Forecasting and comparing the volatility of global government bond indices during 1998-2008

• Main Author: Quesang, William Wayne Tiu
• Format: text
• Language: English
• Published: Animo Repository 2009
• Subjects: Forecasting; Government bonds; Stock market; Stock exchange; Foreign exchange rates; Banks and banking, Central; Finance and Financial Management
• Online Access: https://animorepository.dlsu.edu.ph/etd_masteral/3859; https://animorepository.dlsu.edu.ph/context/etd_masteral/article/10697/viewcontent/CDTG004726_P__1_.pdf
• Institution: De La Salle University

Central Banks seek ways to maximize the returns on its increasing foreign reserves. As an offshoot, most of them create investment policies and select an appropriate index or indices to be used as a benchmark for its excess portfolios performance. While investment-grade sovereign and supranational debts are the preferred asset classes, and different large financial institutions provide indices on these assets, the indices performance somewhat differ due to the set of rules, source of pricing and market compositions. These differences highlighted the need to understand the global government bond indices clearly to be able to use them more effectively. Central Banks often face political and reputational risks on losses that may incur in their portfolios, hence, there is the need to estimate the volatility of the indices they use. Central Banks must also investigate the relationship between the index performance and its volatility for a better understanding of the risk-return profile of the index. To address these concerns, the study utilized the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model and Exponential Generalized Autoregressive Conditional Heteroskedasticity (EGARCH) model to analyze and forecast the volatility of different government bond indices. The GARCH-in-Mean (GARCH-M) model was also used in an effort to explain the relationship between the return of the indices and their volatility. A set of six (6) global government bond indices were selected based on the scope and limitations set by the study. Daily, weekly and monthly return frequencies of the indices covering the period from 1 January 1998 to 1 July 2008 of the indices were used as data for the models. Further, the GARCH and GARCH-M models were used on data series that have significant ARCH effects while an additional asymmetric test was done for the estimation of the EGARCH model. To test the efficiency of the models, the analyzed volatility of the GARCH and EGARCH models were compared to the realized volatility during the same period using the mean absolute error (MAE) and the mean absolute percentage error (MAPE). The models were further tested using the same methods, by comparing the forecasted volatility of the GARCH and EGARCH models and the realized volatility from July 1, 2008 to December 31, 2008. The study found that the estimation of the GARCH and EGARCH models on all indices were significant on the daily return series, while the EGARCH model was significant on the monthly return series and in only two indices on the weekly return series. Comparing the efficiency of the models, the study has found that based on the MAE, the EGARCH model was better in analyzing the volatility while the GARCH model was better than the EGARCH model in forecasting the volatility based on using the MAPE and the MAE. The study also found that the GARCH-M model sufficiently explained the relationship on four indices in using the monthly return series, supporting the financial theory that the assets return required by investors depends on the riskiness of the financial assets over time. The study suggests that the Merrill Lynch Global Government G7 Index, although found to be the least volatile on the daily return series, Central Bank may use the Merrill Lynch Global Sovereign Broad Market Index, being the second least volatile on the daily and monthly frequency while being the most tractable. The index may also be used to sufficiently explain the relationship between the return of the index and its volatility. It may also help diversify the portfolios risk and exposures in the bond markets due to the breadth of its coverage. Moreover, it is advisable to use the conditional variance estimated by the GARCH model in evaluating the volatility of the government bond indices, as it gives a more updated performance rather than just averaging it.