Humboldt-Universität zu Berlin - High Dimensional Nonstationary Time Series

Is Scientific Performance a Function of Funds?

Alona Zharova
Wolfgang K. Härdle
Stefan Lessmann

Abstract:
The management of universities requires data on teaching and research performance.
While teaching quality can be measured via student performance and teacher evaluation
programs, the connection of research outputs and their antecedents is much harder to
check, test and understand. To inform research governance and policy making at
universities, the paper clarifies the relationship between grant money and research
performance. We examine the interdependence structure between third-party expenses (TPE),
publications, citations and academic age. To describe the relationship between these
factors, we analyze individual level data from a sample of professorships from a leading
research university and a Scopus database for the period 2001 to 2015. Using estimates
from a PVARX model, impulse response functions and a forecast error variance
decomposition, we show that an analysis at the university level is inappropriate and
does not reflect the behavior of individual faculties. We explain the differences in
the relationship structure between indicators for social sciences and humanities,
life sciences and mathematical and natural sciences. For instance, for mathematics and
some fields of social sciences and humanities, the influence of TPE on the number of
publications is insignificant, whereas the influence of TPE on the number of citations
is significant and positive. Corresponding results quantify the difference between
the quality and quantity of research outputs, a better understanding of which is
important to design incentive schemes and promotion programs. The paper also proposes
a visualization of the cooperation between faculties and research interdisciplinarity via
the co-authorship structure among publications. We discuss the implications for policy
and decision making and make recommendations for the research management of universities.

Keywords:
causal inference, sample splitting, cross-fitting, sample averaging, machine learning,
simulation study

JEL Classification:
C01, C14, C31, C63

Dynamic Semiparametric Factor Model with a Common Break

Likai Che
Weining Wang
Wei Biao Wu

Abstract:
For change-point analysis of high dimensional time series, we consider a semiparametric model with dynamic structural break factors. The observations are described by a few low dimensional factors with time-invariate loading functions of covariates. The unknown structural break in time models the regime switching effects introduced by exogenous shocks. In particular, the factors are assumed to be nonstationary and follow a Vector Autoregression (VAR) process with a structural break. In addition, to account for the known spatial discrepancies, we introduce discrete loading functions. We study the theoretical properties of the estimates of the loading functions and the factors. Moreover, we provide both the consistency and the asymptotic convergence results for making inference on the common breakpoint in time. The estimation precision is evaluated via a simulation study. Finally we present two empirical illustrations on modeling the dynamics of the minimum wage policy in China and analyzing a limit order book dataset.

Keywords:
high dimensional time series, change-point analysis, temporal and cross-sectional dependence, vector autoregressive process

JEL Classification:
C00

Realized volatility of CO2 futures

Thijs Benschop
Brenda López Cabrera

Abstract:
The EU Emission Trading System (EU ETS) was created to reduce the CO2 and other greenhouse gas emissions at the lowest economic cost. In reality market participants are faced with considerable uncertainty due to price changes and require price and volatility estimates and forecasts for appropriate risk management, asset allocation and volatility trading. Although the simplest approach to estimate volatility is to use the historical standard deviation, realized volatility is a more accurate measure for volatility, since it is based on intraday data. Besides the stylized facts commonly observed in financial time series, we observe long-memory properties in the realized volatility series, which motivates the use of Heterogeneous Autoregressive (HAR) class models. Therefore, we propose to model and forecast the realized volatility of the EU ETS futures with HAR class models. The HAR models outperform benchmark models such as the standard long-memory ARFIMA model in terms of model fit, in-sample and out-of-sample forecasting. The analysis is based on intraday data (May 2007-April 2012) for futures on CO2 certificates for the second EU-ETS trading period (expiry December 2008-2012). The estimation results of the models allow to explain the volatility drivers in the market and volatility structure, according to the Heterogeneous Market Hypothesis as well as the observed asymmetries. We see that both speculators with short investment horizons as well as traders taking long-term hedging positions are active in the market. In a simulation study we test the suitability of the HAR model for option pricing and conclude that the HAR model is capable of mimicking the long-term volatility structure in the futures market and can be used for short-term and long-term option pricing.

Keywords:
EU ETS, Realized Volatility, HAR, Volatility Forecasting, Intraday Data, CO2 Emission Allowances, Emissions Markets, Asymmetry, SHAR, HARQ, MC Simulation

JEL Classification:
C00

Spatial Functional Principal Component Analysis with Applications to
Brain Image Data

Yingxing Li
Chen Huang
Wolfgang K. Härdle

Abstract:
This paper considers a fast and effective algorithm for conducting functional principle component analysis with multivariate factors. Compared with the univariate case, our approach could be more powerful in revealing spatial connections or extracting important features in images. To facilitate fast computation, we connect Singular Value Decomposition with penalized smoothing and avoid estimating a huge dimensional covariance operator. Under regularity assumptions, the results indicate that we may enjoy the optimal convergence rate by employing the smoothness assumption inherent to functional objects. We apply our method on the analysis of brain image data. Our extracted factors provide excellent recovery of the risk related regions of interests in human brain and the estimated loadings are very informative in revealing the individual risk attitude.

Keywords:
Principal Component Analysis, Penalized Smoothing, Asymptotics, functional Magnetic Resonance Imaging (fMRI)

JEL Classification:
C00

Penalized Adaptive Method in Forecasting with Large Information Set and
Structure Change

Xinjue Li
Lenka Zbonakova
Wolfgang Karl Härdle

Abstract:
In the present paper we propose a new method, the Penalized Adaptive Method (PAM), for a data driven detection of structure changes in sparse linear models. The method is able to allocate the longest homogeneous intervals over the data sample and simultaneously choose the most proper variables with help of penalized regression models. The method is simple yet flexible and can be
safely applied in high-dimensional cases with different sources of parameter changes. Comparing with the adaptive method in linear models, its combination with dimension reduction yields a method which selects proper significant variables and detects structure breaks while steadily reduces the forecast error in high-dimensional data. When applying PAM to bond risk premia modelling, the locally selected variables and their estimated coefficient loadings identified in the longest stable subsamples over time align with the true structure changes observed throughout the market.

Keywords:
SCAD penalty, propagation-separation, adaptive window choice, multiplier bootstrap,
bond risk premia

JEL Classification:
C13, C20, E37

Penalized Adaptive Method in Forecasting with Large Information Set and
Structure Change

Xinjue Li
Lenka Zbonakova
Wolfgang Karl Härdle

Abstract:
In the present paper we propose a new method, the Penalized Adaptive Method (PAM), for a data driven detection of structure changes in sparse linear models. The method is able to allocate the longest homogeneous intervals over the data sample and simultaneously choose the most proper variables with help of penalized regression models. The method is simple yet flexible and can be
safely applied in high-dimensional cases with different sources of parameter changes. Comparing with the adaptive method in linear models, its combination with dimension reduction yields a method which selects proper significant variables and detects structure breaks while steadily reduces the forecast error in high-dimensional data. When applying PAM to bond risk premia modelling, the locally selected variables and their estimated coefficient loadings identified in the longest stable subsamples over time align with the true structure changes observed throughout the market.

Keywords:
SCAD penalty, propagation-separation, adaptive window choice, multiplier bootstrap,
bond risk premia

JEL Classification:
C13, C20, E37

The systemic risk of central SIFIs

Cathy Yi-Hsuan Chen
Sergey Nasekin

Abstract:
Systemic risk quantification in the current literature is concentrated on market-based methods such as CoVaR(Adrian and Brunnermeier (2016)). Although it is easily implemented, the interactions among the variables of interest and their joint distribution are less addressed. To quantify systemic risk in a system-wide perspective, we propose a network-based factor copula approach to study systemic risk in a network of systemically important financial institutions (SIFIs). The factor copula model offers a variety of dependencies/tail dependencies conditional on the chosen factor; thus constructing conditional network. Given the network, we identify the most “connected” SIFI as the central SIFI, and demonstrate that its systemic risk exceeds that of non-central SIFIs. Our identification of central SIFIs shows a coincidence with the bucket approach proposed by the Basel Committee on Banking Supervision, but places more emphasis on modeling the interplay among SIFIs in order to generate systemwide quantifications. The network defined by the tail dependence matrix is preferable to that defined by the Pearson correlation matrix since it confirms that the identified central SIFI through it severely impacts the system. This study contributes to quantifying and ranking the systemic importance of SIFIs.

Keywords:
factor copula, network, Value-at-Risk, tail dependence, eigenvector centrality

JEL Classification:
C00, C14, C50, C58

Pricing Green Financial Products

Awdesch Melzer
Wolfgang K. Härdle
Brenda López Cabrera

Abstract:
With increasing wind power penetration more and more volatile and weather dependent energy is fed into the German electricity system. To manage the risk of windless days and transfer revenue risk from wind turbine owners to investors wind power derivatives were introduced. These insurance-like securities (ILS) allow to hedge the risk of unstable wind power production on exchanges like Nasdaq and European Energy Exchange. These products have been priced before using risk neutral pricing techniques. We present a modern and powerful methodology to model weather derivatives with very skewed underlyings incorporating techniques from extreme event modelling to tune seasonal volatility and compare transformed Gaussian and non-Gaussian CARMA(p; q) models. Our results indicate that the transformed Gaussian CARMA(p; q) model is preferred over the non-Gaussian alternative with Lévy increments. Out-of-sample backtesting results show good performance wrt burn analysis employing smooth Market Price of Risk (MPR) estimates based on NASDAQ weekly and monthly German wind power futures prices and German wind power utilisation as underlying. A seasonal MPR of a smile-shape is observed, with positive values in times of high volatility, e.g. winter months, and negative values, in times of low
volatility and production, e.g. in summer months. We conclude that producers pay premiums to
insure stable revenue steams, while investors pay premiums when weather risk is high.

Keywords:
market price of risk, risk premium, renewable energy, wind power futures, stochastic process, expectile, CARMA, jump, Lévy, transform, logit-normal, extreme

JEL Classification:
C00

Racial/Ethnic Differences in Non-Work at Work

Daniel S. Hamermesh
Katie R. Genadek
Michael C. Burda

Abstract:
Evidence from the American Time Use Survey 2003-12 suggests the existence of small but
statistically significant racial/ethnic differences in time spent not working at the
workplace. Minorities, especially men, spend a greater fraction of their workdays not
working than do white non-Hispanics. These differences are robust to the inclusion of
large numbers of demographic, industry, occupation, time and geographic controls. They
do not vary by union status, public-private sector attachment, pay method or age; nor do they arise from the effects of equal-employment enforcement or geographic differences in
racial/ethnic representation. The findings imply that measures of the adjusted wage
disadvantages of minority employees are overstated by about 10 percent.

Keywords:
time use, wage discrimination, wage differentials

JEL Classification:
J22, J15, J31

Investing with cryptocurrencies - A liquidity constrained investment approach

Simon Trimborn
Mingyang Li
Wolfgang Karl Härdle

Abstract:
Cryptocurrencies have left the dark side of the finance universe and become an object of study for asset and portfolio management. Since they have a low liquidity compared to traditional assets, one needs to take into account liquidity issues when one puts them into the same portfolio. We propose use a Liquidity Bounded Risk-return Optimization (LIBRO) approach, which is a combination of the Markowitz framework under the liquidity constraints. The results show that cryptocurrencies add value to a portfolio and the optimization approach is even able to increase the return of a portfolio and lower the volatility risk.

Keywords:
crypto-currency, CRIX, portfolio investment, asset classes, blockchain

JEL Classification:
C01, C58, G11

Adaptive weights clustering of research papers

Larisa Adamyan
Kirill Efimov
Cathy Yi-Hsuan Chen
Wolfgang K. Härdle

Abstract:
The JEL classification system is a standard way of assigning key topics to economic articles in order to make them more easily retrievable in the bulk of nowadays massive literature. Usually the JEL (Journal of Economic Literature) is picked by the author(s) bearing the risk of suboptimal assignment. Using the database of a Collaborative Research Center from Humboldt-Universität zu Berlin and Xiamen University, China we employ a new adaptive clustering technique to identify interpretable JEL (sub)clusters. The proposed Adaptive Weights Clustering (AWC) is available on www.quantlet.de and is based on the idea of locally weighting each point (document, abstract) in terms of cluster membership. Comparison with k-means or CLUTO reveals excellent performance of AWC.

Keywords:
Clustering, JEL system, Adaptive algorithm, Economic articles, Nonparametric

JEL Classification:
C00

Industry Interdependency Dynamics in a Network Context

Ya Qian
Wolfgang Karl Härdle
Cathy Yi-Hsuan Chen

Abstract:
This paper contributes to model the industry interconnecting structure in a network context. General predictive model (Rapach et al. 2016) is extended to quantile LASSO regression so as to incorporate tail risks in the construction of industry
interdependency networks. Empirical results show a denser network with heterogeneous central industries in tail cases. Network dynamics demonstrate the variety of
interdependency across time. Lower tail interdependency structure gives the most accurate out-of-sample forecast of portfolio returns and network centrality-based trading strategies seem to outperform market portfolios, leading to the possible ’too central to fail’ argument.

Keywords:
dynamic network, interdependency, general predictive model, quantile LASSO, connectedness, centrality, prediction accuracy, network-based trading strategy

JEL Classification:
C32, C55, C58, G11, G17

The Economics of German Unification after Twenty-five Years: Lessons for Korea

Michael C. Burda
Mark Weder

Abstract:
This paper reviews the performance of the East German economy in the turbulent quarter-century following reunification and draws some conclusions for the reunification of North and South Korea. In this period, the gap in output per capita between East and West Germany declined at a speed not far from empirical estimates of the
neoclassical growth model, yet systematic total factor productivity
differentials persist despite identical institutional frameworks and
significant investment in the eastern regions. At the same time, regional
disparities in income, well-being, and health are little different from those found within West Germany, and net migration has ceased. On this human metric, German unification has been an unqualified success. For Korea, an effort of this dimension will be costly. A back-of-the- envelope calculation suggests that Korean unification will cost roughly twice as much as its German counterpart.

Keywords:


JEL Classification:
P2, O11, E02

Testing Missing at Random using Instrumental Variables

Christoph Breunig

Abstract:
This paper proposes a test for missing at random (MAR). The MAR assumption is shown to be testable given instrumental variables which are independent of response given potential outcomes. A nonparametric testing procedure based on integrated squared distance is proposed. The statistic’s asymptotic distribution under the MAR hypothesis is derived. In particular, our results can be applied to testing missing completely at random (MCAR). A Monte Carlo study examines finite sample performance of our test statistic. An empirical illustration analyzes the nonresponse mechanism in labor income questions.

Keywords:
Incomplete data, missing-data mechanism, selection model, nonparametric hypothesis testing, consistent testing, instrumental variable, series estimation

JEL Classification:

Tail event driven networks of SIFIs

Cathy Yi-Hsuan Chen
Wolfgang Karl Härdle
Yarema Okhrin

Abstract:
The interdependence, dynamics and riskiness of financial institutions are the key features frequently tackled in financial econometrics. We propose a Tail Event driven Network Quantile Regression (TENQR) model which addresses these three aspects. More precisely, our framework captures the risk propagation and dynamics in terms of a quantile (or expectile) autoregression involving network effects quantified through an adjacency matrix. To reflect the nature and risk content of systemic risk, the construction of the adjacency matrix is suggested to include tail event covariates. The model is evaluated using the SIFIs (systemically important financial institutions) identified by the Financial Stability Board (FSB) as main players in the global financial system. The risk decomposition analysis of it identifies the systemic importance of SIFIs and thus provides measures for the required level of additional loss absorbency. It is discovered that the network effect, as a function of the tail probability, becomes more profound in stress situations and brings the various impacts to the SIFIs located in different geographic regions.

Keywords:
systemic risk, network analysis, network autoregression, tail event

JEL Classification:
C01, C14, C58, C45, G01, G15, G31

An AI approach to measuring financial risk

Lining Yu
Wolfgang Karl Härdle
Lukas Borke
Thijs Benschop

Abstract:
AI artificial intelligence brings about new quantitative techniques to assess the state of an economy. Here we describe a new measure for systemic risk: the Financial Risk Meter (FRM). This measure is based on the penalization parameter (lambda) of a linear quantile lasso regression. The FRM is calculated by taking the average of the penalization parameters over the 100 largest US publicly traded financial institutions. We demonstrate the suitability of this AI based risk measure by comparing the proposed FRM to other measures for systemic risk, such as VIX, SRISK and Google Trends. We find that mutual Granger causality exists between the FRM and these measures, which indicates the validity of the FRM as a systemic risk measure. The implementation of this project is carried out using parallel computing, the codes are published on www.quantlet.de with keyword FRM. The R package RiskAnalytics is another tool with the purpose of integrating and facilitating the research, calculation and analysis methods around the FRM project. The visualization and the up-to-date FRM can be found on hu.berlin/frm.

Keywords:
Systemic Risk, Quantile Regression, Value at Risk, Lasso, Parallel Computing, Financial Risk Meter

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

Dynamic credit default swaps curves in a network topology

Xiu Xu
Cathy Yi-Hsuan Chen
Wolfgang Karl Härdle

Abstract:
Systemically important banks are connected and have dynamic dependencies of their default probabilities. An extraction of default factors from cross-sectional credit default swaps (CDS) curves allows to analyze the shape and the dynamics of the default probabilities. Extending the Dynamic Nelson Siegel (DNS) model, we propose a network DNS model to analyze the interconnectedness of default factors in a dynamic fashion, and forecast the CDS curves. The extracted level factors representing long-term default risk demonstrate 85.5% total connectedness, while the slope and the curvature factors document 79.72% and 62.94% total connectedness for the short-term and middle-term default risk, respectively. The issues of default spillover and systemic risk should be weighted for the market participants with longer credit exposures, and for regulators with a mission to stabilize financial markets. The US banks contribute more to the long-run default spillover before 2012, whereas the European banks are major default transmitters during and after the European debt crisis either in the long-run or short-run. The outperformance of the network DNS model indicates that the prediction on CDS curve requires network information.

Keywords:
CDS, network, default risk, variance decomposition, risk management

JEL Classification:
C32, C51, G17

Multivariate Factorisable Sparse Asymmetric Least Squares Regression

Shih-Kang Chao
Wolfgang K. Härdle
Chen Huang

Abstract:
More and more data are observed in form of curves. Numerous applications in finance, neuroeconomics, demographics and also weather and climate analysis make it necessary to extract common patterns and prompt joint modelling of individual curve variation. Focus of such joint variation analysis has been on fluctuations around a mean curve, a statistical task that can be solved via functional PCA. In a variety of questions concerning the above applications one is more interested in the tail asking therefore for tail event curves (TEC) studies. With increasing dimension of curves and complexity of the covariates though one faces numerical problems and has to look into sparsity related issues. Here the idea of Factorisable Sparse Tail Event Curves (FASTEC) via multivariate asymmetric least squares regression (expectile regression) in a high-dimensional framework is proposed. Expectile regression captures the tail moments globally and the smooth loss function improves the convergence rate in the iterative estimation algorithm compared with quantile regression. The necessary penalization is done via the nuclear norm. Finite sample oracle properties of the estimator associated with asymmetric squared error loss and nuclear norm regularizer are studied formally in this paper. As an empirical illustration, the FASTEC technique is applied on fMRI data to see if individual’s risk perception can be recovered by brain activities. Results show that factor loadings over different tail levels can be employed to predict individual’s risk attitudes.

Keywords:
high-dimensionalM-estimator, nuclear norm regularizer, factorization, expectile regression, fMRI, risk perception, multivariate functional data

JEL Classification:
C38, C55, C61, C91, D87

Dynamic Topic Modelling for Cryptocurrency Community Forums

Marco Linto
Ernie Gin Swee Teo
Elisabeth Bommes
Cathy Yi-Hsuan Chen
Wolfgang K. Härdle

Abstract:
Cryptocurrencies are more and more used in official cash flows and exchange of goods. Bitcoin and the underlying blockchain technology have been looked at by big companies that are adopting and investing in this technology. The CRIX Index of cryptocurrencies hu.berlin/CRIX indicates a wider acceptance of cryptos. One reason for its prosperity certainly being a security aspect, since the underlying network of cryptos is decentralized. It is also unregulated and highly volatile, making the risk assessment at any given moment dicult. In message boards one finds a huge source of information in the form of unstructured text written by e.g. Bitcoin developers and investors. We collect from a popular crypto currency message board texts, user information and associated time stamps. We then provide an indicator for fraudulent schemes. This indicator is constructed using dynamic topic modelling, text mining and unsupervised machine learning. We study how opinions and the evolution of topics are connected with big events in the cryptocurrency universe. Furthermore, the predictive power of these techniques are investigated, comparing the results to known events in the cryptocurrency space. We also test hypothesis of self-fulling prophecies and herding behaviour using the results.

Keywords:
Dynamic Topic Modelling, Cryptocurrencies, Financial Risk

JEL Classification:
C19, G09, G10

Network Quantile Autoregression

Xuening Zhu
Weining Wang
Hangsheng Wang
Wolfgang K. Härdle

Abstract:
It is a challenging task to understand the complex dependency structures in an ultra-high dimensional network, especially when one concentrates on the tail dependency. To tackle this problem, we consider a network quantile autoregression
model (NQAR) to characterize the dynamic quantile behavior in a complex system. In particular, we relate responses to its connected nodes and node specific
characteristics in a quantile autoregression process. A minimum contrast estimation approach for the NQAR model is introduced, and the asymptotic properties are studied. Finally, we demonstrate the usage of our model by
investigating the financial contagions in the Chinese stock market accounting
for shared ownership of companies.

Keywords:
Social Network, Quantile Regression, Autoregression, Systemic Risk,
Financial Contagion, Shared Ownership

JEL Classification:
C12, C22