Emigration Intentions: Mere Words or True Plans? Explaining International Migration Intentions and Behavior

Do people follow up on their intentions? In this paper we confront the emigration intentions formed by inhabitants of the Netherlands during the year 2004-2005 and the emigration steps they took in the subsequent two years. Three results stand out. First, it appears that intentions are good predictors of future emigration: 24 percent of those who had stated an intention to emigrate have actually emigrated within two years time. Second, within the group of potential emigrants, those who have emigrated and those who have not yet emigrated, do not differ much from each other. The potential emigrants who have not yet emigrated are in poorer health. Third, the forces that trigger emigration intentions are also the same forces that make people actually move.


Introduction
Testing theories of international migration is often done by employing one of the two dominant methods in social sciences: by using either revealed preferences or stated preferences. The revealed preferences approach is often preferred by economists (Borjas 1987(Borjas , 1991Williamson 1998, 2004) who interpret the actual movements of people as a reflection of preferences. Social demographers, geographers and psychologists prefer the use of stated preferences, i.e. intentions (De Jong 2000;Van Dalen, Groenewold and Schoorl 2005). However, the interest of economists in social forces and the recognition of the bounded rationality of citizens and the interest of sociologists in rational choice have yielded a more diffuse picture of methodologies employed across the spectrum of the social sciences. Migration research that combines both methods is rare and the few examples that have combined intentions with behavior are restricted to internal migration 1 and not international migration. 2 An important reason for the focus on internal migration is that it is relatively easy to trace such migrants for follow-up surveys and check whether they have realized their intentions. These studies suggest that intentions are good predictors of future behavior, but it remains an open question whether intentions to move abroad lead to actual migration. A growing body of literature uses intention data 3 in both academia and policy circles with the tacit assumption that they approximate actual migration. A priori, one would expect that the gap between intention and action be large, or at least larger than for internal migration data. Prospective migrants not only need the resources to finance their move abroad, but they also need to overcome formal barriers such as obtaining visas, residence permits and/or work permits-all legal documents that are increasingly difficult to obtain.
Besides the simple question of whether intentions are good predictors, there are three supplementary reasons why a check on intentions data, especially with respect to migration, is important.
First of all, validation of a research tool should be at the top of the list for any researcher and this is in no way different for the use of migration intention data. For some researchers, intentions lack credibility, or as Constant and Massey (2002:23) put it: "intentions are notoriously unreliable as guides to eventual behavior." Intentions are by no means perfect determinants of actual behavior. Social psychologists in particular have attempted to tackle the question of why intentions can only be partly considered a suitable indicator of actual behavior (Sutton 1998). In general, behavioral intentions are good predictors of actions if they concern relatively specific behavior in a restricted time span in which individuals have a great freedom of choice. However, the claim that intentions are unreliable guides is perhaps a plausible statement but empirically unfounded. It is more or less common practice in social psychology to use the theory of reasoned action (Ajzen and Fishbein 1980;Fishbein and Ajzen 1975) and the closely related theory of planned behavior (Ajzen 1991) as the basic frame of reference. The theory of reasoned action assumes that most behaviors of social relevance are under volitional control and therefore it is likely that the behavioral intention of a particular act is both the immediate determinant and the single best predictor of behavior.
Secondly, validating intentions helps to put incredible or extreme research results in perspective. It often happens that media report on some spectacular result of researchers, like half of all Dutch youth want emigrate, more than 60 percent of Albanians will probably emigrate (Papapanagos and Sanfey 2001) or 13 percent of the British are thinking about emigrating. 4 The firmness with which those results are reported does not correspond with the looseness of the type of questions used to derive these conclusions. Willingness or dreaming of a life abroad is an entirely different matter than asking people about their plans to emigrate in the near future. In other words, the predictive value of intentions questions enhances those questions as a tool of research.
Thirdly, confronting intentions with behavior helps to understand the selection process underlying the migration process. One clear advantage of using migration intention data is that it allows self-selection theories to be tested without having to deal with the sort of sample selection problems that are associated with host-country data as in the case of Borjas (1991).
Quite a few of the studies that test for self-selection among migrants rely on host-country data and, as Liebig and Sousa-Poza (2004:126) pointed out, this sort of approach can become problematic because specific host-country characteristics, such as migration policy, historical links and geographical proximity, are bound to bias immigration to these countries. By confronting intentions with behavior we are in the unique position to see whether the selection problem changes. If the determinants of emigration intentions are completely different from those that drive actual emigration behavior, or if the profile of the people who have emigrated diverges strongly from those who had intentions but are still in doubt about whether to leave their country, then this result sheds light on why dreams do not come true immediately or why dreams are put off forever.
The arrangement of this paper is as follows. First, we will briefly expound on the theory of emigration (Section 2), which will function as an input into the empirical analysis of intentions and behavior. Next, we will describe the method and data (Section 3). Section 4 contains the empirical analysis. In the present study, we focus on the case of the Netherlands as this country seems to apply well to the conditions of free choice and in addition, this country has shown a strong increase in emigration numbers in the past few years. The upsurge in emigration has surprised both experts and policy makers and left them in wonderment.
Why leave a country where the income level is high, public services are extensive and the standard of living is the envy of immigrants from less developed countries? In a previous study Van Dalen and Henkens (2007) depicted how strong the intentions to emigrate were among the Dutch population and what triggered potential emigrants to state their intentions and plans. However, whatever happened to those emigration plans remains an open question; we examine exactly this question in this paper by using the same database and confront intentions with subsequent behavior. We conclude by summarizing and discussing the main findings in Section 5.

Migration Theory
To understand emigration from a high-income country like the Netherlands, we claim that the standard theory of migration, in which people are triggered by net wage differentials, is a good starting point. This may describe emigration in normal times but it is unlikely to explain the current emigration wave in a country like the Netherlands. To capture the decision process of people contemplating emigration, we extend earlier research on emigration along two lines.
First, this is done by paying attention to individual characteristics that are thought to underlie the economic, social and psychological net benefits of emigration. Besides the traditional forces in which migration can be viewed as a decision influenced by the human capital stock of individuals, we include the forces exerted by social networks and pay attention to the effects of personality traits on the likelihood of emigration, as suggested by psychologists (Berry 2001;Boneva and Frieze 2001) and sociologists (De Jong 2000;De Jong and Fawcett 1981).
Second, to understand emigration from high-income countries we focus not only on factors that refer to individual characteristics, but also on the perceived quality of the public domain in the source country. Several earlier studies have focused on explaining emigration from highly developed countries, but none of the studies pays attention to both the public and private domain of life. Evaluating the quality of the public domain of life involves both the institutions (social security, educational system, law and order) as well as the "public goods" that these institutions produce: social protection, safety, environmental quality, education, integration, etc. We will expand on each of these factors separately and start with the most common driving force: human capital.

Human capital
Human capital theory offers predictions about the intention to migrate that can be tested, like the role education, health and age play in triggering emigration. Starting with the most obvious element-education-we see that the large body of economic migration studies focus on the question of whether migrants are favorably selected or not, where favorable selectivity refers to the tendency that migration flow is disproportionately represented among the skilled or talented of a source country. With a simple model of migration, Chiswick (1999) demonstrates how human capital, approximated by schooling or innate ability, increases the probability of migration. One of the reasons why favorable selectivity occurs is that higherability individuals can recoup the out-of-pocket costs of migration faster and they may also be more efficient in migrating and adapt more easily to the conditions of the destination country (language, norms and rules). With regard to emigration from high-income countries, we expect that the better educated will be more inclined to migrate because their human capital is more internationally transferable, yielding lower transaction costs tied to migration. Furthermore, the opportunities for the higher educated to work in an international labor market are greater, also because the international labor market offers more opportunities to specialize than the internal market. Although the arguments underlying the favorable selectivity of migrants sound persuasive, in the end it remains an empirical question of whether the lesser skilled or the higher skilled are more likely to emigrate. 5 The influence of people's health status on their emigration decisions somewhat follows the logic of the human capital model. Health is part of an individual's human capital and potential migrants can recoup the investment costs of migration only if their health is good. In other words, poor health is associated with a weaker intention to emigrate than good health.
The role of age in migration decision making becomes clear by posing the question: Which moment in the life course is the best time to emigrate? The younger the migrant, the longer the period of time the migrant needs to recoup the "investment," that is, moving the human capital from the source country to the destination country. Other factors might reinforce the age bias in migrating when the age-wage profile is steeper and when personal migration costs are age-related. In short, we would expect the desire to emigrate and actual emigration to be more evident among the young than among the old.

Social networks
Making migration decisions is not a purely individual decision and the influence of networks is bound to affect the decision to emigrate (as stressed by Stark and Bloom 1985). The availability of a social network of friends and family abroad is believed to increase the probability of emigration. The role played by migrant networks is well known (Massey 1999).
Networks not only provide information that may give a more accurate estimate of what potential migrants can earn abroad in real terms, but also, and more importantly, they offer services that reduce the personal costs of migration and integration and thereby can favor the balance to move abroad. This is in fact what may be behind the phenomenon of "chain migration." Most studies therefore assume a relationship between the costs and adjustment costs of migration and the size of the network. In other words, the larger the network of potential migrants, the more set they will be on emigrating. We will restrict ourselves to the theoretical prediction that the larger the number of emigrants in an individual's network, the stronger the intention to emigrate.

Personality
Leaving your home country can give rise to strong feelings of uncertainty, since living and working abroad may mean getting acquainted with a new culture or even adapting one's identity. Psychological characteristics are assumed to predict whether people will enter a new and unfamiliar situation as well as their affective reactions to a novel situation. In this article, we focus on two main personality traits that may contribute toward strong emigration intentions: sensation seeking and the level of self-efficacy. First, sensation seeking (Horvath and Zuckerman 1993) can be of great importance to such daring moves as migration. Risk lovers or sensation seekers have a tendency to take more risks and perceive the world as less threatening. It is therefore likely that the more adventurous or risk loving a person is, the stronger the intention to emigrate will be.
The second personality trait that is deemed important in the decision to emigrate is self-efficacy. Self-efficacy, or the belief that one can effectively cope with a given situation, predicts whether people will enter a new and unfamiliar situation (Sherer et al. 1982). Selfefficacy predicts confidence in the ability to deal with changes and unfamiliar situations.
Given that emigration is a new and uncertain experience, we assume that higher scores on self-efficacy will be associated with stronger intentions to emigrate and subsequent emigration steps. In fact, self-efficacy is central to testing the theory of planned behavior (Ajzen 1991) as this personality trait also helps to understand why behavior differs from intentions.

Public domain
The basic theory of migration tacitly assumes that what drives migrants are differences in utility derived from private consumption financed by private lifetime wealth. However, it makes sense to assume that international migration is not only driven by private goods but also public goods. The hypothesis that public goods trigger migration is essentially the theme put forward by Tiebout (1956) for the competition between local communities in providing local public goods. In Tiebout's traditional system, free migration between localities can deliver the best possible situation in which communities supply the commodities and services individuals want and local governments produce these goods in an efficient manner. By decentralizing the public goods provision, people organize themselves in various communities that best fit their preferences. Needless to say, such an ideal outcome can only exist when everyone is mobile and there are sufficient communities to fit citizens' preferences.
Another line of research is pursued by geographers and economic geographers who focus on local amenities as the driving force for regional migration and development, where the influence of amenities is manifested through wage or land rent compensation (Knapp and Graves 1989). The focus on location-fixed amenities is largely on the externalities of human action. The fact that people are attracted to cities because of the job opportunities there makes the agglomeration itself a source of positive externalities. And the fact that people move away from environmental hazards, like waste sites or polluted cities, is a case of migration in response to negative externalities. The driving force is a situation in which people neglect the effects-positive or negative-their actions can have on the welfare of others. The theory of amenities assumes that the mobility behavior of households or firms is also the mechanism by which location-fixed amenities are capitalized into labor or rent prices (Rosen 1974). The effects of the public domain are effectively translated back into private domain values and the simple migration decision criterion, on which neoclassical theory rests, applies again.
In this study we argue that the public domain may perhaps affect relative prices, but that it is valued primarily for its own sake. In our view, the quality of the public domain depends on how citizens perceive the way in which government institutions function and the goods and services produced by these institutions. Public goods are traditionally defined as goods that are non-excludable-no one can be excluded from the use of the public good in question-and non-rivalrous -access by one user to the good does not in any way diminish other people's capacity to benefit from the good. Public goods should be interpreted broadly since "goods" are, by definition, a response to a situation in which externalities are complex, that is, most citizens benefit from the provision of a good, but every individual is tempted to ride free on the efforts of others. For instance, silence is valued by most citizens but to "produce" silence everyone has to act in accordance with the rules, certainly in densely populated areas or in areas where air, road or rail traffic is heavy. In some cases where externalities are straightforward, private transactions or negotiations can solve the problem, but once it affects the people of a city, region or nation and the transaction costs tied to individual corrective action become excessively high, collective action may become necessary. Perceptions of the quality of the public domain are therefore a reflection of both the "goods" and the governance institutions that try to correct externalities of individual action (cf. Kaul and Mendoza 2004).

Method and data
To answer the central research questions-(1) Are intentions good predictors of behavior?
and (2) What drives those people who actually emigrate and those who still hesitate about their move abroad?-we collected data using a two-step approach. First, an emigration survey was carried out from October 2004 to January 2005 using a targeted sampling method (Watters and Biernacki 1989). This survey was geared toward the over-sampling of potential emigrants to alleviate the problem that the number of potential emigrants was expected to be too low to produce adequate analyses in national representative samples. Visitors to an Expat fair, who saw themselves as potential emigrants 6 , were asked to participate in the survey. A total of 533 potential emigrants received a questionnaire and 214 questionnaires were returned (a response rate of 40 percent). The second step consisted of a nation-wide emigration survey, which was carried out in January 2005. The survey was conducted by the CentERdata survey institute of the University of Tilburg 7 , which maintains a national representative panel of households in the Netherlands. We tested whether respondents with emigration intentions differed significantly in both samples in terms of the socio-demographic variables age, education, income and gender, as significant differences may point to biases in our sampling method. We found no statistically significant correlation. The two samples were combined to analyze emigration intentions in the Netherlands (N = 1,489). The Expat-sample was used to analyze expectations regarding the preferred destination countries, which were not posed in the nation-wide emigration survey. In this article, we focus on the emigration of native-born individuals because they are, a priori, the ones who face the dilemma of leaving their home country.
The emigration intention was measured by an ordered categorical variable representing the response to the question: Do you intend to emigrate in the near future? The responses comprised a five-category ranking of intentions: (1) no, certainly not; (2) no, probably not; (3) maybe; (4) yes, probably; and (5)  (2) When I have to work according to fixed rules, I easily get fed up with them; and (3) People or things that always stay the same, bore me. Respondents could answer on a five-item scale ranging from (1) totally agree to (5) totally disagree (Cronbach's alpha = 0.64).
• The variable Self-efficacy is based on three Likert-type items (cf. Bosscher and Smit 1998): (1) When I make plans, I am convinced that I will succeed in carrying out these plans; (2) When I decide to do something, I firmly cling to that decision; and (3) When unexpected problems occur, I do not handle them well. Answer categories varied from (1) totally agree to (5)  Descriptive statistics for all explanatory variables are presented in Table 1, disaggregated by those who have emigrated and those who did not emigrate or not yet emigrate. The statistics are self-explanatory and give us a clue as to which dimensions distinguish emigrants from the large crowd of non-migrants. Just to state the most prominent differences: emigrants are clearly younger than those staying behind, dominated by men, highly educated (51 percent has a higher education compared to 39 percent of the non-migrants) and their number of network contacts abroad is almost twice as large.

Are intentions good predictors of behavior?
In 2005, three percent of the Dutch population had plans to emigrate in the near future. After two years it appears that these intentions are good predictors of emigration behavior. Table 2 gives a precise overview of how strong the emigration intentions translate into behavior: 36 percent of those respondents who rated their emigration plans for the near future as very probable have actually emigrated after two years. If we take all respondents with clear emigration intentions, then the rate of materialization of plans is 24 percent. Emigration among respondents who claimed to have no emigration plans at all and who subsequently emigrated has hardly occurred. The fact that quite a number of the potential emigrants of 2005 have not emigrated is in no way a definite measurement of the link between intentions and behavior. The time horizon used in planning may differ from one migrant to the other, so these rates can only increase because the intention question was based on the undefined notion of "the near future." Furthermore, personal reactions to our query about the respondents' place of residence suggests that emigration is no simple matter as some have to deal with the administrative procedures or red tape of emigration. In some other cases, the selling of a house or business is a timely affair and people, in their reply to our queries, stated that leaving the Netherlands was a salient issue for them.

What drives intentions and subsequent behavior?
The more fundamental question in this study is of course, why some have taken the initiative to emigrate whereas others initially stated their plans but so far have not taken the step to realize those intentions. Whenever one matches intentions with behavior there are basically four types of groups within the sample that need to be examined. Table 3 summarizes the information that is revealed in Table 2. There are two groups (I and IV) who follow up their plans: the "movers" who had emigration plans and who actually moved two years later, and the "stayers": people who had no emigration plans at all and who also did not migrate. The groups who did not follow up their plans within two years (III and II) are also interesting. The third group (III) constitutes, for lack of a better term, "dreamers": people who had intentions but so far have not realized those plans within two years. The fourth group (II) is the smallest group: people who initially had no plans but after two years had actually emigrated. The characteristics of this group could be potentially informative because it might well be that this is a group of people who are forced or triggered by an exogenous event (e.g., change of job) or perhaps more importantly they changed their minds because other relatives have emigrated and the social network effect triggers so-called "chain migration." Unfortunately, this group is too small to be analyzed confidently within our sample because there are only seven respondents who belong to this rare group. For our purposes we analyze the remaining three groups jointly (indicated by the shaded areas in Table 3) by means of multinomial logit analysis. Because we over-sampled emigrants in our survey, we carried out the statistical analysis using sample weights that are a function of the dependent variable, the intention to emigrate (Winship and Radbill 1994).   (30)] † p<0.10 * p < 0.05 ; ** p < 0.01. (a) Emigrants = those with intentions who have emigrated; Dreamers = those with intentions who have not (yet) emigrated; Stayers = those with no intentions who also have not emigrated. (b) To interpret these results: the higher the score the more negative one is about the quality of the domain in question.
The estimation results reveal a number of illuminating findings about emigration not covered by earlier studies. Starting with actual emigration (Column 1), it becomes quite clear how important age, gender and health status are for moving across borders. Although children are sometimes seen as an impediment to emigration, apparently this does not seem to be the case.
Of course, in actual practice much will depend on the age of a child. The age profile of Dutch emigrants suggests that having children in the age category of 0-6 years does not seem to hinder actual migration, but as soon as children exceed the age of six, emigration drops steeply. The fact that respondents who were unemployed at the time of the interview do not differ significantly from employed respondents may strike the outsider as odd but it may not be so surprising after all. Students who are in the process of finishing their education are a group highly likely to move abroad and in our sample it could also be the case that the respondent is not the head of household. The status of self-employed does, however, exert a large effect on the probability of emigration.
Earlier on, Table 1 showed that emigrants are relatively highly educated compared to those who have not emigrated or not yet emigrated. The results of the multivariate model in Table 4 show that education does not exert an independent force on the emigration process.
What may be behind this result is the fact that the model we present here is far richer in the choice of determinants than most empirical economic models, which rely mostly on a number of social economic characteristics. Our model not only covers those characteristics but it also covers psychological characteristics that may in part explain, or be associated with the level of education. 8 The most noteworthy elements in explaining emigration behavior is the importance of social networks abroad, personality traits such as the reported level self efficacy and the negative evaluation of the public domain of the Netherlands. Networks are important as sources of information or as a resource to settle in the country of destination. However, close inspection of the networks of movers reveals that 42 percent of them did not have any contact abroad. For those not-connected emigrants it would appear that they could fall back on other resources, such as their ability to make plans work or the willingness or preference to seek new adventures and risks. In short, the psychological disposition toward migration matters. In other words, dissatisfaction with the environment and societal problems are of key importance to understanding recent migration flows from the Netherlands. It is a novel finding that these public forces, and not dissatisfaction with private living conditions, are at the forefront in making emigration decisions.
The results from the second column, in which we compare dreamers to stayers, should give us a clue as to why dreamers did not make the final step or have not yet made it. The story is very much like that of the movers. In the third column we tested the differences between dreamers and movers. What does differ across movers and dreamers is their selfreported health status. Health status is the only variable that differs (at a 10 percent level of significance) across the two types. Movers are generally in better health than dreamers. To see how the evaluations of movers and dreamers are very much alike and differ significantly from stayers, one only has to look at the evaluation of living conditions in the Netherlands as summed up in Figure 1. Source: NIDI emigration survey (2005)(2006)(2007) The fact that the results of movers and dreamers are not very much different suggests that the "push" factors may not be very helpful in resolving the question of realizing intentions. The main conclusion that can be deduced from this puzzling result is that it is likely that some emigrants simply need more time than two years to realize their intentions. Whether the country of destination as a "pull" factor plays a role in the realization of intentions remains an open question and the next section will examine the potential of this force.

Choosing a destination country
Every migrant or potential migrant, by definition, has to weigh the pros and cons of two countries. Most studies focus on only one side of the decision and let, for instance, the entire decision be a function of individual characteristics and source country characteristics. This study has so far been no exception to this rule. The above results point out that the profile of emigrants fits the traditional story of emigration. The traditional emigrant is predominantly young, single, male, higher educated, adventurous, has a network in the country of destination and feels competent to execute emigration plans. However, the modern-day emigrant is There is only one exception to this rule and that refers to the income people earn (see Table   5). Surprisingly, the movers are more satisfied with their income than the dreamers, which could be a reflection of the fact that high income households are more likely to move than lower income households. The push factors as a whole, therefore, cannot explain why intentions and behavior diverge.
We are, however, in a position to include the pull factors for the potential emigrants group. These pull factors may well differ across the different types of migrants. If our hypothesis that the speed of emigration is primarily driven by the size of expected improvements in the quality of life is correct, then the following relation should hold: the expected improvement in living conditions by movers should be larger than the improvements expected by dreamers. A check on the evaluation of elements of the preferred country of destination (see last two columns of Table 5) reveals that again movers and dreamers are very much alike. In other words, the dreamers may well materialize their intentions in the future, they simply need more time to realize them, e.g., because their income is not yet sufficient to cover the costs of emigration and the adjustment period in the country of destination or, as the regression analysis of Table 4 suggests, dreamers have a weaker health status thus making the step to emigrate a more risky prospect. (a) The sample for evaluation of preferred country of destination (movers = 42, dreamers = 103) is smaller than of the home country (movers = 45, dreamers = 145) because the evaluation for the preferred country was only asked in the focused Expat sample. (b) The interpretation of evaluation figures for the home country is: 1 = very positive, 2 = positive, 3 = neutral, 4 = negative, 5 = very negative. For the preferred country of destination the interpretation is: 1= much better, 2 = better, 3 = the same, 4 =worse, 5 = much worse. Source: NIDI emigration survey (2005)(2006)(2007) An additional statistic may shed some light on the possibility of why people put off the decision to emigrate and the success rate of reaching the preferred country of destination. By checking the actual country of destination for emigrants, we can see whether people change their mind or stick to their first best choice. The data on country choice are quite clear: 93 percent of emigrants move to the country of their first choice. In other words, people stick to their plan and this is an interesting detail as it could potentially explain why the majority of potential migrants have not realized their intentions so far. People are willing to postpone the date of their emigration until they have made the arrangements for the country of their dreams; they do not switch plans halfway and move to a different country.

Conclusions
Research on international migration increasingly uses the tool of surveys in which people are asked to state their intention to move. However, whether these intentions, or the questions behind these intentions, are valid research tools to predict future migration remains an open question. So far the researchers on international migration remain silent when it comes to corroborating the validity of intentions. This paper is, to our knowledge, the first attempt to correct that state of affairs. By combining a revealed preference (action) and stated preference (intention) approach, one implicitly mixes the insights from two research traditions. The rational choice perspective is implicit in the revealed preference approach; the dominant perspective, within economics, stresses the fact migration flows are primarily inspired by differences in wages or income prospects. The empirics of migration in general confirm this picture as emigration flows from developing countries, or countries in transition, to developed countries are substantial (as summarized by Chiswick and Hatton 2002). However, the stated preference approach is more closely linked to the social psychology of decision making and by implication stresses the importance of attitudes and social norms in making decisions. The theory of reasoned action (Ajzen and Fishbein 1980) and the closely related theory of planned behavior (Ajzen 1991) are pivotal in understanding intentions and subsequent behavior.
Emigration research based on intentions reveals that stating emigration plans is not solely a question of wage differences and expectations of improving one's living standard, it also matters whether the act is in line with prevailing social norms. In developing and developed countries alike, social forces are important in forming emigration intentions (Van Dalen,

Groenewold and Schoorl 2005).
This paper makes two types of contributions: one on method and one on content. The main methodological contribution is that intentions are good predictors of future emigration: 24 percent of those who had stated an intention to emigrate have actually emigrated within two years time. To expect that all respondents would have followed up their intentions would be asking too much of individual rationality and self control. As Manski (1990) makes abundantly clear, "even if individuals have rational expectations and stated intentions are best predictions of future behavior, intentions and behavior need not coincide." A divergence between intentions and behavior may simply occur whenever information available to respondents at the time of stating the intention is more limited than the information they 20 possess at the time when behavior is determined. Especially with such a long-term decision as emigration, 24 percent can be seen as relatively high. This is an important advancement in the literature, which relies heavily on the use of behavioral intentions.
Contributions based on content are equally important as they suggest, first of all, that the dissatisfaction with the public domain at home is of significant importance in triggering not only emigration intentions but also actual emigration. The reigning consensus among migration researchers is that the private domain is of prime importance in understanding international migration, but-as this study shows-the public domain is of substantial importance when we want to understand emigration from a high income country such as the Netherlands. Secondly, within the group of potential emigrants, those who have emigrated and those who have not emigrated or not yet emigrated do not differ much from each other.
The potential emigrants who have not yet emigrated are in poorer health and are more dissatisfied with their income in the Netherlands. In other words, these so-called dreamers may need more time to realize their intentions, thereby increasing the predictive value of intentions in the long run.
Despite these new contributions to the literature of international migration, this type of study has a number of limitations that may inspire future research. The most basic question that we tried to solve is whether intentions are good predictors of future behavior. Despite the high correlation between intention and behavior, the majority of potential emigrants did not move in the two years after we polled their intentions. The most pertinent question is of course, why did people not realize their intentions? Did circumstances change at home and abroad, have differences of opinion arisen within the household about the desirability of migration or is it a matter of time that intentions will be realized? In this paper we ascertained that the characteristics and expectations of emigrants and respondents who have not yet realized their intentions do not substantially differ. However, if one wants to dig deeper into the emigration decision process it becomes necessary to add information on circumstances and expectations between the time of measuring intentions and the time when actual behavior is measured and construct a richer longitudinal database. Until that time, we have to be satisfied with the plain fact that intentions are not mere words, rather they offer information about future behavior. Hence, statements like "intentions are notoriously unreliable as guides to eventual behavior," are in need of revision.

Notes:
1. See De Jong et al. 1985;De Jong 2000;De Groot, Manting and Mulder 2007;Hughes and McCormick 1985;Kan, 1999;and Lu 1999. 2. The only study to our knowledge that captures international migration is a rudimentary report by Gardner et al. (1985) on migration intentions and the behavior of inhabitants of a rural area in the Philippines. The study focuses only on those with migration intentions (that are primarily internally directed) and lacks insight into subsequent movements of those with no intentions. The study also lacks a multivariate analysis of emigration intentions, due to the small numbers, and restricts to cross tabulations. A more recent study by McKenzie, Gibson and Stillman (2007) offers an ingenious data set and an interesting focus on work and income expectations among Tongan citizens who applied for residence applications in New Zealand. This study also confronts those expectations with realizations of those who had emigrated. By design the study is however focussed on those who have made the emigration decision and hence this study does not offer a true comparison for our study. We acknowledge these studies, but for our purposes they do not compare well with the present setup and for that reason we think that ours is the first study to examine the discrepancy between international emigration intentions and behavior.
3. To cite a number of recent studies: Abrams, Hinkle and Tomlins. 1999;Boneva and Frieze 2001;Boneva et al. 1998;Burda et al. 1998;Drinkwater 2003, Drinkwater andIngram 2008;European Commission 2006;Epstein and Gang 2006;Faini 1999;Fidrmuc and Huber 2007;Frieze et al. 2004;Krieger 2004;Lam 2000 5. Katz and Stark (1987: 722) present an example that produces a U-shaped pattern by skill level: only the lower level skills and higher level skills migrate. This pattern is in accordance with recent immigration developments in the US, which according to Schultz (1998) are dominated by the two extremes in skill distribution.
6. The Expat fair was a meeting ground for "those who want to work, live, study, do business and settle abroad." Visitors were asked to indicate whether they considered themselves to be (1) expatriates; (2) potential emigrants; (3) interested in studying abroad; (4) human resource management/multinational; or (5) interested in doing business abroad. We focused exclusively on the second category. 7. For details, consult http://www.centerdata.nl/en/. Center Data of the University of Tilburg maintains a nation-wide panel of households in the Netherlands. The panel is representative of the Dutch population with respect to sex, age, education, religion and regional variation. Respondents are interviewed through an internet connection, and for those who don't have access to the internet, data are collected through a television Netbox system. As such, there is no selectivity with regard to whether people have access to the internet or not. People generally participate for about four years in the panel, during which they are regularly interviewed on several topics. When a respondent leaves the panel, a new respondent is selected on the basis of sociodemographic characteristics so that representativeness will be maintained. Because of the on-going nature of this type of survey, traditional response rates are not reported.
8. To test for the robustness of our conclusion we have left the psychological variables out of the equation. This modeling choice does not lead to a different conclusion.