Search results for: heteroskedasticity

Authors: Salman Mohamadi, Seyed Mohammad Ali Tayaranian Hosseini, Hamidreza Amindavar

Abstract:

In this paper, we provide a procedure to analyze and model EEG (electroencephalogram) signal as a time series using ARIMA-GARCH to predict an epileptic attack. The heteroskedasticity of EEG signal is examined through the ARCH or GARCH, (Autore- gressive conditional heteroskedasticity, Generalized autoregressive conditional heteroskedasticity) test. The best ARIMA-GARCH model in AIC sense is utilized to measure the volatility of the EEG from epileptic canine subjects, to forecast the future values of EEG. ARIMA-only model can perform prediction, but the ARCH or GARCH model acting on the residuals of ARIMA attains a con- siderable improved forecast horizon. First, we estimate the best ARIMA model, then different orders of ARCH and GARCH modelings are surveyed to determine the best heteroskedastic model of the residuals of the mentioned ARIMA. Using the simulated conditional variance of selected ARCH or GARCH model, we suggest the procedure to predict the oncoming seizures. The results indicate that GARCH modeling determines the dynamic changes of variance well before the onset of seizure. It can be inferred that the prediction capability comes from the ability of the combined ARIMA-GARCH modeling to cover the heteroskedastic nature of EEG signal changes.

Keywords: epileptic seizure prediction , ARIMA, ARCH and GARCH modeling, heteroskedasticity, EEG

Procedia PDF Downloads 375

Authors: Andreea Halunga, Chris D. Orme, Takashi Yamagata

Abstract:

This paper proposes a heteroskedasticity-robust Breusch-Pagan test of the null hypothesis of zero cross-section (or contemporaneous) correlation in linear panel-data models, without necessarily assuming independence of the cross-sections. The procedure allows for either fixed, strictly exogenous and/or lagged dependent regressor variables, as well as quite general forms of both non-normality and heteroskedasticity in the error distribution. The asymptotic validity of the test procedure is predicated on the number of time series observations, T, being large relative to the number of cross-section units, N, in that: (i) either N is fixed as T→∞; or, (ii) N²/T→0, as both T and N diverge, jointly, to infinity. Given this, it is not expected that asymptotic theory would provide an adequate guide to finite sample performance when T/N is "small". Because of this, we also propose and establish asymptotic validity of, a number of wild bootstrap schemes designed to provide improved inference when T/N is small. Across a variety of experimental designs, a Monte Carlo study suggests that the predictions from asymptotic theory do, in fact, provide a good guide to the finite sample behaviour of the test when T is large relative to N. However, when T and N are of similar orders of magnitude, discrepancies between the nominal and empirical significance levels occur as predicted by the first-order asymptotic analysis. On the other hand, for all the experimental designs, the proposed wild bootstrap approximations do improve agreement between nominal and empirical significance levels, when T/N is small, with a recursive-design wild bootstrap scheme performing best, in general, and providing quite close agreement between the nominal and empirical significance levels of the test even when T and N are of similar size. Moreover, in comparison with the wild bootstrap "version" of the original Breusch-Pagan test our experiments indicate that the corresponding version of the heteroskedasticity-robust Breusch-Pagan test appears reliable. As an illustration, the proposed tests are applied to a dynamic growth model for a panel of 20 OECD countries.

Keywords: cross-section correlation, time-series heteroskedasticity, dynamic panel data, heteroskedasticity robust Breusch-Pagan test

Procedia PDF Downloads 403

Authors: Margarita Grushanina

Abstract:

After several prominent central banks including European Central Bank (ECB), Federal Reserve System (Fed), Bank of Japan and Bank of England employed unconventional monetary policies in the aftermath of the financial crisis of 2008-2009 the problem of identification of the effects from such policies became of great interest. One of the main difficulties in identification of shocks from unconventional monetary policy measures in structural VAR analysis is that they often are anticipated, which leads to a non-fundamental MA representation of the VAR model. Moreover, the unconventional monetary policy actions may indirectly transmit to markets information about the future stance of the interest rate, which raises a question of the plausibility of the assumption of orthogonality between shocks from unconventional and conventional policy measures. This paper offers a method of identification that takes into account the abovementioned issues. The author uses factor-augmented VARs to increase the information set and identification through heteroskedasticity of error terms and rank restrictions on the errors’ second moments’ matrix to deal with the cross-correlation of the structural shocks.

Keywords: factor-augmented VARs, identification through heteroskedasticity, monetary policy, structural VARs

Procedia PDF Downloads 322

Authors: Shay Kee Tan, Kok Haur Ng, Jennifer So-Kuen Chan

Abstract:

This paper extends the conditional autoregressive range (CARR) model to multivariate CARR (MCARR) model and further to the two-stage MCARR-return model to model and forecast volatilities, correlations and returns of multiple financial assets. The first stage model fits the scaled realised Parkinson volatility measures using individual series and their pairwise sums of indices to the MCARR model to obtain in-sample estimates and forecasts of volatilities for these individual and pairwise sum series. Then covariances are calculated to construct the fitted variance-covariance matrix of returns which are imputed into the stage-two return model to capture the heteroskedasticity of assets’ returns. We investigate different choices of mean functions to describe the volatility dynamics. Empirical applications are based on the Standard and Poor 500, Dow Jones Industrial Average and Dow Jones United States Financial Service Indices. Results show that the stage-one MCARR models using asymmetric mean functions give better in-sample model fits than those based on symmetric mean functions. They also provide better out-of-sample volatility forecasts than those using CARR models based on two robust loss functions with the scaled realised open-to-close volatility measure as the proxy for the unobserved true volatility. We also find that the stage-two return models with constant means and multivariate Student-t errors give better in-sample fits than the Baba, Engle, Kraft, and Kroner type of generalized autoregressive conditional heteroskedasticity (BEKK-GARCH) models. The estimates and forecasts of value-at-risk (VaR) and conditional VaR based on the best MCARR-return models for each asset are provided and tested using Kupiec test to confirm the accuracy of the VaR forecasts.

Keywords: range-based volatility, correlation, multivariate CARR-return model, value-at-risk, conditional value-at-risk

Procedia PDF Downloads 67

Authors: Xu Wang

Abstract:

This research builds linear regressions of U.S. macroeconomic condition and volatility measures in the investment grade and high yield Credit Default Swap index spreads using monthly data from March 2009 to July 2016, to study the relationship between different dimensions of macroeconomy and overall credit risk quality. The most significant contribution of this research is systematically examining individual and joint effects of macroeconomic condition and volatility on CDX spreads by including macroeconomic time series that captures different dimensions of the U.S. economy. The industrial production index growth, non-farm payroll growth, consumer price index growth, 3-month treasury rate and consumer sentiment are introduced to capture the condition of real economic activity, employment, inflation, monetary policy and risk aversion respectively. The conditional variance of the macroeconomic series is constructed using ARMA-GARCH model and is used to measure macroeconomic volatility. The linear regression model is conducted to capture relationships between monthly average CDX spreads and macroeconomic variables. The Newey–West estimator is used to control for autocorrelation and heteroskedasticity in error terms. Furthermore, the sensitivity factor analysis and standardized coefficients analysis are conducted to compare the sensitivity of CDX spreads to different macroeconomic variables and to compare relative effects of macroeconomic condition versus macroeconomic uncertainty respectively. This research shows that macroeconomic condition can have a negative effect on CDX spread while macroeconomic volatility has a positive effect on determining CDX spread. Macroeconomic condition and volatility variables can jointly explain more than 70% of the whole variation of the CDX spread. In addition, sensitivity factor analysis shows that the CDX spread is the most sensitive to Consumer Sentiment index. Finally, the standardized coefficients analysis shows that both macroeconomic condition and volatility variables are important in determining CDX spread but macroeconomic condition category of variables have more relative importance in determining CDX spread than macroeconomic volatility category of variables. This research shows that the CDX spread can reflect the individual and joint effects of macroeconomic condition and volatility, which suggests that individual investors or government should carefully regard CDX spread as a measure of overall credit risk because the CDX spread is influenced by macroeconomy. In addition, the significance of macroeconomic condition and volatility variables, such as Non-farm Payroll growth rate and Industrial Production Index growth volatility suggests that the government, should pay more attention to the overall credit quality in the market when macroecnomy is low or volatile.

Keywords: autoregressive moving average model, credit spread puzzle, credit default swap spread, generalized autoregressive conditional heteroskedasticity model, macroeconomic conditions, macroeconomic uncertainty

Procedia PDF Downloads 139

Authors: Farrukh Javed, Krzysztof Podgórski

Abstract:

We propose a new model that accounts for the asymmetric response of volatility to positive ('good news') and negative ('bad news') shocks in economic time series the so-called leverage effect. In the past, asymmetric powers of errors in the conditionally heteroskedastic models have been used to capture this effect. Our model is using the gamma difference representation of the generalized Laplace distributions that efficiently models the asymmetry. It has one additional natural parameter, the shape, that is used instead of power in the asymmetric power models to capture the strength of a long-lasting effect of shocks. Some fundamental properties of the model are provided including the formula for covariances and an explicit form for the conditional distribution of 'bad' and 'good' news processes given the past the property that is important for the statistical fitting of the model. Relevant features of volatility models are illustrated using S&P 500 historical data.

Keywords: heavy tails, volatility clustering, generalized asymmetric laplace distribution, leverage effect, conditional heteroskedasticity, asymmetric power volatility, GARCH models

Procedia PDF Downloads 358

Authors: Wilmer Osvaldo Martinez, Manuel Dario Hernandez, Juan Manuel Julio

Abstract:

We propose to assess the performance of k forecast procedures by exploring the distributions of forecast errors and error losses. We argue that non systematic forecast errors minimize when their distributions are symmetric and unimodal, and that forecast accuracy should be assessed through stochastic loss order rather than expected loss order, which is the way it is customarily performed in previous work. Moreover, since forecast performance evaluation can be understood as a one way analysis of variance, we propose to explore loss distributions under two circumstances; when a strict (but unknown) joint stochastic order exists among the losses of all forecast alternatives, and when such order happens among subsets of alternative procedures. In spite of the fact that loss stochastic order is stronger than loss moment order, our proposals are at least as powerful as competing tests, and are robust to the correlation, autocorrelation and heteroskedasticity settings they consider. In addition, since our proposals do not require samples of the same size, their scope is also wider, and provided that they test the whole loss distribution instead of just loss moments, they can also be used to study forecast distributions as well. We illustrate the usefulness of our proposals by evaluating a set of real world forecasts.

Keywords: forecast evaluation, stochastic order, multiple comparison, non parametric test

Procedia PDF Downloads 51

Authors: Longqing Li

Abstract:

The paper addresses the inefficiency of the classical model in measuring the Value-at-Risk (VaR) using a normal distribution or a Student’s t distribution. Specifically, the paper focuses on the one day ahead Value-at-Risk (VaR) of major stock market’s daily returns in US, UK, China and Hong Kong in the most recent ten years under 95% confidence level. To improve the predictable power and search for the best performing model, the paper proposes using two leading alternatives, Extreme Value Theory (EVT) and a family of GARCH models, and compares the relative performance. The main contribution could be summarized in two aspects. First, the paper extends the GARCH family model by incorporating EGARCH and TGARCH to shed light on the difference between each in estimating one day ahead Value-at-Risk (VaR). Second, to account for the non-normality in the distribution of financial markets, the paper applies Generalized Error Distribution (GED), instead of the normal distribution, to govern the innovation term. A dynamic back-testing procedure is employed to assess the performance of each model, a family of GARCH and the conditional EVT. The conclusion is that Exponential GARCH yields the best estimate in out-of-sample one day ahead Value-at-Risk (VaR) forecasting. Moreover, the discrepancy of performance between the GARCH and the conditional EVT is indistinguishable.

Keywords: Value-at-Risk, Extreme Value Theory, conditional EVT, backtesting

Procedia PDF Downloads 296

Authors: Rancy Chepchirchir Kosgey, John Olukuru

Abstract:

This study examines the impact of introduction of futures trading on the spot price volatility in the commodity market. The paper considers the United States of America, South Africa and Ethiopian economies. Three commodities i.e. coffee, maize and wheat from New York Merchantile Exchange, South African Futures Exchange and Ethiopian Commodity Exchange are analyzed. ARCH LM test is used to check for heteroskedasticity and GARCH and EGARCH are used to check for the behavior of volatility between the pre- and post-futures periods. For all the three economies, the results indicate presence of the ARCH effect in the log returns. For conditional and unconditional variances; spot price volatility for coffee has decreased after futures trading in all the economies and the EGARCH has also shown reduction in persistence of volatility in the post-futures period in the three economies; while that of maize has reduced for the Ethiopian economy while there has been an increase in both the US and South African economies. For wheat, the conditional variance has been found to rise in the post-futures period in all the three economies.

Keywords: derivatives, futures exchange, agricultural commodities, spot price volatility

Procedia PDF Downloads 406

Authors: Manamba Epaphra

Abstract:

This paper analyzes the causal effect between domestic private investment, public investment, foreign direct investment and economic growth in Tanzania during the 1970-2014 period. The modified neo-classical growth model that includes control variables such as trade liberalization, life expectancy and macroeconomic stability proxied by inflation is used to estimate the impact of investment on economic growth. Also, the economic growth models based on Phetsavong and Ichihashi (2012), and Le and Suruga (2005) are used to estimate the crowding out effect of public investment on private domestic investment on one hand and foreign direct investment on the other hand. A correlation test is applied to check the correlation among independent variables, and the results show that there is very low correlation suggesting that multicollinearity is not a serious problem. Moreover, the diagnostic tests including RESET regression errors specification test, Breusch-Godfrey serial correlation LM test, Jacque-Bera-normality test and white heteroskedasticity test reveal that the model has no signs of misspecification and that, the residuals are serially uncorrelated, normally distributed and homoskedastic. Generally, the empirical results show that the domestic private investment plays an important role in economic growth in Tanzania. FDI also tends to affect growth positively, while control variables such as high population growth and inflation appear to harm economic growth. Results also reveal that control variables such as trade openness and life expectancy improvement tend to increase real GDP growth. Moreover, a revealed negative, albeit weak, association between public and private investment suggests that the positive effect of domestic private investment on economic growth reduces when public investment-to-GDP ratio exceeds 8-10 percent. Thus, there is a great need for promoting domestic saving so as to encourage domestic investment for economic growth.

Keywords: FDI, public investment, domestic private investment, crowding out effect, economic growth

Procedia PDF Downloads 256

Authors: Herbert Ntuli, Edwin Muchapondwa

Abstract:

This study used a sample of 336 households and community level data from 30 communities around the Gonarezhou National Park in Zimbabwe to analyse the association between ability to self-organize or cooperation and institutions on one hand and the relationship between success of biodiversity outcomes and cooperation on the other hand. Using both the ordinary least squares and instrumental variables estimation with heteroskedasticity-based instruments, our results confirmed that sound institutions are indeed an important ingredient for cooperation in the respective communities and cooperation positively and significantly affects biodiversity outcomes. Group size, community level trust, the number of stakeholders and punishment were found to be important variables explaining cooperation. From a policy perspective, our results show that external enforcement of rules and regulations does not necessarily translate into sound ecological outcomes but better outcomes are attainable when punishment is rather endogenized by local communities. This seems to suggest that communities should rather be supported in such a way that robust institutions that are tailor made to suit the needs of local condition will emerge that will in turn facilitate good environmental husbandry. Cooperation, training, benefits, distance from the nearest urban canter, distance from the fence, social capital average age of household head, fence and information sharing were found to be very important variables explaining the success of biodiversity outcomes ceteris paribus. Government programmes should target capacity building in terms of institutional capacity and skills development in order to have a positive impact on biodiversity. Hence, the role of stakeholders (e.g., NGOs) in capacity building and government effort should complement each other to ensure that the necessary resources are mobilized and all communities receive the necessary training and resources.

Keywords: institutions, self-organize, common pool resources, wildlife, conservation, Zimbabwe

Procedia PDF Downloads 249

Authors: R. Bhattu Babajee, Marie Sandrine Estelle Benoit

Abstract:

The past decades have witnessed a considerable institutional shift towards Central Bank Independence across economies of the world. The motivation behind such a change is the acceptance that increased central bank autonomy has the power of alleviating inflation bias. Hence, studying whether Central Bank Independence acts as a significant factor behind the price stability in the African economies or whether this macroeconomic aim in these countries result from other economic, political or social factors is a pertinent issue. The main research objective of this paper is to assess the relationship between central bank autonomy and inflation in African economies where inflation has proved to be a serious problem. In this optic, we shall measure the degree of CBI in Africa by computing the turnover rates of central banks governors thereby studying whether decisions made by African central banks are affected by external forces. The purpose of this study is to investigate empirically the association between Central Bank Independence (CBI) and inflation for 10 African economies over a period of 17 years, from 1995 to 2012. The sample includes Botswana, Egypt, Ghana, Kenya, Madagascar, Mauritius, Mozambique, Nigeria, South Africa, and Uganda. In contrast to empirical research, we have not been using the usual static panel model for it is associated with potential mis specification arising from the absence of dynamics. To this issue a dynamic panel data model which integrates several control variables has been used. Firstly, the analysis includes dynamic terms to explain the tenacity of inflation. Given the confirmation of inflation inertia, that is very likely in African countries there exists the need for including lagged inflation in the empirical model. Secondly, due to known reverse causality between Central Bank Independence and inflation, the system generalized method of moments (GMM) is employed. With GMM estimators, the presence of unknown forms of heteroskedasticity is admissible as well as auto correlation in the error term. Thirdly, control variables have been used to enhance the efficiency of the model. The main finding of this paper is that central bank independence is negatively associated with inflation even after including control variables.

Keywords: central bank independence, inflation, macroeconomic variables, price stability

Procedia PDF Downloads 344

Authors: Janne Engblom, Elias Oikarinen

Abstract:

A panel dataset is one that follows a given sample of individuals over time, and thus provides multiple observations on each individual in the sample. Panel data models include a variety of fixed and random effects models which form a wide range of linear models. A special case of panel data models are dynamic in nature. A complication regarding a dynamic panel data model that includes the lagged dependent variable is endogeneity bias of estimates. Several approaches have been developed to account for this problem. In this paper, the panel models were estimated using the Arellano-Bover/Blundell-Bond Generalized method of moments (GMM) estimator which is an extension of the Arellano-Bond model where past values and different transformations of past values of the potentially problematic independent variable are used as instruments together with other instrumental variables. The Arellano–Bover/Blundell–Bond estimator augments Arellano–Bond by making an additional assumption that first differences of instrument variables are uncorrelated with the fixed effects. This allows the introduction of more instruments and can dramatically improve efficiency. It builds a system of two equations—the original equation and the transformed one—and is also known as system GMM. In this study, Finnish housing price dynamics were examined empirically by using the Arellano–Bover/Blundell–Bond estimation technique together with ordinary OLS. The aim of the analysis was to provide a comparison between conventional fixed-effects panel data models and dynamic panel data models. The Arellano–Bover/Blundell–Bond estimator is suitable for this analysis for a number of reasons: It is a general estimator designed for situations with 1) a linear functional relationship; 2) one left-hand-side variable that is dynamic, depending on its own past realizations; 3) independent variables that are not strictly exogenous, meaning they are correlated with past and possibly current realizations of the error; 4) fixed individual effects; and 5) heteroskedasticity and autocorrelation within individuals but not across them. Based on data of 14 Finnish cities over 1988-2012 differences of short-run housing price dynamics estimates were considerable when different models and instrumenting were used. Especially, the use of different instrumental variables caused variation of model estimates together with their statistical significance. This was particularly clear when comparing estimates of OLS with different dynamic panel data models. Estimates provided by dynamic panel data models were more in line with theory of housing price dynamics.

Keywords: dynamic model, fixed effects, panel data, price dynamics

Procedia PDF Downloads 1431

Authors: Dimakatso Machetele, Kowiyou Yessoufou

Abstract:

Whilst several studies indicated that road accident events might be more complex than thought, we have a limited scientific understanding of this complexity in South Africa. The present project proposes and tests a more comprehensive metamodel that integrates multiple causality relationships among variables previously linked to road accidents. This was done by fitting a structural equation model (SEM) to the data collected from various sources. The study also fitted the GARCH Model (Generalized Auto-Regressive Conditional Heteroskedasticity) to predict the future of road accidents in the country. The analysis shows that the number of road accidents has been increasing since 1935. The road fatality rate follows a polynomial shape following the equation: y = -0.0114x²+1.2378x-2.2627 (R²=0.76) with y = death rate and x = year. This trend results in an average death rate of 23.14 deaths per 100,000 people. Furthermore, the analysis shows that the number of crashes could be significantly explained by the total number of vehicles (P < 0.001), number of registered vehicles (P < 0.001), number of unregistered vehicles (P = 0.003) and the population of the country (P < 0.001). As opposed to expectation, the number of driver licenses issued and total distance traveled by vehicles do not correlate significantly with the number of crashes (P > 0.05). Furthermore, the analysis reveals that the number of casualties could be linked significantly to the number of registered vehicles (P < 0.001) and total distance traveled by vehicles (P = 0.03). As for the number of fatal crashes, the analysis reveals that the total number of vehicles (P < 0.001), number of registered (P < 0.001) and unregistered vehicles (P < 0.001), the population of the country (P < 0.001) and the total distance traveled by vehicles (P < 0.001) correlate significantly with the number of fatal crashes. However, the number of casualties and again the number of driver licenses do not seem to determine the number of fatal crashes (P > 0.05). Finally, the number of crashes is predicted to be roughly constant overtime at 617,253 accidents for the next 10 years, with the worse scenario suggesting that this number may reach 1 896 667. The number of casualties was also predicted to be roughly constant at 93 531 overtime, although this number may reach 661 531 in the worst-case scenario. However, although the number of fatal crashes may decrease over time, it is forecasted to reach 11 241 fatal crashes within the next 10 years, with the worse scenario estimated at 19 034 within the same period. Finally, the number of fatalities is also predicted to be roughly constant at 14 739 but may also reach 172 784 in the worse scenario. Overall, the present study reveals the complexity of road accidents and allows us to propose several recommendations aimed to reduce the trend of road accidents, casualties, fatal crashes, and death in South Africa.

Keywords: road accidents, South Africa, statistical modelling, trends

Procedia PDF Downloads 129

Authors: Somnath Mukhuti, Prem Kumar Ghosh

Abstract:

Background and significance of the study Indian stock market has undergone sudden changes after the current China crisis in terms of turnover, market capitalization, share prices, etc. The average returns on equity investment in both markets have more than three and half times after global financial crisis owing to the development of industrial activity, corporate sectors development, enhancement in global consumption, change of global financial association and fewer imports from developed countries. But the economic policies of both the economies are far different, that is to say, where Indian economy maintaining a conservative policy, Chinese economy maintaining an aggressive policy. Besides this, Chinese economy recently lowering its currency for increasing mysterious growth but Indian does not. But on August 24, 2015 Indian stock market and world stock markets were fall down due to the reason of Chinese stock market. Keeping in view of the above, this study seeks to examine the influence of Chinese stock on Indian stock market. Methodology This research work is based on daily time series data obtained from yahoo finance database between 2009 (April 1) to 2015 (September 28). This study is based on two important stock markets, that is, Indian stock market (Bombay Stock Exchange) and Chinese stock market (Shanghai Stock Exchange). In the course of analysis, the daily raw data were converted into natural logarithm for minimizing the problem of heteroskedasticity. While tackling the issue, correlation statistics, ADF and PP unit root test, bivariate cointegration test and causality test were used. Major findings Correlation statistics show that both stock markets are associated positively. Both ADF and PP unit root test results demonstrate that the time series data were not normal and were not stationary at level however stationary at 1st difference. The bivariate cointegration test results indicate that the Indian stock market was associated with Chinese stock market in the long-run. The Granger causality test illustrates there was a unidirectional causality between Indian stock market and Chinese stock market. Concluding statement The empirical results recommend that India’s stock market was not very much dependent on Chinese stock market because of Indian economic conservative policies. Nevertheless, Indian stock market might be sturdy if Indian economic policies are changed slightly and if increases the portfolio investment with Chinese economy. Indian economy might be a third largest economy in 2030 if India increases its portfolio investment and trade relations with both Chinese economy and US economy.

Keywords: Indian stock market, China stock market, bivariate cointegration, causality test

Procedia PDF Downloads 344