<section name="raw">
    <SEQUENTIAL>
      <record key="001" att1="001" value="184575" att2="184575">001   184575</record>
      <field key="037" subkey="x">englisch</field>
      <field key="050" subkey="x">Forschungsbericht</field>
      <field key="076" subkey="">Ökonomie</field>
      <field key="079" subkey="y">http://www.ihs.ac.at/publications/eco/es-276.pdf</field>
      <field key="079" subkey="z">Costantini, Mauro - et al., On the Usefulness of the Diebold-Mariano Test in the Selection of Prediction Models (pdf)</field>
      <field key="079" subkey="y">http://ideas.repec.org/p/ihs/ihsesp/276.html</field>
      <field key="079" subkey="z">Institute for Advanced Studies. Economics Series; 276 (RePEc)</field>
      <field key="100" subkey="">Costantini, Mauro</field>
      <field key="103" subkey="">BWZ, University of Vienna, Vienna, Austria</field>
      <field key="104" subkey="a">Kunst, Robert M.</field>
      <field key="107" subkey="">Department of Economics and Finance, Institute for Advanced Studies, Vienna, Austria and Department of Economics, University of</field>
      <field key="Vie" subkey="n">na, Austria</field>
      <field key="331" subkey="">On the Usefulness of the Diebold-Mariano Test in the Selection of Prediction Models</field>
      <field key="335" subkey="">Some Monte Carlo Evidence</field>
      <field key="403" subkey="">1. Ed.</field>
      <field key="410" subkey="">Wien</field>
      <field key="412" subkey="">Institut für Höhere Studien</field>
      <field key="425" subkey="">2011, November</field>
      <field key="433" subkey="">18 pp.</field>
      <field key="451" subkey="">Institut für Höhere Studien; Reihe Ökonomie; 276</field>
      <field key="451" subkey="h">Kunst, Robert M. (Ed.) ; Fisher, Walter (Assoc. Ed.) ; Ritzberger, Klaus (Assoc. Ed.)</field>
      <field key="461" subkey="">Economics Series</field>
      <field key="517" subkey="c">from the Table of Contents: Introduction; The theoretical background; The simulations; Summary and conclusion; References;</field>
      <field key="542" subkey="">1605-7996</field>
      <field key="544" subkey="">IHSES 276</field>
      <field key="700" subkey="">C22</field>
      <field key="700" subkey="">C52</field>
      <field key="700" subkey="">C53</field>
      <field key="720" subkey="">Forecasting</field>
      <field key="720" subkey="">Time series</field>
      <field key="720" subkey="">Predictive accuracy</field>
      <field key="720" subkey="">Model selection!</field>
      <field key="753" subkey="">Abstract: In evaluating prediction models, many researchers flank comparative ex-ante prediction experiments by significance</field>
      <field key="tes" subkey="t">s on accuracy improvement, such as the Diebold-Mariano test. We argue that basing the choice of prediction models on such</field>
      <field key="sig" subkey="n">ificance tests is problematic, as this practice may favor the null model, usually a simple benchmark. We explore the validity</field>
      <field key="of" subkey="t">his argument by extensive Monte Carlo simulations with linear (ARMA) and nonlinear (SETAR) generating processes. For many</field>
      <field key="par" subkey="a">meter constellations, we find that utilization of additional significance tests in selecting the forecasting model fails to</field>
      <field key="imp" subkey="r">ove predictive accuracy.;</field>
    </SEQUENTIAL>
  </section>