Residual-Based Procedures for Prediction and Estimation in a Nonlinear Simultaneous System (Q3339151)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 3872534
Language Label Description Also known as
default for all languages
No label defined
    English
    Residual-Based Procedures for Prediction and Estimation in a Nonlinear Simultaneous System
    scientific article; zbMATH DE number 3872534

      Statements

      Residual-Based Procedures for Prediction and Estimation in a Nonlinear Simultaneous System (English)
      0 references
      0 references
      0 references
      0 references
      1984
      0 references
      nonlinear simultaneous system
      0 references
      static nonlinear econometric model
      0 references
      Monte Carlo predictor
      0 references
      residual-based predictor
      0 references
      asymptotic mean squared prediction error
      0 references
      misspecification analysis
      0 references
      This paper deals with procedures for obtaining forecasts with a static nonlinear econometric model: \[ (1)\quad f(y_ t,x_ t;\theta)=u_ t\quad (t=1,...,T),\quad f:{\mathbb{R}}^ n\times {\mathbb{R}}^ m\times {\mathbb{R}}^ p\mapsto {\mathbb{R}}^ n. \] The stochastic disturbances \(u_ t\) are i.i.d. and \(E(u_ t)=0\), \(V(u_ t)=I\). (1) is assumed to define implicitly: \(y_ t=g(u_ t,x_ t;\theta)\), \(y_ t\in L^ 2\). The aim of the paper is basically to compare two predictors of y for given \(x_*\) and estimator \({\hat\theta }\): -A Monte Carlo predictor: \(\hat y{}_*^{(m)}=\sum^{S}_{s=1}g(\tilde u_ s,x_*;{\hat\theta })/S\), \(\tilde u{}_ s (s=1,...,S)\) being independent random draws from some completely specified distribution of \(u_*.\) -A residual-based predictor: \(\hat y{}_*^{(r)}=\sum^{T}_{t=1}g(\hat u_ t,x_*;{\hat\theta })/T\) where \(\hat u{}_ t=f(y_ t,x_ t;{\hat\theta }).\) Under convenient assumptions, both are asymptotically unbiased. According to the asymptotic mean squared prediction error, \(\hat y{}_*^{(r)}\) is more efficient than \(\hat y{}_*^{(m)}\) provided that S and T are similar. Possible applications in the area of misspecification analysis are indicated.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references