投資國內公債與外匯避險之美國公債在FRTB標準法下兩者風險計提之比較：以日本為例

Abstract

本文志在探討於FRTB市場風險計提之標準法下，以日本銀行角度出發為例，並以美日利差尚未擴大之2011年為背景，考慮投資十年期日本公債及經XCCY Basis Swap或 FX Swap避險後的十年期美國公債投資組合，進行試算比較報酬率、風險計提及風險計提使用效率（每單位風險計提可換到的報酬率）之差異，用上述試算結果分析兩投資路線在市場風險計提新標準法下之優劣勢，並檢視現今（2024年）美日利差已擴大到近二十年高點下，給予日本銀行兩種投資路線中具較佳風險使用效率之建議。本研究發現在FRTB風險計提標準法下，雖日本公債報酬率不及避險後的美國公債，但因日本銀行投資日本公債具有較低風險計提之優勢，因此日本公債具有不亞於避險後美國公債之風險計提使用效率，此研究結果與單就報酬率考量之直覺相左，而此現象在利差擴大的現今，此效果更為明顯。此外，本研究也發現若是具有現金流可相互抵消特性之投資組合，一起放入交易簿相較於分開放置於銀行簿及交易簿能更加節省資本計提。期望能供同樣具台美利率差距大現象的台灣銀行進行投資參考。

This study explores the application of the Fundamental Review of the Trading Book (FRTB) standard approach for market risk provisioning from the perspective of Japanese banks. Using 2011 as a backdrop, a period before the significant widening of the US-Japan interest rate differential, it compares the returns, risk provisioning, and risk provisioning efficiency (return per unit of risk provisioned) of ten-year Japanese Government Bonds and ten-year US Treasury Bonds hedged through XCCY Basis Swaps or FX Swaps. The analysis evaluates the strengths and weaknesses of these investment strategies under the new FRTB standards and offers recommendations for Japanese banks in 2024, where the US-Japan interest rate differential has reached a near twenty-year high. The results indicate that, under the FRTB standard approach, although Japanese Government Bonds yield lower returns than hedged US Treasuries, they offer a lower risk charge, making their risk provisioning efficiency comparable to or greater than that of hedged US Treasuries. This finding contrasts with intuitive expectations based solely on returns and is more pronounced given the current widened interest rate differentials. Additionally, the study finds that portfolios with offsetting cash flows are more capital-efficient when consolidated in the trading book rather than separated into the banking and trading books. These insights provide valuable guidance for Taiwanese banks, which face similar US-Taiwan interest rate differentials, in their investment strategies.