Dreaming Another Life. The Role of Foreign Media in Migration Decisions. Evidence from Albania

Tags: television, Albania, media exposure, standard errors, regional controls, migrants, individual, Dependent Variable, individuals, Italian television, Greece Italy, international migration, Bocconi University, variable, television transmitter, electromagnetic waves, information sources, population, private station, Albanian media, internal migration, migration type, private stations, destination country, individual choice, mass media
Content: Dreaming Another Life. The Role of Foreign Media in Migration Decision. Evidence from Albania. Michela Braga Bocconi University 31 July 2007 Work in progress Abstract Using data from Albanian Panel Living Standard Measurement Survey we examine the e¤ect of exposure to foreign media on individual migration decision. We model the probability to watch foreign television as a function of the distance from the nearest foreign transmitter. The identi...cation strategy is based on the physics model of electromagnetic and radio wave propagation explaining how radio waves behave when they are transmitted (or propagated) from one point on the earth to another. The results suggest that individuals exposed to foreign media are more likely to migrate internationally. 1 Introduction This paper examines the existing link between foreign media exposure and individual migration decision. We investigate whether individuals who have been exposed to foreign media are more prone to move abroad and to which extent their location choice depends on the received information. Two main literature streams consider migration choice. Starting from Todaro (1969) seminal paper, traditional neoclassical theory explains individual migration decision through a cost-bene...t analysis based on di¤erentials in economic conditions between receiving and sending country. The "new economics of migration" extends neoclassical theory modelling individual migration as a household decision not only to maximize expected income gain but also to minimize risks of potential market failures (Stark and Bloom (1985)). Both approaches implicitly assume that potential migrants optimally decide whether to move or not using information on economic conditions and opportunities in 1
destination country. Before migrating individuals gather information from different sources. Relatives and friends, previous migrants or network abroad have been widely analyzed by economic theory. In particular, the ...ndings suggest that wider family and friend networks of previous migrants enhances migration (Massey and Espinosa (1997), Orrenious (1999), Zahniser (1999), Davis and Winters (2001), Munshi (2003)). Providing information on the migration process, on the economic opportunities at destination or helping integration once arrived, migration networks are crucial in individual migration decision. Nevertheless, little attention is given to information sources extremely common nowadays: television, radio, newspapers and internet. The impact of media on individual choice or behavior has been studied by some economic ...elds. Olken (2006) in a recent paper identify the impact of exposure to television and radio on social capital in Indonesian villages while Boeri and Tabellini (2005) ...nd that individuals are more willing to accept reforms when they are more informed about them. Suphachalasai (2005) investigates the relationship between development, environmental policy determination, and mass media stressing the role of mass media as a channel through which the level of development inuence environmental policy making. Some researches have been made in the political economy ...eld. Stromberg (2001) for example uses a simple model to analyze the e¤ect of mass media provision of news on di¤erent policy issues such as redistribution, the size of the government sector, rents and corruption, the e¤ectiveness of lobby groups and political business cycles. The relationship between media and migration has been investigated by sociologists. Some studies highlight that images of wealth and a free and relaxed lifestyle in the northern and western economies are commonplace in developing and transition countries. Foreign media system represents an important source of information for potential migrants and somehow it tends to reinforce the idea of migration as a trip towards El Dorado. The received information could be more or less precise, complete and accurate, di¤erent individual could understand the same information in very di¤erent way but, generally speaking, media help potential migrants in constructing images of their future life. Whether and to which extent foreign media a¤ect individual migration project has not been explored in a systematic way. As far as we know this is the ...rst attempt to quantify the impact of foreign media in migration decision. Countries that represent a good benchmark for analysis are the totalitarian ones in which, for years, the free access to information has been forbidden. Among them, the Albanian case is emblematic. From 1941 to the late eighties Albania has been one of the most oppressive and isolated communist regimes: both economic and political contacts with the rest of the world, even communist, were absent. International migration was forbidden and severely punished, but also internal mobility was accurately monitored. The country's isolation was exacerbated by the lack of expression freedom and by the control of media and broadcasting system. Serving the interests of the Communist party, press was strictly controlled and censorship was extremely strict. The same happened with radio and television. Until 1990, Albanian media system had only one 2
national television channel broadcasting every evening from 6 to 10 p.m., the supply of programmes included only one ...lm per week and a lot of politicized and propagandistic documentaries. Although foreign broadcasts were forbidden, starting from the early 1960s Albanians could easily watch Italian television due to the geographical proximity between the two countries. At the beginning only Italian public service television could be received but, with the increase of private broadcast system, all Italian televisions reached Albanian households. Italian television allows Albanian to construct an image of the west and to imagine their life after migration. After the collapse of communist regime in 1990, political, economic and social changes lead to massive migratory waves de...ning one of the most relevant migration ows in the last decade. Do foreign media play a role in Albanian migration projects? Does the perceived foreign reality induce people to migrate? Does foreign television attract Albanian? We answer these questions using data from the Living Standard Measurement Study (LSMS) carried out by the World Bank and the Albanian Institute of Statistics (INSTAT) from 2002 to 2004. The data set contains a lot of information about individual experience abroad. Complete migration history, both internal and international, from 1990 to 2004 is available: information about migration length, country of destination, occupation in the host country, monetary aids received to migrate, legal/illegal status abroad. We model individual migration choice through a standard probit model including traditional explanatory variables (individual speci...c characteristics, household characteristics, geographic characteristics) and a variable capturing the degree of exposure to foreign media. The key identi...cation assumption is the following: the probability to watch foreign television depends on the distance from the nearest foreign transmitter. The identi...cation strategy is based on the physics model of electromagnetic and radio wave propagation explaining how radio waves behave when they are transmitted (or propagated) from one point on the earth to another. In free space, all electromagnetic waves follow the inverse squared law which states that the signal's strength is proportional to the inverse of the squared distance from the source. Therefore, for all television transmitters in Italy, we collect data on their location (latitude, longitude, altitude) and using the greatcircle formula we compute the shortest distance between Italian transmitters and Albanian cities. For each individual we reconstruct the place of residence before 1990 and we assign the corresponding media exposure measure. Nevertheless, waves propagation is also a¤ected by other factors as climate and weather (temperature, pressure, rain intensity, cloud cover), ground characteristics and presence of obstructions. Due to the di¤raction phenomenon, the presence of obstructions does not avoid per se the signal reception: signals diffract around the obstacle, change their trajectory and reach the ground although with a lower strength. To exploit the di¤raction of waves due to obstacles we consider the topography of the area between the nearest transmitter and the receiver. Therefore, we construct a variable that is the highest altitude of the earth surface between Italian transmitters and Albanian cities. In addition, we 3
consider also a variable capturing the transmitter, the receiver and the obstacle altitude given that the degree of di¤raction and the strength of the received signal depends crucially on the di¤erence in height between the obstacle and the receiving point. The empirical analysis shows that migration probability decreases as the distance from the Italian transmitter increases and as the altitude of the obstacle blocking signal line increases. Foreign media play a crucial role in individual migration decision: people that have been exposed to foreign television are more likely to move. Other three crucial variables are included in the model: distance from the coast, distance from the nearest border cross and distance from the nearest border cross to Italy and Greece, the main destination countries of Albanian migrants. These variables should be a proxy for migration costs and should allow to check whether the distance from Italian television transmitters does not actually measure individual migration cost. Our media exposure variable remains highly signi...cant in all speci...cations. Distance from the coast a¤ects positively migration probability capturing the fact that migration ows are higher for internal regions, more depressed and poor. As expected, distance from Italian and Greek frontier captures usual migration cost: individuals living far from the border are less likely to migrate internationally. In the basic speci...cation only the media exposure variable and regional ...xed e¤ects are included while in the more complete one all distance measures are added. The results are robust also including individual and household di¤erent controls. Through a multinomial logit speci...cation, we also test whether exposure to Italian media a¤ects the choice of the destination. Having been exposed to Italian media increases the migration probability tout court. Italian television does not attract Albanian only to Italy but induces them to move towards di¤erent western countries: Italian television was simply a door on the world, a way to know and maybe to idealize a di¤erent culture and lifestyle. This paper consider one of the most relevant temporary migration ows in the last decade: the Albanian one. The Albanian population have had huge changes during the last 15 years of transition to a market based economy. Migration patters during the 1990s have been critical aspects of this transition period. After the communist regime collapse, due to political and economic changes in the country, controls of people movements were abolished so that people start to move both internally and internationally. According to 2001 Census data population in rural area decreased by 13% . Albanian migration is mostly temporary: after having spent some months abroad migrants come back in their country of origin. Despite the motivations behind returning in the home country, this decision in potentially positive from the point of view of the country of origin: an individual stayed abroad for a while may bring skills and capital to the home economy and contribute to the economic growth of the country through the after migration economic activities. Temporary migration has many advantages. For host countries temporary labour migration can help meet speci...c labour - force needs, when there is disequilibrium in demand and supply 4
in some speci...c periods or area temporary migrants are a useful resource able to ...ll the existing gap reducing some problems coming from permanent migration (reactions and negative attitudes towards immigrants). For home country temporary worker outows can reduce unemployment, contribute to growth and development through remittances, reduce brain drain given the subsequent return of worker and increase brain gain when migrant come back with augmented and additional skills (Galor and Stark (1990)). Most temporary movements are from developing to developed countries, signi...cant ows are registered also between developed countries (World Economic and Social Survey (2004)). Gains from more extensive use of temporary migrants are relatively larger for unskilled workers than for skilled ones: the wage di¤erential between developed and developing countries is wider for unskilled workers (Winters (2002)). The paper is organized as follows. Section 2 gives an overview of Albanian migration phenomenon. Section 3 presents Albanian media and broadcasting system. Section 4 introduces the dataset used for the empirical analysis that is discussed in Section 5. Finally, Section 6 concludes. 2 Migration in the Albanian context Since 1990 Albania faces the big challenge to become a market economy and a more open society passing from totalitarianism to democracy. Thanks to a cumulative real economic growth of about 40% between 1990 and 2004, the country is among the most successful transition economies. Starting from extremely low income levels and very poor infrastructure, when the communist regime collapsed in 1991 there were a signi...cant decrease in output and a rise in ination. In the ...rst two years of the transition, between 1990 and 1992, GDP decreased by 39%. In the following years a macroeconomic stabilization program was implemented to reduce ination and the budget de...cit together with the liberalization of prices and foreign trade, the privatization of agricultural land making possible an high average growth rate of 9.3%. The economic growth was led by agricultural sector, the service sector and the construction sector. This recovery period was interrupted in 1997 by the crisis of the pyramid investment scheme. Furthermore, the country su¤ered from the social and economic shocks accompanying the Kosovo crisis in 1999 when more than half a million of Kosovo - Albanian refugees arrived in the northern regions. Because of the ...nancial crisis, GDP decreased by 7% and with this growth track real GDP in 1999 reached the 1990 level. After some problems in energy supply and bad weather conditions for the agricultural sector, economic growth slowed down to 3.4% again in 2002 but in the last years the real growth rate went up to 6%. Despite the shocks hitting the economy, starting from a very low income level, Albanian economy has been able to reach a sustained growth, even though it remains one of the poorest countries in Europe (with GDP per capita at around 2,700 US$). The poverty is high and pervasive: more than ten years 5
after the transition 29.2% of Albanians households was poor and 28.8% was very poor (De Soto et al. (2002)). One of the consequences of this transition period has been huge migration ows. However, it is important to underline that Albania has a long history of emigration, migration ows are presents throughout all the centuries in response to political, social or economic events (King et al. (2003), Piperno (2002)). In particular during the 20th century we can observe three di¤erent phases: before 1944, from 1945 to 1990 and from 1990 on (UNDP (2000)) In the ...rst wave the main destination countries were US and Latin America and almost all the people left the country because of economic push factors. The problems faced by the industry and agriculture after the war, the absence of modern technology and the inadequate exploitation of natural resources were the main driving force of individual mobility. During this period Albanian governments were almost indi¤erent towards these migration ows. In the second wave migration was directed both towards US, Latin America and Australia and towards near countries including Italy, Greece, Bulgaria, Egypt, Romania and Serbia. Albanians left their country for political reasons related to the communist regime, a huge part of emigrants were opponents of the regime. Oў cially migration was forbidden and punished: political and legal barriers were established, migration was considered a crime. The third phase started in 1990 after the collapse of the state - socialist regimes in Eastern Europe. All policies and measures restricting the free movement of citizens were removed. Without control on internal and external migrations, single individuals and entire households started to move internally form rural area towards urban area, and internationally. It is possible to identify three regions that drives migration ows. The north (districts of Diber, Mat, Puke,Tropoje) was characterized by both internal and international migration. internal migration was directed towards central richer regions with the Tirana, Durres, Kruje axis and towards southern regions that although very poor were relatively richer with respect to the north, the poorest part of the country with few employment opportunities, exclusively in agriculture, and low income level. The main destinations for international migration were Italy, Greece, Germany and UK. The Tirana - Durres central area was the main destination for internal migration from all the decentralized Albanian area, but especially from the north, and experienced a signi...cant outow towards the main international destinations. The central area is the one that also experienced the positive inow of return migrants. In the south of the country (districts of Vlore, Berat, Korce) three quarters of the migration outow was directed abroad, mainly to Greece, and one quarter moved internally from poor rural area towards urban centers. This area experienced a huge depopulation in particular in the ethnic Greek regions although in some cases internal migration from the north reduced the problem. 6
Figure 1: Albanian administrative division Surely Albania during the 1990s has had one of the largest outows of people relative to population size in Europe and the migration outow has increased steadily since 1991 but the exact magnitude of this phenomenon is diў cult to be known also because oў cial data sources are scarce and inadequate. According to estimates by the Greek Ministry of Foreign A¤airs from 1990 to 1999 migrants towards Greece have increased by four times while towards Italy have doubled (Table A). Signi...cant is also the magnitude of internal migration from rural towards urban areas and from small urban towns towards big cities. This form of migration is the natural consequence of diў culties faced in the agriculture sector and is often seasonal. The internal mobility phenomenon is not simple to be measured but according to 2001 Census 5,7% of total population in 1989 moved from one region to another, changing place of residence between 1989 to 2001. Internal mobility can be de...ned as a "one - way" ow: almost 91% of the intra - country movement were directed to the central and coastal districts. In 1989, approximately 60% of migrants lived in the North, 32% in the South and 8% 7
in the centre and on the coast. As a result, in the decade after the transition, despite the emigration ows central and costal areas experienced a signi...cant population increase (Table B). Migration had a signi...cant impact on population structure of the country. The sex-age impact of emigration during the 1990s can be easily intuitively quanti...ed considering the age - sex pyramids in 1989 and 2001 from the 2001 Census (Figure 2). Figure 2: Age - sex pyramid (1989 - 2001) It is evident that during the 1990s there has been a signi...cant loss of males aged between 15 and 35 and a slightly smaller loss of woman in the age-band 18 32. Analyzing the pyramid, it is reasonable to expect that in the following years there will be other signi...cant migration ows of people actually in the ageband 10 15 even though in the long run the migration ow should decrease given the 0 5 years old low - birth cohort due to lower fertility rates in the last years and to the migration of potential parents. In these years migration from Albania continues and is transforming from illegal and sporadic to legal and organized. However, once abroad, Albanian migrants tend to work more in sectors that are avoided by the local population. They are used as a regulatory mechanism in the receiving labour market. However there is also intellectual elite that leave the country towards more promising countries, but this migration type presents di¤erent features form other emigrating groups. 8
3 Albanian media system The Radio Television of Albania was monopolistic until 1995, when a private station started operating. Up to that date for most Albanians electronics media meant the State radio and television network. During the communist regime and until 1990, the Albanian national television had only one channel broadcasting only few hours in the evening, between 6 and 10 p.m.. Both television and radio was used by the communist regime for propaganda purposes. The supply of programmes was very scarce and constantly monitored by the regime. Only one ...lm per week was broadcast, having mainly an educational and pedagogic aim. Films were Albanian or imported from other socialist countries although some of them come from Western and non communist countries if considered ideologically neutral and non harmful for national culture (Dores (1991)). The majority of the time was dedicated to documentaries showing the history of the country and to programmes celebrating the successes reached by the Albanian Party of Labour. The regime considered the media system as a way to educate and to give political and patriotic messages to the population, to indoctrinate the audience and to inuence its choices having no entertainment aims. As it was the case in East Germany, watching foreign programs was a natural alternative of the communism propaganda. Watching foreign television although not strictly forbidden by the law, was discouraged in order to avoid the inuence of western capitalist culture until the early 70s. A di¤erent behavior was followed when the dissatisfaction grew in the population for the culture di¤used and the economic policies implemented by the regime, so that oў cial campaigns against listening and watching external broadcasts were made (See Logoreci (1977) for more details). Despite what strictly prescribed by the party, Albanians watch foreign television, mainly Italian, Greek and Jugoslave. In particular, the television signal of the ...rst Italian national channel (Rai 1 - VHF) could reach Albanian regions starting from the early 60s thanks to the geographical proximity and the absence of natural obstacles (i.e. mountains). The government made a lot of e¤orts to curtail the reception of foreign broadcasts and to jam their signals but it became particular diў cult when the number of Italian private stations and networks increase in the 80s. Oў cially an Italian public television transmitter were located of the Dajt mountain over Tirane and it was active only in the evening from 8 to 9 p.m. for the news. In the last years of the regime, although controls became less stringent, watching foreign television or listening radio programmes from abroad were considered a politically seditious activities punishable with a period of reclusion while discussions, both private or public, about foreign programmes were considered activities of subversive propaganda (Dores (1991)). Despite all the e¤orts made by the authorities, foreign culture, especially the Italian one, entered Albanian household through ...lms, game-shows, news and advertisements. During all the communist period and in the ...rst years of transition Albanian people through foreign televisions were able to have access to a di¤erent life style characterized by entertainment and abundance in which freedom of expression was guaranteed (Mai (2004)). 9
After the collapse of the communist regime, the media sector underwent fundamental and far-reaching changes. Di¤erent from the past when there were no private media alternatives to state broadcasting services, after 1991 there was an explosion of electronic and printed media. In Albania, there are at present 54 private radio and 64 TV stations, of which the majority are local and some national. With regard to written media, the data is almost accurate. A large number of newspapers and magazines started publication after 1991. Three television channels broadcast at national level while others operate at local level. Out of 64 private TV stations available in Albania, only 20 are located in Tirana, while the rest are in districts so that di¤erent districts have their local private TV stations. However, the geography of TV stations is not always in harmony with the quality of the programs they o¤er. Given the television medium's rapid, almost chaotic growth rate and lack of applicable legal framework, electronic piracy is prevalent both within and outside Tirana. The same happen with radio:out of 54 radio station country wide, 28 are located in Tirana, and the rest in local districts. During the early 90s in all cities of Albania were installed transmitters with low power to broadcast the programs by Italian Rai and Mediaset channels and some German stations reached some areas. Greece television, both public and private, strongly appears in the south of the country, in Saranda in all Ionian seaside due to geographical proximity with Corfщ. Also in Tirana, Greek television broadcast for a while from Dajt Mountain but the transmitter was shortly uninstalled for license absence reasons. In some cities of the south east, like Gjirokaster, some greek repeaters were installed and some signals from Macedonian television appear in that area. Regarding other sources of information, the access to the Internet is increasing in the big towns. Private businessmen and state institutions mainly rely on this system of information. The role of the independent media is tangible for the public at large, providing consumers with various sources of information and a diversity of social viewpoints. All transformations were accompanied by changes in regulation and legislation. In particular, the development of legislation for media has been a focus in recent years. Work on legislation started in 1993 with the law of freedom of the press and the right of press to provide information. In 1997 came the law for private TV and radios. The National Council of Radios and Televisions was born in 1998 as a body that has the competence to license media and protect children from it. After decades in which the free access to information has been forbidden, the 1998 Constitution guarantees freedom of expression, freedom of the press, radio and television, freedom of broadcasting and prior censorship of means of communication is prohibited. 4 The data The data used for the analysis come from the Living Standard Measurement Study (LSMS) carried out by the World Bank and the Albanian Institute of 10
Statistics (INSTAT) from 2002 to 2004. The LSMS is part of a bigger strategy aimed to improve the data quality in Albania. For its history as a communist country data in Albania are few and their quality is quite low. According to recent surveys carried out by INSTAT1, it is crucial to have accurate measures of household welfare in line with well accepted standards in order to monitor trends on a regular basis. Following the Poverty Reduction Strategy Paper, the Government of Albania reinforced its commitment to strenghtening its ability to collect and analyze, on a regular basis, the information necessary to inform policy - making through the Population and Housing Census, the Living Standard Measurement Study every three year and the annual panel surveys. The LSMS was established by the World Bank in the 1980 to explore ways of improving the type and quality of household data collected by government statistical oў ces in developing countries. The objectives of the LSMS were to develop new methods for monitoring progress in raising levels of living, to identify the consequences for households of current and proposed government policies and to improve communications between survey statisticians, analysts and policymakers. Data are collected on may dimensions of household well - being including employment, income, saving, consumption, migration, education, fertility, housing. The Albanian panel survey sample was selected from households interviewed on the 2002 LSMS. The selected panel component is designed to provide a nationally representative sample of household and individual within Albania and to minimize the variability in households'selection probabilities. The sample size for the panel is: - 1,782 interviewed households (891 urban, 850 rural) and 7973 household members including children aged under 15 in Wave 1 (2002); - 1,780 interviewed households (2,155 selected households, 375 not interviewed), 900 urban and 880 rural, and 8110 household members including children aged under 15 in Wave 2 (2003). The majority of the non interviewed households (348) were due to split - o¤ moves out of the country while the other 4 had moved but could not be traced; - 1,797 interviewed households and 7,476 household members including children aged under 15 in Wave 3 (2004), of which 7,212 already sampled in Wave 1 or 2 and 264 new members. The ...nal sample is composed by 23,748 individuals belonging to 5,356 households, 50. 29 % are male, 49.71 % are female. The Albanian Panel Living Standard Measurement Study contains a lot of information about the experience abroad, data are collected for all household's members. For individual present in all waves, complete migration history, both internal and international, from 1990 to 2004 is available. If some migration occurred in the last 20 months we know its length, country of destination, occupation in the host country, monetary aids received to migrate, legal/illegal status abroad. Almost same data are available for migration history from 1990. 1 The 1998 Living Condition Survey (LCS) and the 2000 Household Budget Survey (HBS). 11
In particular we know period, country and length of ...rst migration, total number of migrations and working status abroad. Given that individual were asked whether they migrated internationally from 1990 to the date of the interview we are able to de...ne a temporary migrants as an individual spending some time abroad and being in Albania at the time of the interview. More than 8 % of individuals is an international temporary migrant, 14. 78% of male and only 1.62% of female. This ...gures are perfectly in line with Albanian patriarchal family structure in which the household head man play a central role. In addition, is clear that Albania is a relatively young sending country and therefore is characterized by a huge male outow. It could be expected that in the future the share of migrants woman would sharply increase for family reuni...cation or similar reasons. Considering the household perspective, temporary migration seems a more pervasive phenomenon hitting around 19% of them (Table 1).On average, migrants are 36 years old: males are relatively younger than females (35.43 against 39.21) Considering age at ...rst migration, males migrate more than 6 years before females. Male temporary migrants are extremely concentrated in the central class of age, while females are more equally distributed across all classes (Table 2). Education level for migrants is relatively low: more than half only completed primary 8 years school but female are more educated. In particular, while the share of graduated males is 4:23% while the share of female with the same education level is 16:67%: Migration pattern by year of ...rst migration, in our sample, is completely in line with Albania history2. After the collapse of communist and the invasion of Western embassies in Tirana in summer 1990, a huge migration outow were registered, both legal (thanks to liberalization of passport issuing) and illegal, steadily increasing up to the end of 90s. Years 1997 and 1998 were characterized by extremely signi...cant migration phenomena due to the collapse of the system of pyramid investment scheme in which around half of Albanian Households had invested3. The same dynamics emerges from ALSMS ...gures: share of ...rst experience temporary migrants increases from 1990 to 1993, is almost constant and slightly declining in the following 3 years when Albanian economy stabilized, is increasing in 1997 and 1998 returning to its previous level in the following years when there was a gradual return to normality (Table 4). Almost all ...rst migration phenomena were due to job reasons, in particular for males while females migrate also for study (3:66%), health reasons (4:71%) or family reuni...cation (7:33%) (Table 5.) Among migrants, more than 80% ...nd a job during his/her ...rst migration although around two third of them did not enter legally in the host country: their migration motivation was satis...ed once abroad. The share of individual who legally worked once migrated is lower any more: less than 20 % found a regular and legal job. The legality share is extremely high for females with 2 See King, R. and Vullnetari, J. "Migration and Development in Albania" for a detailed review. 3 According to World Bank estimates the lost of savins was around 1.2 billion US dollar that account for half of the Albanian GDP in 1996 (Olsen, 2002) . 12
respect both to the entry in the host country and the position in the labour market (Table 6.-8.). Looking the legality pattern from 1990 on, we observe that the share of legal workers and legal migrants increases over time. While in the ...rst year of market economy only 15% of individuals migrate legally, in 2003 exactly 50% of migrants were legal. As regard migrants position in the receiving labour market, the legality share decreases over time, with only 20% of individuals having their ...rst migration spell in 2003 being legal worker (Table 9). In migrating, individual are helped by friend and family, with a transfer of information about the place where to go to ...nd a work and with a real monetary help (Table 10). Migration network is crucial seem to be crucial in all phases of migration (Carletto et al. (2005)) From the data analysis we observe that in the year just before the survey almost 5% of the population went abroad for reasons di¤erent from family visits, 8:78% of the male population and 1% of the female one. On average temporary migrants stay abroad 5:1 months, and the length spell for men is relatively longer than for women ( 5:24 months against 3:98). From 1997 to 2001 almost 6% of surveyed individuals went abroad for at least three months: on average they stay abroad 2:65 months in 1997, 3:23 months in 1998, 3:1 months in 1999; 3:08 months in 2001 and 2:8 months in 2002 5 Econometric analysis 5.1 Empirical strategy We model individual migration choice through a standard binary outcome model including a set of variables variable capturing the degree of exposure to foreign media and, where possible, traditional explanatory variables (individual speci...c characteristics, household characteristics, geographic characteristics). Our prior is that exposure to foreign media has a direct positive e¤ect on individual migration probability because it reduces the cost of migration thank to received information on potential destination countries. The empirical analysis proceeds in di¤erent steps. In the ...rst step we evaluate the role of foreign media, in particular television, in the ...rst individual migration decision. In the second step, we focus on the more recent migration, during the last year, and we consider the e¤ect of di¤erent source of information. In the third step we explore whether received information by foreign media a¤ect the choice of the ...nal destination, during the ...rst and the last migration experience. Formally, in order to study the impact of foreign media on individual migration decision we estimate the following equation: 13
yi = 0 + 1Mi + 2Xi + i
(1)
where i s N ormal(0; 2)
f or i = 1; 2; :::; n:
The vector yi is a binary variable de...ning individual migration status, Mi is our foreign media exposure measure, Xi is a vector of exogenous explanatory variables at individual and household level and i is the stochastic error term. No oў cial data are available about the signals of foreign televisions in Albania and about their power, therefore we construct our measure of media exposure in an indirect way. The key identi...cation assumption is the following: the probability to watch foreign television depends on the distance from the nearest foreign transmitter. The identi...cation strategy is based on the physics model of electromagnetic and radio wave propagation explaining how radio waves behave when they are transmitted (or propagated) from one point on the earth to another (Barclay, L. (2003), Ellington et al. (1980)). In free space, all electromagnetic waves follow the inverse squared law which states that the signal's strength is proportional to the inverse of the squared distance from the source so that doubling the distance from the transmitter leads to a reduction in the signal strength to nearly one quarter. If the medium in which they are propagated is the same everywhere, the waves will spread out uniformly in all directions. Nevertheless, electromagnetic and radio waves interact with the objects and the media in which they travel. In particular, waves propagation is also a¤ected by factors as climate and weather (temperature, pressure, rain intensity, cloud cover), ground characteristics and presence of obstructions from point to point. These interactions causes the signals to change direction and to reach areas which would not be possible if the waves travelled in the direct line. Three di¤erent phenomena are possible: reection, refraction and di¤raction. For our analysis we focus on the third one. When signals encounter some obstacles they tend to travel around them so that they may be received from a transmitter even though it may be shaded by a large object. Di¤raction is more pronounced when the obstacle becomes sharper and more like a knife - edge (we have the so called Knife - edge di¤raction). As represented in Figure 3, due to the di¤raction phenomenon, the presence of obstructions does not avoid per se the signal reception: even though there will be a shadow zone immediately behind the obstacle, signals di¤ract around it, change their trajectory and reach the ground although with a lower strength.
14
Figure 3: The di¤raction phenomenon over an obstacle The wave propagation model suggests us to consider distance from foreign television transmitters and Albanian topographic characteristics as an exogenous source of variation in exposure to foreign media in order to investigate their causal impact on individual migration decision. In particular we focus on Italian television. Our choice is based on three di¤erent considerations. The ...rst one is related to the geographical proximity of the two countries and the absence of natural barriers. Only 150 Km of sea separate Italy's southern regions from Albania's coasts so that programs broadcast from television repeaters located in that area could be easily seen also in Albania. The same does not happen with other neighboring countries, all borders are in mountain areas with few foreign television transmitters faraway. The second reason is related to economic theory explaining individual migration decision. Starting from traditional neoclassical models (Todaro (1969), Harris and Todaro (1970)) migration is modelled as the result of a cost - bene...t analysis in which before moving individuals compare the expected income di¤erentials between the home and the receiving country. Among Albanian neighboring countries, the U.E. members (Italy and Greece) are the ones having a signi...cant wage and wealth di¤erential (Table C) and therefore we expect that information received from these countries could signi...cantly inuence movement decision. Finally our choice is based on anecdotal evidence and sociological studies (Mai (2004), Mai (2001), Mai and King (2002)) suggesting that Italian television played a crucial role in Albanians migration ows. Interviews to young people, aged between 15 and 30 years, conducted in Albania during a ...eldwork in 1998 and 1999, allows Mai (2001) to conclude that Italian media and in particular television helped to construct and to shape their migration project in a broad sense. From a sociological point of view, the migration project is not related to the actual geographical displacement but it refers "to the wider discursive process by means of which 15
Albanians have come to perceive, describe and situate themselves with respect to their wider social and cultural environment - whether this be Albania or Italy4". In order to construct our indirect measure of media exposure we ...rst make a list of all television transmitters. Then we restrict to the 83 located in Italy's costal southern regions (Abruzzo, Molise and Puglia), belonging to both public (RAI 1, RAI2 and RAI 3) and private (Canale 5, Rete 4 and Italia 1) national networks and active in 1980. For each one we collect data on their location (latitude, longitude, altitude). For each individual we have information on geographical location (city, municipality and district) at the date of the interview but our data set contains also retrospective life history data back to 1990. If an individual has moved to current location in the previous decade, we have information about his/her previous location within the country at district level or abroad. Therefore we are able to recover individual placement in 1990. The idea behind the importance of this information for our analysis is that before that date watching foreign television was oў cially forbidden and no contact with other countries were allowed, therefore we can actually investigate whether the image of the west received by foreign media a¤ected individual migration decision. Using the great-circle formula5 we compute the shortest distance between Italian transmitters and Albanian locations. 5.2 First migration decision We begin our analysis evaluating the role of foreign media, in particular television, in the ...rst individual migration decision. In our baseline speci...cation the dependent variable takes value 1 for those individuals who migrated internationally. Column 1 of Table 13 reports the estimates of the basic speci...cation. As expected, foreign media exposure has a positive e¤ect on individual migration decision: individuals living faraway from Italian television transmitter are less likely to watch foreign television and therefore less prone to move, being less informed or having no overoptimistic expectations. In order to check whether the distance from Italian television transmitters does not actually measure migration costs we include in the model other three distance variables. In term of magnitude, ceteris paribus, the probability to migrate decreases by 0.3 percentage point as the distance from Italian television transmitters increases by 1 kilometer. In Column 2 we add distance from the coast, in Column 3 distance from the nearest border cross and in Column 4 also distance from the nearest border cross to Italy and Greece, the main destination countries of Albanian migrants. All the three variables are referred to individual place of residence before migrating that we recover from 4 Mai (2001) 5 Having geographic coordinates of two points A and B on the earth surface (latitudeA, latitudeB, longitudeA, longitudeB) in order to compute the shortest distance between them the formula is d=3963.0 * arccos[sin(lat1/57.2958) * sin(lat2/57.2958) + cos(lat1/57.2958) * cos(lat2/57.2958) * cos(lon2/57.2958 -lon1/57.2958)]. 16
retrospective information contained in the survey. These variables should be a valid proxy for migration costs and should allow to check whether this information is contained in our variable of interest, the distance from Italian television transmitters. In all speci...cations our media exposure variable remains highly statistical signi...cant. Distance from the coast a¤ects positively migration probability capturing the fact that migration ows are higher for internal regions, more depressed and poor. As expected, distance from frontier captures usual migration cost: individuals living far from the border are less likely to migrate internationally. Once controlling for regional characteristics (Columns 5 - 8) our results still hold. In Table 14, we exploit the di¤raction phenomenon of waves due to obstacles including in our model topographic characteristics of the area between the nearest Italian transmitter and the receiver. As expected, the probability of moving decreases as the elevation of the obstacle increases (variable altitude_obstacle_90 is the highest altitude of the earth surface between Italian transmitters and Albanian cities constructed using the software Google - Earth) and the di¤erence in altitude between the obstacle and individual place of residence increases (variable di¤ erence_altitude_90 ). We include also the interaction between distance and altitude and we ...nd a lightly positive e¤ect on migration decision only controlling for regional characteristics. In our previous speci...cations we cannot control for district ...xed e¤ect given that all measures of media exposure are constructed at that level of disaggregation. Therefore, we construct all previous variables at the level of 350 Albanian cities contained in the sample for all individuals who do not migrate within the country from 1990 on and for who we are sure that their current residence is the same of the 1990 one. We run previous regression for this subsample of individuals (Table 15 and 16). Our media exposure variable is highly statistical signi...cant in all speci...cation and it has a positive e¤ect on individual migration decision. Distance from the coast is always positively correlated with the probability of moving; while, di¤erent from previous estimates, when controlling for district ...xed e¤ects we ...nd that individuals living faraway from Italian and Greek cross - borders are less likely to move. In addition also all measures of topographic characteristics are highly statistical signi...cant and have the expected sign. 5.2.1 The omitted variable problem The main concern in identifying the link between the migration choice and the exposure to foreign media is related to the consistency and eў ciency of previous estimates due to the fact that we omit individual variables (Wooldridge (2002)). The omitted variable problem can be easily seen assuming that X is a matrix of included variables and Z is a matrix of variables not included in our analysis. The full model would be: y=X X +Z Z +" 17
while the estimated model is:
y = X X+ = Z Z +" The parameter vector estimated when only the variables included in the matrix X are included will be a generic bX for which:
E(bX ) = E[(X0X) 1X0y]
= E[(X0X) 1X0X X + (X0X) 1X0 ]
= X + E[(X0X) 1X0Z Z ] + E[(X0X) 1X0"]
=
X+
0 ZX Z
where
0 ZX
is
a
matrix
of
coeў
cients
from
regression
of
columns
of
Z on
the
observable variables. The bias will depend on the correlation between each
observable and unobservable.
Considering an additive e¤ect of the omitted individual factors (as income
or employment status before migration decision), our model could be written
as:
E(yjdist_tv; dist_coast; dist_border; X) = 0 + 1dist_tv + 2dist_coast +
3dist_border + q
(2)
where q is the omitted variable that include any unobservable individual characteristic. Equation (2) can be considered a structural model that can be written in error form as:
y = 0 + 1dist_tv + 2dist_coast + 3dist_border + q + " (3)
E("jdist_tv; dist_coast; dist_border; X; q) = 0 where " is the structural error. Being unobservable q can be put into the error term assuming, without loss of generality, E(q) = 0 because an intercept is included in our model (3):The error form model becomes:
y = 0 + 1dist_tv + 2dist_coast + 3dist_border + = q+" For the distribution assumption, " has zero means and is uncorrelated with ist_tv; dist_coast; dist_border and q. Also q has zero mean, by normalization. Therefore, the zero mean assumption for the error term is guaranteed. The orthogonality condition instead, holds if and only if the omitted variable q is 18
uncorrelated with each of the observable regressors. The linear projection of q onto the observable explanatory variables in our model is:
q = 0 + 1dist_tv + 2dist_coast + 3dist_border +
(4)
where by de...nition E( ) = 0 and Cov(dist_tv; ) = 0; Cov(dist_coast; ) = 0; Cov(dist_border; ) = 0: Substituting the linear projection (4) into the error form model (3) we obtain:
y = ( 0 + 0) + ( 1 + 1)dist_tv + ( 2 + 2)dist_coast + ( 3 + 3)dist_border + " +
where again the zero mean and the orthogonality conditions hold. Therefore,
^
X3
de...ning plim k = k +
3 we can determine the sign and the magnitude
j=1
of the inconsistency, if any.
Our variable of interest is the distance from foreign transmitters for which
we can write:
^ plim 1 =
1+ 1+ 2+ 3
Cov(dist_tv; q) Cov(dist_coast; q) Cov(dist_border; q)
1+
+ V ar(dist_tv)
+ V ar(dist_coast)
V ar(dist_border)
By construction, distance from Italian television transmitters represent an exogenous source of variation uncorrelated with individual speci...c characteristic. The same is true for distance from the coast and from the border. Therefore we can conclude that 1 is consistently estimated by our regression that omits unobservable individual characteristics.
5.3 Last migration decision We next model individual migration decision in the year before the survey. Therefore, we can model individual choice including all observable speci...c characteristics before moving that are contained in the previous wave of the survey. Table 17 shows the estimates for the probability to migrate internationally in the previous 12 months based on the physics model of electromagnetic and radio wave propagation, with and without district controls. Our measure of media exposure is always highly signi...cant. Compared with results of previous section, the order of magnitude increases. When controlling for regional ...xed e¤ects, the probability to migrate decreases by more than 2 percentage point as the distance from Italian television transmitters increases by 1 kilometer. As before, distance from the coast a¤ect positively the migration decision, in almost
19
all the speci...cations, while other orographic variables, are no more statistical signi...cant. In table 18, we estimate a complete model for individual moving decision, including observable individual characteristics. In Column 1 we have our baseline model. Our analysis suggests that male are more likely to migrate abroad and that the probability of moving decreases as age increases. As expected, civil status matters in the sense that people having no family are less likely to move. Household size per se does not inuence migration decision while the percentage of children less than 13 years old has a positive e¤ect on migration behavior. Individual coming form urban areas are less prone to move but surprisingly the statistical correlation between household income and migration probability is positive. Our media exposure variable is signi...cant at 10% level but when including other distance measures (Column 2 and 3) it turns out to be signi...cant at 1% level. As in our previous estimates, topographic variable are not signi...cant when all other controls are included. In Column 4, we include a dummy variable assuming value 1 if the individual had others international migrations in previous years. As expected, on average, the probability to move increases as an individual had other migration experiences. Also controlling from repeated migrations, distance from Italian television repeaters is highly statistical signi...cant. In Column 5, we include also a dummy for purchasing a television and we ...nd that individuals having a television are less likely to move. This result is not against our theory: what matters for individual migration project is foreign media exposure, as con...rmed by our distance measure while national television has a negative e¤ect on the probability of moving. Another question we address is whether information received through internet has an impact on migration. In Table 19 we present results from a probability model for migration in which a dummy for the internet use is included (our internet variable assumes value 1 if in the last year internet has been used). In all the sample, including both urban and rural area, the internet use is not statistically correlated with migration decision but in urban area it is positively correlated at 1% level. In rural areas instead the access to internet predict perfectly the probability to move: other things constants, the ones using internet are the ones who migrate. Having a television has a negative e¤ects on migration in all sample and in rural area but has no e¤ect in urban areas. 5.4 Media and destination choice In the last step of our study we explore whether received information by foreign media a¤ect the choice of the ...nal destination, during the ...rst and the last migration experience. We perform a multinomial logit analysis of the decision of migration in Greece, Italy or other Countries versus the non migration choice. 20
In Table 20 and 20.1 we model destination choice in the ...rst migration episode. Having no other controls than district ...xed e¤ect and distances from the coast and cross - borders, exposure to Italian media increases the probability of moving in Greece, decreases the probability of moving in other countries but the e¤ect on migration in Italy is not statistically di¤erent from zero. For what concerns migration in the last year, our estimates suggest that Italian media increases the probability of moving both in Greece and in Italy but it has no statistical signi...cant e¤ect on migration towards other destinations (Table 21 and 21.1). [TO BE ADDED: Marginal e¤ects] 6 Conclusion Economic theory suggests that potential migrants takes their decision comparing costs and bene...ts. Potential migrants make conjectures about economic condition in destination counties using available information. Traditional information sources as relatives and friends, previous migrants or network abroad have been widely analyzed by economic theory. This paper sheds light on sources of information extremely di¤used nowadays:television and internet. In particular we consider the role played by foreign media in individual migration decision. We consider Albanian migration towards di¤erent countries and we focus on the role played by Italian media in moving decision. Using variation in the position of foreign television transmitters, we model the probability of watching foreign television as a function of distance from the repeaters and topographical and orographic characteristics. Using these measure of exposure to foreign media we ...nd that migration probability is higher for individual who are exposed to foreign media. We then show that Italian media are crucial both for ...rst and last migration experience. We also test, through a multinomial logit speci...cation, whether exposure to Italian media a¤ects the choice of the destination. In particular, having been exposed to Italian media increases the migration probability tout court. Italian television does not attract Albanian only to Italy but induces them to move towards di¤erent western countries: Italian television was simply a door on the world, a way to know and maybe to idealize a di¤erent culture and lifestyle. Italian television was an easy way to know all the world during the communist period: having been exposed to Italian television increases openness towards other cultures and therefore individual probability to migrate internationally. 21
References [1] Barclay, L. (2003), Propagation of Radio Waves, IEE Press, London. [2] Boeri, T. and Tabellini, G. (2005), "Does Information Increase Political Support for Pension Reform?", Bocconi University, mimeo. [3] Carletto, G., Davis, B. and Stampini, M. (2005), "Familiar Face, Familiar Places: the Role of family networks and Previous Experience for Albanian Migration", ESA Working Paper No. 05 - 03. [4] De Soto, H., Gordon, P., Gedeshi, I. and Sinoimeri, Z. (2002), "Poverty in Albania: a Qualitative Assessment", World Bank Technical Paper. [5] Djajiic, S. and Milbourne, R. (1988), "A general model of Guest- Worker Migration: A source country perspective", Journal of International Economics, 25: 335 - 351. [6] Dores, P. (1991), "Guardando all'Italia: Inuenza delle TV e delle Radio Italiane sull'Esodo degli Albanesi", RAI - VQPT, Rome. [7] Dustmann, C. (1997), "Return Migration: The European Experience", Economic Policy, 22: 215-249. [8] Ellington, H.I, Addinall, E. and Hately, M.C. (1980), "The Physics of Television Broadcasting", Physics Education, 15: 222-228. [9] Galor, O. and Stark, O. (1991), "The Probability of Return Migration,Migrants' Work E¤ort, and Migrants' Performance", Journal of Development Economics, 35: 339 - 405. [10] Galor, O. and Stark, O. (1990), "Migrations' Saving, The Probability of Return Migration, and Migrants' Performance", International Economic Review, 31 (2): 463 - 467. [11] Harris, J.R. and Todaro, M.P. (1970), "Migration, Unemployment and Development a 2 - Sector Analysis", American Economic Review, 60 (1): 126 - 142 [12] King, R. and Vullnetari, J. (2003), "Migration and Development in Albania", Development Research Centre on Migration, Globalisation and Poverty Working Paper. [13] Logoreci, A. (1977), "The Albanians: Europe's Forgotten Survivors", Victor Gollancz, London. [14] Mai, N. (2004), "'Looking for a More Modern Life:...': The Role of Italian Television in the Albanian Migration to Italy", Westminster Papers in Communication and Culture, Vol. 1, Issue 1, 3 - 22. 22
[15] Mai, N. and King, R. (2002), "Of Myths and Mirrors: Interpretations of the Albanian Migration to Italy.", Studi Emigrazione, 39, no. 145: 161-200. [16] Mai, N. (2001), "'Italy is Beautiful": The Role of Italian Television in the Albanian Migratory Flow to Italy." In Media and Migration: Constructions of Mobility and Di¤ erence. Edited by King, R., Wood, N., Routledge, pp. 95-109. [17] Massey, D. and Espinosa, K. (1997), "What's Driving Mexico U.S. Migration? A Theoretical, Empirical and Policy Analysis", American Journal of Sociology, 102 (4): 939 - 999. [18] Munshi, K. (2003), "Identi...cation of Network E¤ects: Mexican Migrants in the U.S. labor market", Quarterly Journal of Economics, 118 (2): 549 - 597. [19] Olken, B. (2006), "Do Television and Radio Destroy Social Capital? Evidence from Indonesian Villages" , NBER Working Paper Series. [20] Olsen, N., (2000), Albania, Oxford: Oxfam Country Pro...le. [21] Orrenious, P.M. (1999), "The Role of Family Networks, Coyote Prices and Rural Economy in Migration from Western Mexico: 1965 - 1994", Federal Reserve Bank of Dallas Working Paper 9910. [22] Piperno, F. (2002), "From Albania to Italy", CEME - CeSPI Working Paper. [23] SOPEMI, (2004), Continuous Reporting System on Migration, OECD, Paris. [24] Stromberg, D. (2001), "Mass media and public policy", European Economic Review, 45 (4-6): 652 - 663. [25] Stark, O. and Bloom, D. (1985), "On Migration and Risk in LDCs", Economic Development and Cultural Change, 31 (1):191 - 196. [26] Suphachalasai, S. (2005), "Development, Environmental Policy, and Mass Media: Theory and Evidence", Environmental Economy and Policy Research Discussion Paper Series. [27] Todaro, M. (1969), "A Model of Labor Migration and Urban Unemployment in Less Developed Countries", American Economic Review, 59 (1): 138 - 148 [28] UNDP (2000), Albanian Human Development Report. [29] ]Wooldridge, J. M. (2002), Econometric analysis of cross section and panel data, Massachusetts Institute of Technology. 23
[30] Zahniser, S. (1999), Mexican Migration to the United States: the Role of Migration Networks and Human Capital Accumulation, Garland Publishing, New York. 24
Table A: Emigration from Albania (thousands)
1991
1992
Greece
121
218
Italy
78
88
Other
44
45
Total
243
351
Source: Greek Ministry of Foreign Affairs
1993 241 90 48 381
1994 208 95 50 353
1995 263 98 52 413
1996 312 110 53 476
1997 379 136 55 571
1998 470 162 58 690
Table B: Intra - country migration flows (1989 - 2001)
North - East
Region of arrival Centre/Coast South - East
Total
North - East Percent
Region of Centre/Coast
arrival
Percent
South - East Percent Total Source: INSTAT - Census 2001
2,491 17.5 1,273 2.1 3,764
107,433 98.7 58,256 97.9 165,689
1,465 108,898
1.3
100
11,721 82.5
14,212 100
13,186
59,529 100 182,639
Table C: GDP per capita, US dollar
year 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Albania 686.31 770.76 777.30 769.60 744.38 744.69 807.59 785.67 759.16 817.91 642.35 385.05 248.99 434.36 694.34 835.28 917.82 660.72 813.28 1018.20 1083.99 1194.08 1297.95 1637.11 2131.16 2504.37 2704.18
Italy 8057.9 7291.4 7389.0 7369.0 7360.2 7568.5 10712.0 13419.2 14860.4 15405.8 19472.4 20504.6 21734.1 17438.1 17945.2 19157.6 21487.7 20328.6 20836.4 20546.1 18734.5 18973.1 20708.5 25619.1 29218.8 31873.6 33080.6
Macedonia
Greece
Serbia
Former
Montenegro Republic of
Yugoslav
5077.8
4644.4
4813.9
4329.8
4197.6
4146.6
4857.0
5626.1
6522.9
6729.3
8282.0
8888.9
9747.2
1201.3
9080.2
1315.1
9658.6
1734.2
11253.7
2267.8
11797.1
2234.7
11416.4
1870.1
11371.6
1708.8
1782.0
11611.2
1219.8
1817.5
10451.3
1031.3
1760.9
10743.5
1389.4
1682.0
12197.0
1865.9
1837.8
15811.4
2484.1
2257.9
18721.5
2893.0
2294.8
21016.6
3142.1
2404.1
22273.6
3382.9
2527.5
Source: IMF, World Economic Outlook
US dollars
Table1: Temporary migration experience
Frequency Percent
Individual
Total
Male
1,956
1,765
8.24
14.78
Household
Female
Total
191
1,008
1.62
18.82
Table 2: Temporary migration experience by gender and class of age
Age 0-25 26-35 36-45 46-55 56-65 > 65 Average age Migration age
Total
Male
Female
19.63
19.32
22.51
31.34
32.01
25.13
31.85
33.14
19.9
11.81
11.44
15.18
4.14
3.34
11.52
1.23
0.74
5.76
35.81
35.43
39.21
28.79
28.33
33.14
Table 3: Temporary migration experience by gender and education level
Age None Primary 8 years Secondary general Vacational 2 Years Vacational 4/5 Years University Post-graduate
Total
Male
Female
0.33
0.18
1.61
54.05
55.83
38.17
19.39
18.79
24.73
2.27
2.84
1.61
17.76
17.82
17.2
5.49
4.23
16.67
0.27
0.3 -
Table 4: Year of first migration
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Male Frequency 26 127 168 193 154 136 150
Percent 1.47 7.2 9.52 10.9 8.73 7.71 8.5
Female Frequency
1 6 24 9 3 6 14
Percent 0.52 3.14 12.6 4.71 1.57 3.14 7.33
196 211 107
11.1 12 6.06
38 38
9
19.9 19.9 4.71
115 65 110 6.52 3.68 6.23 11 10 21 5.76 5.24 10.99
7 0.4 1 0.52
Total Frequency 27 133 192 202 157 142 164 234 249 116 126 75 131
8
Percent 1.38 6.8 9.82 10.3 8.03 7.26 8.38 11.96 12.7 5.93 6.44 3.83 6.7 0.41
Table 5: Migration reason
To start a new job/business To look for a better paid job Study Security Health Poor quality land Not enough land To joint family Job reasons
Total
Male
Female
91.46
94.45
63.87
0.66
0.74 -
1.99
1.18
3.66
0.77
0.34
4.71
1.23
0.85
4.71
1.07
0.4
7.33
1.38
0.62
8.38
1.43
0.79
7.33
94.57
96.21
79.58
Table 6: Occupation during first migration experience
Frequency
Total
Male
Female
1,508
1,413
97
Percent
83.68
84.11
77.89
Table 7: Occupation during first migration by education level
None Primary 8 years Secondary general Vacational 2 Years Vacational 4/5 Years University Post-graduate
Total
Male
Female
0.14
0.15 -
55.32
56.56
37.63
19.03
18.48
26.88
2.89
2.87
3.23
17.69
18.1
11.83
4.72
6.62
20.43
0.21
0.23 -
Table 8: Legality Legal migration Legal work
Frequency Percent Frequency Percent
Total
Male
Female
639
489
141
32.67
28.22
73.22
286
233
53
18.97
16.49
55.9
Table 9: Legality by migration year
Year Enter Legally Work Legally
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
14.81 29.32 30.73 26.73 21.02 31.69 28.05 33.33 44.98 30.17 37.3 36 42.75 50
26.09 17.21 31.88 14.19 17.42 11.43 17.19 18.18 16.94 21.74 11.22 31.25 26.6 20
Table 10: Migrating help
Information on how to find
Monetary help
Family
15.21
Family
66.35
Friends
66.76
Friends
11.94
Neighbours 3.11
Neighbours 1.59
Yourself 14.54
Yourself
17.3
Other
0.93
Other
2.81
Help during migration
Family
14.09
Friends
36.3
Acquaintances
32.36
Strangers
3.72
Other (NGO /religious) 13.19
Table 11: Number of migrations
Total 1 2 3 4 5-7 8-21
Male 44.98 14.18 10.15 7.83 14.8 8.05
Female
42.88 66.46
14.36 12.42
10.49
6.83
7.68
9.32
15.99
3.11
6.86
1.86
Table 11.1: First migration destination (%)
Greece
Italy
1990-1993
85.2
1994-1996
78.62
1997
86.32
1998-2000
74.5
2001-2003
65.42
Total
79.88
Other
13.18
1.62
20.73
0.65
12.39
1.28
25.25
30.84
3.37
19.84
1.18
Table 12: Months abroad in the last 20 months (%)
Number of Months 1 2 3 4 5 6 7 8 9 10 11 - 12 More than 13
Total 9.7 9.9 14.95 9.9 20.4 13.94 2.42 5.66 2.83 6.06 2.83 2.21
Male Female
8.82 9.28 14.25 7.92 21.72 15.38 2.71 5.66 2.49 6.33 2.13 2.5
16.98 15.09 20.75 18.87 9.43 1.89 5.66 5.66 3.77 1.89
Table 13 : First migration decision and exposure to foreign media - district level
Dependent Variable =1 if an individual migrated internationally for at least three months
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Distance_90 -0.0032** [0.0016] Distance_coast Distance_border
-0.0073*** [0.0022] 0.0087*** [0.0014]
Distance_itagre
Costal
Central
Mountain
Constant
-1.8085***
[0.1334]
Observations
6673
Pseudo R-squared 0.0022
-1.5152*** [0.2405] 6673 0.0071
-0.0078*** [0.0014] 0.0100*** [0.0010] -0.0099*** [0.0036] -1.2788*** [0.1498] 6673 0.0088
-0.0088*** [0.0015] 0.0092*** [0.0003] -0.0119*** [0.0032] 0.0045** [0.0019] -1.1972*** [0.1569] 6673 0.0094
-0.0046** [0.0021] 0.3933*** [0.0988] 0.6948*** [0.1057] 0.6463*** [0.1022] -2.0841*** [0.3599] 6673 0.0095
-0.0058*** [0.0019] 0.0048*** [0.0012] 0.4214*** [0.0763] 0.5595*** [0.1184] 0.5200*** [0.0611] -2.0260*** [0.2985] 6673 0.0103
-0.0066*** [0.0010] 0.0052*** [0.0014] -0.0088** [0.0037] 0.2200*** [0.0421] 0.4997*** [0.0942] 0.3959*** [0.0496] -1.6183*** [0.0587] 6673 0.0112
-0.0067*** [0.0011] 0.0054*** [0.0011] -0.0113*** [0.0025] 0.0049 [0.0041] 0.2543*** [0.0329] 0.4334*** [0.1363] 0.159 [0.1547] -1.6839*** [0.0838] 6673 0.0115
Robust Standard errors in brackets * significant at 10%; ** significant at 5%; *** significant at 1% For regional controls, the excluded region is Tirana.
Table 14: First migration decision and exposure to foreign media - district level altitude variables included
Dependent Variable =1 if an individual migrated internationally for at least three months
(1)
(2)
Distance_90
-0.0085***
-0.0071***
[0.0019]
[0.0015]
Distance_coast
0.0157***
0.0154***
[0.0022]
[0.0022]
Distance_border
-0.007
0.0025
[0.0065]
[0.0046]
Distance_itagre
-0.0011
-0.0048
[0.0036]
[0.0053]
Altitude_obstacle_90 -0.0023**
-0.0036***
[0.0010]
[0.0013]
Difference_altitude_90 -0.0025**
-0.0036***
[0.0010]
[0.0012]
Altitude*distance_90
0.0069
0.0125*
[0.0051]
[0.0069]
Costal
0.4542***
[0.0428]
Central
0.5339**
[0.2443]
Mountain
0.5267*
[0.2836]
Constant
-1.2669***
-1.8880***
[0.2282]
[0.0932]
Observations
6673
6673
Pseudo R-squared
0.0129
0.0149
Robust standard errors in brackets
* significant at 10%; ** significant at 5%; *** significant at 1%
For regional controls, the excluded region is Tirana.
Table 15: First migration decision and exposure to foreign media - city level
Dependent Variable =1 if an individual migrated internationally for at least three months
Distance_tv Distance_coast Distance_border Distance_itagre Berat
(1) -0.0022*** [0.0008]
(2) -0.0080*** [0.0021] 0.0121*** [0.0034]
(3)
(4)
-0.0077*** -0.0081**
[0.0026] [0.0033]
0.0117*** 0.0116***
[0.0040] [0.0036]
0.0052 0.0043
[0.0072]
[0.0091] 0.0015 [0.0038]
Diber
Durres
Elbasan
Fier
Gjirokaster
Korce
Kukes
Lezhe
Shkroder
Vlore
Constant
-1.9555*** -1.5232*** -1.6999*** -1.6647***
[0.2556] [0.2258] [0.3219] [0.3884]
Observations
4492
4492
4492
4492
Pseudo R-squared 0.0011 0.0108 0.0113 0.0114
Robust standard errors in brackets
* significant at 10%; ** significant at 5%; *** significant at 1%
For regional controls, the excluded district is Tirana.
(5) -0.0058** [0.0023]
(6) -0.0076*** [0.0018] 0.0084** [0.0039]
0.9775*** 0.9391***
[0.0950] [0.0910]
0.5414*** 0.3367***
[0.1442] [0.1173]
0.1076 0.1959
[0.1655] [0.1599]
1.1413*** 0.9182***
[0.1368] [0.1585]
0.5674*** 0.5590***
[0.1165] [0.0782]
0.7114*** 0.4775***
[0.0899] [0.1256]
1.1718*** 0.6254***
[0.1398] [0.2252]
1.2252*** 1.0194***
[0.2604] [0.1492]
0.4320*** 0.5346***
[0.1167] [0.1187]
0.6503*** 0.5709***
[0.1435] [0.1105]
0.4282*** 0.4656***
[0.1399] [0.0768]
-2.0532*** -2.0069***
[0.3561] [0.1714]
4492
4492
0.0219 0.0228
(7) -0.0072*** [0.0019] 0.0078** [0.0039] 0.002 [0.0043] 0.9362*** [0.0966] 0.3659*** [0.1152] 0.2183 [0.1651] 0.9143*** [0.1660] 0.5896*** [0.1133] 0.4742*** [0.1351] 0.6705*** [0.2037] 1.0379*** [0.1301] 0.5354*** [0.1170] 0.5672*** [0.1161] 0.5146*** [0.1574] -2.1128*** [0.3203] 4492 0.0229
(8) -0.0066*** [0.0022] 0.0165*** [0.0055] 0.0062 [0.0042] -0.0143*** [0.0023] 0.9492*** [0.0986] 0.5876*** [0.1022] 0.1822 [0.1703] 0.9968*** [0.1778] 0.6041*** [0.1392] 0.4814*** [0.1383] 0.0293 [0.3378] 1.2946*** [0.1819] 0.5129*** [0.1183] 0.6372*** [0.1311] 0.5258*** [0.1974] -2.1660*** [0.4172] 4492 0.0243
Table 16: First migration decision and exposure to foreign media - city level altitude variables included
Dependent Variable =1 if an individual migrated internationally for at least three months
(1)
(2)
Distance_tv
-0.0101***
-0.0074***
[0.0033]
[0.0015]
Distance_coast
0.0156***
0.0163**
[0.0033]
[0.0073]
Distance_border
0.0076 0.0079***
[0.0068]
[0.0019]
Distance_itagre
-0.0015 -0.0147***
[0.0033]
[0.0032]
Altitude_obstacle
-0.0031***
-0.0020***
[0.0008]
[0.0003]
Difference_altitude -0.0032***
-0.0021***
[0.0009]
[0.0005]
Altitude*distance
0.0132***
0.0090***
[0.0032]
[0.0016]
Berat
0.8865***
[0.0973]
Diber
0.5753***
[0.1863]
Durres
0.1642
[0.1609]
Elbasan
1.0181***
[0.1534]
Fier
0.5674***
[0.0790]
Gjirokaster
0.5447***
[0.1632]
Korce
0.117
[0.3375]
Kukes
1.0724***
[0.1501]
Lezhe
0.4988***
[0.1161]
Shkroder
0.6512***
[0.1254]
Vlore
0.5431***
[0.1156]
Constant
-1.3737***
-2.0069***
[0.3620]
[0.2110]
Observations
4492
4492
Pseudo R-squared
0.0148
0.0255
Robust standard errors in brackets
* significant at 10%; ** significant at 5%; *** significant at 1%
For regional controls, the excluded district is Tirana.
Table 17: Migration decision in the last year and exposure to foreign media
Dependent Variable =1 if an individual migrated internationally for at least three months
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Distance_tv
-0.0029 -0.0123*** -0.0114*** -0.0173*** -0.0208*** -0.0113*** -0.0212*** -0.0194*** -0.0201*** -0.0227***
[0.0020] [0.0029] [0.0030] [0.0031] [0.0035] [0.0040] [0.0056] [0.0057] [0.0060] [0.0061]
Distance_coast
0.0185*** 0.0173*** 0.0141*** 0.0096
0.0289*** 0.0266*** 0.0207*
0.0144
[0.0029] [0.0030] [0.0041] [0.0065]
[0.0073] [0.0081] [0.0114] [0.0128]
Distance_border
0.0153** -0.0017 0.0068
0.0086 0.0063 0.011
[0.0063] [0.0065] [0.0077]
[0.0090] [0.0097] [0.0108]
Distance_itagre
0.0265*** 0.0250***
0.009 0.0098
[0.0058] [0.0058]
[0.0121] [0.0123]
Altitude_obstacle
-0.0022
-0.0018
[0.0014]
[0.0018]
Difference_altitude
-0.0023*
-0.0018
[0.0014]
[0.0018]
Altitude*distance
0.0000**
0
[0.0000]
[0.0000]
Berat
1.0351*** 0.6838** 0.6801** 0.6563** 0.6510*
[0.3008] [0.3241] [0.3240] [0.3285] [0.3559]
Diber
2.2993*** 1.7447*** 1.8658*** 1.7565*** 1.5587***
[0.3976] [0.4218] [0.4574] [0.4976] [0.5560]
Durres
0.4977 0.8290* 0.9443** 0.9609** 0.9283**
[0.4155] [0.4241] [0.4540] [0.4559] [0.4536]
Elbasan
1.7728*** 1.0379*** 1.0239*** 0.9769*** 1.0163***
[0.2757] [0.3242] [0.3234] [0.3346] [0.3423]
Fier
0.6746*
0.4713 0.6156 0.588 0.5201
[0.3515] [0.3617] [0.3873] [0.3913] [0.4002]
Gjirokaster
1.2547*** 0.3789 0.3701 0.344 0.3514
[0.3902] [0.4473] [0.4495] [0.4567] [0.5345]
Korce
1.3508*** -0.4323 -0.2474 0.1972 0.2076
[0.3554] [0.5356] [0.6087] [0.8467] [0.8714]
Kukes
1.8167*** 1.4689** 1.5292** 1.4247**
1.058
[0.5980] [0.6194] [0.6293] [0.6522] [0.8001]
Lezhe
0.9529** 1.3647*** 1.3591*** 1.3793*** 1.3348**
[0.4845] [0.5065] [0.5051] [0.5074] [0.5259]
Shkroder
0.3503 -0.1173 -0.1899 -0.2168 -0.2102
[0.7541] [0.7736] [0.7894] [0.7893] [0.8170]
Vlore
-0.3067 -0.4285 -0.196 -0.2301 -0.3078
[0.5070] [0.5350] [0.5784] [0.5843] [0.5875]
Constant
-2.6812**-1.9379*** -2.4623*** -1.8381*** -1.4018*** -2.2279*** -1.4350* -1.9422** -1.8439* -1.4203
[0.3424] [0.3985] [0.4522] [0.4396] [0.4674] [0.7039] [0.8364] [0.9562] [0.9817] [1.0019]
Observations
4790
4790
4790
4790 4790
4790
4790
4790
4790
4790
Pseudo R-squared 0.0013 0.0199 0.0239 0.0374 0.0408 0.0462 0.0543 0.0549 0.0553 0.0575
Robust standard errors in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
Table 18: Migration decision in the last year individual and HH variables included
Dependent Variable =1 if an individual migrated internationally during the last year
Distance_tv Sex Age Years of school No family % child <13 % HH membrs > 64 HH size Migration network HH income Employed before migration Urban area Distance_coast
(1) -0.0169* [0.0094] 2.7357*** [0.2973] -0.0621*** [0.0063] -0.0734 [0.0508] -1.3227*** [0.3669] 0.7417*** [0.2632] 0.6149 [0.7558] -0.0089 [0.0586] -1.1872*** [0.1772] 0.3347*** [0.1133] -0.6102*** [0.2234] -1.2034*** [0.2666]
Distance_border
Distance_itagre
Other international migration
(2) -0.0291*** [0.0048] 2.7188*** [0.2894] -0.0606*** [0.0059] -0.0678 [0.0480] -1.3001*** [0.3438] 0.7639*** [0.2164] 0.7508 [0.7421] -0.0157 [0.0557] -1.2107*** [0.1381] 0.3990*** [0.1349] -0.6278** [0.2525] -1.2460*** [0.3246] 0.0362*** [0.0062]
(3) -0.0296*** [0.0065] 2.7215*** [0.2878] -0.0605*** [0.0058] -0.0687 [0.0480] -1.2990*** [0.3477] 0.7488*** [0.2373] 0.7579 [0.7350] -0.0165 [0.0580] -1.2085*** [0.1396] 0.4083*** [0.1274] -0.6263** [0.2625] -1.2542*** [0.3410] 0.0317*** [0.0054] -0.0019 [0.0124] 0.007 [0.0057]
Television
Altitude_obstacle
Difference_altitude
Altitude*distance
Constant
-2.8246
-2.5397
[2.5453]
[2.4623]
District Dummies Observations
Yes 4399
Yes 4399
Pseudo R-squared
0.2913
0.3008
Robust standard errors in brackets
* significant at 10%; ** significant at 5%; *** significant at 1%
-2.5582 [2.6399] Yes 4399 0.301
(4) -0.0262*** [0.0050] 1.6888*** [0.4292] -0.0384*** [0.0063] -0.1164*** [0.0229] -0.1768 [0.2657] 0.5584*** [0.1272] 0.8905 [0.6785] -0.0262 [0.0877] -0.6962*** [0.2652] 0.3298*** [0.1045] -0.7144*** [0.2620] -0.9149*** [0.3122] 0.0208*** [0.0033] -0.0067 [0.0123] 0.0194*** [0.0063] 2.6271*** [0.4607]
(5) -0.0262*** [0.0052] 1.6898*** [0.4327] -0.0382*** [0.0065] -0.1152*** [0.0228] -0.1842 [0.2609] 0.4966*** [0.1119] 0.8304 [0.6877] -0.0257 [0.0861] -0.6861*** [0.2657] 0.3393*** [0.0998] -0.7238*** [0.2611] -0.9035*** [0.2945] 0.0209*** [0.0031] -0.0071 [0.0125] 0.0187*** [0.0066] 2.6504*** [0.4731] -0.8340** [0.3782]
-2.9051 [2.3760] Yes 4399 0.4014
-2.1579 [2.4063] Yes 4399 0.4028
(6) -0.0285*** [0.0041] 1.6887*** [0.4294] -0.0382*** [0.0063] -0.1150*** [0.0228] -0.1829 [0.2621] 0.5009*** [0.1134] 0.8322 [0.6461] -0.0262 [0.0791] -0.6901*** [0.2602] 0.3338*** [0.0914] -0.7267*** [0.2689] -0.8905*** [0.3221] 0.0177** [0.0083] -0.0048 [0.0176] 0.0192*** [0.0059] 2.6481*** [0.4603] -0.8139** [0.3490] -0.0017 [0.0023] -0.0017 [0.0023] 0.01 [0.0110] -1.7339 [2.1635] Yes 4399 0.4035
Table 19: Migration decision in the last year and all media
Dependent Variable =1 if an individual migrated internationally during the last year
All sample
Urban
Rural
(1)
(2)
(3)
Distance_tv Television Internet Constant Individual controls Regional dummies Distance controls Observations Pseudo R-squared
-0.0287*** [0.0041] -0.8237** [0.3470] 0.9706 [0.7665] -1.6448 [2.0784] Yes Yes Yes 4399 0.4048
-0.0434*** [0.0095] -2.0125 [4.8500] 1.3782*** [0.3149] -2.2926 [9.6134] Yes Yes Yes 1819 0.3358
-0.0211*** [0.0047] -0.7985*** [0.2590] -2.1922 [2.5395] Yes Yes Yes 2200 0.4356
Robust standard errors in brackets * significant at 10%; ** significant at 5%; *** significant at 1%
Table 20: First migration decision and media by destination country
Dependent variable = 0 no migration, 1 migration to Greece, 2 migration to
Italy, 3 migration to other countries
Distance_90
Greece
Italy
-0.012** -0.003
Other 0.010*
Greece -0.011**
Italy 0.003
Other 0.010**
[0.002] [0.003] [0.004]
[0.003]
[0.006]
[0.003]
Distance_coast 0.015** -0.048** -0.028*
0.011**
-0.052** -0.029**
[0.002] [0.010] [0.015]
[0.004]
[0.004]
[0.004]
Distance_border -0.017** 0.007
0.004
-0.013**
0.011**
-0.030*
[0.005] [0.011] [0.013]
[0.002]
[0.003]
[0.014]
Distance_itagre Costal
0.004 0.033** [0.003] [0.012]
0.029* [0.015]
0.003 [0.007] 0.275**
0.049* [0.025] 1.000**
0.041** [0.005] -1.150**
[0.025]
[0.242]
[0.346]
Central
0.587*
0.345
-0.443**
[0.231]
[0.262]
[0.118]
Mountain
0.464*
-0.741
-1.548**
[0.278]
[0.640]
[0.177]
Constant
-1.063** -3.348** -6.658** [0.257] [0.469] [0.780]
-1.488** [0.226]
-5.317** [0.548]
-5.686** [0.704]
Robust standard errors in brackets
* significant at 10%; ** significant at 5%; *** significant at 1% The base category is "no migration"
For regional controls, the excluded region is Tirana.
Table 20.1: First migration decision and media by destination country altitude variables included
Dependent variable = 0 no migration, 1 migration to Greece, 2 migration to Italy, 3
migration to other countries
Greece
Italy
Other
Greece
Italy
Other
Distance_90
-0.011**
-0.003
0.015*
-0.012**
0.003
0.014**
[0.003]
[0.005]
[0.007]
[0.003]
[0.005]
[0.005]
Distance_coast
0.016**
-0.023**
-0.013
0.019**
-0.022
-0.021**
[0.004]
[0.007]
[0.021]
[0.002]
[0.021]
[0.008]
Distance_border
-0.016*
0.012
0.006
-0.001
0.037*
-0.023*
[0.006]
[0.012]
[0.016]
[0.008]
[0.016]
[0.014]
Distance_itagre
0.002
0.014
0.014
-0.005
0.024
0.039*
[0.004]
[0.009]
[0.017]
[0.012]
[0.032]
[0.019]
Altitude_obstacle_90 -0.000
-0.005
-0.001
-0.003
-0.007
-0.000
[0.002]
[0.006]
[0.006]
[0.003]
[0.009]
[0.005]
Difference_altitude_90 -0.001
-0.005
-0.001
-0.003
-0.007
-0.001
[0.002]
[0.006]
[0.006]
[0.002]
[0.009]
[0.005]
Altitude*distance_90 -0.000
0.015
-0.008
0.013
0.027
-0.006
[0.009]
[0.028]
[0.027]
[0.013]
[0.041]
[0.021]
Costal
0.438**
1.429**
-0.882*
[0.063]
[0.303]
[0.469]
Central
0.698
0.546*
-0.864*
[0.439]
[0.248]
[0.347]
Mountain
0.794
-0.153
-2.186*
[0.640]
[0.628]
[0.852]
Constant
-1.170**
-3.395**
-7.315**
-1.577**
-5.818**
-6.577**
[0.329]
[0.630]
[1.119]
[0.186]
[0.621]
[1.439]
Robust standard errors in brackets
* significant at 10%; ** significant at 5%; *** significant at 1%
The base category is "no migration"
For regional controls, the excluded region is Tirana.
Table 21: Migration decision in the last year and media by destination country
Dependent variable = 0 no migration, 1 migration to Greece, 2 migration to Italy, 3 migration to other countries
Distance_tv Distance_coast Distance_border Sex
Greece -0.019* [0.010] 0.030** [0.006] 0.018** [0.006]
Italy -0.046** [0.004] 0.043** [0.009] -0.009** [0.003]
Other -0.006 [0.008] 0.009 [0.026] -0.009 [0.033]
Age
Years of school
No family
% child <13
% HH membrs > 64
HH size
Migration network
HH income
Employed before migration
Urban area
Other international migration
Television
Constant
-3.553* 0.500
-4.548*
District Dummies
[1.497] Yes
[0.509] Yes
[1.895] Yes
Robust standard errors in brackets
* significant at 10%; ** significant at 5%; *** significant at 1%
The base category is "no migration"
Greece -0.024* [0.011] 0.032** [0.011] 0.010 [0.007] 2.251** [0.601] -0.064** [0.018] -0.122** [0.038] -0.592 [0.454] 1.034** [0.234] 0.081 [0.752] 0.056 [0.117] -0.532** [0.176] 0.219 [0.159] -0.871** [0.237] -1.148** [0.080] 2.765** [0.475] -0.740 [0.565] -2.331 [4.303] Yes
Italy -0.066** [0.008] 0.075** [0.015] -0.034** [0.002] 1.975* [0.778] 0.001 [0.010] -0.087 [0.102] 0.493 [0.509] -1.844** [0.302] -1.848 [2.049] -0.133 [0.140] -34.311** [0.840] 0.707* [0.275] 0.269* [0.115] -0.396 [0.599] 1.733** [0.219] 16.048** [3.695] -19.480 [0.000] Yes
Other -0.006 [0.006] 0.019 [0.065] -0.017 [0.096] 0.138 [0.632] -0.010 [0.026] -0.077 [0.096] 0.203 [0.599] 0.316 [0.957] 2.568* [1.416] -0.120 [0.173] -0.421 [0.368] 1.738** [0.629] -1.288* [0.522] -0.609 [0.923] 3.258** [0.862] -2.096** [0.609] -19.760** [4.199] Yes
Table 21.1 : Migration decision in the last year and media by destination country altitude variables included
Dependent variable = 0 no migration, 1 migration to Greece, 2 migration to Italy, 3 migration to other countries
Distance_tv Distance_coast Distance_border Altitude_obstacle Difference_altitude Altitude*distance Sex
Greece -0.024* [0.010] 0.020* [0.009] 0.027** [0.005] -0.002 [0.002] -0.002 [0.002] 0.018* [0.009]
Italy -0.040** [0.006] 0.075** [0.016] -0.031* [0.014] 0.005* [0.002] 0.005* [0.002] -0.042* [0.020]
Other -0.003 [0.003] 0.005 [0.020] 0.021 [0.032] -0.009** [0.003] -0.011** [0.003] 0.034** [0.009]
Age
Years of school
No family
% child <13
% HH membrs > 64
HH size
Migration network
HH income
Employed before migration
Urban area
Other international migration
Television
Constant
-2.833* -0.368 -5.389**
District Dummies
[1.621] Yes
[1.151] Yes
[1.116] Yes
Robust standard errors in brackets
* significant at 10%; ** significant at 5%; *** significant at 1%
The base category is "no migration"
Greece -0.029** [0.010] 0.026 [0.018] 0.015 [0.011] -0.003 [0.003] -0.003 [0.003] 0.018 [0.011] 2.246** [0.591] -0.065** [0.017] -0.123** [0.034] -0.604 [0.448] 1.033** [0.242] 0.093 [0.730] 0.054 [0.101] -0.546** [0.171] 0.210 [0.128] -0.884** [0.249] -1.101** [0.115] 2.771** [0.443] -0.762 [0.488] -1.400 [3.831] Yes
Italy -0.055** [0.014] 0.108** [0.028] -0.061* [0.025] 0.006* [0.003] 0.006* [0.003] -0.054 [0.034] 1.887* [0.748] -0.001 [0.010] -0.084 [0.099] 0.390 [0.551] -2.025** [0.635] -1.662 [2.039] -0.132 [0.143] -37.300** [0.833] 0.707** [0.260] 0.250* [0.126] -0.470 [0.703] 1.736** [0.227] 16.019** [4.516] -20.819 [0.000] Yes
Other -0.008 [0.007] 0.028 [0.030] -0.015 [0.065] -0.004 [0.005] -0.004 [0.006] -0.000 [0.021] 0.174 [0.689] -0.011 [0.030] -0.066 [0.096] 0.096 [0.601] 0.107 [1.144] 2.622 [1.616] -0.150 [0.191] -0.547* [0.329] 1.891* [0.751] -1.401** [0.477] -0.761 [1.021] 3.366** [0.892] -3.050** [0.878] -19.827** [5.486] Yes

File: dreaming-another-life-the-role-of-foreign-media-in-migration.pdf
Published: Thu Sep 20 10:37:13 2007
Pages: 44
File size: 1.03 Mb


, pages, 0 Mb
Copyright © 2018 doc.uments.com