以貨幣危機預警模型探討不同國家主權評等與地理位置下之匯率風險

Discussing Exchange Rate Risk of Different Ratings and Geographic Areas with Currency Crisis Warning System

楊志偉(Zhi-Wei Yang)

指導教授 ： 沈中華

國立臺灣大學/管理學院/財務金融學研究所/碩士(2009年)

https://doi.org/10.6342/NTU.2009.00281

摘要

本研究試圖找出顯著影響貨幣危機之總體經濟因子，並以這些總體經濟因子為主軸建立貨幣危機預警模型，同時另一方面，再以地理區域與國家主權評等，將樣本國家加以細分至數小樣本，探討是否其顯著影響貨幣危機之總體經濟因子是否有所不同，及其背後之關係。因此，本研究採IMF的IFS資料庫與World Bank的WDI以及國內外文獻為基礎蒐集59個樣本國家，研究期間自1980年至2007年為止的資料作為樣本，資料頻率為年資料。

從實證結果可以發現經濟成長率只有在以國家主權評等在BBB+以上之研究樣本內才有與假說相同之顯著解釋能力。再者，消費者物價指數成長率對於全部國家皆有相當不錯且與假說相同之解釋能力，其中又以在國家主權評等在BBB+以上之國家為研究樣本時有最佳之解釋能力。其次，消費比率不論在哪一種樣本下，皆為相當不顯著之總體經濟因子。過度投資率在不同樣本下，皆有不錯之解釋能力，其中又以美洲國家為研究樣本下，有相當顯著之解釋能力。關於資本產出增量比，則出現與假說相反之影響方式。其次，關於金融機構放款給私部門增量，本研究發現在以國家主權評等BBB+以上之國家中，會得到與假說相反之結論。另一方面，經常帳比率在不同樣本下，皆出現與假說相同之實證結果，此外，此假說效果在國家主權評等在BBB以下國家之研究樣本中最為顯著。接著，關於外資反轉流出，只有在金融體系較不健全之國家，假說之影響方式才會成立。另一方面，關於出口比率，本研究得到與假說相反之實證結果。再者，財政收支餘額比率在不同樣本下皆有與假說相同之影響效果，其中又以在美洲國家、國家主權評等在BBB以下國家為研究樣本時，有極為顯著之實證結果。

以Pseudo判定係數而言，本研究發現當國家經過適度的細分時，會出現較佳之Pseudo判定係數。而就Probit Model與Logit Model來說，本研究發現，不論哪種樣本下，在採用Probit Model時所得到的實證結果，皆比採用Logit Model來的稍微顯著一點。

關鍵字

貨幣危機 ； 匯率風險 ； 預警模型 ； 國家主權評等 ； 總體經濟因子

並列摘要

In this thesis, I try to find the substantial macro-factors which result in currency crisis, and to build a currency crisis warning model based in these macro-factors. Furthermore, I dividend all of the sample countries into several groups, according to geographic zones and ratings, and discuss whether the influential macro-factors would be different and the relationship. Therefore, all of the data are downloaded from IFS database, IMF, and WDI, World Bank. Meanwhile, I select 59 countries as samples by referring to local and international theses. The period is from 1980 to 2007. The frequency of data is yearly.

In the empirical results, I discover that the GDP Growth Rate is substantial and the same as the hypothesis merely in the sample of the countries with rating above BBB+. What is more, the Consumer Price Index Growth Rate has decent capacity of explaining in all kinds of the samples. Among them, the sample of countries above BBB+ shows the best capability of explaining. Also, the Consumption to GDP is an unsubstantial macro-factor whichever I adopt. However, the Difference between Saving to GDP and Investment to GDP holds good explaining abilities whichever I use, and the sample of America shows the most powerful one. In terms of Incremental Capital Output Ratio, the empirical result is opposite to the hypothesis. Next, in view of the Loan for Private Sector, this empirical research discovers the sample of the countries below BBB appears an adverse effect, compared with the hypothesis. On the other hand, Current Account to GDP presents the same empirical outcome as the hypothesis in all samples. Moreover, in terms of Anti-Flow Foreign Capital, the hypothesis is tenable only in the countries with unstable financial systems. What is more, the Export to GDP appears reverse conclusions. Furthermore, the influence of Financial Surplus or Deficit is identical.

In view of the Pseudo R-square, this research will detect better ones if all of the countries are divided into several small groups. In terms of the model, the outcomes of the Probit models are more substantial than the ones of the Logit models slightly.

並列關鍵字

Currency Crisis ； Exchange Rate Risk ； Waning System ； Ratings ； Macroeconomic Factors

參考文獻

15. 蕭文宜(2003)，「亞洲單一貨幣的中心匯率機制與貨幣危機預警模型之關聯性研究」，中原大學企業管理研究所。

1. 王盈傑(2003) ，「具學習能力之專家系統在貨幣危機預警模型上之應用」，靜宜大學會計所碩士論文。

Berg, A. and C. Pattillo (1999), “Are currency crises predictable? A test,” IMF Staff Papers, Vol. 46, pp.107-138.

Bosworth, B. P. and S. M. Collins (1999), “Capital Flows to Developing Economies: Implications for Saving and Investment," Brookings Papers on Economic Activity, 1, pp.143-180.

Diebold, F. X. and G. D. Rudebusch (1989), “Scoring the Leading Indicators,” Journal of Business, 62, pp.369-391.