Abstract
Estimating nonlinear effects of continuous covariates by penalized splines is well established for regressions with cross-sectional data as well as for panel data regressions with random effects. Penalized splines are particularly advantageous since they enable both the estimation of unknown nonlinear covariate effects and inferential statements about these effects. The latter are based, for example, on simultaneous confidence bands that provide a simultaneous uncertainty assessment for the whole estimated functions. In this paper, we consider fixed effects panel data models instead of random effects specifications and develop a first-difference approach for the inclusion of penalized splines in this case. We take the resulting dependence structure into account and adapt the construction of simultaneous confidence bands accordingly. In addition, the penalized spline estimates as well as the confidence bands are also made available for derivatives of the estimated effects which are of considerable interest in many application areas. As an empirical illustration, we analyze the dynamics of life satisfaction over the life span based on data from the German Socio-Economic Panel. An open-source software implementation of our methods is available in the R package pamfe.
Similar content being viewed by others
Notes
Socio-Economic Panel (SOEP), data of the years 1984–2011, version 28, SOEP, 2012, doi:10.5684/soep.v28.
For notational simplicity, we refrain from adding stochastic covariates and covariates with strictly parametric effects. However, as can be seen in Sect. 5, semiparametric partially linear models can also be handled easily within our framework.
One way to obtain such a decomposition is described in Wood (2006, pp. 316–317).
\(\varvec{\varPsi }\) can be obtained from \(\varvec{\varOmega }^{-1}\) with the help of the Cholesky factorization and matrix inversion.
We observe similar problems for other functions, e.g., \(f(x)=(0.5-x)^{3}\); see the online supplement.
The corresponding question in the SOEP survey is: “How satisfied are you with your life, all things considered?”.
Incorporating leads and lags results in a smaller sample size. In our case, we require an individual to be observed in at least six consecutive periods corresponding to at least one observation for estimation after building two leads and two lags and taking first differences. Albeit the loss of observations, this modeling procedure allows us to investigate whether effects on life satisfaction are long-lasting or just temporary; see for instance Lucas (2007) for a discussion on this issue.
In fact, the estimated derivative was obtained in a new estimation with spline degree five and third-order difference penalty, leading to a smoother curve.
References
Antoniadis, A., Gijbels, I., Verhasselt, A.: Variable selection in additive models using P-splines. Technometrics 54(4), 425–438 (2012)
Baltagi, B.H., Li, D.: Series estimation of partially linear panel data models with fixed effects. Ann. Econ. Financ. 3(1995), 103–116 (2002)
Chen, J., Gao, J., Li, D.: Estimation in partially linear single-index panel data models with fixed effects. J. Bus. Econ. Stat. 31(3), 315–330 (2013a)
Chen, J., Li, D., Gao, J.: Non- and semi-parametric panel data models: a selective review. http://ecgi.ssrn.com/delivery.php?ID=http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2313431 (2013b)
Claeskens, G., Krivobokova, T., Opsomer, J.D.: Asymptotic properties of penalized spline estimators. Biometrika 96(3), 529–544 (2009)
Claeskens, G., Van Keilegom, I.: Bootstrap confidence bands for regression curves and their derivatives. Ann. Stat. 31(6), 1852–1884 (2003)
Crainiceanu, C.M., Ruppert, D., Carroll, R.J., Joshi, A., Goodner, B.: Spatially adaptive Bayesian penalized splines with heteroscedastic errors. J. Comput. Graph. Stat. 16(2), 265–288 (2007)
De Boor, C.: A Practical Guide to Splines. Springer, Berlin (2001)
Eilers, P.H.C., Marx, B.D.: Flexible smoothing with B-splines and penalties. Stat. Sci. 11(2), 89–102 (1996)
Eubank, A.R.L., Speckman, P.L.: Confidence bands in nonparametric regression. J. Am. Stat. Assoc. 88(424), 1287–1301 (1993)
Fahrmeir, L., Kneib, T., Lang, S., Marx, B.: Regression. Models, Methods and Applications. Springer, Berlin (2013)
Ferrer-i Carbonell, A., Frijters, P.: How important is methodology for the estimates of the determinants of happiness? Econ. J. 114(497), 641–659 (2004)
Frijters, P., Beatton, T.: The mystery of the U-shaped relationship between happiness and age. J. Econ. Behav. Organ. 82(2–3), 525–542 (2012)
Hajargasht, G.: Nonparametric panel data models: a penalized spline approach. https://sites.google.com/site/ghgasht/NonpPanelSpline.pdf?attredirects=0 (2009)
Härdle, W., Huet, S., Mammen, E., Sperlich, S.: Bootstrap inference in semiparametric generalized additive models. Econ. Theory 20(02), 265–300 (2004)
Henderson, D.J., Carroll, R.J., Li, Q.: Nonparametric estimation and testing of fixed effects panel data models. J. Econ. 144(1), 257–275 (2008)
Kauermann, G., Krivobokova, T., Fahrmeir, L.: Some asymptotic results on generalized penalized spline smoothing. J. R. Stat. Soc. Ser. B Stat. Methodol. 71(2), 487–503 (2009)
Krivobokova, T., Kauermann, G.: A note on penalized spline smoothing with correlated errors. J. Am. Stat. Assoc. 102(480), 1328–1337 (2007)
Krivobokova, T., Kneib, T., Claeskens, G.: Simultaneous confidence bands for penalized spline estimators. J. Am. Stat. Assoc. 105(490), 852–863 (2010)
Laporte, A., Windmeijer, F.: Estimation of panel data models with binary indicators when treatment effects are not constant over time. Econ. Lett. 88(3), 389–396 (2005)
Li, G., Peng, H., Tong, T.: Simultaneous confidence band for nonparametric fixed effects panel data models. Econ. Lett. 119(3), 229–232 (2013)
Loader, C.R., Sun, J.: Robustness of tube formula based confidence bands. J. Comput. Graph. Stat. 6(2), 242–250 (1997)
López Ulloa, B.F., Møller, V., Sousa-Poza, A.: How does subjective well-being evolve with age? A literature review. J. Popul. Ageing 6(3), 227–246 (2013)
Lucas, R.E.: Adaptation and the set-point model of subjective well-being: does happiness change after major life events? Curr. Dir. Psychol. Sci. 16(2), 75–79 (2007)
Mammen, E., Støve, B., Tjøstheim, D.: Nonparametric additive models for panels of time series. Econ. Theory 25(02), 442 (2009)
Neumann, M.H., Polzehl, J.: Simultaneous bootstrap confidence bands in nonparametric regression. J. Nonparametr. Stat. 9(4), 307–333 (1998)
Pinheiro, J.C., Bates, D.M.: Mixed Effects Models in S and S-PLUS. Springer, Berlin (2000)
Qian, J., Wang, L.: Estimating semiparametric panel data models by marginal integration. J. Econo. 167(2), 483–493 (2012)
Ruppert, D., Wand, M.: Semiparametric Regression. Cambridge University Press, Cambridge (2003)
Su, L., Ullah, A.: Profile likelihood estimation of partially linear panel data models with fixed effects. Econ. Lett. 92, 75–81 (2006)
Su, L., Ullah, A.: Nonparametric and Semiparametric Panel Econometric Models: Estimation and Testing. Handbook of Empirical Economics and Finance, pp. 455–497. Taylor & Francis Group, Abingdon (2011)
Sun, J., Loader, C.R.: Simultaneous confidence bands for linear regression and smoothing. Ann. Stat. 22(3), 1328–1345 (1994)
Wagner, G.G., Frick, J.R., Schupp, J.: The German Socio-Economic Panel Study (SOEP)—scope, evolution, and enhancements. Schmollers Jahrbuch 127(1), 139–169 (2007)
Wang, J., Yang, L.: Polynomial spline confidence bands for regression curves. Stat. Sin. 19, 325–342 (2009)
Wang, X., Shen, J., Ruppert, D.: On the asymptotics of penalized spline smoothing. Electron. J. Stat. 5, 1–17 (2011)
Weyl, H.: Weyl—on the volume of tubes.pdf. Am. J. Math. 61(2), 461–472 (1939)
Wiesenfarth, M., Krivobokova, T., Klasen, S., Sperlich, S.: Direct simultaneous inference in additive models and its application to model undernutrition. J. Am. Stat. Assoc. 107(500), 1286–1296 (2012)
Wood, S.N.: Generalized Additive Models: An Introduction with R. Chapman & Hall/CRC, Boca Raton (2006)
Wood, S.N., Pya, N., Säfken, B.: Smoothing parameter and model selection for general smooth models. J. Am. Stat. Assoc. 1459(8), 1–45 (2016)
Wooldridge, J.M.: Econometric Analysis of Cross Section and Panel Data. MIT Press, Cambridge (2002)
Wunder, C., Wiencierz, A., Schwarze, J., Küchenhoff, H.: Well-being over the life span: semiparametric evidence from British and German longitudinal data. Rev. Econ. Stat. 95(1), 154–167 (2011)
Yang, L.: Confidence band for additive regression model. J. Data Sci. 6, 207–217 (2008)
Yoshida, T., Naito, K.: Asymptotics for penalized additive B-spline regression. J. Jpn. Stat. Soc. 42(1), 81–107 (2012)
Yoshida, T., Naito, K.: Asymptotics for penalized splines in generalized additive models. J. Nonparametr. Stat. 26(2), 269–289 (2014)
Zhang, J., Feng, S., Li, G., Lian, H.: Empirical likelihood inference for partially linear panel data models with fixed effects. Econ. Lett. 113(2), 165–167 (2011)
Author information
Authors and Affiliations
Corresponding author
Electronic supplementary material
Below is the link to the electronic supplementary material.
Appendix: Serial correlation in the first-difference errors
Appendix: Serial correlation in the first-difference errors
Consider Eq. (7): If the error terms \(u_{it},\; i=1,\ldots ,N,\, t=1,\ldots ,T_{i}\) are homoscedastic and independent with expectation zero, then \(\text {E}(u_{it}u_{i,t-1})=0 \text { and }\text {E}(u_{it}u_{it})=\sigma _{u}^{2}.\) It follows for the errors \(\triangle u_{it}=u_{it}-u_{i,t-1}\) in Eq. (9):
and
The correlation of two consecutive error terms for the same individual after applying first differences is then given by
Rights and permissions
About this article
Cite this article
Pütz, P., Kneib, T. A penalized spline estimator for fixed effects panel data models. AStA Adv Stat Anal 102, 145–166 (2018). https://doi.org/10.1007/s10182-017-0296-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10182-017-0296-1