#### DMCA

## Mobility and the return to education: Testing a Roy Model with multiple markets (2002)

Venue: | ECONOMETRICA |

Citations: | 173 - 0 self |

### Citations

3261 | Sample selection bias as a specification error. - Heckman - 1979 |

2682 | The Central Role of the Propensity Score in Observational Studies for Causal Effects - Rosenbaum, Rubin - 1983 |

2295 | Conditional logit analysis of qualitative choice behaviour, - McFadden - 1974 |

1957 |
Limited-dependent and qualitative variables in econometrics
- Maddala
- 1983
(Show Context)
Citation Context ... could think about extending Lee's approach so the retention probability could be part of the selection correction function just as Heckman's method (1979) has been extended to higher dimensions (see =-=Maddala 1983-=-, pp. 278-283). For example, the researcher could specify the inverse cumulative distribution functions corresponding to the first-best migration probability and the retention probability to both be i... |

1354 |
Specification tests in econometrics
- Hausman
- 1978
(Show Context)
Citation Context ...nder the alternative, OLS is biased but the selection corrected estimates are still consistent. Therefore, the variance of the difference in the estimates is equal to the difference in the variances (=-=Hausman, 1978-=-). When testing for differences in individual coefficients, one should be aware of a multiple comparisons problem. That is, the covariance matrix for the full set of coefficients appearing in Table II... |

1077 | Changes in relative wages, 1963–1987: supply and demand factors.
- Murphy
- 1992
(Show Context)
Citation Context ... time. One possible interpretation is that the rise in the return to education was a nationwide increase which affected state-specific labor markets in a similar fashion (see Bound and Johnson, 1992; =-=Katz and Murphy, 1992-=-). How do the corrected estimates in a state depend on the number of observations? Consistency of the corrected estimates requires sufficient variability in the cell migration probabilities (see Secti... |

1058 | Matching as an econometric evaluation estimator: Evidence from evaluating a job training program.” - Heckman, Ichimura, et al. - 1997 |

694 | Characterizing Selection Bias using Experimental Data - Heckman, Smith, et al. - 1998 |

590 |
Alternative Methods for Evaluating the Impact of Interventions’, in J. J.Heckman and B. Singer (eds) Longitudinal Analysis
- Heckman, Robb
- 1985
(Show Context)
Citation Context ... pursue an alternative approach motivated by the observation that in single-index selection models, the selectivity bias can be written as a function of the probability of selection given covariates (=-=Heckman and Robb, 1985-=-; Ahn and Powell, 1993). This form for the correction term follows from the fact that in latent index models, the selected mean of the error term in the outcome equation is an invertible function of t... |

586 |
Costs and Returns of Human Migration",
- Sjaastad
- 1962
(Show Context)
Citation Context ... cell assignment; therefore, I assume that educational attainment and family circumstances define the cells for a given birth state. 17 For a detailed explanation of the determinants of migration see =-=Sjaastad, 1962-=-; Bowles, 1970; Chiswick, 1974; and Robinson and Tomes, 1982. For a breakdown of mobility rates by age, education, race, sex, and marital status see Geographical Mobility, U.S. Bureau of the Census, s... |

565 |
Wages, Rents and the Quality of Life.
- Roback
- 1982
(Show Context)
Citation Context ...n. A few examples include moving costs, differences in nonwage local labor market characteristics, the relative cost of living, and differences in public services, taxation, climate, and crime rates (=-=Roback, 1982-=-, 1988). To complicate matters, many of the variables which belong in Vjk may be unobservable or poorly measured. In this paper, I attempt to overcome some of the challenges inherent in an extended Ro... |

554 | Some Thoughts on the Distribution of Earnings", - Roy - 1951 |

545 |
Changes in the Structure of Wages in the 1980s: An Evaluation of Alternative Explanations.” American Economic Review,
- Bound, Johnson
- 1992
(Show Context)
Citation Context ... education has risen over time. One possible interpretation is that the rise in the return to education was a nationwide increase which affected state-specific labor markets in a similar fashion (see =-=Bound and Johnson, 1992-=-; Katz and Murphy, 1992). How do the corrected estimates in a state depend on the number of observations? Consistency of the corrected estimates requires sufficient variability in the cell migration p... |

532 |
Does school quality matter? Returns to education and the characteristics of public schools in the United States.
- Card, Krueger
- 1992
(Show Context)
Citation Context ...ividual's state of birth may contain additional information about the earnings process. For example, differences in school quality from state to state may affect the measured return to education (see =-=Card and Krueger, 1992-=-). Equation (1) could be relaxed to allow state of birth to affect earnings; identification would then be achieved through demographic variables not appearing in the wage equations or by restricting t... |

515 |
Self-selection and the Earnings of Immigrants.
- Borjas
- 1987
(Show Context)
Citation Context ...(Gronau, 1974), union versus nonunion employment (Lee, 1978), choice of schooling (Willis and Rosen, 1979), internal and international migration (Nakosteen and Zimmer, 1980; Robinson and Tomes, 1982; =-=Borjas, 1987-=-), training program participation (Ashenfelter and Card, 1985; Ham and LaLonde, 1996), and choice of industry (Heckman and Sedlacek, 1990). In each application, the researchers replace the choice of "... |

490 | Root-N-consistent semiparametric regression. - Robinson - 1988 |

462 | A method of simulated moments for estimation of discrete response models without numerical integration. - McFadden - 1989 |

421 | Integrated Public Use Microdata Series: Version 2.0. Historical Census Projects, - Ruggles, Sobek - 1997 |

410 |
Estimation and Inference in Two-Step Econometric Models,
- Murphy, Topel
- 1985
(Show Context)
Citation Context ...ficients are consistent but the estimated covariance matrix is biased. Naive standard errors are likely to be understated in the second step of a model which does not account for such sampling error (=-=Murphy and Topel, 1985-=-). I correct for the extra sampling variability arising from the imputed migration probabilities to obtain asymptotically correct standard errors. 22 The adjustment turns out to increase the estimated... |

392 | Reducing bias in observational studies using subclassification on the propensity score - Rosenbaum, Rubin - 1984 |

337 | An Econometric Analysis of Residential Electric Appliance Holding - Dubin, McFadden - 1984 |

292 | Using the longitudinal structure of earnings to estimate the effect of training programs.
- Ashenfelter, Card
- 1985
(Show Context)
Citation Context ...t (Lee, 1978), choice of schooling (Willis and Rosen, 1979), internal and international migration (Nakosteen and Zimmer, 1980; Robinson and Tomes, 1982; Borjas, 1987), training program participation (=-=Ashenfelter and Card, 1985-=-; Ham and LaLonde, 1996), and choice of industry (Heckman and Sedlacek, 1990). In each application, the researchers replace the choice of "occupation" in Roy's original paper with a parallel choice of... |

268 | The Sensitivity of an Empirical Model of Married Women’s Hours of Work to Economic and Statistical Assumptions.” Econometrica 55 - Mroz - 1987 |

265 | Generalized econometric models with selectivity. - Lee - 1983 |

251 | Estimating Models with Sample Selection Bias: A Survey’, The - Vella - 1998 |

238 |
Econometric analysis of qualitative response models.
- McFadden
- 1984
(Show Context)
Citation Context ...of the selectivity bias. 6sThe most popular procedure for estimating the selection equations in a multi-choice Roy model has been the conditional logit model or its extension, the nested logit model (=-=McFadden, 1984-=-; Trost and Lee, 1984; Falaris, 1987). While this formulation results in convenient expressions for the selection probabilities, it has the undesirable property referred to as "independence of irrelev... |

238 | Education and Self-Selection.
- Willis, Rosen
- 1979
(Show Context)
Citation Context ...general framework has been applied to a variety of labor market settings, including female labor force participation (Gronau, 1974), union versus nonunion employment (Lee, 1978), choice of schooling (=-=Willis and Rosen, 1979-=-), internal and international migration (Nakosteen and Zimmer, 1980; Robinson and Tomes, 1982; Borjas, 1987), training program participation (Ashenfelter and Card, 1985; Ham and LaLonde, 1996), and ch... |

218 | Characterizing selection bias. - Heckman, Ichimura, et al. - 1997 |

192 | Convergence rates and asymptotic normality for series estimators. - Newey - 1997 |

182 |
Semiparametric estimation of censored selection models. In Nonlinear Econometric Inference,
- Powell
- 2001
(Show Context)
Citation Context ...y which combines Lee's (1983) parametric maximum order statistic approach to reduce the dimensionality of the error terms with more recent work on semiparametric estimation of selection models (e.g., =-=Ahn and Powell, 1993-=-). The resulting semiparametric correction is easy to implement and can be adapted to a variety of other polychotomous choice problems. The empirical work, which uses 1990 U.S. Census data, confirms t... |

178 |
Asymptotic variance of semiparametric estimators.
- Newey
- 1994
(Show Context)
Citation Context ...sufficiency. This is because the rate of convergence for the nonparametric control functions is not affected by the number of correction functions included in the regression (Andrews and Whang, 1990; =-=Newey, 1994-=-b). For a Roy model with five or perhaps as many as ten sectors, a separate correction function could be included for each sending sector j. However, in the current application, estimating 51 differen... |

164 | An Efficient Semiparametric Estimator of the Binary Response Model,
- Klein, Spady
- 1993
(Show Context)
Citation Context ...ke advantage of a latent index framework to characterize the conditional mean of the error term in the outcome equation (see for example Manski, 1985; Newey, Powell, and Walker, 1990; Cosslett, 1991; =-=Klein and Spady, 1993-=-; Ahn and Powell, 1993). A survey of methods to deal with sample selection bias can be found in Vella (1998). 4Notable exceptions are Ichimura and Lee (1991) and Lee (1995a), who extend the semiparame... |

158 | The empirical content of the Roy model - Heckman, Honoré - 1990 |

156 | Semiparametric analysis of discrete response: asymptotic properties of the maximum score estimator
- Manski
- 1985
(Show Context)
Citation Context ... in Heckman, 1981. Most of the approaches of the last decade take advantage of a latent index framework to characterize the conditional mean of the error term in the outcome equation (see for example =-=Manski, 1985-=-; Newey, Powell, and Walker, 1990; Cosslett, 1991; Klein and Spady, 1993; Ahn and Powell, 1993). A survey of methods to deal with sample selection bias can be found in Vella (1998). 4Notable exception... |

144 |
The Effect of Sample Selection and Initial Conditions in Duration Models: Evidence from Experimental Data on
- Ham, Lalonde
- 1996
(Show Context)
Citation Context ...ooling (Willis and Rosen, 1979), internal and international migration (Nakosteen and Zimmer, 1980; Robinson and Tomes, 1982; Borjas, 1987), training program participation (Ashenfelter and Card, 1985; =-=Ham and LaLonde, 1996-=-), and choice of industry (Heckman and Sedlacek, 1990). In each application, the researchers replace the choice of "occupation" in Roy's original paper with a parallel choice of which market or sector... |

141 | Analysis of a Generalized Regression Model,” - Han - 1987 |

123 | Measuring the effect of subsidized training programs on movements in and out of employment.” - Card, Sullivan - 1988 |

115 | Alternative Methods For Solving The Problem of Selection Bias in Evaluating The Impact of Treatments on Outcomes - Heckman, Robb - 1986 |

113 |
Wage Comparisons - A Selectivity Bias,”
- Gronau
- 1974
(Show Context)
Citation Context ...resulting effects on the distribution of earnings in different occupations. Roy’s general framework has been applied to a variety of labor market settings, including female labor force participation (=-=Gronau, 1974-=-), union versus nonunion employment (Lee, 1978), choice of schooling (Willis and Rosen, 1979), internal and international migration (Nakosteen and Zimmer, 1980; Robinson and Tomes, 1982; Borjas, 1987)... |

109 | Shadow Prices - Heckman - 1974 |

109 | Estimation of semiparametric models. - Powell - 1994 |

91 |
Asymptotic Normality of Series Estimators for Nonparametric and Semiparametric Regression Models
- Andrews
- 1991
(Show Context)
Citation Context ... a suitable method for estimating the unknown correction functions. A variety of nonparametric techniques exist to estimate unknown functions; I employ series expansions for the correction functions (=-=Andrews, 1991-=-; Newey, 1988, 1997). 15 Consider 15 Estimating the unknown function semiparametrically can be compared to a parametric approach, such as Lee's, which specifies the joint distribution of the error ter... |

91 |
Migration and Income: The Question of Self- Selection",
- Nakosteen, Zimmer
- 1980
(Show Context)
Citation Context ... settings, including female labor force participation (Gronau, 1974), union versus nonunion employment (Lee, 1978), choice of schooling (Willis and Rosen, 1979), internal and international migration (=-=Nakosteen and Zimmer, 1980-=-; Robinson and Tomes, 1982; Borjas, 1987), training program participation (Ashenfelter and Card, 1985; Ham and LaLonde, 1996), and choice of industry (Heckman and Sedlacek, 1990). In each application,... |

86 |
Some approaches to the correction of selectivity bias.
- Lee
- 1982
(Show Context)
Citation Context ...). Lee then assumes that this newly created random variable has a joint distribution (bivariate normal, for example) with the error term in the outcome equation of interest. Building on earlier work (=-=Lee, 1982-=-), Lee points out that distributions other than the normal can be used for the transformation, providing a method for generating a large class of models with selectivity. Different transformations all... |

82 | Self-Selection and Internal Migration in the United States", NBER Working Papers 4002, - Borjas, Bronars, et al. - 1992 |

78 | Two-step series estimation of sample selection models, MIT Working Papers 99-04
- Newey
- 1999
(Show Context)
Citation Context ...hod for estimating the unknown correction functions. A variety of nonparametric techniques exist to estimate unknown functions; I employ series expansions for the correction functions (Andrews, 1991; =-=Newey, 1988-=-, 1997). 15 Consider 15 Estimating the unknown function semiparametrically can be compared to a parametric approach, such as Lee's, which specifies the joint distribution of the error terms and hence ... |

71 | Semiparametric least squares estimation of multiple index models: single equation estimation,” - Ichimura, Lee - 1991 |

67 |
Abnormal selection bias,
- Goldberger
- 1983
(Show Context)
Citation Context ...triction that the correlation between choices in the same nest be positive. Selectivity bias corrections can also be quite sensitive to departures from the true joint distribution of the error terms (=-=Goldberger, 1983-=-; Arabmazar and Schmidt, 1981, 1982). Estimators relying on joint normality, for example, can perform poorly when the true selectivity effect is nonlinear and nonmonotonic. The bias arising from incor... |

58 | An investigation of the robustness of the Tobit estimator to nonnormality - Arabmazar, Schmidt - 1982 |

55 |
Semiparametric Estimation of Selection Models: Some Empirical Results’, American Economic Review 80(2): 324–8
- Newey, Powell, et al.
- 1990
(Show Context)
Citation Context ...981. Most of the approaches of the last decade take advantage of a latent index framework to characterize the conditional mean of the error term in the outcome equation (see for example Manski, 1985; =-=Newey, Powell, and Walker, 1990-=-; Cosslett, 1991; Klein and Spady, 1993; Ahn and Powell, 1993). A survey of methods to deal with sample selection bias can be found in Vella (1998). 4Notable exceptions are Ichimura and Lee (1991) and... |

50 | Nonparametric identification and estimation of polychotomous choice models,” - Matzkin - 1993 |

42 |
Semiparametric Estimation of a Regression Model with Sample Selectivity
- Cosslett
- 1991
(Show Context)
Citation Context ...e last decade take advantage of a latent index framework to characterize the conditional mean of the error term in the outcome equation (see for example Manski, 1985; Newey, Powell, and Walker, 1990; =-=Cosslett, 1991-=-; Klein and Spady, 1993; Ahn and Powell, 1993). A survey of methods to deal with sample selection bias can be found in Vella (1998). 4Notable exceptions are Ichimura and Lee (1991) and Lee (1995a), wh... |

42 | A Least Squares Correction for Selectivity Bias - Olsen - 1980 |

42 |
Wages, Rents, and Amenities: Differences among Workers and Regions.
- Roback
- 1988
(Show Context)
Citation Context ...the wages of comparably educated labor across space. One explanation for such variation is that differences in the returns to a college education are equalizing differentials for amenity differences (=-=Roback, 1988-=-). If amenities remain largely unchanged over time, permanent differentials in the return to education could exist from state to state. Figure 4 examines the stability in the corrected returns by comp... |

41 |
Semiparametric Estimation of Bivariate Latent Variable Models
- Powell
- 1987
(Show Context)
Citation Context ...person with a college education. A simple differences estimator would yield a consistent estimate of the return to a college education, since the selection bias term is the same for both individuals (=-=Powell, 1987-=-). The polynomial series expansions for the mover and stayer probabilities perform a similar role in separating out the selection bias. Table II summarizes the overall variability in the cell migratio... |

40 |
Further Evidence on the Robustness of the Tobit Estimator to Heteroscedasticity
- Arabmazar, Schmidt
- 1981
(Show Context)
Citation Context ...correlation between choices in the same nest be positive. Selectivity bias corrections can also be quite sensitive to departures from the true joint distribution of the error terms (Goldberger, 1983; =-=Arabmazar and Schmidt, 1981-=-, 1982). Estimators relying on joint normality, for example, can perform poorly when the true selectivity effect is nonlinear and nonmonotonic. The bias arising from incorrectly assuming joint normali... |

34 |
Self-Selection and the Distribution of Hourly Wages
- Sedlacek
- 1990
(Show Context)
Citation Context ...ternational migration (Nakosteen and Zimmer, 1980; Robinson and Tomes, 1982; Borjas, 1987), training program participation (Ashenfelter and Card, 1985; Ham and LaLonde, 1996), and choice of industry (=-=Heckman and Sedlacek, 1990-=-). In each application, the researchers replace the choice of "occupation" in Roy's original paper with a parallel choice of which market or sector to enter. In this section I develop a Roy model for ... |

34 |
Self-selection and interprovincial migration in Canada
- Robinson, Tomes
- 1982
(Show Context)
Citation Context ...labor force participation (Gronau, 1974), union versus nonunion employment (Lee, 1978), choice of schooling (Willis and Rosen, 1979), internal and international migration (Nakosteen and Zimmer, 1980; =-=Robinson and Tomes, 1982-=-; Borjas, 1987), training program participation (Ashenfelter and Card, 1985; Ham and LaLonde, 1996), and choice of industry (Heckman and Sedlacek, 1990). In each application, the researchers replace t... |

29 | Conditioning on the Probability of Selection to Control Selection Bias - Angrist - 1995 |

28 |
Income Inequality: Regional Analyses Within a Human Capital Framework. National Bureau of Economic Research
- Chiswick
- 1974
(Show Context)
Citation Context ... assume that educational attainment and family circumstances define the cells for a given birth state. 17 For a detailed explanation of the determinants of migration see Sjaastad, 1962; Bowles, 1970; =-=Chiswick, 1974-=-; and Robinson and Tomes, 1982. For a breakdown of mobility rates by age, education, race, sex, and marital status see Geographical Mobility, U.S. Bureau of the Census, series P-20, various years. 18 ... |

26 |
Technical training and earnings: A polychotomous choice model with selectivity
- Lee, Trost
- 1984
(Show Context)
Citation Context ...ty bias. 6sThe most popular procedure for estimating the selection equations in a multi-choice Roy model has been the conditional logit model or its extension, the nested logit model (McFadden, 1984; =-=Trost and Lee, 1984-=-; Falaris, 1987). While this formulation results in convenient expressions for the selection probabilities, it has the undesirable property referred to as "independence of irrelevant alternatives." Fo... |

25 |
A Nested Logit Migration Model with Selectivity
- Falaris
- 1987
(Show Context)
Citation Context ...opular procedure for estimating the selection equations in a multi-choice Roy model has been the conditional logit model or its extension, the nested logit model (McFadden, 1984; Trost and Lee, 1984; =-=Falaris, 1987-=-). While this formulation results in convenient expressions for the selection probabilities, it has the undesirable property referred to as "independence of irrelevant alternatives." For two alternati... |

22 | Occupational choice and earnings determination: The role of sample selection and non-pecuniary factors - Dolton, Makepease, et al. - 1989 |

20 |
School Finance Reform, the Distribution
- Card, Payne
- 1998
(Show Context)
Citation Context ...he state-wide average performance on the test declines. The reasoning usually given is that as the fraction of test-takers increases, the quality of the marginal student declines (see Dynarski, 1987; =-=Card and Payne, 1998-=-). 27swith the significant changes in the return to college, these tests indicate that self-selected mobility plays an important role in earnings determination for many states. One possible explanatio... |

20 | Selectivity Bias Correction Methods in Polychotomous Sample Selection Models - Schmertmann - 1994 |

19 |
Additive interactive regression models: circumvention of the curse of dimensionality. Econometric Theory 6
- Andrews, Whang
- 1990
(Show Context)
Citation Context ... the assumption of index sufficiency. This is because the rate of convergence for the nonparametric control functions is not affected by the number of correction functions included in the regression (=-=Andrews and Whang, 1990-=-; Newey, 1994b). For a Roy model with five or perhaps as many as ten sectors, a separate correction function could be included for each sending sector j. However, in the current application, estimatin... |

17 | Semiparametric Maximum Likelihood Estimation of Polychotomous and Sequential Choice Models - Lee - 1995 |

15 |
The Scholastic Aptitude Test: Participation and Performance
- Dynarski
- 1987
(Show Context)
Citation Context ...SAT increases, the state-wide average performance on the test declines. The reasoning usually given is that as the fraction of test-takers increases, the quality of the marginal student declines (see =-=Dynarski, 1987-=-; Card and Payne, 1998). 27swith the significant changes in the return to college, these tests indicate that self-selected mobility plays an important role in earnings determination for many states. O... |

15 | Measuring the unidentified parameter of the extended Roy model of selectivity - Vijverberg - 1993 |

11 |
Series Estimation of Regression Functionals
- Newey
- 1994
(Show Context)
Citation Context ...sufficiency. This is because the rate of convergence for the nonparametric control functions is not affected by the number of correction functions included in the regression (Andrews and Whang, 1990; =-=Newey, 1994-=-b). For a Roy model with five or perhaps as many as ten sectors, a separate correction function could be included for each sending sector j. However, in the current application, estimating 51 differen... |

7 | A note on aggregation of skills and labor quality - Rosen - 1983 |

7 | Generating Conditional Expectations from Models with Sample Selectivity Bias." Economic Letters - Vella - 1988 |

6 | Occupational Choice and Occupational Earnings: Selectivity Bias in a Simultaneous LogitOLS Model. Unpublished - Hay - 1980 |

5 |
Unionism and relative wage rates: A simultaneous equations model with qualitative and limited dependent variables
- Lee
- 1978
(Show Context)
Citation Context ... in different occupations. Roy’s general framework has been applied to a variety of labor market settings, including female labor force participation (Gronau, 1974), union versus nonunion employment (=-=Lee, 1978-=-), choice of schooling (Willis and Rosen, 1979), internal and international migration (Nakosteen and Zimmer, 1980; Robinson and Tomes, 1982; Borjas, 1987), training program participation (Ashenfelter ... |

4 | Computation of opportunity costs in polychotomous choice models with selectivity - Lee - 1995 |

3 |
Migration as Investment: Empirical Tests of the Human Capital Approach to Geographic Mobility
- Bowles
- 1970
(Show Context)
Citation Context ...; therefore, I assume that educational attainment and family circumstances define the cells for a given birth state. 17 For a detailed explanation of the determinants of migration see Sjaastad, 1962; =-=Bowles, 1970-=-; Chiswick, 1974; and Robinson and Tomes, 1982. For a breakdown of mobility rates by age, education, race, sex, and marital status see Geographical Mobility, U.S. Bureau of the Census, series P-20, va... |

2 |
Consistent Estimation of Index Model Coefficients
- Ichimura
- 1987
(Show Context)
Citation Context ...butional assumptions. Some of these methods use nonparametric regression to estimate the unknown distribution function of the selection errors and the accompanying regression parameters (for example, =-=Ichimura, 1987-=-; Newey, Powell, and Walker 1990; Cosslett, 1991; Klein and Spady, 1993). Such an approach would still require modeling the determinants of utility and a correct specification of how these conditionin... |

1 | Identification and Estimation of Nonparametric and Semiparametric Sample Selection Models." Doctoral Dissertation - Choi - 1992 |

1 | Source for latitude and longitude of state capitals - Munro - 1988 |

1 | Source for variables (13)-(16 - S |