Following upon the work at the Cowles commission in the late forties on simultaneous equations systems (intended to represent the macroeconomic system as a set of equations to be used for policy scenario analyses), CORE contribution in the late sixties was in introducing Bayesian estimation of such systems. This brought in flexibility by allowing for the inclusion of prior knowledge about the economic structure.
The Bayesian tradition has been kept alive in the research at CORE, including in financial econometrics as mentioned below. The failure of simultaneous equation systems to take into account the occurrence of structural changes, such as those resulting from the first oil shock in 1974, led to question the exogeneity status of certain variables for statistical inference. The CORE team contributed to this debate with the much quoted concepts of weak and strong exogeneity.
Financial econometrics became an active area of research at CORE after 1990. Topics dealt with have included the microstructure of financial markets and volatility models. Each topic has required the development of new models and econometric tools. For microstructure, dynamic duration, count, and intensity models are complementary approaches to model the dynamics of the trading processes of securities on stock markets. Empirical evidence has been shed on the issue of whether fully electronic markets based on order books are viable in periods of stress.
The quick expansion of semiparametric and nonparametric methods in the eighties led to new issues in structural econometrics. Our team is also contributing to this growing topic, with a particular focus on identification and inference in conditional models of high or infinite dimension.
Current Research Areas and People
- ARCH models: development of Markov-switching and change-point models, to take account of changing levels of the volatility of financial markets
- Realized volatility: robust measures, separation of continuous and jump components. Link between news and jumps.
- Models for realized covariance matrices in large dimension
- Modeling the investment behavior of private equity and venture capital funds
- Applications to risk measurement, option pricing, portfolio allocation in financial assets, energy markets...
Time Series Econometrics
- Local stationarity and structural breaks
- Multi and High dimensional time series, factor models
- Multiscale (wavelet based) models and inference
- Forecasting: The impact of structural break on forecasting methods
- Forecasting long memory processes through autoregressive models
Complex Data Analysis and Stochastic Models
- High dimensional and functional models
- Network and spatial analysis
- Identification and inference in donditional models
- Treatment effects
- Resampling methods
- MCMC algorithms for dynamic models featuring path dependence
- Use of Bayesian inference to account for model uncertainty and structural breaks in time series analysis and forecasting