by Lauren Taylor TP_LaurenTaylor
 Background Information:
The process of migration is largely affected by spatial differences in economic opportunities. In the typical model explaining the movement of migrants,
key factors include wage levels, unemployment rates, distance, origin, destination, quality of life, housing cost disparities, and personal characteristics
of migrants such as age and education. It is also important to recognize the effects of racial and ethnic background on migration in the United States
today. One such model by William H. Frey and Kao-Lee Liaw analyzes the role that race-ethnicity plays in migration within the United States.
[1.1] Focus of Frey and Liaw’s study:
Frey and Liaw’s study focuses on analyzing interstate migration. In their model, Frey and Liaw assess the role of cultural constraints in migrants’
departure and destination choices. Cultural constraints have an impact on migrants’ choices because of migrants’ needs for social support networks, kinship
ties, and access to informal employment opportunities that arise from within cultural communities. Frey and Liaw additionally analyze the impact of
low-skilled immigration on domestic out-migration from urbanized, high-immigrant states.
[1.2] Initial conclusions:
In their analysis of interstate migration of young adults between 1985 and 1990, Frey and Liaw conclude that the presence of communities of same-race
residents reduces out-migration and attracts Hispanics, Asians, and blacks, particularly those who are foreign-born, who are choosing a destination to move
to. This cultural constraints argument implies that states with high same-race concentrations should be major destinations for migrants. Furthermore, in
considering the recent dispersion of minorities, Frey and Liaw suggest that the theory of spatial assimilation may also be at play in addition to the
cultural constraints theory.
[1.3] Evidence of cultural constraints and spatial assimilation from Frey and Liaw’s previous studies:
Immigrant minorities typically follow well-worn migration patterns that are shaped by racial and ethnic attachments and migration networks. These
traditional patterns are often motivated by employment opportunities and social support provided by networks of individuals of similar backgrounds.
Interestingly, research also shows that the attractiveness of same-race neighbors is decreasing among more well-educated and well-off blacks. Instead,
these individuals choose to relocate to more prosperous cities instead of traditional areas with large black concentrations. In such instances, spatial
assimilation may be occurring. Spatial assimilation involves individuals choosing to move to a new location that is more economically advantaged and has a
better quality of life instead of an area with large residential concentrations of those individuals’ same ethnic-race group. It is also interesting to
note that more assimilated, long-term residents show lower segregation in their housing decisions. Frey and Liaw state that they anticipate the effects of
cultural constraints on migration to be reduced somewhat by spatial assimilation considerations. In particular, the presence of large, same race-ethnic
residential areas in a state will reduce residents’ tendencies to leave the area; however, these patterns will be weaker for more highly educated and
native born, rather than foreign-born, members of these groups.
[1.4] Other factors in migration:
In previous studies, Frey and Liaw find that the influx of low-skilled immigrants into a state causes an outflow of low-skilled domestic migrants to
neighboring states. Migrants’ decisions to leave a state when there is an influx of immigrants is more common for low-skilled and poor whites than for any
another race and ethnic group. Increasing housing prices in high-immigration states also causes out-migration and middle-class flight. Figure 1 shows that
states that gain the most foreign immigrants are also the states that lose the most domestic migrants. Conversely, states that gain the most domestic
migrants are not among the states that gain the most immigrants. Additionally, Hispanic and Asian populations are much more concentrated in the United
States than those of whites and blacks. 70% and 60% of the nation’s population of Hispanics and Asians, respectively, reside in five states, while only
about 30% of whites’ and blacks’ total populations live in five states.
[1.5] Education as an indicator of migration:
College graduates are more likely to migrate than individuals who are less educated. Additionally, college graduates choose to move to different locations
than those without college degrees. For example, California is a top destination for college graduates while neighboring Arizona and Nevada are top gainers
of individuals with high-school educations or lower. Frey and Liaw also noted that college graduates have more focused destinations while people with
lesser educations are far more dispersed across the United States. However, members of both groups generally tend to choose southeastern destinations.
 Modeling Interstate Migration:
To conduct their research, Frey and Liaw fit a two-level nested logit model to a highly disaggregated table generated from the 5 percent Public Use
Microsamples of the 2000 U.S. Census. Using a two-level model enabled them to separately determine how origin area and personal characteristics affect
departure decisions and how destination area and personal characteristics affect migrants’ destination choices. When looking at area attributes of a given
location, Frey and Liaw put particular emphasis on observing the impacts of cultural constraints, spatial assimilation, and middle-class flight. The
overall goal of their analysis is to determine the role that race-ethnicity plays in out-migration and destination selection by observing interactions, or
lack of interactions, between race-ethnic categories and different area attributes.
[2.1] Area attributes considered include:
· low-skilled, foreign-born immigration rate
· total employment growth rate
· service employment growth rate
· unemployment rate
· racial similarity
· median housing value
· distance to next state
· contiguity to next state
· population size of origin and destination location
[2.2] Personal characteristics considered include:
- poverty status
- place of birth
· state of residence in 1995
· state of residence in 2000
 The model:
A two-level nested logit model was used in Frey and Liaw’s analysis. In this model, a given potential migrant has personal attributes, s, and lives in
state, i. Migration behavior depends on departure probability, P(s, i), in the departure submodel and a set of destination choice probabilities, P(j|s, i)
for all j ≠ i, in the destination choice submodel. These probabilities become functions of observable explanatory variables in the two submodels
the inclusive variable defined in (3) reflects the attractiveness of the rest of the system from the perspective of the potential migrant with personal
attributes, s, and living in state, i, in 1995.
[3.1] Model details:
In this model, unknown coefficients in equations (1) and (2) are estimated by the maximum quasi-likelihood method. The dummy variables that represent
personal characteristics are entered into each submodel as interactions with the variables that represent area attributes. The farther away from zero,
either negatively or positively, the resulting coefficient is, the more responsive the personal characteristic is to the area attribute in question.
Furthermore, these interactions are used to determine if a given variable has a significant effect on each of the racial categories of migrants. The two
submodels described above indicate the probability of departure and destination selection given certain personal and area characteristics. Of equal, if not
greater, importance in this study are the Rho-square calculations. These Rho-squares are used to determine the goodness of fit of a given specification of
the submodel and the equation as follows:
Rho-squares are particularly useful in determining the importance of a given subset of explanatory variables against another subset. To do this, each
subset is deleted in turn from the specification and the resulting decreases in Rho-squares are then compared. A greater decrease in Rho-square indicates
that the deleted subset of variables, for example racial similarity, is of greater importance.
 Importance of findings and implications:
This model is important because it helps develop an understanding of the effect of race-ethnicity on individuals’ movement choices across labor markets.
Immigration and its effects on the U.S. economy are important to study now more than ever because of the increased immigration of individuals to the United
States in the past several decades. Furthermore, individuals across different races and ethnicities have different preferences and thus may act differently
than the population as a whole. Thus, traditional labor market migration models no longer adequately explain migration across labor markets. With the
United States becoming even more racially and ethnically diverse, a model such as Frey and Liaw’s that incorporates race-ethnicity in its analysis of
domestic migration patterns is necessary.
Frey and Liaw’s model found that cultural constraints reduce out-migration from places with large same-race concentrations, but contrary to Frey and Liaw’s
hypothesis, this result is not affected by the education level of migrants. Therefore, the departure submodel does not follow the
expected spatial assimilation model. Additionally, it was determined from the data that both high low-skilled immigration rates and median housing values
promoted out-migration for all race-ethnic groups but particularly affected those who had less than a college-level education. Furthermore, individuals who
are older, less-educated, non-poor, and born in the state they live in are less likely to move. Lastly, the Rho-squares indicate that migrant departure can
be explained more by the racial similarity of a given area than by the labor market variables or housing value there.
Frey and Liaw found that both the cultural constraints model and spatial assimilation affect migrants’ destination selection. In other words, migrants of a
particular race-ethnic group are more attracted to areas with large concentrations of others in their ethnic group, and especially attracted if these
migrants’ have only a high-school education or less. Furthermore, high low-skilled immigration rates and housing values act to dissuade worse-educated
migrants when selecting destinations. Additionally, areas with increasing total employment growth rates and income per capita attract migrants. The
Rho-square calculations indicate that areas of racially similar composition (Rho-square of 0.0059) play a more important role than the roles of climate or
housing prices. However, together immigration and labor market variables have a greater impact than the effect of racial similarity (the joint Rho-square
value is 0.0101). Thus, when determining a location to move to, migrants largely value not only racial composition but also immigration rates and other
labor market variables in their decision.
 Possible extensions:
The racial composition of the United States has changed quite drastically in the past several decades. Thus, it is important to adjust traditional economic
models, such as the one describing migration patterns, to incorporate the effect that race-ethnicity plays in these models. Furthermore, the immigration
debate has emerged in the news and in public policy as an important issue in the past few years. Frey and Liaw’s study of the impact of race-ethnicity, in
addition to other factors, on migration patterns in the United States provides important additions to traditional models of migration. However, because
their study used data from 1995-2000, it is important to conduct further studies which incorporate data from more recent years, particularly because annual
immigration rates have increased since 2000 (Migration Policy Institute, 2007). It is also important to analyze the impact of race-migration patterns on
the economies and social structures of both the communities that migrants leave and move to. Studies analyzing the effects of race-ethnicity on local
economies and social structures will be useful in this discussion of migration within the United States.
Frey, William H. and Kao-Lee Liaw. “Migration within the United States: Role of Race-Ethnicity”. Brookings-Wharton Papers on Urban Affairs, 2005,
Migration Policy Institute. “Annual Immigration to the United States: The Real Numbers”. 2007.