Humboldt-Universität zu Berlin - High Dimensional Nonstationary Time Series

TENET: Tail-Event driven NETwork risk

Wolfgang Karl Härdle
Weining Wang
Lining Yu

Abstract:
A system of risk factors necessarily involves systemic risk. The analysis of systemic risk is in the focus of recent econometric analysis and uses tail event and network based techniques. Here we bring tail event and network dynamics together into one context. In order to pursue such joint effects, we propose a semiparametric measure to estimate systemic interconnectedness across financial institutions based on tail-driven spillover effects in a high dimensional framework. The systemically important institutions are identified conditional on their interconnectedness structure. Methodologically, a variable selection technique in a time series setting is applied in the context of a single-index model for a generalized quantile regression framework. We could thus include more financial institutions into the analysis to measure their tail event interdependencies and, at the same time, being sensitive to non-linear relationships between them. Network analysis, its behaviour and dynamics, allows us to characterize the role of each industry group in the U. S. financial market 2007 - 2012. The proposed TENET - Tail Event driven NETwork technique allows us to rank the systemic risk contributions of publicly traded U.S. financial institutions.

Keywords:
Systemic Risk, Systemic Risk Network, Generalized Quantile, Quantile Single-Index
Regression, Value at Risk, CoVaR, Lasso

JEL Classification:
G01, G18, G32, G38, C21, C51, C63

Volatility Modelling of CO2 Emission Allowance Spot Prices with Regime-Switching GARCH Models

Thijs Benschop
Brenda López Cabrera

Abstract:
We analyse the short-term spot price of European Union Allowances (EUAs), which is of
particular importance in the transition of energy markets and for the development of new
risk management strategies. Due to the characteristics of the price process, such as
volatility persistence, breaks in the volatility process and heavy-tailed distributions,
we investigate the use of Markov switching GARCH (MS-GARCH) models on daily spot market
data from the second trading period of the EU ETS. Emphasis is given to short-term forecasting of prices and volatility. We find that MS-GARCH models distinguish well between
two states and that the volatility processes in the states are clearly different. This
finding can be explained by the EU ETS design. Our results support the use of MS-GARCH
models for risk management, especially because their forecasting ability is better than
other Markov switching or simple GARCH models.

Keywords:
CO2 Emission Allowances, CO2 Emission Trading, Spot Price Modelling, Markov Switching GARCH Models, Volatility Forecasting

JEL Classification:
C53, G17, Q49, Q53, Q59

Adaptive Order Flow Forecasting with Multiplicative Error Models

Wolfgang K. Härdle
Andrija Mihoci
Christopher Hian-Ann Ting

Abstract:
A flexible statistical approach for the analysis of time-varying dynamics of transaction data on financial markets is here applied to intra-day trading strategies. A local adaptive technique is used to successfully predict financial time series, i.e., the buyer and the seller-initiated trading volumes and the order flow dynamics. Analysing order flow series and its information content of mini Nikkei 225 index futures traded at the Osaka Securities Exchange in 2012 and 2013, a data-driven optimal length of local windows up to approximately 1-2 hours is reasonable to capture parameter variations and is suitable for short-term prediction. Our proposed trading strategies achieve statistical arbitrage opportunities and are therefore beneficial for quantitative finance practice.

Keywords:
multiplicative error models, trading volume, order flow, forecasting

JEL Classification:
C41, C51, C53, G12, G17

TEDAS - Tail Event Driven ASset Allocation

Wolfgang Karl Härdle
Sergey Nasekin
David Lee Kuo Chuen
Phoon Kok Fai

Abstract:
Portfolio selection and risk management are very actively studied topics in quantitative finance and applied statistics. They are closely related to the dependency structure of portfolio assets or risk factors. The correlation structure across assets and opposite tail movements are essential to the asset allocation problem, since they determine the level of risk in a position. Correlation alone is not informative on the distributional details of the assets. By introducing TEDAS -Tail Event Driven ASset allocation, one studies the dependence between assets at different quantiles. In a hedging exercise, TEDAS uses adaptive Lasso based quantile regression in order to determine an active set of negative non-zero coefficients. Based on these active risk factors, an adjustment for intertemporal correlation is made. Finally, the asset allocation weights are determined via a Cornish-Fisher Value-at-Risk optimization. TEDAS is studied in simulation and a practical utility-based example using hedge fund indices.

Keywords:
portfolio optimization, asset allocation, adaptive lasso, quantile regression, value-at-risk

JEL Classification:
C00, C14, C50, C58

Forecasting Generalized Quantiles of Electricity Demand: A Functional Data Approach

Brenda López Cabrera
Franziska Schulz

Abstract:
Electricity load forecasts are an integral part of many decision-making processes
in the electricity market. However, most literature on electricity load forecasting concentrates on deterministic forecasts, neglecting possibly important
information about uncertainty. A more complete picture of future demand can be obtained by using distributional forecasts, allowing for a more efficient decision-making. A predictive density can be fully characterized by tail measures
such as quantiles and expectiles. Furthermore, interest often lies in the accurate estimation of tail events rather than in the mean or median. We propose
a new methodology to obtain probabilistic forecasts of electricity load, that is based on functional data analysis of generalized quantile curves. The core of the methodology is dimension reduction based on functional principal components of tail curves with dependence structure. The approach has several
advantages, such as flexible inclusion of explanatory variables including meteorological forecasts and no distributional assumptions. The methodology
is applied to load data from a transmission system operator (TSO) and a balancing unit in Germany. Our forecast method is evaluated against other models including the TSO forecast model. It outperforms them in terms of mean absolute percentage error (MAPE) and achieves a MAPE of 2:7% for the TSO.

Keywords:
Electricity, Load forecasting, FPCA

JEL Classification:
G19, G29, G22, Q14, Q49, Q59

Dynamic semi-parametric factor model for functional expectiles

Petra Burdejová
Wolfgang K. Härdle

Abstract:
High-frequency data can provide us with a quantity of information
for forecasting, help to calculate and prevent the future risk based on extremes. This tail behaviour is very often driven by exogenous
components and may be modelled conditional on other variables.
However, many of these phenomena are observed over time, exhibiting non-trivial dynamics and dependencies. We propose a functional
dynamic factor model to study the dynamics of expectile curves. The complexity of the model and the number of dependent variables are reduced by lasso penalization. The functional factors serve as a low-dimensional representation of the conditional tail event, while the time-variation is captured by factor loadings. We illustrate the model with an application to climatology, where daily data over years on temperature, rainfalls or strength of wind are available.

Keywords:
factor model, functional data, expectiles, extremes

JEL Classification:
C14, C38, C55, C61, Q54

Data Science & Digital Society

Cathy Yi-Hsuan Chen
Wolfgang Karl Härdle

Abstract:
Data Science looks at raw numbers and informational objects created by different disciplines. The Digital Society creates information and numbers from many scientific disciplines. The amassment of data though makes is hard to Hind structures and requires a skill full analysis of this massive raw material. The thoughts presented here on DS2 - Data Science & Digital Society analyze these challenges and offers ways to handle the questions arising in this evolving context. We propose three levels of analysis and lay out how one can react to the challenges that come about. Concrete examples concern Credit default swaps, Dynamic Topic modeling, Crypto currencies and above all the quantitative analysis of real data in a DS2 context.

Keywords:
Data Science, Digital Society, social networks, herding, sentiments

JEL Classification:

The Cross-Section of Crypto-Currencies as Financial Assets: An Overview

Brenda López Cabrera
Franziska Schulz

Abstract:
Crypto-currencies have developed a vibrant market since bitcoin, the first crypto-currency, was created in 2009. We look at the properties of cryptocurrencies as financial assets in a broad cross-section. We discuss approaches of altcoins to generate value and their trading and information platforms. Then we investigate crypto-currencies as alternative investment assets, studying their returns and the co-movements of altcoin prices with bitcoin and against each other. We evaluate their addition to investors' portfolios and document they are indeed able to enhance the diversification of portfolios due to their little co-movements with established assets, as well as with each other. Furthermore, we evaluate pure portfolios of crypto-currencies: an equally weighted one, a value-weighted one, and one based on the CRypto-currency IndeX (CRIX). The CRIX portfolio displays lower risk than any individual of the liquid crypto-currencies. We also document the changing characteristics of the crypto-currency market. Deepening liquidity is accompanied by a rise in market value, and a growing number of altcoins is contributing larger amounts to aggregate crypto-currency market capitalization.

Keywords:
Crypto-currenciesA, ltcoins, Investment assets, Bitcoin, Blockchain, Alternative
investments, Financial risk and return

JEL Classification:
G11, G15, F31

Time-Adaptive Probabilistic Forecasts of Electricity Spot Prices with
Application to Risk Management

Brenda López Cabrera
Franziska Schulz

Abstract:
The increasing exposure to renewable energy has amplied the need for risk management in electricity markets. Electricity price risk poses a major challenge to market participants. We propose an approach to model and forecast
electricity prices taking into account information on renewable energy production. While most literature focuses on point forecasting, our methodology
forecasts the whole distribution of electricity prices and incorporates spike risk, which is of great value for risk management. It is based on
functional principal component analysis and time-adaptive nonparametric density estimation techniques. The methodology is applied to electricity market data from Germany. We find that renewable infeed effects both, the location and the shape of spot price densities. A comparison with benchmark methods and an application to risk management are provided.

Keywords:
electricity prices; residual load, probabilistic forecasting, value at risk,
expected shortfall, functional data analysis

JEL Classification:
C1, Q41, Q47

Estimation of NAIRU with Inflation Expectation Data

Wei Cui
Wolfgang K. Härdle
Weining Wang

Abstract:
Estimating natural rate of unemployment (NAIRU) is important for understanding the joint dynamics of unemployment, inflation, and inflation expectation. However,
existing literature falls short of endogenizing inflation expectation together with NAIRU in a model consistent way. We estimate a structural model with forward and backward looking Phillips curve. Inflation expectation is treated as a function of state variables and we use survey data as its noisy observations. Surprisingly, we find that the estimated NAIRU tracks unemployment rate closely, except for the high inflation period (late 1970s). Compared to the estimation without using the survey data, the estimated Bayesian credible sets are narrower and our model leads to better inflation and unemployment
forecasts. These results suggest that monetary policy was very effective and there was not much room for policy improvement.

Keywords:
NAIRU, New Keynesian Phillips Curve, Inflation Expectation

JEL Classification:
C32, E31, E32