Chapter 2 Migration Concepts

Migration is a process that involves a change in the place of residence. There are two key dimensions that distinguish migration from other types of human movements and these are time (duration) and space (between places). That is, the process of migration involves moving to a place for an amount of time that is considered long and across a distance that is considered far – both aspects make the move significant and life changing. By life changing, we mean the person who has moved must alter they way they interact with the society they moved away from. In other words, migrations are movements that change the social circumstances of the individual. It an event that is distinguishable from other types of moves that are short duration (such as tourism or family visits) or short distance (residential mobility). Of course, both the duration and spatial dimensions of migration may be culturally defined, meaning that the processes of migration may differ between countries at a given spatial / time scale.

In terms of practical measurement, the United Nations recommends that migration be linked to population change, where the place where one lives is defined as a place where the person ‘usually resides’ withing a given year. the measurement often excludes some groups, such as nomads, who do not have a fixed place of residence, or seasonal migrants, who live in two or more places during the course of a year. Further, migration is often measured as the crossing of a boundary (as opposed to a threshold of distance), whether it be an internal administrative bountry or an international border ((UnitedNations1983?)). Moreover, migration can be measured at various levels, including individuals and households, and domestic migration may be measured at different geographic scales depending on the types of administrative units that are available. Different individual / household groupings and spatial scales can have large invluences on the patterns we see.

Thus, both spatial and temporal dimensions are required in the definition of migration. Moreover, unlike other demographic processes, migration is not a one-time event, but rather a process that can involve multiple moves over the course of a lifetime.

2.1 Spatial

2.1.1 Usual Residence

Central to the spatial dimension of defining a migration event is the concept of place of residence, used to determine the migrants origin and destination locations. The Principles and Recommendations for Population and Housing Censuses (UN Statistics Division 2008: 102, para. 1.463) defines usual residence as follows:

  1. The place at which the person has lived continuously for most of the last 12 months (that is, for at least six months and one day), not including temporary absences for holidays or work assignments, or intends to live for at least six months

  2. The place at which the person has lived continuously for at least the last 12 months, not including temporary absences for holidays or work assignments, or intends to live for at least 12 months.”

The use of two alternative criteria leaves some area of ambiguity, where the subtle differences in each could have significant implications for the measurement of migration. Further, within either of these criteria they might be additional ambiguity. For example, persons on long work assignments might have intentions to stay for only a few months that might turn into many months.

When defining migration, restrictions on distance are often included ((Lee1966?)). In this situation, any move could be a migration including, for example, a move from one apartment to another in the same building or a move to another country. We advocate distinguishing between short disance / local moves, such as within cities or labor markets, and those that are considered longer distance (between cities or labor markets). We recognise that this distinction can be problematic given data availability and different contexts. However, we believe there are important distinctions to be made between local mobility and migration that make their separation worthwhile.Local mobility is primarily driven by household factors while migration is driven by economic and social network factors. Note, the methods covered in this manual can be used to analyse and estimate both types of movements.

When analyzing migration patterns, it is useful to have information on the distance involved in a relocation. If address information on points of origin and destination is available, it is possible to tabulate moves by the distance covered. However, in many countries without population registers, this is not possible, and it may be time-consuming and of little policy relevance. Instead, administrative or political units into which the country is divided are often used. Therefore, migration is operationally defined as a change of residence from one civil division to another, and the volume of migration is then function of the size of areas chosen for compilation, where typically larger administrative units, such as a state or country, have a higher volume of migration compared to smaller units like cities or counties. Most countries typically have hierarchies in their administrative units and provide some internal migration data to reflect one or more of these geographies. Using larger geographic units can result in a loss of detail and accuracy in migration data. For instance, if a state is used as the unit of compilation, it may not capture migration patterns between different cities within the state, which could be important for local policy-making. On the other hand, using smaller geographic units such as census tracts may result in too much detail, making it difficult to draw broader conclusions about migration trends at the regional or national level.

2.1.2 Migration Data Types

There are two main sources of migration flow data and they, to a large extent, determine the type of migration that are analysed and used in demographic estimation and projections. The two main sources are administrative registers and census / survey questionnaires. Administrative registers measure migration events. They record when a person relocates in continuous time, much like administrative registers of births or deaths. Thus, the number of persons who have migrated within a year can be calculated as the sum of all recorded moves that occured in that year. Often there are no duration qualifications so potentially persons can have multiple or return moves within a given year. Censuses and surveys, on the other hand, typically have questions on place of usual residence that can be cross tabulated with questions on place(s) of residence at a previous time, usually one or five years prior to the census / survey. Cross tabulations between place of residence one or five years ago with place of current residence represent migrant transitions. With these data, we do not know the exact date of the move, only that a move has occurred between two points in time. We do not know whether the move occurred 10 months ago or one week ago. Other types of migration data are occasionally collected that do not involve the origin location of a migration, including the duration at residence, number of moves over a given interval, and country of citizenship.

Lifetime migration data is another common type of migration data. Similar to migrant transition data described above, lifetime migration data is cross tabulation between current place of usual residence and place of birth. However, these data are not considered a type of flow data. Instead, they are considered migrant population stock data. With these data, we do not know a time or period when the person migrated, only that they have migrated at least once since birth. The migration data literature often distinguishes between stock and flow data. For international migration, stock data have been gathered and tabulated from most countries in the world by the United Nations and World Bank and are widely used to study the cumulative effects of international migration.

2.1.3 Migrant Transition Data

Migrant transition data are typically collected in national censuses, which identify migrants by comparing their place of usual residence at the time of enumeration (\(t\)) with that at a specified earlier date (\(t-n\)). This type of data provides information on the movements of migrants over a given time period, which is usually either 1 year (e.g. UK) or 5 years (e.g. USA). In terms of limitations, migrant transition data fail to identify multiple and return moves, which can lead to an underestimation of the true level of migration. Additionally, migrants who are born or who die during the measurement period are not counted. For example, a baby born three years ago would not have a place of usual residence five years ago. Surveys may also be used to gather migration data but the sample size must be sufficent to capture what is usually a small proportion of the population making a residential transition. In any given year, the vast majority of the population do not migrate domestically, let alone internationally.

In migrant transition data, a migrant is a person who has experienced one or more migrations during a period measured retrospectively. It is important to note that persons who moved during the measurement interval and subsequently died before its end should technically be counted as migrants and their moves as migrations. However, in practice, such cases are usually unknown or excluded, as the information is usually obtained at the end of the time interval and with reference to persons who are alive at that time. Similarly, with international migration, censuses can be used to measure immigrant transitions but not emigrant transitions. The reason is that the current place of residence is in country not gathering the census data. Censuses have been used to ask questions about family members who have emigrated but if the whole family migrated, then again, censuses would not capture them.

2.1.4 Migation Event Data

Migration event data record all moves made by individual in an administrative system. These include multiple and return moves, newborn moves, and moves immediately before death. Ideally, moves come from a centralised population register, however, other registers that do not necessarily cover the whole population may be used. For instance, some countries require persons to register with local health clinics when moves are made, others may use tax file reports with updated address information as sources of moves. Other administrative registers include those are used for education or voting registrations. Outside population registers, it is important that the migration event data are linked to the usual residence population before analysis of patterns or use in demographic estimation models.

While registers tend to provide a more complete record of migration over time, they may not include the whole population or there may be instances where people do not register or de-register. For example, a common problem with health registration data is young adults (especially males) only register when they see a doctor (which may be several years after the move) or they do not register and continue to visit the health clinic linked to their previous residential location if, say, their parents still live in that location or it is not too far away. Another issue is that geographical units used in administrative registers are generally coarser, and registers often fail to capture within-region moves. Additionally, administrative registers often do not include information about the characteristics of migrants outside age and sex. Some groups may be omitted altogether, such as prisoners and military personnel.

2.2 Temporal

2.2.1 Migration Interval

Migration is a continuous process that occurs over time. To study the propensity or rate of migration, data must be compiled with reference to specific periods of time. These time periods can be either definite or indefinite. Definite interval data is typically collected over fixed-term periods such as one year, five years, ten years, or intercensal periods. Indefinate interval data such as lifetime migration measures or data based on place of last residence lack a definite time reference as age or time at the current residence varies by each individual migrant. The comparability of migration data with different definite time intervals can be prohibitively complicated. Commonly described as the one-year / five-year problem, observed migration data consistently shows the number of migrants recorded over a five-year interval is far less than five times the number recorded over a one-year interval. In addition, the ratio of migrants between a five-year period and a one-year period is not constant, where variations occur depending on multiple factors such as the intensity and type of migration both over time and in each origin and destination. Consequently, there is no straightforward algebraic solution to comparing one-year and five-year migration probabilities (Rogerson1990?).

With international migration, each country measures migration differently and for their own purposes. Often, the rules of the administrative register determine the measure of migration. While the United Nations recommends measures that are aligned with changes in the country of usual residence, the reality is that countries may differ greatly, ranging from, say, no defined time period, three months, six months, or 12 months. Some countries with strong border control statistics are able to track all entries and exits. In Australia, an immigrant is someone who was previously living abroad and stays in the country for 12/16 months. Similarly, an emigrant is someone who has been outside Australia for 12/16 months. Some countries, especially in Eastern Europe, have population registers that rely on the notion of a ‘permanent’ residence, implying that persons leaving (entering) will never return (leave). Countries who use types of systems grossly undercount the true levels of migration. (Cite MIMOSA / IMEM papers)

2.3 Migration Measures

Migration measures are used to quantify the magnitude and direction of population movements between places or regions. These measures can provide important insights into the demographic and social dynamics of populations. There are several different types of migration measures that are commonly used in research and policy analysis, each with its own strengths and limitations.

One of the most common migration measures is the migration rate or intensity, which is defined as the number of migrants observed during a period of time divided by the mid-year population (or average population ‘at risk’ of migration). The migration rate can be calculated for different migration types discussed above, such as one year or five years, and can be used to compare migration across different places or regions. Note, a probability of migration is different from a rate in how the ‘at risk’ population is defined. In a rate, the population at risk is the mid-year (average) population, whereas in probability, the population at the beginning of the time interval is used as the population at risk.

Other migration measures include the count of the number of migrants and the migrant proportion. The migrant proportion is share of the population population that migrated. These measures can be useful for identifying patterns in migration behavior, such as the prevalence of long-distance migration or the likelihood of migration among certain demographic groups. In demography, some analysts would argue that there is no such thing as a in-migration or immigration rate since the the population at risk is outside the population of interest. Rather they are simply ratios that compare the number of migrants to the size of the receiving population. This problem does occur for out-migration rates or emigration rates since both have clear populations at risk of migration.

All migration measures can be affected by data quality, such as underreporting of migrants or errors in place of residence information. These problems tend to increase as the data become more detailed. Additionally, different measures may be more appropriate for different research questions or policy applications. For example, the migration rate may be more useful for understanding the overall magnitude of migration in the population, while the count of migrants can provide a basic understanding of the scale of migration patterns over time and between different spatial units.

Migration measures can be defined at different levels of detail, ranging from region-to-region measures, to region totals, to system totals or index measures. Region-to-region measures capture the flow of migrants between two specific regions, while region totals capture the total number of migrants coming in or going out of a specific region. System totals, or index measures, provide an overall picture of migration within a given system or country, which we will discuss in more detail in the next chapter.

2.3.1 Region-to-region

Region-to-region migration measures also known as bilateral migration, migration streams or origin-destination migration, refer to a migration measure that cross-classified by region of origin and region of destination, forming a matrix of \(n \times (n-1)\) streams along each origin-destination combination, where \(n\) represents the number of regions. The set of region-to-region migration measures can be represented by \(m_{ij}\), where the sub-scripts \(i\) and \(j\) represent the same set of regions for each origin-destination combination. The set of bilateral migration flows provide a basis to asses the compartive volumnes and directions of migration between a set of regions.

The gross interchange represents the total number of migrants moving between a particular pair of regions, i.e. \(m_{ij} + m_{ji}\). The net migration steam or bilateral net migration represents the difference in migration between a pair of region i.e. \(m_{ij} - m_{ji}\). For a pair of streams that are of unequal size, where the net migration stream is not close to zero, there exists a dominant stream which is far large than the reverse or counter stream in the opposite direction.

2.3.2 Region Totals

Every migration event can be considered an out-migration in relation to the region of origin and an in-migration in relation to the region of destination. When migration events involve changes of countries, migration events are typically described as emigration and immigration, rather than out-migration and in-migration. Totals on in- or out-migration for each region are typically used to evaluate the volume of migration to or from a particular set of regions. In some countries, data is collected or aggregated without reference to the place of origin for in-migration totals or destination for out-migration totals. Consequently the migration totals provide the most detailed measure of regional migration but with little information on the direction of the migration flows between each region. A summary of the common terms for migration totals are shown in Table 1. The in-migration (or immigration) totals can be represented by replacing the origin \(i\) index with a \(+\); \(m_{+j}\). Similarly, the out-migration (or emigration) totals can be represented by replacing the destination \(j\) index with \(+\); \(m_{+j}\).

Scale Area Event Term Migrant Term
Internal Origin out-migration out-migrant
Destination in-migration in-migrant
International Origin emigration emigrant
Destination immigration immigrant

The sum of the in-migration and out-migration totals (\(m_{i+} + m_{+j}\)) provides the turnover of each region. Net migration totals provides a balance of movements in opposing directions from the difference between in-migration and out-migration (\(m_{+j} - m_{i+}\)). Net migration measures are more typically obtained via demographic accounting, as a residual from the differences in population change, births and deaths over a period in each region. As this calculation does not require expensive migration data collection systems, net migration measures are one of the most common forms of migration measures. However, net migration measures have a number of notable drawbacks, as highlighted by (rogers1990rnm?). In particular, net migration does not enumerate migrants themselves, but instead follows a residual of in-migrants and out-migrants. Consequently, the dynamics related to the observed migration patterns can be missed. For example, an net migration of -100 might involve a region receiving no in-migrants and sending 100 out-migrants or receiving 1,000,000 migrants and sending 1,000,100 out-migrants. Further migration dynamics are also missed when looking at net migration rates (discussed in the next section) and regularities in age profiles of migration (discussed in Chapter X) are often preculded when using age-specicifc net migration measures.

2.3.3 Rate measures

Migration rates are important indicators for understanding the dynamics of population movement. Out-migration or emigration rates are calculated by dividing the number of out-migrants or emigrants during a specific period by the population exposed to the likelihood of migration. This is represented by the formula:

\[ e^{[t, t+1]} = \frac{E^{[t, t+1]}}{P}k \]

Here, \(e^{[t, t+1]}\) represents the out or emigration rate, \(E\) is the number of out-migrants or emigrants during the period, \(P\) is the population exposed to the likelihood of migration, and \(k\) is a constant, often set as 1000. The exposure population can be the population at the mid-interval, assuming migration is evenly distributed, or the population at the start or end of the interval if migration has a negligible effect on population change. Additionally, out-migration rates can be further decomposed by subsets of the population, such as age or sex:

\[ e^{[t, t+1]}_i = \frac{E^{[t, t+1]}_i}{P_i}k \]

On the other hand, in-migration or immigration ratios are calculated by dividing the number of in-migrants or immigrants by the population not exposed to the risk of migrating into the region. The formula for in-migration ratio is:

\[ i^{[t, t+1]} = \frac{I^{[t, t+1]}}{P}k \]

Similarly, net migration ratios are calculated by dividing the net migration (difference between in-migration and out-migration) by the population not exposed to migration risk:

\[ m^{[t, t+1]} = \frac{M^{[t, t+1]}}{P}k \]

As mentioned previously, in-migration and net migration ratios are different from demographic rates because they use the resident population in the denominator. This approach satisfies the needs of the demographic balancing equation, as rates of gain and loss are measured relative to the same population. The demographic balancing equation is expressed as:

\[ P^{t+1} = P^t \left(1 + b^{[t, t+1]} - d^{[t, t+1]} + i^{[t, t+1]} - e^{[t, t+1]} \right) \]

where \(P^{t+1}\) is the population at the next time point, \(b^{[t, t+1]}\) and \(d^{[t, t+1]}\) represent births and deaths during the period, and \(i^{[t, t+1]}\) and \(e^{[t, t+1]}\) denote in-migration and out-migration rates. Net migration (\(M^{[t, t+1]}\)) can be substituted with the difference between in-migration and out-migration (\(I^{[t, t+1]} - O^{[t, t+1]}\)). The equation can be simplified as:

\[ P^{t+1} = P^t \left(1 + b^{[t, t+1]} - d^{[t, t+1]} + i^{[t, t+1]} - o^{[t, t+1]} \right) \]

This formulation allows for the analysis of population change considering the effects of births, deaths, in-migration, and out-migration.

2.4 References