Tesi etd-05152026-140719
Link copiato negli appunti
Tipo di tesi
Corso Ordinario Secondo Livello
Autore
CARDOSI CARRARA, IACOPO
URN
etd-05152026-140719
Titolo
Indirect inference estimation of the rough Heston model
Struttura
Classe Scienze Sociali
Corso di studi
SCIENZE ECONOMICHE E MANAGERIALI - SCIENZE ECONOMICHE E MANAGERIALI
Relatori
tutor Prof. GIACHINI, DANIELE
relatore Prof. CORSI, FULVIO
relatore Prof. CORSI, FULVIO
Parole chiave
- double Heston model
- Heston model
- high-frequency data
- indirect inference estimation
- LHAR model
- RAUX model
- rough Heston model
Data inizio appello
15/06/2026;
Disponibilità
completa
Riassunto analitico
This thesis investigates, within an indirect inference framework, the extent to which continuous-time stochastic volatility models of the Heston family can reproduce the empirical stylized facts of realized volatility, as summarized by discrete-time HAR-RV-type auxiliary models. Following the methodological stance of Corsi and Renò (2012), a continuous-time model is judged by how closely the auxiliary moments it generates match the empirical ones that is, by its ability to reproduce the heterogeneous, multi-horizon persistence and the multi-horizon leverage of log-realized variance. Five estimations are carried out on the same SPY high-frequency sample using THE indirect inference EMSM Efficient estimator: the Heston, the Double Heston and the rough Heston with the LHAR criterion, and the rough Heston and the double Heston with the RAUX criterion, a new and richer auxiliary criterion introduced in this work that augments the persistence block with roughness statistics based on the log-realized-variance variogram, leverage statistics based on realized semivariances, and return-distribution shape statistics. The results show that a single variance factor, whether Markovian or rough, cannot reproduce the stylized facts, whereas the two-factor double Heston does so on an over-identified metric, with implications for the empirical debate on rough volatility. Throughout, the analysis concerns the reproduction of the auxiliary stylized facts and not the identification of the individual structural parameters: a small minimized distance certifies that a model can match the auxiliary moments, but does not establish that those moments pin down the underlying parameters, a question left to future research.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|