Soutenance de thèse de Christian Nguenang : 25 juillet

25 Juillet 2018 Recherche

Monsieur Christian Nguenang soutiendra sa thèse de doctorat en Sciences économiques le 25 juillet 2018 à 10h30 sur le sujet « Essays in Financial Econometrics: Interlinked Assets and High-Frequency Data »
Salle  MS 001  (Manufacture des Tabacs)

Directeur de thèse: Nour MEDDAHI, Professeur, Toulouse School of Economics

Les membres du jury sont :

  • Professeur Serge DAROLLES, Université Dauphine
  • Professeur Andres RAHBEK, Université de Copenhague
  • Professeure Sophie MOINAS, Professeur TSE
  • Professeur Nour MEDDAHI, Professeur TSE

Résumé (en anglais):
Institutional changes in markets regulation in recent years have enhanced the multiplication of markets and the cross listing of assets simultaneously in many places. The prices for a security on those interrelated markets are strongly linked by arbitrage activities. This is also the case for one security and its derivatives: Cash and futures, CDS and Credit spread, spot and options. In those multiple markets settings, it is interesting for regulators, investors and academia to understand and measure how each market contributes to the dynamic of the common fundamental value. At the same time, improvement in ITC fueled trading activity and generated High frequency data. My thesis develops new frameworks, with respect to the data frequency, to measure the contribution of each market to the formation of prices (Price discovery) and to the formation of volatility (Volatility discovery).
In the first chapter, I show that existing metrics of price discovery lead to misleading conclusions when using High-frequency data. Due to uninformative microstructure noises, they confuse speed and noise dimension of information processing. I then propose robust-to-noise metrics, that are good at detecting “which market is fast”, and produce tighten bounds. Using Monte Carlo simulations and Dow Jones stocks traded on NYSE and NASDAQ, I show that the data are in line with my theoretical conclusions. 

In the second chapter, I propose a new way to define price adjustment by building an Impulse Response measuring the permanent impact of market's innovation and I give its asymptotic distribution. The framework innovates in providing testable results for price discovery measures based on innovation variance. I later present an equilibrium model of different maturities futures markets and show that it supports my metric: As the theory suggests, the measure selects the market with the higher number of participants as dominating the price discovery. An application on some metals of the London Metal Exchange shows that 3-month futures contract dominates the spot and the 15-month in price formation. 

The third chapter builds a continuous time comprehensive framework for Price discovery measures with High Frequency data, as the literature exists only in a discrete time. It also has advantages on the literature in that it explicitly deals with non-informative microstructure noises and accommodates a stochastic volatility. We derive a measure of price discovery evaluating the permanent impact of a shock on a market’s innovation. Empirics show that it has good properties.

In the fourth chapter, I develop a framework to study the contribution to the volatility of common volatility. This allows answering questions such as: Does volatility of futures markets dominate volatility of the Cash market in the formation of permanent volatility? I build a VECM with Autoregressive Stochastic Volatility estimated by MCMC method and Bayesian inference. I show that not only prices are cointegrated, their conditional volatilities also share a permanent factor at the daily and intraday level. I derive measures of market's contribution to Volatility discovery. In the application on metals and EuroStoxx50 futures, I find that for most of the securities, while price discovery happens on the cash market, the volatility discovery happens in the Futures market. Lastly, I build a framework that exploits High frequency data and avoid computational burden of MCMC. I show that Realized Volatilities are driven by a common component and I compute contribution of NYSE and NASDAQ to permanent volatility of some Dow Jones stocks. I obtain that volatility of the volume is the best determinant of volatility discovery, but low figures suggest others important factors.