In any act of household migration, there are movers (the migrant) and stayers (those left behind), and both of these two groups have expectations. The movers expect to make some benefits at the destination while the stayers expect the migrant to send or do something at home of origin. Some work, though limited, has been done to improve our understanding of how potential migrants form their expectations of what they can get from the destination country in studies involving determinants of individual migration. But for those left behind very little is known about how they form their expectations of what they can get from the migrant. The few studies that have been done on this have only used observed flow of remittances to estimate what people left behind expect from migration. Hence these studies equate observed flow of remittances to expected flows. And by this equation, these studies also assume perfect information flow between migrants and relations left behind as well as perfect knowledge to help those left behind to form realistic expectations: expectations that reflects exactly what can be sent to them. Obviously these assumptions are not tenable. These untenable assumptions also leave a hole in our ability to explain why a household will choose to either continue supporting members for migration or not. This is because we cannot tell from observed data alone whether or not the desire to continue to support migration of a household or a family member is as a result of well-informed subjective expectations or not. The crust of the problem here is therefore that by relying on observed data alone we fail to account for the important role subjective expectations or beliefs of those left behind play in decisions for further migration movements, especially within the family. To be able to unravel this problem we need elicitation of subjective expectations of remittance flows from those left behind. Using data from a specially designed survey in two districts in Ghana, I construct time-adjusted subjective remittance expectations of migrant families at home of origin and analyse the factors that determine the formation of these expectations and how formation of these expectations can help us explain perpetuation of migration within a household. The key analytical models employed in these investigations are summarised below In order to understand the exogenous determinants of remittance expectations of migrant households, I first of all estimate factors that influence performance of migrant at home of origin and general flow of information between the migrants and the household members left behind. In order to see the effect of remittances on formation of subjective expectations, remittance flow was measured in terms of migrant performance by adjusting the flows to the time period during which the migrant could do what he or she has done. The items were limited to the popular ones people receive: money for living expenses, establishment of a house and business investment. The theoretical explanation for this adjustment is that if the observed trend in remittance flow has any effect on expectations it would be through individual household’s evaluation of what migrants have achieved within a certain number of years. In other words, all things being equal, families whose migrants took much longer period to achieve certain things would have lower levels of expectations than a comparable family whose migrant took relatively shorter period. This is because taking a long time to achieve something at home of origin would breed some kind of skepticism and uncertainty among those left behind as to what they can get from migration. And this skepticism can lead to low levels of expectations. This is also in line with the reference people left behind often make when talking about achievements of migrants at home of origin as they always point to what XYZ has done. Ordinary least squared regression is then used to estimate factors determining level of migrant performance at home of origin after the transformation of the dependent variable: migrant performance. Heckman selection model is also applied to control for possible effect of bias since some households have migrants who have done nothing at home. Kinship ties are the major factors under this investigation. To determine the main factors influencing information flow, ordinary least squared estimates are used while a generalised ordered logit model, with maximum likelihood method, is used to estimate the factors influencing the likelihood of a household getting higher categories of private/dedicated information from the migrant. Major factors for this investigation are kinship ties and performance of migrant at home of origin. Since information flow and remittance flows are suspected to have endogenous relationship, instrumental variables (IV) technique is employed to estimated impact of remittance flow on both private and public information flows. This is important for us to understand how information flow act as exogenous determinant of subjective remittance expectations, and resultant effect on perpetuation of migration. Once current information flow and performance of migrants have been examined and effects of their exogenous factors estimated, the next stage of the analysis is the examination of effects of these past performance and information flow on household subjective remittance expectations while controlling for other major exogenous factors such as kinship ties, level of education and household wealth. Ordinary least square regression technique is used to estimate major determinants of these levels of expectations. However, to control for possible bias resulting from the fact that a select group of households may not expect anything, Heckman selection model is applied. The final analysis is the estimation of impact household subjective remittance expectations on migration-support intentions. Due to the problem of endogenous relationship between expectations and migration decisions, ordinary maximum likelihood estimates would not be very effective in identifying the real impact expectations have on migration decisions. Hence I use maximum likelihood with endogenous repressors to estimate or identify the influence of expectation on potential migration decisions, applying the probit model with selection model (heckprob) technique. Ordered probit analysis is also used to investigate what determines household’s desire to support more than one person for migration. The results are summarised below. Summary of Findings Economics and sociology literature makes us aware that in order to understand formation of expectations of any kind we first have to investigate two important factors: past events and current information flow, because these are the two factors that hugely influence expectations. Hence, for us to understand remittance expectations, we first have to understand two issues: observed past flows of remittances and current flow of information between the migrant and relations left behind at home of origin. If remittance flows should influence household or family’s (including the extended family members) subjective expectations and the support to move abroad, it should largely do so in terms of what has been observed in the past. In Chapter Six, I investigated the influence of kinship ties on receipts of remittances. As expected, closer migrant relations such as spouse and head of family stand a much better chance of having better performance from migrant than distant kinship ties such as friendship. However when it comes to performance in individual items such as house or business investment, a household cannot rely only on kinship ties with migrant. It should also have some wealth. Specifically, among the kinship ties only spousal relationship was found to have positive effect on migrant performance in areas such as housing and business investment. Thus the influence of kinship ties on observed flow of remittances is mostly limited to money for living expenses, unless the family left behind is wealthy enough to enable allocation of what is sent into other things such as investment in housing and business. With kinship ties being very influential in the determination of past performance of migrants one would expect that these ties would also influence information flow if the assumption of remittance and information flow being together holds. It has always been assumed by cumulative causation theories of migration that together with the flow of remittances from migrant to relations back at home is the flow of information that connects migrant, potential migrants and those left behind (Massey et al, 1993). If this is the case then relationship should be a key factor in determining information flow from the migrants, because these ties influence flow of remittances. Results from the 2SLS model show that remittance flow has impact only at the lower levels of private information flow, reinforcing the point that information that comes with remittance flow may just be social issues such as size of family, marital status, and not economic ones. In spite of their strong effect on remittance flow or migrant performance, all the types of kinship ties generally have negative effects on private information flow. Thus kinship ties are not enough for those left behind to get more private information from the migrant relations residing abroad. It should not be surprising that remittance flows do not lead to higher levels of information flow from the migrants to those left behind. This is because remittances are mostly made up of monetary transfers for living expenses which may not carry much information with it as, in most cases, migrants do not require monitoring. And with electronic transfers of these days, it becomes more implausible to assume that remittance flows, which are mostly limited to monetary transfers, would generate private information as the interpersonal exchanges in these transfers become more and more reduced. But since the lower levels of private information flows only contain pieces of information such as marital status, household size and education levels, it follows that remittance flow may not be the best channel through which relations get important information about the socioeconomic conditions of the migrant. Perhaps this assumption was more plausible about 30 years ago when migrants mostly relied on methods such as using other migrants going home. Families left behind have to rely on their wealth or good level of education to be able to source information from the migrants. On the other hand, remittance flow or migrant performance has highly significant and positive influence on public information flow, suggesting that what migrants do at home influence some perceived knowledge of the migrants’ socioeconomic conditions. It is also interesting to note that factors such as average household education and wealth that have significant positive effect on private information flow have negative effect on public information flow. One can therefore deduce that the more families are able to access information from the migrants themselves, the less they rely on migration information from nonmigrant sources or the general public in the community of origin. Unfortunately remittance flow is unable to help those left behind to get more information from the migrant. Hence most of them will have to rely on public information. With the flow of crucial information such as economic conditions of migrants lacking or being inadequate, it can be concluded that there would be some level of uncertainty about conditions. And this level of uncertainty may lead to some guess-work or reliance on information from other sources in the formation of remittance expectations. That is, would their inability to access crucial information on economic conditions of the migrants “push” them to rely on information reaching them from other sources in the formation of expectations? Also if the wealthy and the more educated families are more likely to know more about the migrants, and if knowing more about the migrant is most likely to temper high expectations with realism as hypothesized in this study, would it be fair to conclude that wealthier and more educated families may have ambivalent, if not negative expectation levels? Results from Chapter Seven show that families would use their experience of what migrants have done at home of origin as a starting point in the formation of their remittance expectations in terms of whether or not they should expect something. But once their expectation status is assured, families are much more influenced by other factors than migrant performance in the formation of their subjective remittance expectation levels. In other words at lower levels of information, remittance expectations seem to be more adaptive to past trends of observed remittance flows. Kinship ties become very significant in this respect in spite of its insignificant influence on information flow. This raises a question of whether or not the effect of kinship ties on formation of remittance expectations is informed by information from the migrants. All the results point to the contrary. The effects of kinship ties on subjective remittance expectations are informed more by past experience of remittance receipts than current dedicated/ private flow of information between the families and the migrants. When kinship ties are interacted with private information their effects on remittance expectations are, however, significantly reduced, indicating that when people take private or dedicated information into consideration their high expectations are very much checked. What are the implications of subjective remittance expectations form under low levels of dedicated information flow for migration decisions? Chapter Eight sought to provide the answer to this question. The results confirmed the hypothesis that subjective remittance expectations formed under inadequate flow of dedicated information would lead to increasing desire to support more migration from the family and the opposite should also true. That is under inadequate information flow, subjective remittance expectations have highly positive effect on desire to perpetuate migration more than the demonstrative effect of migrant performance, emphasizing the importance of expectations in perpetuation of migration. However, the strong effect of expectations and kinship ties on desire to support migration could be reduced if high levels of dedicated information are taken into consideration. Further investigation into why some families with remittance expectations would still not want to support members to migrate revealed that, in addition to private or dedicated information flow, average household education level is a major factor that discourages families with remittance expectations from further supporting members to migrate. This is in sharp contrast with the generally accepted view that education selects families and individuals into migration, especially international migration. This is true in the general population. When only migrant families are sampled, as in this study, the effects of education on migration are tempered with information flow. Education allows the family to access more and more of private/dedicated information which has negative effect on remittance expectations. It is therefore not surprising that education may discourage families with expectations to continue supporting migration. But since most people do not get the private information or do not even consider it as, expectations which are hugely informed by past performance, public information and mere kinship ties would continue to drive perpetuation of migration, at least, at the household level.