Figure 1
Map of greater regions and metropolitan cities of Brazil.
Table 1
Migrants between metropolitan and non-metropolitan microregiões between 2009 and 2010
| Destination | ||||
|---|---|---|---|---|
| Non-metropolitan | Metropolitan | |||
| Origin | N | % | N | % |
| Non-metropolitan | 380,627 | 46.9 | 167,781 | 20.7 |
| Metropolitan | 162,647 | 20.1 | 99,143 | 12.2 |
[i] Total N = 810,196, using survey weights.
Figure 2
Out-migration rate from metropolitan cities from 2004 to 2009.
Table 2
Characteristics of metropolitan and non-metropolitan microregiões in 2010
| Metropolitan | Non-metropolitan | |||
|---|---|---|---|---|
| Mean | Coeff. of variation | Mean | Coeff. of variation | |
| Population | 2,679,687 | 1.11 | 213,680 | 0.79 |
| Room rent (R$, median) | 72.47 | 0.23 | 45.22 | 0.42 |
| Hourly wage (R$) | 12.11 | 0.22 | 7.23 | 0.29 |
| Share of | ||||
| Unskilled workers | 0.37 | 0.09 | 0.37 | 0.14 |
| Skilled workers | 0.31 | 0.11 | 0.40 | 0.14 |
| High-skilled workers | 0.24 | 0.17 | 0.16 | 0.23 |
| Formally employed | 0.58 | 0.11 | 0.40 | 0.36 |
| Unemployed | 0.06 | 0.29 | 0.05 | 0.41 |
| Share of workers in | ||||
| Agriculture | 0.09 | 0.36 | 0.30 | 0.39 |
| Industry | 0.21 | 0.23 | 0.18 | 0.37 |
| Services | 0.53 | 0.08 | 0.35 | 0.23 |
| Public services | 0.11 | 0.25 | 0.12 | 0.24 |
| People living in | ||||
| Adequate living conditions | 0.57 | 0.28 | 0.36 | 0.67 |
| Other measures | ||||
| GDP growth 2005–2010 | 0.16 | 0.31 | 0.18 | 0.79 |
| Health facilities (per 100,000) | 16.40 | 0.42 | 41.86 | 0.35 |
| Health quality index (0–1) | 0.82 | 0.09 | 0.79 | 0.11 |
| Education quality index (0–1) | 0.77 | 0.14 | 0.73 | 0.14 |
| Homicide rate (per 100,000) | 38.00 | 0.54 | 18.58 | 0.77 |
Table 3
Characteristics of migrants and non-migrants 2010
| Non-metropolitan residents | Metropolitan out-migrants | Metropolitan residents | |
|---|---|---|---|
| Number of observations | 4,184,904 | 19,318 | 1,598,869 |
| Age | 40.25 | 36.85 | 40.22 |
| Female | 0.41 | 0.37 | 0.45 |
| White | 0.51 | 0.51 | 0.51 |
| Education level | |||
| None, primary incomplete | 0.47 | 0.29 | 0.29 |
| Primary, secondary incomplete | 0.16 | 0.16 | 0.17 |
| Secondary, higher incomplete | 0.26 | 0.33 | 0.36 |
| Higher complete | 0.11 | 0.21 | 0.19 |
[i] Proportions and means are computed using survey weights.
Table 4
Labor market characteristics of migrants and non-migrants 2010
| Non-metropolitan residents | Metropolitan out-migrants | Metropolitan residents | |
|---|---|---|---|
| Unemployed | 0.05 | 0.12 | 0.06 |
| Log (monthly wages) | 6.59 | 6.95 | 6.98 |
| Sector | |||
| Formal private | 0.40 | 0.43 | 0.56 |
| Formal public | 0.06 | 0.08 | 0.06 |
| Informal | 0.26 | 0.23 | 0.21 |
| Self-employed | 0.02 | 0.02 | 0.01 |
| Small business | 0.26 | 0.24 | 0.15 |
| Industry, ISIC | |||
| Agriculture | 0.26 | 0.14 | 0.09 |
| Industry | 0.22 | 0.27 | 0.21 |
| Services | 0.38 | 0.44 | 0.54 |
| Public services | 0.15 | 0.17 | 0.17 |
[i] Proportions and means are computed using survey weights.
[ii] Industries include extractive industry processing industry, electricity/gas, sanitation/sewage, construction.
[iii] Services include commerce, transport, housing/food, information/communication, financial services, real estate, professional consulting, science and technology, administrative services, Arts/culture/sports, domestic services, and other services.
[iv] Public services include public administration, security, education, health and social services, and international organizations/foreign institutions.
Table 5
Difference between non-metropolitan destination and metropolitan origin comparing chosen destination to alternative destinations
| Difference between destination and origin in | Chosen destination | Alternative destinations | t-statistic, difference in mean |
|---|---|---|---|
| Expected hourly wages (log) | −0.53 | −0.61 | −24.7 |
| Matched expected wages (log) | −2.73 | −2.79 | −16.2 |
| Rent per room (log) | −0.55 | −0.64 | −21.9 |
| IV (wages in neighboring MRs, log) | −0.10 | −0.14 | −23.5 |
| Population in thousands | −5,605 | −6,326 | −17.1 |
| Homicide rate | −17.66 | −14.11 | 18.2 |
| Health facilities (per 100,000) | 25.49 | 26.46 | 8.1 |
| Health provision quality index (0–1) | −0.03 | −0.05 | −24.0 |
| Education provision quality index (0–1) | −0.00 | −0.04 | −32.8 |
| Distance to origin (km) | 573 | 1,295 | 108.6 |
| Other state than origin | 0.45 | 0.92 | 202.4 |
Table 6
Destination choice conditional on migration, alternative specific logit
| (1) | (2) | (3) | |
|---|---|---|---|
| Wage measure: | Expected wages (log) | Matched expected wages (log) | |
| Price measure: | Rent per room (log) | Wages in neighboring locations (log) | |
| Difference in: | |||
| Wages | 0.054 (0.175) | −0.041 (0.234) | −0.069 (0.244) |
| Prices | −0.173 (0.213) | −0.805 (0.730) | −0.822 (0.631) |
| Population (log) | −0.019 (0.068) | −0.040 (0.064) | 0.011 (0.080) |
| Homicide rate | 0.004 (0.004) | ||
| Health facilities | 0.008*** (0.003) | ||
| Health quality index | −1.292* (0.758) | ||
| Education quality index | 1.569* (0.929) | ||
| Destination specific: | |||
| Distance to origin (log) | −0.524*** (0.085) | −0.523*** (0.087) | −0.521*** (0.087) |
| Other state | −1.800*** (0.265) | −1.850*** (0.258) | −1.853*** (0.250) |
| Observations | 5730782 | 5730782 | 5730782 |
| Wald chi2 | 742 | 1222 | 1367 |
| Number of cases | 14509 | 14509 | 14509 |
| Number of alternatives | 514 | 514 | 514 |
Standard errors are clustered at the metropolitan microregião of origin. Estimator is alternative specific conditional logit. In each column, the first set of regressors is the difference between destination and origin for each destination alternative. The second set, indicated as Destination specific, is measured at destination relative to the origin. Prices are measured with the rent per room. Columns 2 and 3 use expected wage differences based on past migrants at the destination and matched residents at origin as explained in Section 4.1. Significance levels
Table 7
Destination choice conditional on migration by the education of migrant, alternative specific logit
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| Level of education: | None or primary | Lower secondary | Upper secondary | Higher |
| Difference in: | ||||
| Matched expected wages (log) | −0.279 (0.255) | −0.161 (0.328) | −0.023 (0.238) | 0.336 (0.238) |
| Prices (IV) | −1.360* (0.700) | −1.205* (0.691) | −0.686 (0.638) | 0.202 (0.563) |
| Population (log) | −0.028 (0.093) | 0.049 (0.112) | 0.022 (0.076) | 0.036 (0.069) |
| Homicide rate | 0.004 (0.004) | 0.003 (0.003) | 0.006 (0.004) | 0.000 (0.004) |
| Health facilities | 0.009*** (0.004) | 0.010** (0.004) | 0.008** (0.003) | 0.002 (0.003) |
| Health quality index | −0.787 (0.950) | −0.754 (0.953) | −1.668* (0.886) | −2.184** (0.852) |
| Education quality index | 1.684 (1.128) | 1.238 (1.114) | 1.977* (1.019) | 1.055 (0.693) |
| Destination specific: | ||||
| Distance to origin (log) | −0.496*** (0.086) | −0.574*** (0.081) | −0.515*** (0.102) | −0.538*** (0.104) |
| Other state | −1.963*** (0.253) | −1.808*** (0.261) | −1.881*** (0.281) | −1.663*** (0.282) |
| Observations | 1871193 | 954109 | 1840023 | 1065457 |
| Wald chi2 | 765 | 1255 | 1658 | 2850 |
| Number of cases | 4835 | 2425 | 4598 | 2651 |
| Number of alternatives | 514 | 514 | 514 | 514 |
Standard errors are clustered at the metropolitan microregião of origin. Estimator is alternative specific conditional logit. Expected wage differences are based on past migrants at destination and matched residents at origin as explained in Section 4.1. Prices are measured with average municipality wages from neighboring mciroregiões. In each column, the first set of regressors is the difference between destination and origin for each destination alternative. The second set, indicated as Destination specific, is measured at destination relative to the origin.
Significance levels
Table 8
Elasticities of significant covariates by sub-sample
| Education | No or primary | Lower secondary | Upper secondary | Higher |
|---|---|---|---|---|
| Distance (log) | −3.4 | −4.0 | −3.6 | −3.7 |
| Other state | −1.8 | −1.7 | −1.7 | −1.5 |
| Health facilities | 0.2 | 0.3 | 0.2 | 0.1 |
| Prices (IV) | 0.2 | 0.2 | 0.1 | 0.0 |
| Education quality | −0.1 | −0.1 | −0.1 | 0.0 |
| Health quality | 0.0 | 0.0 | 0.1 | 0.1 |
Table 9
Differences in actual and predicted wages for metropolitan out-migrants, after matching
Table 10
Differences in actual and predicted wages for metropolitan out-migrants, by education level
| High-educated | ||
| Log (nominal hourly wages) | N | Mean |
| Observed | 3,107 | 2.846 |
| Predicted | 3,107 | 2.930 |
| Difference | −0.084*** | |
| Log (real hourly wages) | N | Mean |
| Observed | 3,107 | −1.270 |
| Predicted | 3,107 | −1.544 |
| Difference | 0.274*** | |
| Low-educated | ||
| Log (nominal hourly wages) | N | Mean |
| Observed | 12,317 | 1.556 |
| Predicted | 12,317 | 1.851 |
| Difference | −0.295*** | |
| Log (real hourly wages) | N | Mean |
| Observed | 12,317 | −2.481 |
| Predicted | 12,317 | −2.611 |
| Difference | 0.130*** | |
Figure 3
Kernel density plots of actual and predicted nominal wages of metropolitan out-migrants with matching.
Figure 4
Kernel density plots of actual and predicted real wages of metropolitan out-migrants with matching.
Table 11
Differences in actual and predicted real wages for metropolitan out-migrants using hedonic prices as a denominator, after matching
| Log (real hourly wages) | N | Mean |
|---|---|---|
| Observed | 15,424 | −2.105 |
| Predicted | 15,424 | −2.155 |
| Difference | 0.050*** |
Figure A1
Price instrument (average wages in contiguous microregiões) and prices (average municipality rents in microregião).
Table A1
Balancing statistics before matching
| Multivariate L1 distance: | 0.83541258 | ||||||
|---|---|---|---|---|---|---|---|
| Univariate imbalance: | L1 | Mean | Min | 25% | 50% | 75% | Max |
| Age at migration | 0.10634 | −2.2895 | −1 | −1 | −3 | −3 | −1 |
| Sex | 0.0979 | −0.0979 | 0 | 0 | 0 | 0 | 0 |
| Education level | 0.07352 | −0.12922 | 0 | −1 | 0 | 0 | 0 |
| Race | 0.00576 | 0.00576 | 0 | 0 | 0 | 0 | 0 |
| City of origin | 0.14503 | 0.72656 | 0 | 0 | 4 | 0 | 0 |
| Marital status | 0.01909 | −0.0098 | 0 | 0 | 0 | 0 | 0 |
| Sector of activity | 0.15388 | −0.96689 | 0 | −3 | −1 | 0 | −1 |
Table A2
Balancing statistics after matching
| Multivariate L1 distance: | 0.78689126 | ||||||
|---|---|---|---|---|---|---|---|
| Univariate imbalance: | L1 | Mean | Min | 25% | 50% | 75% | Max |
| Age at migration | 0.04019 | −0.15014 | 0 | 0 | 0 | −1 | 0 |
| Sex | 0.06926 | −0.06926 | 0 | 0 | 0 | 0 | 0 |
| Education level | 0.03117 | −0.03117 | 0 | 0 | 0 | 0 | 0 |
| Race | 0.00467 | −0.00467 | 0 | 0 | 0 | 0 | 0 |
| City of origin | 0.00479 | 0.00479 | 0 | 0 | 0 | 0 | 0 |
| Marital status | 0.02571 | −0.02571 | 0 | 0 | 0 | 0 | 0 |
| Sector of activity | 0.00355 | 0.00355 | 0 | 0 | 0 | 0 | 0 |
Table A3
Matching summary
| Number of strata: | 9,796 | |
|---|---|---|
| Number of matched strata: | 3,785 | |
| Non-migrants | Migrants | |
| All | 683,517 | 16,172 |
| Matched | 587,346 | 15,401 |
| Unmatched | 96,171 | 771 |
Table A4
Differences of actual and predicted wages for metropolitan out-migrants, before matching
Table A5
Observed and predicted real wage differences using different measures of living costs
| Log (real hourly wages) | ||
|---|---|---|
| High skilled | ||
| Skill-specific mean rents | N | Mean |
| Observed | 2,702 | −1.587 |
| Predicted | 2,702 | −2.098 |
| Difference | 0.510*** | |
| Skill-specific median rents | N | Mean |
| Observed | 2,702 | −1.508 |
| Predicted | 2,702 | −1.948 |
| Difference | 0.439*** | |
| Median hedonic prices | N | Mean |
| Observed | 2,702 | −1.161 |
| Predicted | 2,702 | −1.469 |
| Difference | 0.308*** | |
Significance levels
*** 1% for t-test of difference in means between observed and predicted wages. Predicted wages are based on a matched sample. Rent for the room is aggregated at the micro-region level. Skill-specific rents are the rent per room aggregated only for the high or low skilled observations respectively in a micro-region applying population survey weights. Once the mean is aggregated, in another case the median. Lastly, the author also uses the median to aggregate the hedonic housing price measure. These different price measures are used as denominator to compute real hourly wages.
Table A6
Observed and predicted real wage differences using different measures of living costs
| Log (real hourly wages) | ||
|---|---|---|
| Low skilled | ||
| Skill-specific mean rents | N | Mean |
| Observed | 11,393 | −2.456 |
| Predicted | 11,393 | −2.691 |
| Difference | 0.235*** | |
| Skill-specific median rents | N | Mean |
| Observed | 11,393 | −2.365 |
| Predicted | 11,393 | −2.589 |
| Difference | 0.224*** | |
| Median hedonic prices | N | Mean |
| Observed | 11,393 | −2.374 |
| Predicted | 11,393 | −2.555 |
| Difference | 0.181*** | |
Significance levels
*** 1% for t-test of difference in means between observed and predicted wages. Predicted wages are based on a matched sample. Rent for the room is aggregated at the micro-region level. Skill-specific rents are the rent per room aggregated only for the high or low skilled observations respectively in a micro-region applying population survey weights. Once the mean is aggregated, in another case the median. Lastly, the author also uses the median to aggregate the hedonic housing price measure. These different price measures are used as denominator to compute real hourly wages.
Table A7
Regression of housing prices on housing characteristics, OLS estimates
| log(rent per room) | |
|---|---|
| Urban area | 0.256*** (0.005) |
| Type of dwelling (Base = House) | |
| Townhouse/condominion | 0.146*** (0.003) |
| Flat | 0.396*** (0.002) |
| Hut | 0.196*** (0.006) |
| Wall material (Base = Bricks coated) | |
| Bricks not coated | −0.160*** (0.002) |
| Wood | −0.265*** (0.003) |
| Plaster coated | −0.461*** (0.015) |
| Plaster not coated | −0.521*** (0.020) |
| Wood unprepared | −0.344*** (0.010) |
| Straw | −0.073 (0.155) |
| Others | −0.146*** (0.015) |
| Bathroom (Base = none) | |
| 1 | −0.213*** (0.006) |
| 2 | −0.095*** (0.006) |
| 3 | 0.047*** (0.007) |
| 4 | 0.220*** (0.012) |
| 5 | 0.355*** (0.027) |
| 6 | 0.517*** (0.054) |
| 7 | 0.430*** (0.119) |
| 8 | 1.046*** (0.237) |
| 9 or more | 0.356*** (0.083) |
| Sanitation (Base = General sanitation network) | |
| Septic sump | −0.089*** (0.002) |
| Rudimentary Sump | −0.200*** (0.002) |
| Ditch | −0.225*** (0.005) |
| River, lake or sea | −0.152*** (0.004) |
| Other | −0.212*** (0.009) |
| Waste water (Base = General distribution network) | |
| Well on property | 0.007** (0.003) |
| Well outside property | −0.088*** (0.005) |
| Carro-pipa | −0.072*** (0.014) |
| Rainwater cistern | −0.074*** (0.028) |
| Rain water other | −0.097 (0.068) |
| Rivers, lakes, etc. | −0.081*** (0.023) |
| Other | −0.155*** (0.010) |
| Well in village | 0.165** (0.066) |
| Canalization access (Base = Yes, in min. 1 room) | |
| Yes, only on the property | −0.052*** (0.004) |
| No | −0.148*** (0.006) |
| Garbage collection (Base = Collected directly) | |
| Collected in collective | −0.054*** (0.002) |
| Burnt | −0.229*** (0.008) |
| Buried | −0.017 (0.043) |
| Tossed in a public area | −0.229*** (0.008) |
| Tossed in river, lake, or sea | −0.195*** (0.036) |
| Other | 0.005 (0.027) |
| Electricity provision (base = Yes by the company) | |
| Yes, other | −0.094*** (0.010) |
| No electricity | −0.238*** (0.021) |
| Constant | 4.235*** (0.016) |
| Microregion dummies | Yes |
| Observations | 927,192 |
| R-squared | 0.539 |
Table A8
Coefficients and t-statistics of prediction of wages for migrants based on past migrants at the destination, OLS
| Log(hourly wage) | ||
|---|---|---|
| Coefficient | t-statistic | |
| Age | 0.048 | 18.250 |
| Age squared | −0.045 | −13.939 |
| Female | −0.368 | −61.567 |
| White | 0.123 | 19.385 |
| Education (Base = none) | ||
| Primary, secondary incomplete | 0.240 | 28.118 |
| Secondary, higher incomplete | 0.530 | 71.485 |
| Higher complete | 1.441 | 154.831 |
| Mean (Log (hourly wage)) = | −0.754 | |
[i] OLS estimates weighted with population weights. The samples were all migrants who moved more than one year ago to the destinations.
Table A9
Observed and predicted real wage differences using different measures of living costs, metropolitan in-migrants
| Log (real hourly wages) | ||
|---|---|---|
| High skilled | ||
| Skill-specific mean rents | N | Mean |
| Observed | 1,068 | −1.931 |
| Predicted | 1,068 | −1.974 |
| Difference | 0.043* | |
| Skill-specific median rents | N | Mean |
| Observed | 1,068 | −1.795 |
| Predicted | 1,068 | −1.894 |
| Difference | 0.099*** | |
| Low skilled | ||
| Skill-specific mean rents | N | Mean |
| Observed | 7,357 | −2.775 |
| Predicted | 7,357 | −2.501 |
| Difference | −0.274*** | |
| Skill-specific median rents | N | Mean |
| Observed | 7,357 | −2.680 |
| Predicted | 7,357 | −2.394 |
| Difference | −0.286*** | |
Significance levels
*** 1% for t-test of difference in means between observed and predicted wages. Predicted wages are based on a matched sample. Rent for the room is aggregated at the micro-region level. Skill-specific rents are the rent per room aggregated only for the high or low skilled observations respectively in a micro-region applying population survey weights. Once the mean is aggregated, in another case the median. Lastly, the author also uses the median to aggregate the hedonic housing price measure. These different price measures are used as denominator to compute real hourly wages.
Table A10
Variables and data sources
| Variable | Description | Source |
|---|---|---|
| Variables for descriptive statistics and destination choice model on microregião level | ||
| Wages (IV) | Average monthly wages in neighboring microregião | RAIS* |
| Housing prices | Average rent on microregião level | Census, IBGE |
| Education provision quality index | Index from 0 to 1, computed based on: Subscription rate of pre-school children, dropout rate | FIRJAN** |
| Rate of teachers with higher education, average daily teaching hours, results of the IDEB (Indicator of development of education in Brazil) | ||
| Health provision quality index | Index from 0 to 1, computed based on: Number of pre-natal consultations, deaths due to mal-defined causes, child-deaths due to evitable causes | FIRJAN** |
| Number of health care facilities | Per 100,000 inhabitants; include general hospitals, day hospitals, polyclinics, health point, general emergency, pharmacy, basic health center. | CNES*** |
| Homicide rate | Per 100,000 inhabitants in 2008 | Ipeadata |
| Distance to the state capital | Indicator for market access (Fally et al. 2010) | Ipeadata |
| GDP | Log of GDP in 2009 | Ipeadata |
| Distance between origin and destination | Geodesic distance as an indicator for fixed moving costs, author's calculation from coordinates | Census, IBGE |
| Additional variables for wage regression, on an individual level | ||
| Partner participation | Dummy whether the partner is working | Census, IBGE |
| The proportion of children in the household | Census, IBGE | |
| Marital status | Separated/divorced/widowed, single, married | Census, IBGE |
| Sector | Public, private, informal, self-employed | Census, IBGE |
| Industry | 21 industries according to International Standard Industrial Classification of all Economic Activities (ISIC) | Census, IBGE |
| Federal state | 27 states | Census, IBGE |
| Variables for matching, on an individual level | ||
| Age | At the time of migration, i.e., one year ago | Census, IBGE |
| Race | White and non-white | Census, IBGE |
| Education level | Primary, middle, high-school, college | Census, IBGE |
| Micro-region of origin/residency | City of origin for migrants and city of residency for comparison group of non-migrants | Census, IBGE |