Naslov
Can popular VAR Models be trusted?: A case of EU new member states

Autori
Žiković, Saša

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Achieving Competitive Advantage Through Managing Global Resources, Readings Book / Delener, Nejdet ; Fuxman, Leonora ; Victor, Lu F. ; Rivera-Solis, Luis Eduardo ; Su, Che-Jen - : Global Business and Technology Association (GBATA), 2007, 825-831

ISBN
1-932917-03-9

Skup
Ninth International Conference Global Business and Technology Association

Mjesto i datum
Taipei, Tajvan, 03.07.2007. - 07.07.2007

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
VaR models; market risk; GARCH; Bulgaria; Romania

Sažetak
The author in this paper examines different ways of calculating VaR in transitional economies of EU new member states – Romania and Bulgaria. Nine VaR models are tested on official stock indexes from Romania and Bulgaria. Performance of analyzed VaR models is tested by Kupiec test, Christoffersen unconditional coverage test, Christoffersen independence test, Christoffersen conditional coverage test, Lopez test, Blanco-Ihle test, Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE). The obtained results show that ARMA-GARCH volatility forecasts are superior to simpler models of volatility forecasting and provide among best VaR forecasts. The findings show that common VaR models that are widely used in mature markets, such as historical simulation with shorter observation windows, BRW historical simulation, variance-covariance model and RiskMetrics system are not well suited to these capital markets.

Izvorni jezik
Engleski

Znanstvena područja
Ekonomija

POVEZANOST RADA