b). "A Profile of Immigrants from China in Canada," Ottawa from http://www.ci.gc.ca/english/ref/profile/9608chne.html

Citizenship and Immigration Canada (1996c). "A Profile of Immigrants from India in Canada," Ottawa from http:/Iwww.ci.gc.ca/english/ref/profile/9608inde.html

DeWind, Josh and David H. Kinley 111(1988): Aiding Migration: The Impact of International Development Assistance an Haiti, (London: Westview Press).

Diaz-Briquets, Sergio and Sidney Weintraub, eds. (1991): Migration Impacts of Trade and Foreign Investment: Mexico and Caribbean Basin Countries (Boulder: Westview Press).

The Economist (1994): Democracy and Growth: Why voting is good for you, August 27.

Ghosh, B. (1992): Migration Development Linkages: Some Specific Issues and Practical Policy Measures,Ó International Migration 30(3/4): 423-456.

The Globe and Mail, January 12, 1998, p. B7.

Grandea, Nona (1996): Uneven Gains. Filipina Domestic Workers in Canada. (Ottawa: The Philippines Human Resource Development Program with The North-South Institute).

International Monetary Fund (1997): Guidelines Regarding Governance Issues. News Brief http ://www.imf.org/external/np/sec/nb/ 1 997/NB9715.htm (August 4).

Kang, Su Dol (1996): "Typology and Conditions of Migrant Workers," Asian and Pacific Migration Journal. 5:2-3.

Kim, Won-Bae (1996): "Economic Interdependence and Migration Dynamics in Asia," Asian and Pacific Migration Journal. 5:2-3.

Laroque, P. (1987): Conclusions, International Migration Review XXI( 1): 145-154.

Lim, Lin Lean (1996): "The Migration Transition in Malaysia," Asian and Pacific Migration Journal. 5:2-3.

Lim, Lin Lean and Nana Oishi (1996): "International Labour Migration of Asian Women:

Distinctive Characteristics and Policy Concerns," ILO. February.

Martin, P. et al (1996): "Introduction," Asian and Pacific Migration Journal. 5:2-3.

Martin P. (1997): "Mexico-U.S. Migration Three Years after NAFTA," http ://migration.ucdavis.edu/mm21/PLM 1 0-6-97.html

a

Massey, D. and A. Joaquin, G, Hugo, A. Kouaouci, A. Pellegrino, J.E. Taylor (1993): "Theories of International Migration: A Review and Appraisal," Population and Development Review 19(3), 431-466.

Mauro, P. (1995): Corruption and Growth, Quarterly Journal of Economics CXI: 681-712.

Mehmet, Ozay (1997): "Labour Mobility Issues in APEC with Special Reference to the Asia Pacific," in Heather Gibb (editor). Canadian Perspectives on Labour Mobility in APEC Ottawa: The North-South Institute with the Asia-Pacific Foundation.

Miller, M. and P. Martin (1996): "Prospects for Cooperative Management of International Migration in the 2lst Century,". Asian and Pacific Migration Journal. 5:2-3.

Molle, W., J de Koning, and C. Th. Zandvliet (1993) in Bohning, W. and M. Schloeter-Paredes, eds.(1994): Aid in Place of Migration?: Selected Contributions to an ILO-UNHCR Meeting (Geneva: ILO).

Mosley, P., Hudson and Horrell (1987)...citation incomplete.

The North-South Institute (1996). Canadian Development Report 1996-97. Ottawa.

Olesen, H. (1995): "Migration...Into the 2lst Century," paper presented to Citizenship and Immigration Canada Staff, May 23, 1995.

Olson, M. (1997): Big Bills Left on the Sidewalk: Why Some Nations are Rish, and Others Poor, Journal of Economic Perspectives 10(2): 3-24.

Oualalou, Fathalla (1992): Migration for Employment. International Aid to offset the ban on immigration to Europe. Working Paper, Employment Programme (Geneva: International Labour Office) April 1992.

Peace Initiatives (1997): Good Governance 3(3): 1-22.

Przeworski, A. and F. Limongi (1993): Political regimes and Economic Growth, Journal of Economic Perspectives 7(3).

Ranis, U. (1992): International Migration and Foreign Assistance: Concepts and Application to the Philippines ILO Working Paper 92-63 (Geneva: ILO)

Rowlands, Dane, and Ann Weston (1996): How Aid, Trade and Development Affect Migration, Canadian Foreign Policy IV(1): 1-47.

Russell, 5. and M. Teitlbaum (1992): International Migration and International Trade (Washington D.C.: World Bank).

Skeldon, Ronald (1997): "The Migration of Women in the Context of Globalization in the ESCAP Region". mimeo.

Stahl, Charles (1997): "Some Observations on Trade and the Export of Labour Services," Working Paper No. 2. Asia-Pacific Centre for Human Resource and Development Studies, University of Newcastle, Australia.

Stark, O. and J.E. Taylor (1991): "Relative Deprivation and Migration: Theory, Evidence, and Policy Implications" in S. Diaz-Briquets and S. Weintraub eds. Determinants of Emigration from Mexico, Central America, and the Caribbean (Boulder Co.: Westview Press), 122-144.

Status of Women (1996): Gender-based Analysis. A guide for policy-making. Working Document. (Ottawa).

Tyner, James A. (1996): "The Gendering of Philippine International Labor Migration," The Professional Geographer. 48:4, November.

Weintraub, S. and S. Diaz-Briquets (1992): The Use of Foreign Aid to Reduce Incentives to

Emigrate from Central America, World Employment Programme Working Paper (Geneva:

International labour Organization).

World Bank (1993): Governance: The World Bank’s Experience (Executive Summary). http ://w'ww.worldbank.org/html/hcovp/pubsec/governan.html.

World Bank (1989): Sub-Saharan Africa: From Crisis to Sustainable Development (Washington D.£.: The World Bank).

Zlotnik, Hania (1995): "The South-to-North Migration of Women," International Migration Review. xxxix: 1

---

16 As underlined above, we do not consider refugee flows here, although there are particular gender aspects as women form a large share of refugees. Rather we are focusing on economic migrants.

17 Ghosh (1992) estimates that only 1.2% of DAC ODA is targeted towards population-related activities.

18 Data were taken from the Canadian International Development Agency's Performance Report for 1997. The data represent country specific disbursements excluding financing provided by international financial agencies (such as the World Bank) and technical cooperation.

---

1 The research assistance of Vicky Edgecombe is acknowledged.

2 See CIDA (1993, 1996).

3 Japan's reports were not compatible, though the numbers of official migrants was low. There were some problems with the New Zealand numbers as well. Finally, no numbers were available for potential recipient countries such as Singapore or other recently industrialized countries that could be clearly be classified in the ranks of the 'North'.

4 Japan is, of course, a major provider of ODA.

5 Obviously large outward migration from a small population indicates stronger migration pressure than the same level of emigration from a large population. Thus absolute emigration numbers need to be scaled by population to provide a reasonable measure of emigration pressure. The ANOVA analysis was run on total emigration as well and, not surprisingly, most of the results were much less significant statistically. The most important measure explaining migration numbers in total was population itself; a result confirmed by the regression analysis. In other words, it is unreasonable to try to explain emigration from a country without taking the population base into account.

6 Few East European or former Soviet states are in the sample due to poor data. Furthermore, the sample year of 1990 was one of great turbulence in these countries, and including them in the sample may have introduced a substantial bias.

7 The expected effect of population growth is probably the opposite: high rates of population growth, at least in the past, may be thought of as contributing to greater competition for employment and public services, and perhaps more pressure on the environment. The inversion of the relationship may be the result of an absence of correlation between past and current growth rates, or because low population growth rates reflect poor living conditions.

8 Not ah of the data were available for ah of the countries in the sample. Hence, certain combinations of explanatory variables in a regression equation led to a substantial number of countries being removed from the sample for that particular estimation.

9 The range is determined by solving the equation 0 = -0.000336x² + 2.21x - 1032 for x, which here represents UDP per capita. The actual numbers should be interpreted with some caution since the parameter estimates used are only approximate. Oleson (1995) reported numbers of $350-$1500 (US) per capita as the range. A direct comparison, however, is not possible since the UNDP numbers used here are adjusted to reflect purchasing power parity and other factors, thus making them more comparable to actual $US figures.

10 Only Babrain, Bangladesh, Hong Kong, Korea, Mauritius, and Singapore had density levels in excess of 4255.

11 This makes income data more internationally comparable by taking into account factors such as purchasing power parity.

12 Some elements of this section have benefited from the work by Belnye, Dang, Ivascanu, Ketcheson, Shulman and Slavens (1998).

13 See Bardhan, 1997 for a recent review of the linkages between corruption and development. Mauro (1995) is an important study detailing the empirical relationship. It should be noted, as discussed in Bardhan, that earlier researchers had theorized that corruption could actually assist the development process in some instances. This view seems less persuasive now.

14 Recent research on the linkages between democracy, human rights, and development include World Bank (1993), Przeworski and Limongi (1993), and the Economist (1994). Olson (1997) provides strong theoretical and prima facie evidence of the importance of public institutions and policy.

15 Ranis (1992) emphasizes status as a specific factor in the migration decision in the Philippines. Re also cites the more traditional studies of the major 19^th migrations from Europe in which Britain - the most developed country - was also the major source of migrants.