銀行危機實證研究：併購、銀行治理與預警模型

Three Essays on Banking Crisis: M&As, Bank Governance and Early Warning Models

許倖華(Hsing-Hua Hsu)

指導教授 ： 沈中華

共同指導教授 ： 陳彥行(Yan-Shing Chen)

國立臺灣大學/管理學院/財務金融學研究所/博士(2019年)

https://doi.org/10.6342/NTU201901109

摘要

本論文由三篇與銀行危機相關的文章組成。第一篇與主併銀行進行併購的時間點有關。第二篇研究公司治理和銀行治理對銀行在危機期間風險承擔行為的影響。第三篇文章則是建立一個考量門檻效果和固定效果的銀行危機預警模型，該模型尚未被過去的實證研究所探討。此三篇分別為本論文之第一至第三章。以下是這三篇文章的重點摘要。
第一篇：「銀行危機與入市時機：銀行業併購的實證探討」
本文旨在探討在銀行危機期間進行併購(M&As)的銀行，其收益是否大於非危機時期。本篇在學術上的貢獻主要在於使用1994−2009年期間、來自106個國家（地區）的1,984筆銀行併購樣本。實證結果顯示，若銀行於危機期間進行併購，主併者的綜效會大於在非危機時期進行併購。而當主併銀行係來自已開發經濟體或屬於國內併購時，這種併購綜效會更加彰顯。我們的研究證實：銀行危機是銀行進行併購的適當時機。
第二篇：「銀行治理是否會降低風險承擔行為？2007−2009全球金融危機的實證」
藉由全球23個國家、105家銀行於2004−2013年的資料，本文發現：若銀行更關注公司治理，往往會冒更大的風險；但若他們更關注銀行治理，便會採取較少的風險承擔行為。具體而言，這種效應主要發生在2007−2009年全球金融危機期間。實證結果顯示，銀行過度關注股東保護但忽視其他利益相關者會給銀行本身帶來更多風險，特別是在危機期間。我們的發現不僅能解釋過去研究中公司治理與銀行績效間的負向關係，也揭露了銀行在不穩定的金融環境下面臨動盪時，「股東─債權人衝突」在其中所扮演的重要角色。
第三篇：「亞洲銀行危機早期預警訊號的決定-面板門檻羅吉斯模型的應用」
本文旨在建立一種新的計量方法：面板門檻羅吉斯模型 (PTLM)，它是由 Hansen (1999) 的面板門檻模型所延伸。在Hansen (1999) 的模型中，依變數為連續變數；在我們的模型則是二元變數。本文將說明如何在高門檻和低門區間中估計和測試參數。接著，我們將這種新方法應用到10個亞洲經濟體的總經數據，探討9個預警指標對銀行危機的影響，樣本期間涵蓋1987年至2016年。我們先使用短期債務對外匯存底比率作為門檻變數，並使用政府債務比率來檢驗結果的穩健性。實證結果顯示，門檻效果確實存在於亞洲銀行危機的預測。此外，我們也發現大部分的預警指標在兩個門檻區間內表現不同。因此，進行危機預測時，使用傳統的羅吉斯模型可能會讓我們遺漏重要訊息。

關鍵字

銀行危機 ； 合併與收購 ； 預警訊號 ； 門檻效果 ； 條件羅吉斯迴歸 ； 銀行治理 ； 風險承擔

並列摘要

This dissertation consists of three essays on the issue of banking crisis. The first is related to the profitable time point in conducting M&As for an acquiring bank. The second investigates the impacts of corporate governance and bank governance on a bank’s risk-taking behavior during the crisis period. The third essay is about constructing an early warning model for banking crises considering threshold and fixed effects, which has not been empirically explored by prior studies. The three essays are summarized below and are organized into Charter 1 to Chapter 3 respectively.
Essay 1. Banking Crises and Market Timing: Evidence from M&As in the Banking Sector
This study investigates whether the gains are greater for banks that conduct mergers and acquisitions (M&As) during banking crises than during non-crisis periods. It devotes to the literature by examining 1,984 M&As using global banking sample from 106 countries (areas) during 1994−2009. The empirical results show that the synergistic gains of acquiring banks during banking crises are larger than those during non-crisis periods. Such gains are greater when acquiring weak targets, for acquirers in developed countries, and for acquirers in domestic M&As. This study confirms that a banking crisis is an appropriate time for banks to conduct M&As.
Essay 2. Does bank governance lower risk-taking behavior? Evidence from the 2007−2009 global financial crisis
Using international data during 2004−2013 from 23 countries of 105 banks, this study finds that banks tend to be more risk-taking if they focus more on corporate governance, but engage in less risk-taking behavior if they pay more attention to bank governance. Specifically, this effect mainly exists during the 2007−2009 global financial crisis. The results indicate that focusing too much on shareholder protection but neglecting other stakeholders brings more risk to the banks, especially during the crisis. These findings can not only explain the negative relation between corporate governance and bank performance which has been found in prior studies, but also reveal the important role that shareholder-debtholder conflict plays when the bank is facing turmoil under the unstable financial environment.
Essay 3. Determinants of an Early Warning Signal for Asian Banking Crises − Application of the Panel Threshold Logit Model
This paper develops a novel econometric method, a panel threshold logit model (PTLM), which is a direct extension of Hansen’s (1999) panel threshold model that the dependent variable is a continuous variable, whereas, in our model, it is a binary number. We show how to estimate and test the parameters in high and low threshold regimes of such model. Then, we apply this new method on the data of ten Asian economies to investigate the effect of nine early warning indicators on banking crises. Our sample periods cover 1987 to 2016. We first use a ratio of short-term debt to foreign reserves as our threshold variable and then use the government debt ratio to examine the robustness of the results. The empirical results confirm the existence of the threshold effect. We also find that most of the early warning indicators perform differently in the two debt regimes. When using a conventional logit model, important information may be missed in making a crisis prediction.

並列關鍵字

banking crisis ； mergers and acquisitions ； early warning signal ； threshold effect ； conditional logit regression ； bank governance ； risk taking

參考文獻

Acharya, V., & Naqvi, H. (2012). The seeds of a crisis: A theory of bank liquidity and risk taking over the business cycle. Journal of Financial Economics, 106(2), 349-366.

Adhikari, B. K., & Agrawal, A. (2016). Does local religiosity matter for bank risk-taking?. Journal of Corporate Finance, 38, 272-293.

Aghion, P., Bacchetta, P., Ranciere, R., & Rogoff, K. (2009). Exchange rate volatility and productivity growth: The role of financial development. Journal of monetary economics, 56(4), 494–513.

Akter, R., & Roy, J. K. (2017). The impacts of non-performing loan on profitability: An empirical study on banking sector of Dhaka stock exchange. International Journal of Economics and Finance, 9(3), 126-132.

Alesina, A., Prati, A., & Tabellini, G. (1989). Public confidence and debt management: A model and a case study of Italy (No. w3135). National Bureau of Economic Research.