Anomalies in the financial market

Introduction

Anomalies in the financial market draw attention among market players; this is because they indicate evidence of ineffectiveness. Any form of prediction on asset returns, becomes a debate for market ineffectiveness. Additionally, the anomalies are evident in developed as well as emerging markets. Nevertheless, the significance is varying for each market. Financial marketing anomalies for emerging markets are significant for investors (both domestic and international) generally this is as a result of the certainty of returns. Since this is likely to influence expansion and global integration. With respect to efficient market assumption, all the necessary information should be integrated in the price of stocks. A price that presents a particular increase should change based on the arbitrage (Bekaert, et al. 2006).

As a result, investors should not realize profit from trading due to market forces or opportunities for particular trading estimations or even technical assessment. Anomalies can be at a natural phase for all types of financial markets while emerging markets can be at the commencement of such a phase. Essentially, developed financial markets received significant attention. Moreover, anomalies such as size effect were found to be vital. International markets as well received significant attention though small in size. Previous studies indicate that majority of these anomalies disappear or drop considerably in scale. Experts (Campbell, et al, 1997) have urged that it logically contradictory as well as statistically ineffective to employ volatility evaluations that focus on supposition of constant volatility over a given time frame.

Volatility is a field that has attracted attention in research, but the main unresolved issue is the connection between the condition of the economy and fiscal volatility. Several models have been formulated to help in forecasting volatility, however models that integrate economic variables are difficult to get. Utilizing different techniques, associations are discernible though they are usually weaker. For instance, popularly it is acknowledged that during recession volatility is extremely high, however these effects seems small component of the estimated volatility.

Scholars (Bekaert, et al. 2006) have attempted to address high volatility in the 30s using leverage as well as volatility of industrial production. In addition, they sought to demonstrate the relations between financial instability and macro volatility although they indicated that the mystery emphasized by stock instability is not related to variables of economic unpredictability. A different approach evaluates the impacts of announcements on returns. Using elaborate regression techniques, contemporary announcement measure are incorporated in stock returns. It is important to note that these models only depict a small part of volatility and current edition utilize intraday information with relatively related results (Chib,et al. 2008). In this paper, GARCH model will be used to forecast the volatility in Borsa Istanbul January 2000 to April 30, 2013. This model will used to present high frequency financial information. Accordingly, this will help in forecasting volatility in the financial market. Consequently, the hypothesis that volatility is a mean changing to a stable level that triggers will be modeled using GARCH technique (Ling and McAleer 2003).

Financial volatility in a Macroeconomic background

The popular is the log return, d the log dividend and t-I be time, then the equation will be

Unanticipated returns may be decomposed as projected cash flow shocks or even shocks for estimated returns. Shocks have a positive impact on the dividends whereas on interest rate they have a negative impact. Various news scenarios have varying effects on stock market depending on time frame – short term or long term horizon (Hafner, 2003). To effectively describe the size of such shocks, previous studies have decomposed unanticipated stocks into news elements.

In this case k denotes news sources; e signifies news events that could be variation between expected value and declared values. It is evident that announcements are not the only source, because of the continuing accrual of proof before announcement can greatly affect stock prices. However, this impact depends on condition of the economy as represented by z_i. For instance, during recession bad news about certain stocks are influential compared to expansion phase. As such, this model is applicable only at time when news is discernible. If this is not the case, then

This equation has a distinctive innovation that demonstrates all sorts of news. Additionally, the discrepancy of such innovation will in turn rely on macro components, partially due to the size of news will rely on such conditions and partially as a result of news intensity will rely on macro economic variables. This can be represented as

Based on this equation, macroeconomic factors z are considered as a deterministic or even the variance is estimated as a condition variable of macro economics (Chib et al. 2008). However, innovation u has a temporary reliance that is not caused but macro economic variables. When these is modeled by GARCH technique the equation will be

Where g and ε contain unconditional opportunities

It is evident that the impact of macro economics on volatility of stocks depends on news as well as news multiplier factor; nonetheless, it is difficult to separately categorize these unless news is evident. In reality, estimation is not an appropriate as macro economic factors are not described for each given date. While, using quarterly values will result to breaks that will be insignificant economically. Rather, non-parametric estimation is used to calculate macro variables. It demonstrates the notion that they are gradually varying. In essence this is beneficial because it can be applied for any give series without necessarily specifying economic framework (Bekaert, et al. 2006).

A model for conditional and unconditional Volatility

In this scenario, GARCH model will be used, the model was initiated in 1986 by Bollerslev to present new flexible requirements for unconditional volatility via a semi-parametric structure (Bollerslev, 1986). In spite of the accomplishment of GARCH model in explaining the dynamics of volatility in economic, its significance in continuous volatilities are restricted in a manner that the model is only applicable to constant estimated volatility (Kawakatsu, 2006). As a result, this feature is not consistent with time series performance of realized volatilities of stocks. Additionally, it is imperative to formulate a model that is adequately flexible to produce expected volatility that captures long term trends detected from the data. To meet this objective, GARCH model will be used in generating a pattern in the volatility procedure.

In particular, this pattern is modeled non-parametrically via exponential that produces a smooth curve explaining the long term volatility aspect centered entirely on data (Domowitz, et al. 2001). Subsequently, GARCH model entails decomposition of volatility procedure into different elements. One explains the short term dynamics of restricted volatility related with temporary impacts of volatility improvements. On the other hand, it demonstrates the slower changes in volatility procedure related with extra permanent impacts (Campbell, et al. 1997). An additive decomposition is triggered by substituting σ with a stochastic aspect indicating the long memory characteristic of volatility procedure. In turn, the long memory aspect establishes unconditional volatility and can be assessed as a pattern via which conditional volatility changes. For recognition purposes, this element is considered as a slower mean reverting frequency compared to temporary element. Based on this, GARCH model ensures that unconditional volatility and the velocity of mean restrictions are constant. Additionally, large data are required to accurately classify different elements. Nevertheless, (Wurgler, 2000) the slow moving pattern is mean reversion to a constant value and the notion that volatility procedure ultimately changes to a fixed level. In this scenario, it is important to note that slow moving pattern in volatility, changes to a fixed level. Consequently, non-parametric technique helps in generating data sets in functional forms. Also instead of utilizing additive decomposition, it is necessary to categorize the components (low as well as high frequency) elements of volatility procedure by way of multiplicative decomposition inspired by economic approach of volatility (Maheu, 2002). This pattern demonstrates low frequency aspect of volatility connected with gradually changing deterministic conditions in the financial market, or even random variables which are greatly constant and move gradually. The unobserved pattern is evaluated non-parametrically by a GARCH model that produces a smooth curve depicting low frequency volatility element primarily centered on data sets (Wurgler, 2000). In this case, exponential functional assures that the low frequency aspect is regularly positive. The quadratic feature is triggered by the need to get smoothness via connection of minimum derivative cost effectively. GARCH model also contributes to time changing unconditional volatility which varies based on the volatility system. But, the evaluation procedure may be intricate as well as demanding (Bekaert and Harvey 1997).

Data returns

This assessment evaluates stock markets revenues. Utilizing XU300 index related to the daily stock returns of Borsa Istanbul from January 1997 to April 2013. Moreover, yearly realized volatility is used to demonstrate a constant evaluation of unconditional volatility related to evaluated pattern. To begin with, long term periods are associated with yearly intervals. Therefore, for every variable, yearly averages are calculated. For each year, it is vital to match yearly low rate volatility with different macro economic conditions. This data can be summed up in a linear structure forecasting annual stock volatility. The mean for XU300 cases is positive. With regards to skewness, the means are not normally distributed: In this section, we illustrate the analysis of anticipated long-term market volatilities. Before illustrating the collective set-up, it is pertinent to note that some statistics intricacies and rules. To begin with, we equally associate to long-term periods with progressive yearly intermissions. Therefore for each of the variables brought forward above, yearly aggregates are developed. In totality, what is correlated in yearly low frequency volatility time series with Istanbul macroeconomic time series? This approach leads to session-centered sample windows.

Table 1 show the GARCH model summary of the statistics used in this analysis. Fluctuations in this data are related to the period and volatility trend.

Table 1: Shows the GARCH Model Summary

Model: GARCH(1,1) [Bollerslev] (Normal)*

Dependent variable: CLOSE__USD_

Sample: 1950-10089 (T = 8140), VCV method: Robust

Conditional mean equation

Coefficient std. error z p-value

—————————————————–

const 1014.63 13.1347 77.25 0.0000 ***

Conditional variance equation

coefficient std. error z p-value

———————————————————

omega 461.733 118.156 3.908 9.31e-05 ***

alpha 0.943619 0.0456440 20.67 6.01e-095 ***

beta 0.0549742 0.0466353 1.179 0.2385

Llik: -62390.80918 AIC: 124789.61836

BIC: 124817.63654 HQC: 124799.20020

Table 2: Shows Heteroskedasticity-corrected Model

Model 2: Heteroskedasticity-corrected, using observations 1700:01-2378:04 (T = 8140)

Dependent variable: SESSION

	Coefficient	Std. Error	t-ratio	p-value
CLOSE__USD_	0.000653846	4.06568e-06	160.8208	<0.00001	***

Statistics based on the weighted data:

Sum squared resid	21843.29	S.E. of regression	1.638225
R-squared	-1.388994	Adjusted R-squared	-1.388994
F(1, 8139)	-4732.126	P-value(F)	NA
Log-likelihood	-15567.67	Akaike criterion	31137.34
Schwarz criterion	31144.35	Hannan-Quinn	31139.74
rho	0.353239	Durbin-Watson	1.293448

Statistics based on the original data:

Mean dependent var	1.498280	S.D. dependent var	0.500028
Sum squared resid	6270.844	S.E. of regression	0.877763

Graph 1: Shows time plot series of Close_USD

As an emerging market, Turkey demonstrates more cycles based on successive years. However, to investigate the probable variations in the dependence arrangement of GARCH model Mean values indicate slightly less persistence in the GARCH representation a figure that indicates how different the model is. By and large, the results indicate that GARCH model demonstrates a slightly shorter memory ARMA structure in the squared innovations, an aspect shared by other GARCH class representation that relax the value restrictions for unconditional variance, like component GARCH representation.

To substantiate the enhanced performance of the GARCH representation over the simple GARCH, BIC is employed. This fit static approach indicates that the GARCH model is the best representations used when forecasting volatility in Turkey especially when the optimal number of sessions is larger than one. In addition, in the event of a one-knot case, we’re likely to expect more difficulties in dismissing the hypothesis of mean degeneration in volatility to a fixed value; we therefore reject the GARCH for all the years.

Economic determinants of low frequency volatility

To adequately fit time series of stock returns GARCH model is applied. In every case, low frequency volatility is obtained, therefore it can be explained as the average of the daily return over a given time frame-annually. Using financial theory and past empirical analysis are used to choose the appropriate variables (Jorion, 2000). Changes in financial markets and levels are the min themes in this scenario. These aspects largely impact on the uncertainty of cash flow; risk premiums and effect stock volatility rely on the economic conditions. In line with this model, past studies (Hafner et, al.2005) have indicated a connection between business cycle and volatility; for instance, they show that financial recession as the main aspect affecting stock volatility.

On the contrary, under certain conditions, realized volatility can be considered as estimated awareness of volatility. Volatility as well as uncertainty fundamentals are potential aspects influencing financial market. For instance, experts (source) gain both risk premium and stock volatility according to stochastic volatility techniques; the main sources of stock volatility are earning uncertainty and inflation. Subsequently, empirical data shows a close relationship between stock and macro economic volatility. To account for such kind of uncertainties it is necessary to consider macro economic variables of volatility.

Based on the above assumption, the analysis integrates other factors for annual stock uncertainty. Specifically, GARCH technique is used to estimate exchange rates. The exchange rate is estimated in USD and Euros. Both predictors of economic measures or future condition of financial market are vital explanatory measures for low rate volatility (Bekaert and Harvey 1997). For instance, variables related to the financial policy and projected economic developments are important in assessing projected uncertainties regarding interest rates. In this scenario, inflation level is a main aspect used by market players to assess financial institutions’ credibility, particularly in developed economies where many macroeconomic policies aim at enhancing institutional management of inflation.

Nevertheless, a group of experts (Taylor, 1986 and Pesaran, 2006) have indicated that financial effects and inflation costs are different since they relatively depend on diverse institutional frameworks. Additionally, empirical data has evaluated the relationship between expansion and volatility of stocks. For instance, level as well as volatility of stocks has distinct important effect on expansion. Some annual empirical analysis has demonstrated that financial market growth is an essential aspect in describing disparities in stock volatility. For instance, data shows huge volatility and high possibility of extreme conditions, in emerging economies compared to developed ones.

Conclusion

Volatility is a major complication facing stock markets world over. In this paper, we used GARCH model to predict the performance of Istanbul’s stock exchange. This forecast is the rationale behind market incompetence. Whereas it is anticipated that these complications fade away as they are report, this is barely the case for Istanbul Stock Exchange. We also had to experiment that GARCH model for all the indexes of XU300, from January 1997 to April 2013 and we conclude that the GARCH model fails to substantiate any negative returns. Experiments on mean disparities and disparities in variation equally substantiated the outcome. When it comes to market competence presumption, we can wrap-up that the Istanbul Stock Market competence is questionable with regards to return forecasts that stretch for years. GARCH was relevant in this study, for purposes of proficiency of the estimation. Models are evaluated differently for each index permitting for different GARCH cases. Because the Istanbul Stock Exchange uses XU300 index, this makes it possible for other market intricacies to subsist. Bearing in mind the fact that XU300 is the market anomaly that attained most of the concentration for ISE and continued other complications such as January impact, USCLOSE impact and Session Impact should be an area that requires serious future research. Analogous to other emerging markets, Istanbul Stock Exchange also went through reasonable transformation in the previous years. This is to say any market complexity is likely to be mitigated for session specific outcomes. In the end, Borsa Istanbul is a large economy and greatly complex. This being the case, it has extremely levered and this results to higher equity volatility of stocks. Again, there is a possibility of diversification consequences that are likely to decrease equity volatilities.

References

Bekaert, G., and C. Harvey, (1997), “Emerging Equity Market Volatility,” Journal of Financial Economics, 43, 29-77.

Bekaert, G., C. Harvey, and C. Lundblad, (2006), “Growth Volatility and Financial Liberalization,” Journal of International Money and Finance, 25, 370-403.

Bollerslev, T., (1986), “Generalized Autoregressive Conditional Heteroskedasticity,” Journal of Econometrics, 31, 307-327.

Campbell, J. Y., Lo, A. W., and MacKinlay, A. C. (1997), The Econometrics of Financial Markets, Princeton, New Jersey: Princeton University Press.

Chib, S., Y. Omori, and M. Asai (2008): “Multivariate stochastic volatility,” in Handbook of Financial Time Series, Springer, New York.

Domowitz, I., J. Glen, and A. Madhavan, (2001), “Liquidity, Volatility and Equity Trading Costs Across Countries and Over Time,” International Finance, 4, 221-255.

Hafner, C. M. (2003): “Fourth moment structure of multivariate GARCH models,” Journal of Financial Econometrics, 1, 26–54.

Hafner, C. M., D. van Dijk, and P. H. Franses (2005): “Semi-parametric modeling of correlation dynamics,” in Advances in Econometrics, ed. by T. Fomby, C. Hill, and D. Terrell,vol. 20/A, pp. 59–103. Amsterdam: Elsevier Sciences

Jorion, P. (2000) Value at Risk, The New Benchmark for Managing Financial Risk, McGraw- Hill, USA.

Kawakatsu, H. (2006): “Matrix exponential GARCH,” Journal of Econometrics, 134, 95–128.

Ling, S., and M. McAleer (2003): “Asymptotic theory for a vector ARMA–GARCH model, ”Econometric Theory, 19, 280–310.

Maheu, J., (2002), “Can GARCH Models Capture Long-Range Dependence in Financial Market Volatility?” Working Paper, University of Toronto.

Pesaran, H., (2006), “Estimation and Inference in Large Heterogeneous Panels with a Multifactor Error Structure,” Econometrica, 74, 967-1012.

Taylor, S.J. (1986) “Forecasting the Volatility of Currency Exchange Rates.” International Journal of Forecasting3: 159-170.

Wurgler, J., (2000), “Financial Markets and the Allocation of Capital,” Journal of Financial Economics, 58, 187-214.

Appendix 1 shows date time series plot

Appendix 2 shows time plot series of Close_USD

Appendix 3 shows time plot series of Session

Appendix 4: Shows the GARCH Model Summary