Finansal Piyasalarda Etkinlik Hipotezi: OECD Ülkeleri Üzerine Bir Uygulama
Derya ÖzBu çalışmada finansal piyasalarda etkinlik hipotezinin geçerliliğinin test edilmesi amacıyla, geleneksel ve güncel ekonometrik yöntemler kullanılarak, OECD ülkelerine ait borsa endeks değerleri incelenmiştir. Etkinlik hipotezi, menkul kıymet fiyatlarının modellenerek tahmin edilemeyeceği şeklinde tanımlanmaktadır. Etkin piyasalarda fiyatların tahmin edilememesinin nedeni ise fiyatların rastgele yürüyüş sergilemesidir. Finansal zaman serilerinin tanımlayıcı istatistikleri incelendiğinde, genellikle normal dağılıma uygunluk göstermedikleri bilinmektedir. Bu durumda finansal zaman serilerinde normal dağılım varsayımına dayalı birim kök testlerinin kullanımı, hatalı sonuçlar elde edilmesine sebep olabilmektedir. Dolayısıyla çalışmanın ampirik kısmında normal dağılım varsayımına dayalı birim kök testleri ile birlikte Kalıntılarla Arttırılmış En Küçük Kareler (RALS) yöntemine dayanan birim kök testleri de kullanılmıştır. Etkinlik hipotezinin test edilmesi için 19 OECD ülkesine ait borsa endeks değerleri üzerinde; LM tipi bir ve iki yapısal kırılmalı birim kök testleri ile RALS yöntemine dayalı olan RALS-LM birim kök testleri kullanılmıştır. Normal dağılım varsayımına dayanan SP ve LM testi ile RALS-LM testlerinin uygulamadaki sonuçlarına göre; SP ve LM birim kök testlerinin yardımcı regresyonlarına ait kalıntıların normal dağılım göstermediği gözlenmiştir. Dolayısıyla ele alınan finansal serilere ilişkin kalıntıların normallik varsayımını sağlamadığı gösterilmiştir. Ayrıca serilerin birim köklü bir süreç gösterip göstermediği incelendiğinde, RALS-LM birim kök testlerinde, SP ve LM birim kök testlerine göre daha çok ülke piyasa endeksi için birim kök hipotezinin reddedildiği, yani ilgili ülkelerin piyasa fiyatlarının etkin olmadığı tespit edilmiştir. Bu durumda etkin olmayan piyasalarda yatırımcıların fiyatları tahmin edebilmesi bilgisi ile etkin piyasalarda fiyatların tahmin edilemeyeceği bilgisi göz önünde bulundurularak yatırım kararları alınması uygun olacaktır.
The Efficiency Hypothesis in Financial Markets: An Application over OECD Countries
Derya ÖzThis study examines the stock market index values of OECD countries using traditional and current econometric methods to test the validity of the efficiency hypothesis in financial markets. The efficiency hypothesis is defined as stock exchange markets’ inability to be predicted through modeling. The reason why prices cannot be predicted in efficient markets is that prices exhibit random walks. When examining the descriptive statistics of financial time series, they are generally known to not conform to normal distributions. Therefore, the use of unit root tests based on the assumption of normal distribution in a financial time series may lead to incorrect results. As such, the empirical part of the study uses unit root tests based on the residual augmented least squares (RALS) method alongside unit root tests based on the assumption of normal distribution. To test the efficiency hypothesis on the stock market index values of 19 OECD countries, the study uses the Lagrange multiplier (LM) unit root tests with one and two structural breaks and RALS-LM tests, which are unit root tests based on the RALS method. According to the results from the Schmidt and Phillips (SP), LM and RALS-LM tests, the residuals from the auxiliary regressions of the SP and LM unit root tests were observed to not show normal distributions. Therefore, the residuals related to the financial series under consideration have been shown to not support the normality assumption. Additionally, when examining whether the series shows a unit-rooted process the RALS-LM unit root tests reject, the unit root hypothesis for more country market indexes than the SP and LM unit root tests. In other words, the study has determined the market prices of the relevant countries to not be efficient. In this case, making investment decisions would be appropriate by taking into account the knowledge that investors can predict prices in inefficient markets and that prices cannot be predicted in efficient markets.
The statistical feature that distinguishes financial time series from economic time series is that financial time series show a leptokurtic distribution. Volatility clustering and positive excess kurtosis are known to occur in leptokurtic distributions. The number of outliers in financial time series is higher than in normally distributed series. This is why, when examining the probability density function graph, the tail parts of the distribution in these series are thicker than the normal distribution. Based on the reasons for these explanations, financial series are known to not conform to a normal distribution. Therefore, the use of tests based on the assumption of normal distribution regarding analyses of series that do not have a normal distribution (e.g., financial series) will negatively affect the obtained results.As such, other methods are used in cases of normal non-dispersion that also give effective and reliable results. The normality assumption is also important in unit root analyses performed over financial series. In order for the obtained results to provide effective estimates, residuals related to auxiliary regression models are required to have normal distributions. The studies of Meng et al. (2014) and Meng et al. (2016) applied unit root tests based on the residual augmented least squares (RALS) method cases where the normality assumption was not provided, and the residuals related to auxiliary regressions that showed a leptokurtic distribution also deviated from normality as a result of asymmetry or nonlinearity. The RALS method has reasons related to the deviation of the higher-order moments of the residuals belonging to auxiliary regressions that do not show a normal distribution. Therefore, this study uses this method and the second and third moments of the residuals to attempt to resolve the problem of the normality assumption.
This study considers the efficiency hypothesis, which is defined as the inability to use modeling to predict the prices of securities. Therefore, prices can not be predicted because they exhibit a random walk (Champell et al., 1996, p.22). When the prices for securities bought and sold on a securities market reflect all the information and prices respond quickly or without deviation to new information, such markets are generally interpreted as being efficient (Deckman & Dale, 1986, p.5).
This study uses traditional and current econometric methods to examine the stock market index values of the Organisation for Economic Co-operation and Development (OECD) countries and to test the validity of the efficiency hypothesis over financial markets. Due to the leptokurtic distribution of the financial series, the empirical part of the study uses unit root tests based on the RALS method, which is widely used in cases where the residuals do not comply with normal distribution, as well as unit root tests based on the assumption of normal distribution. To test the efficiency hypothesis over the stock market index values of 19 OECD countries, the study uses the Lagrange multiplier (LM) unit root tests with one and two structural breaks and RALS-LM unit root tests, which are unit root tests based on the RALS method. According to the practical results from the Schmidt and Phillips (SP), LM and RALS-LM tests based on the assumption of normal distribution, the residuals from auxiliary regressions of the SP and LM unit root tests were observed to not show normal distributions. Therefore, the residuals related to the financial series under consideration have been shown to not support the normality assumption. Additionally, when examining whether the series shows a unit-rooted process, the RALS-LM unit root tests reject the unit root hypothesis for more country market indexes than the SP and LM unit root tests.
