Abstracts

How to measure lead-lag relationships from high frequency data ?
Frédéric Abergel
lead-lag relationships are an important stylized fact of high frequency financial data, especially in the equity markets. They are related to other important measure of microstructure such as volatility, market depth, turnover... in this talk I will present some recent results, in particular introducing a new indicator, the lead-lag ratio (LLR). This new statistical indicator provides a natural measure of lead-lag relationships, it is based on the assymetry of the cross-correlogram of two financial assets (based on a joint work with N. Huth).

Correlation and Interdependencies in coupled financial networks
Hideaki Aoyama
We explore the relationships of coupled financial networks. We study the foreign exchange and stock market networks for 48 countries from 1999 until 2012. We examine the influences of one network on the dynamics of the other and propose a model, based on complex principal component analysis, for extracting signifi cant relations between these networks. We identify significant component (node) behavior within and between the two networks that cannot be detected by concurrent simple correlation analyses or equal-time principal component analysis.

The Asian Economic Observatory Network (AEON) Proposal on Data-Driven Agent-Based Modeling of the Asian Economies Slides
Siew Ann CHEONG
Because of economic downturns in the US and Europe, hot money has been flowing into Asia since 2007. As a result, Asia as a region is experiencing an unprecedented property bubble that dwarfs the one that precipitated the US Subprime Crisis. Although there are signs that the Korean property bubble may have deflated in 2011 and 2012, the situations in China, Hong Kong, Singapore, Taiwan, and to a less extent India, remain dire. Cooling measures have been imposed in China, Hong Kong, and Singapore, but they have had limited impact on the housing bubbles in these countries. In the mean time, new property bubbles are growing in Indonesia, Malaysia, and the Philippines. By all sensible econometric measures, the bubbles have grown so big that housing in these countries are no longer affordable to large segments of the populations. Some kind of downward correction is expected over the next one to two years, and governments are trying their bests to engineer soft landings. It may just require nothing short of a divine intervention to stave off a regional meltdown. In this talk, I will describe an ongoing effort to develop a large econophysics grant proposal, where economists and econophysicists from around Asia will set up a network of collaborative teams. Our common goal is to have this network of economic observatories in place before the regional housing markets go south, so that we can collect macroeconomic, microeconomic, and financial market data before, during, and after the property bubbles burst. By sharing data and methods, and using the insights gathered to build exploratory and explanatory agent-based models, we hope to understand the main structural problems in the Asian economies that are exposed by the property bubbles. Thereafter, we aim to develop scenarios for different mitigation and intervention strategies targeted at these problems, to arrive at customized economic policies that could be implemented across the region in anticipation of future economic crises.

On the nature of inequalities from socio-economic data Slides
Arnab Chatterjee
I will talk about the nature of inequalities found in socio-economic data, ranging from financial markets, income and wealth, other social indicators, electoral data, crime, academic indicators like citations etc.

Inequality in Societies, Academic Institutions and Science Journals: Gini and k-indices Slides
Asim Ghosh
Social inequality is traditionally measured by the Gini-index (g). Most of the estimates of the income or wealth data indicate the g value to be widely dispersed across the countries of the world: \textit{g} values typically range from 0.30 to 0.65 at a particular time (year). We estimated similarly the Gini-index for the citations earned by the yearly publications of various academic institutions and the science journals. The ISI web of science data suggests remarkably strong inequality and universality (g=0.70±0.07) across all the universities and institutions of the world, while for the journals we find g=0.65±0.15 for any typical year. We define a new inequality measure, namely the k-index, saying that the cumulative income or citations of (1−k) fraction of people or papers exceed those earned by the fraction (k) of the people or publications respectively. We find, while the k-index value for income ranges from 0.60 to 0.75 for income distributions across the world, it has a value around 0.75±0.05 for different universities and institutions across the world and around 0.77±0.10 for the science journals.

Relation between Total Factor Productivity and Patents of Firms Slides
Shouji Fujimoto
It is one of important issues to understand what determines the productivity of firms. In economics, generally, production is described as a function of capital and labor. This is called production function. The total factor productivity is introduced as the residual of the production function that can not be explained by capital and labor. The total factor productivity is probably determined by various factors. In the available database, it is thought that patents owned by a firm related with the total factor productivity of the firm. In this study, we discuss relation between the total factor productivity and the patents of firms. We analyse the world wide firms dataset ORBIS supplied by Bureau van Dijk. We adopt sales, tangible fixed assets and number of employee as product, capital and labor, respectively. As a result, it is confirmed that sales correlate with number of patent applications even after removing the dependence of tangible fixed assets and number of employee.

Financial Distress Propagation in Japanese Credit Network Slides
Yoshi Fujiwara
We present an analysis of the lending/borrowing relationship between Japanese banks and Japanese firms, which form a bipartite credit network. We introduce distress to some initial nodes (banks or firms) and allow it to propagate and contaminate other nodes in this network according to the relative exposure. First, by choosing the initial node to be a bank and taking the weighted average of the resulting distress distribution, with the weight proportional to the size (total assets) of each node, we identify the bank's importance to the whole network at the time of crisis. This is a straightforward extension of what is called DebtRank, invented by Battiston et al.(2012), to a bipartite network. By introducing the initial distress to firms in certain industrial sector(s), we evaluate the vulnerability of banks and firms in other sectors due to the distress in the initial sectors. This work is based on Aoyama, Battiston and Fujiwara (2013).

Agent Based modelling of Housing Asset Bubble: a habit or subsistence utility function based investigation Slides
Kausik Gangopadhyay and Kousik Guhathakurta
The housing asset bubble and mortgage crisis of 2007-08 in the US market poses a challenge to understanding of market and hypotheses related to market efficiency. In Gangopdhyay and Guhathakurta (2013) we explored the literature on agent based models and presented a felicity function based framework which displayed the power of irrational expectation in bringing about an artificial and unintended boost in demand for investment of housing asset. In this paper we extend our model to understand the role of interest rate and credit facility in the form of leverage in the housing asset bubble and corresponding crash. Investor chooses between savings and Housing asset. The decision making process is not affected by increasing the choice set of assets since essentially we are modelling the changes in investor behaviour towards housing asset only which is independent of the other asset selection decision. We modify the modelling of consumer with bounded rationality from the previous case by introducing random income shocks. Interest plays a role in investor behaviour. The ability to invest in housing asset increases with low interest rates as leverage plays a role. Since there is asymmetry in income, the low income group can switch to housing assets only when interest rates are low, the model incorporates the repayment of loans. When the interest rates are raised, the default starts. This behaviour is examined by introducing a subsistence or habit model as the utility function.

Nonlinear dynamics of stock markets during critical periods Slides
Kousik Guhathakurta
Stock market crashes have always been a subject of intimate study in financial economics literature. Starting from (Fisher, 1930) to Lauterbauch et al. (2012), the reasons, nature and impact of stock market crashes have been analysed in a various ways by various authors ranging from econometric to behavioural and physics based models to explain the phenomena. Mostly, these recent works have shown an analogy between crashes and phase transition. Using a technique evolved from nonlinear dynamics and physics, it is possible to graphically represent the dynamic evolution of a system. This technique known as Recurrence Plot (RP) can detect critical phases in the system and changes in the same. Inspired by this several authors used this technique to try and detect bubbles and crashes including Guhathakurta et al. (2010). The present work extends the findings of the same work. Using the recurrence statistics, we show that it is possible to detect critical periods in advance for all the cases where there was a known bubble building up in the market. RP alone can not predict cashes but definitely, this tool may be used to identify changes in market dynamics and can serve as a warning bell.

Probabilistic flows of inhabitants in urban areas and self-organization in housing markets Slides
Jun-ichi Inoue
We propose a very simple probabilistic model to explain the spatial structure of the rent distribution of housing market in city of Sapporo. Here we modify the mathematical model proposed by Gauvin et. al. In their mathematical modeling, they utilized several assumptions to describe the decision making of each inhabitant in Paris. Namely, they assumed that the intrinsic attractiveness of a city depends on the location and there exists a single peak at the center. They also used the assumption that each inhabitant tends to choose the place where the other inhabitants having the similar or superior income to himself/herself are living. In order to find the best possible (desirable) place to live, each buyer in the system moves from one place to the other according to the transition (aggregation) probability described by the above two assumption, and he/she makes a deal with the seller who presents the best possible condition for the buyer. They concluded that the resultant self-organized rent distribution is consistent with the corresponding empirical evidence in Paris. However, it is hard for us to apply their model directly to the other cities having plural centers (not only a single center as in Paris). Hence, here we shall modify the Gauvin's model to include the much more detail structure of the attractiveness by taking into account the empirical data concerning the housing situation in city of Sapporo. We also consider the competition between two distances, namely, the distance between house and center, and the distance between house and office. Computer simulations are carried out to reveal the self-organized spatial structure appearing in the rent distribution. We also compare the resulting distribution with empirical rent distribution in Sapporo as an example of cities designated by ordinance.

Statistically significant fits of Hawkes processes to FX data
Mehdi Lallouache
Most studies fitting Hawkes processes to financial data do not obtain statistically significant results. Significance on the activity of EBS limit order books cannot be achieved on mid-price changes. We argue that this is due to the limited time resolution (0.1s) of such data that prevents many events to be detected. This limitation can be self-consistently lifted for transactions on the bid and ask sides by using the volumes of transactions on both sides during a time slice. Fitting one hour of activity yields fits that pass Kolmogorov-Smirnov tests provided that at least two exponentials are used; all hours of the day have an endogeneity factor of about 0.7. We are able to fit accurately single days by accounting for the intra-day variability of activity, which however suggest larger endogeneity factor (0.8). We therefore argue that seasonalities and limited time resolution are major obstacles when fitting Hawkes processes to financial market data.

Financial market reactions to exogenous shocks Slides
Takayuki Mizuno
We mine raw texts of more than 24 million news records provided by Thompson Reuters and examine their impact on market activity in stocks of the 205 firms listed in the S&P 500 US stock index for each of which there were more than 5,000 news records over the period from January 2003 to November 2013. I also use tick data for US stock market and focus on "volatility", "number of transaction" and "trading volume" in order to measure market activity. In this presentation, we first define novelty and topicality of news using techniques of NLP (Natural language processing). Second, we investigate relationships between change of market activity and news impact. Change of market activities (volatility, number of transaction, trade volume) depend on size of novelty and topicality of news. When news has high novelty, market activities respond immediately after the news. Relationship between the response size of market activities and topicality of the news follows a nonlinear function. News with low topicality hardly increases market activities. In this conference, we will also introduce about definition of credibility of news, and will show you characteristics of news article which affect market activities and sign of price change.

Testing Contagion in Financial Time Series Slides
Rituparna Sen
Financial contagion indicates a process through which transmission of shock originating in the financial market of one economy spreads to others. Although the study of causes and prevention of contagion is popularized by economists, very few quantitative studies exist on detection of contagion. This paper provides a new idea of Residual and Recurrence Times (RRT) of high or low values for multivariate time series to detect contagion. In presence of financial contagion, the distributions of residual and recurrence times are not the same. We examine the equality of two distributions using the permutation test. In comparison to other methods in multivariate extreme value theory, our proposed method does not need the i.i.d. assumption. We derive asymptotic results under the GARCH model. Our method can handle the situation where the extremes for different components do not occur at the same time. We justify our methods in two ways: first using thorough simulation studies and then applying the proposed method to real data on weekly stock indices from seventeen markets.

Newton's Revenge: the Inverse Square Law of Price Fluctuations in the Bitcoin Market
Sitabhra Sinha
The recent advent of digital currencies such as Bitcoin has been suggested by many as a revolutionary transition to a new phase of economic life. To the physicist however, the trading in such virtual crypto-currencies provide a rare opportunity to study in detail the dynamics of financial markets in a relatively simple setting, providing an ideal testing ground for the various theories of market dynamics proposed so far. An important feature of the Bitcoin market is that, unlike equities or foreign exchange markets, the value of the traded asset is decided purely by the collective perception of agents trading in the market - thus, governed by a completely endogenous process - and is not tied to any underlying quantities external to the market (such as the value of the company for equities or the state of an economy for foreign exchange). We investigate the empirical features of the Bitcoin dynamics by an exhaustive investigation of high-frequency trading data over the 2-year period 2011-2013. We observe that the fluctuations in Bitcoin price - as measured by log return over intervals measured in ``tick time" - appear to be described by a cumulative distribution having a power-law tail. This tail is described well by an exponent of -2 suggesting an ``inverse square law" of finance, which is distinct from the ``inverse cubic law'' reported earlier in equities and foreign exchange markets. We theoretically explore plausible reasons for this apparent departure from universality of the Bitcoin market dynamics using a modified version of a mean-field theory of market activity proposed earlier by us [1] that reproduces all the known stylized facts of financial markets. Reference: [1] S V Vikram and S Sinha, "Emergence of universal scaling in financial markets from mean-field dynamics", Phys Rev E 83, 016101 (2011)

Wealth Exchange Models with preference in interaction Slides
Sanchari Goswami
A set of conservative models is studied in which agents exchange wealth with a preference in the choice of interacting agents in several ways. The choice depends on the difference of their status, a common feature in all the models. The choice can also depend on past interactions and other factors. Wealth distribution, network properties, activity are the main quantities which have been studied. Evidence of phase transition and other interesting features are presented. The results show that certain observations of real economic system can be reproduced by the models.

NETWORK ANALYSIS OF CRISIS IN THE GLOBAL FINANCIAL NETWORK Slides
*Sunil Kumar* and Nivedita Deo
The study of financial crisis and finding the organizational changes in clusters during a financial crisis is useful and interesting as similar changes may occur during other crisis, leading to innovative ways of prevention and control. We study the dynamics of networks of global financial indices before, during, and after the global financial crisis of 2008. To study the interaction between global financial indices, we construct networks of financial indices at different correlation thresholds. The Fruchterman-Reingold layout is used to obtain different clusters in these networks and provides useful information about the interactions in financial indices in different periods of crisis. Further, the average linkage hierarchical clustering algorithm is used to investigate the clearer cluster structure. We find that the cophenetic correlation coefficients increase significantly during the period of crisis.

Can selfish rational agents achieve co-operation ? Slides
V. Sasidevan
The origin and maintenance of co-operation among selfish individuals is an intriguing question. The Iterated Prisoner's Dilemma (IPD) has become the paradigm problem in this regard for understanding co-operative behavior among selfish individuals. In this work, we analyse IPD using the solution-concept of co-action equilibrium proposed in [1]. We show that rational agents following the solution concept will adopt a win-stay lose-shift strategy in two-player IPD and mutual co-operation is the steady state. The win-stay lose-shift strategy for IPD was considered first by Nowak and Sigmund in [2] where they showed that it has two advantages over the widely considered tit-for-tat strategy [3]; it can correct for occasional mistakes and also can exploit unconditional cooperators. Here, we show that rational agents under co-action equilibrium will follow such a strategy. We also analyse $N$-player IPD where each agent plays with everybody else in a pair-wise fashion in each round, using the solution concept. We show that interesting non-trivial steady states may be reached where a group of cooperators who are the majority and a group of defectors may co-exist depending upon the ratio of the temptation payoff $T$ to the reward payoff $R$. For other parameter regimes, system goes to all cooperation. The results are true even for a finitely repeated game. [1]. V. Sasidevan and Deepak Dhar, Physica A 402, 306 (2014). [2] M. Nowak and K. Sigmund, Nature 364 56 (1993). [3] R. Axelrod, The evolution of cooperation, New York: Basic Books (1984).

Patterns of development: Identifying critical factors behind urban growth in the Indian context
S. Sridhar
Why do some places evolve into large sprawling metropolitan settlements over time, while other initially similar sites decay into obscurity ? Identifying the factors underlying the phenomenon of urban growth has sparked the curiosity of scientists ever since Walter Christaller proposed the Central Place Theory in order to explain the observed number, sizes and locations of settlements in southern Germany. However, lack of availability of sufficient empirical data has hampered progress in developing a quantitative understanding of this process. In order to initiate a data-driven approach to answer questions on the growth of settlements, we have undertaken the compilation of a large database of economic, demographic and infrastructural factors associated with different sites of habitation in different states of India. I will be presenting preliminary results of our analysis of such data for settlements in Tamil Nadu, which is one of the most rapidly urbanizing states and provides a window into the fast-changing rural-urban landscape of India. Using information gleaned from surveys done by our field teams as well as from government census reports (both population and economic), we have investigated the growth of populations across different regions in the state - identifying growth "hotspots" and decay "coldspots" using statistical & visualization techniques. Further, using tools of network analysis, we have identified inter-relations between key infrastructural factors affecting the development of these settlements.

Long-term evolution of the interaction structure in the New York Stock Exchange 1925-2012 Slides
K A Chandrashekar
Financial markets are complex systems that comprise of many agents interacting with each other as well as responding to external information. Earlier studies on the cross-correlations of price movements of different stocks have revealed the interaction structure of various financial markets - which has resulted in the intriguing speculation that the evolution of a market from emerging or developing to developed status is accompanied by systematic changes in its interaction structure. Using a very large data-base of daily price changes of equities listed in the New York Stock Exchange we have investigated the long-term changes that this financial market has undergone over a period of nearly nine decades (1925-2012). We have used spectral analysis of the daily log-return cross-correlations in order to reveal the network of significant interactions between equities. We find that the distribution of interaction strengths varies with the state of the economy. In particular, the skewness of the distribution shows a remarkable increase in recent years. We have investigated the strength distribution over the network in different periods by treating the network as resulting from a percolation process where the threshold value of interaction strength for deciding whether to connect a pair of nodes is varied. We find that the formation of the giant component can occur very differently in different periods - which reflects the micro-structure of the interactions between the equities. By comparison with surrogate ensembles of randomized networks we have identified the significant features in the topological organization of the empirical networks.

INDIAN SCHOOL OF ECONOPHYSICS Slides
Kishore C. Dash
There has been a long journey right from Chanakya to Mahalnobis, from Meghnad Saha to Bikas Chakrabarti, from Artha Sastra to Econophysics. In this article, I have tried to capture the developments in the field of Eonophysics in Indian soil by Indian scientists. Kolkata, the birth place of the new word ‘Econophysics’ in dictionary, could spread the fragrance of econophysics to different parts of the country in all directions. I have discussed at length the contributions of Indian Institutions, scientists how econophysics has been nurtured in this land in the form of publication of papers, books, conducting conferences etc.

Econophysics and Random Market analysis
Subhamoy Singha Roy
Economic market dynamics is thoroughly study via the exact widespread Langevin equation. presumptuous market Random self‐resemblance, the market return rate reminiscence and autoassociation functions are derived, which exhibit an oscillatory manners with a long‐time tail, similar to empirical explanation. Creature stocks are also described using the generalized Langevin equation.