Empirisk analyse af en nk DSGE-model bestående af 16 ligninger ved brug af CVAR-metoden: Taking a medium sized DSGE-model to the data

Empirical analysis of a DSGE-model consisting of 16 equations by using the CVAR-method: Abstract Criticism has been raised against dynamic stochastic equilibrium models (DSGE). One critique of the DSGE-models is that they fail to present an adequate representation of the financial sectors interaction with the rest of the economy. Furthermore, critics state that the DSGE-models are built on implicit and explicit assumptions that are not supported in by the data It is concluded that the CVAR approach is a powerful and more empirically founded tool that enables taking the economic models to the data that includes calculations and hypothesis testing but does not include fixed prior distributions. and to a greater extent is more empirically founded.

Laugesen, Thor

4. semester

Økonomi (cand.oecon), Kandidat

2020

2020-06-02

30

Denne afhandling undersøger en ny-keynesiansk DSGE-model med 16 ligninger, der inkluderer prisrigiditeter og en finansiel sektor via en finansiel accelerator, ved hjælp af den kointegrerede vektorautoregressive metode (CVAR). Motiveret af kritik af DSGE-modeller – herunder deres behandling af den finansielle sektor og brug af antagelser, der ikke understøttes af data – oversætter studiet modellens strukturelle antagelser til testbare hypoteser som kointegrationsrestriktioner i et data-rigt miljø med elleve makrofinansielle variabler. Modellen log-lineariseres og sættes i state-space form for at muliggøre empirisk evaluering, og der foretages også tests relateret til svag eksogenitet. Resultaterne tyder på, at en række af modellens centrale strukturelle antagelser ikke er i overensstemmelse med data, hvilket gør den analyserede nk DSGE-model stærkt kritisabel i lyset af empirien. Afhandlingen konkluderer, at CVAR-tilgangen er et robust, empirisk funderet værktøj til at tage økonomiske modeller til data uden afhængighed af faste priorfordelinger, og den muliggør systematisk hypotesetestning af modellernes antagelser.

This thesis examines a 16‑equation New Keynesian DSGE model that embeds price rigidities and a financial sector via a financial accelerator using the cointegrated vector autoregressive (CVAR) approach. Motivated by critiques of DSGE models—particularly their treatment of the financial sector and reliance on assumptions not supported by data—the study translates the model’s structural assumptions into testable hypotheses as cointegration restrictions within a data‑rich setting spanning eleven macro‑financial variables. The model is log‑linearized and cast in state‑space form to enable empirical evaluation, and tests related to weak exogeneity are conducted. The results indicate that several central structural assumptions are not consistent with the data, rendering the analyzed NK DSGE model highly criticisable when confronted with empirical evidence. The thesis concludes that the CVAR approach is a robust, empirically grounded framework for taking economic models to the data without fixed prior distributions, facilitating systematic hypothesis testing of model assumptions.

[Dette resumé er genereret med hjælp fra AI direkte fra projektet (PDF)]

DSGE ; CVAR

Download
Vis denne rapport i AAU Studenterprojekter