Abstract
This paper analyzes transitions into and out of social assistance, unemployment, and employment. We estimate a dynamic multinomial logit model, controlling for endogenous initial condition and unobserved heterogeneity, using a large representative Swedish panel data set. The empirical results suggest that particularly refugee immigrants display a greater degree of “structural” state dependence than natives. The high welfare participation rates among refugee immigrants may be due to the existence of a “welfare trap”, while participation among natives and non-refugee immigrants is largely due to permanent unobserved characteristics. These results suggest that welfare reforms may have differential effects on refugee immigrants and natives.
Similar content being viewed by others
Notes
Our definition of unemployment in our analysis is based on receipt of unemployment compensation and hence differs from the official Statistics Sweden definition of unemployment.
According to the national norm in 1998, a single person would receive 2,884 SEK per month in SA, while a couple with two children would receive about 7,500 SEK per month (depending on the age of the children). These amounts are intended to cover expenses for so-called necessary consumption, such as food, basic clothing, leisure, health, newspapers, telephone, and fees for TV, and partially for housing. Additional assistance for housing is also available, known as “bostadsbidrag”. In 1998, the exchange rate was roughly 8 SEK per U.S. Dollar.
Until 1993, the UB replacement rate was 90% of earnings up to a maximum level determined by the government. In July 1993, the replacement rate was reduced to 80%, and in January 1996, it was further reduced to 75%. The replacement rate was raised back to 80% in September of 1997.
Under the UA program, an unemployed worker receives approximately 200 SEK/day, corresponding to roughly 30% of average earnings, and the maximum benefit period is 150 working days.
We lack information about welfare use before 1990.
The countries defined by the Swedish Immigration Board as refugee countries: Ethiopia, Afghanistan, Bulgaria, Bangladesh, Bosnia, Chile, Sri Lanka, Cuba, Iraq, Iran, India, Yugoslavia, China, Croatia, Lebanon, Moldavia, Peru, Pakistan, Poland, Russia, Soviet Union, Romania Somalia, Syria, Togo, Turkey, Ukraine, Uganda, and Vietnam.
All immigrant households included in LINDA, whether defined as refugees or not, have obtained residence permits. This means, for instance, that asylum seekers who have not yet obtained a residence permit are not included in LINDA. Furthermore, the data does not allow us to identify the exact year of arrival for immigrants who arrived in 1968 or earlier.
The possibility of nonrandom return migration is another reason to define the immigrant sample in this way. Edin et al. (2000) find that that return migration among refugees is low, less than 10% within 5 years since arrival, and if an immigrant is to leave Sweden, it is most likely to take place within the first few years after arrival. By excluding the most recent immigrants, we may decrease the potential effects of return migration on our estimates. We also find it comforting that the results do not change very much between the samples with and without the years since migration restriction.
If the person is participating in an active labor market program, which may include training, and receives unemployment compensation of more that 18,100 Swedish kronor, and did not receive income from SA, he is defined to be unemployed in the given year. Note that our definition of unemployment differs from the official definition used by Statistics Sweden. Nonetheless, for simplicity, we refer to unemployment in this paper utilizing the above definition. Lastly, we are not able to distinguish between unemployment compensation from UA or UB.
The seemingly arbitrary value of annual earnings chosen to indicate employment, 36,200 SEK, refers to the so-called “basic amount”. Statistics Sweden defines individuals as employed during a year if they earned this amount.
As we need to rely on the income sources to classify individuals into different labor market states, in any given year, approximately 3% of our sample does not satisfy the criteria for the above three states. These individuals appear to be predominantly students living with family, individuals on disability or early retirement. We have excluded this group in the subsequent analysis.
Note that the data does not allow us to distinguish when during the year the person received income from a particular source. It is hence possible that someone can be defined to be unemployed directly after a year when the person was defined to be on welfare, despite both UA’s and UB’s work requirements.
It should also be noted that the labor supply decision of one spouse is likely to depend on the labor supply decision of the other spouse. Hence, both the employment and unemployment states may be considered as household states and not individual states.
Note that in this case, it is inappropriate to use an unbalanced panel as this would underestimate the number of spells. This is a problem, in particular, for refugee immigrants, as many arrived during the period analyzed and consequently cannot have as many spells as individuals who were observed the entire period 1990–1996.
It is also possible that long-term illness or disability is another source of spurious state dependence. This is due to the fact that the data does not permit identification of this state and that we observe individuals from 1990 to 1996 and, hence, time invariant, or permanent, refers to no changes over this period.
Note however that the permanent factor, η, allows for a particular form of serial correlation in ε.
A simple and naïve approach would be to assume that the initial conditions are exogenous (uncorrelated with the unobserved individual-specific effects). However, this is a very strong assumption and unlikely to hold. Alternatively, we could assume that the stochastic process that generates the observed participation sequences is in equilibrium at the beginning of the sample period (see Card and Sullivan 1988). As pointed out by Chay and Hyslop 1998, this assumption is unlikely to hold when the observable covariates are time-varying and important determinants of participation. Finally, the random effects assumption could be relaxed in favor of a fixed effects estimator. In this framework, the unobserved individual-specific effects can be absorbed with a conditioning statement which would circumvent the initial conditions problem (see Arellano and Honore 2001, and Honore and Kyriazidou 2000). However, in dynamic models with observable characteristics, the necessary conditioning statement is somewhat restrictive, as it requires exogenous characteristics to be stationary in the final two periods. This implies, among other things that time dummies and local labor market conditions are ruled out.
Instead of using a dynamic multinomial logit model, we could have estimated a discrete-time competing risks model with correlation across the risks. The left censoring problem that exists in the data could have been addressed by specifying distributions for the initial observations similar to what we do above. However, such a competing risks model implies estimating about 50% more parameters than in the dynamic multinomial logit model (in which we have over 100 parameters for each immigrant group).
A similar reduction in serial persistence when controls for endogenous initial conditions and unobserved heterogeneity are incorporated is reported in Chay and Hyslop (1998). Furthermore, we find that controls for both unobserved heterogeneity and initial conditions contribute the reduction in state dependence, but it appears to be mainly driven by the control for endogenous initial conditions.
We define structural state dependence as the ratio of the persistence probabilities with and without controls for unobserved heterogeneity and initial conditions.
In a previous version of this paper, IZA Discussion Paper No. 360, we also explored the sensitivity of our welfare definition where persons belong to the welfare state if they received social assistance for at least three months during the year. We find that the results do not appear to be sensitive to this alternative definition.
References
Andrén T, Gustafsson B (2004) Income effects from labor market training programs in Sweden During the 80’s and 90’s. Int J Manpow 25(8):688–713
Arellano M, Honore B (2001) Panel data models: some recent developments. In: Heckman J, Leamer E (eds) Handbook of econometrics, vol 5. Elsevier, North Holland, pp 3229–3296
Baker M, Benjamin D (1995) The receipt of transfer payments by immigrants to Canada. J Hum Resour 30(4):650–676
Blank RM (1989) Analyzing the Length of Welfare Spells. J Public Econ 39(3):245–273
Borjas GJ, Hilton L (1996) Immigration and the welfare state: immigrant participation in means-tested entitlement programs. Q J Econ 111(2):575–604
Card D, Sullivan D (1988) Measuring the effect of subsidized training programs on movements in and out of employment. Econometrica 56(3):497–530
Chay KY, Hyslop D (1998) Identification and estimation of dynamic binary response panel data models: empirical evidence using alternative approaches. Center for Labor Economics, UC Berkeley, Working Paper No 5
Edin P, Fredriksson P (2000) LINDA—Longitudinal Individual Data for Sweden. Working Paper, Uppsala University, Department of Economics
Edin P, LaLonde RJ, Aslund O (2000) Emigration of immigrants and measures of immigrant assimilation: evidence from Sweden. Swed Econ Policy Rev 7(2):163–204
Engberg J, Gottschalk P, Wolf DA (1990) A random-effects logit model of work-welfare transitions. J Econom 43(1):63–75
Hansen J, Lofstrom M (2003) Immigrant assimilation and welfare participation: do immigrants assimilate into or out-of welfare. J Hum Resour 38(1):74–98
Hansen J, Lofstrom M (2006) Immigrant-native differences in welfare participation: the role of entry and exit rates. IZA Discussion Paper 2261
Heckman J (1981) The incidental parameters problem and the problem of initial conditions in estimating a discrete time-discrete data stochastic process. In: Manski C, McFadden D (eds) Structural analysis of discrete data with econometric applications. MIT Press, London, pp 114–179
Heckman J, Singer B (1984) A method for minimizing the impact of distributional assumption in econometric models for duration data. Econometrica 52:271–320
Honore B, Kyriazidou E (2000) Panel data discrete choice models with lagged dependent variables. Econometrica 68(4):839–874
Hyslop DR (1999) State dependence, serial correlation and heterogeneity in intertemporal labor force participation of married women. Econometrica 67(6):1255–1294
Lofstrom M, Bean FD (2002) Assessing immigrant policy options: labor market conditions and post-reform declines in welfare receipt among immigrants. Demography 39(4):617–37
McFadden D, Train K (2000) Mixed MNL models for discrete response. J Appl Econ 15:447–470
Riphanh RT (2004) Immigrant participation in social assistance programs: evidence from German guestworkers. Applied Economics Quarterly 50(4):329–362
SOPEMI (2000) Trends in international migration. OECD, Paris
Acknowledgment
We would like to thank three anonymous referees, George Borjas, P-A Edin, Peter Fredriksson, Christopher Worswick, Olof Åslund, seminar participants at UC Irvine, Concordia U, Kansas State U, Louisiana State U, U of Miami, Northeastern U, PPIC, UT Dallas, Uppsala U, Williams College, and participants at the CEA annual meeting and the IZA workshop on “Welfare Transitions” for helpful comments.
Author information
Authors and Affiliations
Corresponding author
Additional information
Responsible editor: Klaus F. Zimmermann
Rights and permissions
About this article
Cite this article
Hansen, J., Lofstrom, M. The dynamics of immigrant welfare and labor market behavior. J Popul Econ 22, 941–970 (2009). https://doi.org/10.1007/s00148-008-0195-6
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00148-008-0195-6