Hüseyin Tatlidil

Forecasting the primary energy demand in Turkey and analysis of cyclic patterns

Abstract The planning and estimation of future energy demand via modern statistical methods have been officially used in Turkey since 1984. However, almost all previous forecasts proved significantly higher than actual observations because of several reasons discussed here. The cycle analysis, which is a semi-statistical technique that makes use of any cyclicity in the historical data of annual additional amounts of energy demand, appears to give better results than the other techniques for forecasting energy demand in Turkey. This method suggests that the energy demand will be around 130 million toe in 2010. This figure is very close to the estimates obtained by the Winters exponential smoothing method. To increase the scientific validity of the method, it should be applied in other similar countries.

Profiling Property Criminals in Turkey

The main aim of this research is to develop a profile of a thief. Through a comprehensive survey of property crime suspects arrested in the city of Ankara between 2004 and 2005, the authors have attempted to determine the socioeconomic qualities of those involved in these forms of property crimes. Results of the study show that property crimes are a consequence of low education, lack of occupational skills, and alcohol and drug addiction on the part of the offenders. Because of these factors, theft becomes a way of life for offenders after their first act of property crime.

A new generalization of skew-T distribution with volatility models

ABSTRACT In this paper, we propose a new generalized alpha-skew-T (GAST) distribution for generalized autoregressive conditional heteroskedasticity (GARCH) models in modelling daily Value-at-Risk (VaR). Some mathematical properties of the proposed distribution are derived including density function, moments and stochastic representation. The maximum likelihood estimation method is discussed to estimate parameters via a simulation study. Then, the real data application on S&P-500 index is performed to investigate the performance of GARCH models specified under GAST innovation distribution with respect to normal, Students-t and Skew-T models in terms of the VaR accuracy. Backtesting methodology is used to compare the out-of-sample performance of the VaR models. The results show that GARCH models with GAST innovation distribution outperforms among others and generates the most conservative VaR forecasts for all confidence levels and for both long and short positions.