Papers & Presentations

A good description of the dynamics of interest rates is crucial to price derivatives and to hedge corresponding risk. Interest rate modelling in an unstable macroeconomic context motivates one factor models with time varying parameters. In this paper, the local parameter approach is introduced to adaptively estimate interest rate models. This method can be generally used in time varying coefficient parametric models. It is used not only to detect the jumps and structural breaks, but also to choose the largest time homogeneous interval for each time point, such that in this interval, the coefficients are statistically constant. We use this adaptive approach and apply it in simulations and real data. Using the three month treasure bill rate as a proxy of the short rate, we find that our method can detect both structural changes and stable intervals for homogeneous modelling of the interest rate process. In more unstable macroeconomy periods, the time homogeneous interval can not last long. Furthermore, our approach performs well in long horizon forecasting.

This paper proposes an affine term structure model in a stochastic volatility setting. It provides a useful modeling tool to bridge the two strands of macroeconomic research: the DSGE-VAR with stochastic volatility and the macro-finance model of term structure. The state vector X_{t }follows a VAR. Its innovations are conditional normal with a time-varying variance-covariance _{t} modeled by a Wishart Autoregression process, which directly drives the risk price in the stochastic discount factor. In this setting, the yield curve under no-arbitrage is determined both by the state vector X_{t} and the stochastic volatility-covolatility _{t}. Simulation of the baseline model shows that: 1) two factors are sufficient to fully reproduce all typical shapes of the yield curve; 2) _{t} has sizable effect on medium to long maturity yields; 3) volatility is a curvature factor of the yield curve, and the net effect of a multivariate variance-covariance matrix is also a curvature factor; 4) expected excess returns are explicitly linked to the volatility-covolatility; 5) the model can well explain the bond yield "conundrum" in 2004-2005, where the long term interest rate remains low while short term rate keeps rising continously. This can happen when volatility is sharply declining as is the case at that time which conteracts the effect of rising short rate.

We propose a new class of autoregressive conditional interval (ACI) models for interval-valued time series data. As an appealing measure for variability and uncertainty, intervals are often of direct interest for forecast. An interval-valued observation in a time period contains more information than a point-valued observation in the same time period. Examples of interval data include the maximum and minimum temperatures in a day, the maximum and minimum GDP growth rates in a year, the maximum and minimum stock prices in a trading day, the ask and bid prices in a trading period, and the 90%-tile and 10%-tile incomes of a cohort in a year, etc. Estimation methods, particularly a minimum distance estimation method, are proposed to estimate the parameters of an ACI model. We establish the consistency and asymptotic normality of the proposed estimators and construct some hypothesis tests for model parameters. Simulation studies show that the use of interval time series data can provide more accurate estimation for model parameters than point-valued observations in terms of mean squared error criterion, even when partial information (e.g., the difference or range between the right and left bounds) of an interval is of interest.

In 1978, China began the reform and opening of its market to the international community. It has achieved substantial economic success with social stability for the past 30 years. However, this success has been at the cost of nature's depletion, resource wastage and environmental pollution. The Chinese government has gradually come to understand the full implications of these worsening conditions. The country is going through a steep learning curve in relation to upholding environmental protection. Along with corporate changes, both social responsibilities and public awareness are maturing. specific jurisprudence and frameworks have also been developed due to the pressure from Chinese non-governmental organizations (NGOs) and governmental institutions. China's environmental innovation policy faces two great challenges. First, there is the need to establish a circular economy that encourages environmental innovation. Second, the implementation gap between the central integral planning departments and regional governments needs to be closed with regard to different regional economies, fragmented environmental targets and methodologies. These gaps need to be addressed. Based on the factual conditions of different districts, a proper custom-made environmental model has to be implemented. Effective Chinese environmental innovation requires the development of laws, systems, and policies that promote public participation in environmental activities as well a transparent governmental decision-making process that invites further social participation.

The Fourier transform approach is an important tool in options pricing, and together with the Fast Fourier transform (FFT) algorithms, real time options pricing can be delivered. The underlying asset price processe can allow for more general realistic structure of asset returns, say, excess kurtosis and stochastic volatility. It is known that once the characteristic function of the risk neutral density is known analytically, the analytic expression for the Fourier transform of the option value can be derived. By treating option price analogous to a probability density function, option prices across the whole spectrum of strikes can be obtained via fast Fourier transform. Fourier transform is an effective tool to compute convolution products. We show how this property of the Fourier transform of a convolution product can be used to value various types of option pricing models. In particular, we show how one can price Bermudan style options under Levy processes using FFT techniques in an efficient manner by reformulating the risk neutral valuation formulation as a convolution. By extending the finite state Markov chain approach in option pricing, we illustrate an innovative FFT-based network tree approach for option pricing under Levy process. Similar to the forward shooting grid technique in the usual lattice tree algorithms, the approach can be adapted to valuation of options with exotic path dependence. Limitations of the FFT techniques are also addressed, like the difficulties in dealing with time dependent parameter functions and barrier type models.

This paper estimates a New Keynesian model augmented by a general formed price reset hazard function. The identification of the hazard function is achieved due to the fact that current inflation consists of current and past reset prices, and its composition depends on the price reset hazard function. The dynamic structure of the generalized NKPC links those compositional effects of reset prices on inflation to the hazard function, and hence estimating coefficients of the generalized Phillips curve can be used to infer the price reset hazards from aggregate data. My empirical result shows that a constant hazard function is easily rejected by the data. The empirical hazard function is generally increasing with the age of prices, but with spikes at the 1st and 5th quarters, indicating that a large fraction of firms adjust their prices either frequently or in yearly frequency. Finally, I show that the flexible hazard model generates empirical impulse responses of inflation and price level to a monetary policy shock which are closer to the SVAR evidence than the Calvo responses.

We discissed the issues in forecasting volatility based on three general catagories: time series model; conditional volatility model and hybrid volatility models and try to find the micromechanism for determination of supuority of the volatility forecasts. The financial application for volatility forecasts are examined in option pricing and market direction prediction. And generally, we find hybrid model works best in forecasting. A lot of interesting and useful emprical findings have been documented.

High-dimensional time series which reveal dynamic periodic behavior occur frequently in many different fields of science and are typically analyzed by time propagation of a few number of factors. In this article we address the problem of inference when the factors and factor loadings are estimated by semiparametric methods when additional periodic effects are present, i.e. instead of the VAR process estimation in time, we incorporate basis in time with the group Lasso-type variable selection technique and choose the basis in space based on the functional principle component method. This more flexible modeling approach makes direct interpretations based on the selected basis in time possible and accompanies with much fewer basis in space with the accessible computation which is also lack in the original dynamic semiparametric factor model (DSFM) approach, as introduced by \cite{RePEc:bes:jnlasa:v:104:i:485:y:2009:p:284-298}. We show the upper bounds on the prediction error and the distance between the estimator and the true regression matrix. But it also poses an important question: Is it justified, from an inferential point of view, to base statistical inference on the estimated time series factors? We show that the difference of the inference based on the estimated time series and ``true'' unobserved time series is asymptotically negligible, which allows one to study the dynamics of the whole high-dimensional system with a low dimensional representation. We confirm this theoretical result by a Monte Carlo simulation study. Also, we apply the method to a study of the dynamic behavior of temperatures and implied volatilities and to the analysis of functional magnetic resonance imaging (fMRI) data.

Multiasset equity options (or basket option) are over-the-counter derivatives. There are various types of such products traded on the market either separately or as a part of more sophisticated structures. The buyers of such derivatives benefit from the portfolio effect that makes possible to buy volatility cheaper in comparison with single stock options. However to price and hedge such derivatives one needs to account for correlation structure of a basket. One of the ways, usually used in practice, is to estimate correlation from historical data of stock prices. However, intuitively, for derivatives pricing the forward-looking implied values would be of more use. In this paper we are concerned about modeling implied correlations. This is a challenging task both in terms of computational burden and estimation error. First it cannot be observed directly and must be recovered from option prices. Second, since it is obtained from implied volatilities, it is is not constant over maturities and strikes. We also expect this object to change over time. It means that, we study the sophisticated high-diminutional object, which has been little explored before. To analyze structure and dynamics of implied correlation surfaces we consider the dynamic semiparametric factor model (DSFM), which assumes nonparametric loading functions and low-dimensional time series of factors. Factors and factor loadings are estimated by semiparametric methods. In such way we study the dynamics of the system in it's low-dimensional representation. We make the inference of the whole system based on low-dimensional time series analysis. We procede as follows. First we introduce the way how implied correlation can be backed out from option prices. Then, based on rather restrictive assumption that the correlation between all parts of equities in the index is constant, we approximate the implied correlation by the implied volatilities of stocks in the basket and implied volatility of the basket. Further we relax the equicorrelation assumption and consider the block structure of the implied correlation matrix. In both cases me model the dynamics of implied correlation surfaces with DSFM. Finally we use obtained estimates to price different types of equity basket options and compare the results with prices of actual products traded on the market. The empirical analysis is made on the dynamics of implied correlation structure of the 30 DAX stocks.

The curve time series framework provides a convenient vehicle to accommodate some nonstationary features into a stationary setup. We propose a new method to identify the dimensionality of curve time series based on the dynamical dependence across different curves. The practical implementation of our method boils down to an eigenanalysis of a finite-dimensional matrix. Furthermore, the determination of the dimensionality is equivalent to the identification of the non-zero eigenvalues of the matrix, which we carry out in terms of some bootstrap tests. Asymptotic properties of the proposed method are investigated. In particular, our estimators for zero-eigenvalues enjoy the fast convergence rate n while the estimators for non-zero eigenvalues converge at the standard root-n rate. The proposed methodology is illustrated with both simulated and real data sets.

Using the 2004 China economic census database, this paper examines the impact of information and communication technologies (ICT) on the geographic concentration of manufacturing industries, controlling for other determinants of industrial agglomeration. Contrary to the argument that ICT lead to more dispersion, we found that ICT promote geographic concentration of industries. The results are robust to different measures of ICT and different geographic levels, the inclusion of other determinants of geographic concentration, and consideration of endogeneity.

Is it possible to develop simplified models to gain insights for large and complex biologic networks? This talk will discuss our attempts to develop statistical methods for this purpose that include network reconstruction by Boolean networks, studies of yeast transcription factors and evolution of the yeast protein interaction network. Future developments regarding this direction will be discussed as well.

Penalized spline smoothing has been shown to connect to linear mixed models in the statistical literature. In this paper, we show that local polynomial regression when expressed in a projection framework also has interesting connections to mixed models.

Exploratory data analysis (EDA, Tukey, 1977) has been introduced and extensively used for more than 30 years yet boxplot and scatterplot are still the major EDA tools for visualizing continuous data into the 21st century. All conventional graphical tools have their own limits. For continuous data: Scatterplot Matrix (SM) is useful for visualizing about only twenty variables; Box-Plot (BP) does not provide interactions between variables; Parallel Coordinate Plot (PCP: Inselberg, 1985) requires extensive conditioning for extracting overall information. Dimension reduction tools such as PCA and MDS also lose effectiveness when visualizing very high dimensional data sets. For categorical data: Mosaic Plots are most popular in practice for visualizing no more than ten variables and MCA type of dimension reduction methods may also lose effectiveness for very high dimensional data sets. Generalized association plots (GAP: Chen, 2002; Tien et al., 2008; Wu et al., 2010) related matrix visualization (MV) tools can simultaneously explore the associations of up to thousands of subjects, variables, and their interactions, without reducing dimension for continuous data. For non-continuous data, SM, BP, and PCP basically cannot provide much visual information while GAP related tools still gives us comprehensive information about individual profiles for subjects and variables together with the interaction patterns of each subject-cluster on every variable-group. We believe GAP related MV techniques have great potential to become major data/information visualization tools for next generation EDA. GAP related information can be obtained at: http://gap.stat.sinica.edu.tw/Software/index.htm References. 1. Tukey,J.W.: Exploratory Data Analysis, 1977, Addison-Wesley. 2. Chen,C.H.: Generalized Association Plots: Information Visualization via Iteratively Generated Correlation Matrices, Statistica Sinica, 2002, 12, 7-29. 3. Tien, Y. J., Lee, Y. S., Wu, H. M., and Chen, C. H., (2008), "Methods for Simultaneously Identifying Coherent Local Clusters with Smooth Global Patterns in Gene Expression Profiles." BMC Bioinformatics. 9:155. 4. Wu, H. M., Tien, Y. J., and Chen, C. H*., (2010), "GAP: A graphical environment for matrix visualization and cluster analysis." Computational Statistics and Data Analysis, 54 (3), 767-778.

It is important to model and estimate the covariance of assets innance With the availability of high-frequency intraday data, realized covariance matrices are constructed, based on which statistical methods can be used to model the dynamic of covariance. However the most available methods often encounter the diculties of over-parameterization and factor interpretation when the number of assets involved is large. This paper proposes a new parsimonious dynamic model to make use of prior information and to extract the useful information when the dimension is high. In particular, realized covariance (RCOV) matrices are modeled by constrained factor models. Real and simulated examples are used to demonstrate the proposed analysis.

In this paper, we analyze the properties of nonparametric estimators of a regression function when the covariates are not directly observed, but have only been estimated by some nonparametric procedure. We provide general results that can be used to establish rates of consistency or asymptotic normality in numerous econometric applications, including nonparametric estimation of simultaneous equation models, sample selection models,treatment effect models, and censored regression models.

This talk is an attempt to address the following:

• What are genetic/transcritpional regulatory interactions (GIs/TIs) and networks?
• Why inferring genetic network is one of the frontier areas in computational biology?
• How statistical modeling and computational algorithms pave the way to infer complex molecular mechanism, e.g. biochecmical pathways central to diseases, from genomics data.
• How to use WebPARE, the web-computing service of the PARE algorithm (Bioinformatics, 2008; Bioinformatics, 2010), to infer GIs/TIs using microarray gene expression data.

Applications to infer transcriptional compensation interactions in yeast and transcriptional regulatory interactions involved in kidney disease in mouse will be illustrated.

Many industries are exposed to weather risk which they can transfer on financial markets via weather derivatives. Equilibrium models based on partial market clearing became a useful tool for pricing such kind of financial instruments. Recently a multi-period equilibrium pricing model of weather derivatives was proposed. In this model agents rebalance their portfolio of commodities, weather derivatives and bonds in each period such that they maximize the expected utility of their incomes constituted by possibly weather dependent profits and payoffs of portfolio positions. We extend the model to a multi-period multi-site version and apply it to pricing rainfall derivatives for Chinese provinces. Rainfall occurrences are modeled in a multi-site setting with first order space-time Markov model with spatially and temporally dependent transitional probabilities. To the rainfall amounts we fit a mixture of exponential distributions with seasonal spatially dependent mixing parameters. By simulating realistic market conditions we obtain equilibrium prices for weather derivatives on cumulative average rainfall. In our simulation we assume three different investor types: farmers represent agents with profits highly exposed to climate risk (rainfall amount), industrial investors hold rainfall derivatives to hedge against minor profit fluctuations due to rainfall conditions and pure financial investors seek to diversify their financial portfolio. Dynamic portfolio optimization under market clearing and utility indifference of these representative agents determines equilibrium quantity and price for rainfall derivatives.

In this paper we analyse the properties of hierarchical Archimedean copulas. This class is a generalisation of the Archimedean copulas and allows for general non-exchangeable dependency structures. We show that the structure of the copula can be uniquely recovered from all bivariate margins. We derive the distribution of the copula value, which is particularly useful for tests and constructing confidence intervals. Furthermore, we analyse dependence orderings, multivariate dependence measures and extreme value copulas. Special attention we pay to the tail dependencies and derive several tail dependence indices for general hierarchical Archimedean copulas.

In this paper, we identify return based style factors for Asia‐focused hedge funds represented by the HFRI
Emerging Market‐Asia exclude Japan index. This hedge fund index has a particularly strong exposure to Asian
equity markets and bond indices. Next to linear fixed income and equity factors, we also include
non linear equity factors based on hypothetical option positions, in order to model the often
reported non-linear exposures of hedge funds. The inclusion of these factors results in a marginal increase
in explanatory power for the considered index indicating that small positions in options provide protection 
against extreme losses of the market. Our model provides a high explanatory power for returns 
of the hedge fund index, both for in sample and out of sample periods. We further conduct a Value‐
at‐Risk analysis using the identified return based style factors. We propose a parametric approach 
using Monte Carlo simulation in combination with a multivariateGARCH BEKK model to forecast 
he Value at Risk. Our results suggest that this approach provides an appropriate quantification
of the risk and yields a more plausible estimatin of Value‐at‐Risk than a simple bootstrap method.   

A large body of literature explains the inferior position of unskilled workers by imposing a structural shift in the labor force skill composition. This paper takes a different approach by emphasizing the connection between cyclical variations in skilled and unskilled labor markets. Using a stylized business cycle model with search frictions in the respective sub-markets, I find that imperfect substitution between skilled and unskilled labor creates a channel for the variations in the sub-markets. Together with a skill-neutral or a skill-biased technology shock, it can generate downward sloping Beveridge curves in skill-specific labor markets. Moreover, the skill-biased shock excels in capturing the cyclical properties when the model is calibrated to US data.

This paper studies split-award procurement auctions where a buyer can either divide full production among multiple suppliers or award the entire production to a single supplier. The literature shows that single sourcing usually dominates multiple sourcing. This paper challenges the “winner-takes-all” argument. In a framework of generalized second-price auctions with pre-auction investment, we show that splitting the award improves the suppliers’ investment incentives, intensifies competition at the bidding stage, and minimizes the buyer’s procurement costs. Finally, in an N-supplier setting, using quadratic investment technology, we illustrate that it is optimal for a buyer to restrict the number of suppliers to two.

Information and Panel Discussion