Nonparametric and semiparametric methods allow for the es-
timation of panel data models that impose relatively few as-
sumptions. This flexibility has made these methods increasingly
popular among applied researchers. An early paper by Li and Sten-
gos (1996) proposes a method for estimating a fixed effects panel
data model that uses standard methods for estimating nonpara-
metric additive models such as the marginal integration method of
Linton and Nielsen (1995) or a backfitting method such as in Op-
somer and Ruppert (1997) or Mammen et al. (1999). However, this
method does not take full advantage of the structure of the model,
and several more recent papers introduce methods that use more
of this structure. Baltagi and Li (2002) propose a method that uses a
series approximation to estimate the regression function. Hender-
son et al. (2008) introduce an iterative nonparametric kernel esti-
mator and conjecture its asymptotic distribution. This conjecture
is confirmed in Li and Liang (2015).
At the same time, parametric dynamic panel models, which al-
low for the inclusion of lagged dependent variables as regressors,
are also becoming more popular. Dynamic panel models are useful