附录
ST公司 上海电气 中国远洋 世荣兆业 威尔泰 华帝股份 海特高新 兔宝宝 传化股份 巨轮股份 *ST 夏新 ST 华光 ST 中葡 *ST 华源 S*ST 黑龙 ST 松江 *ST 昌鱼 ST 中农 *ST博盈 ST 银广夏 *ST创智 ST盛润A *ST 威达 SST 中华A *ST 化工 ST 科健 *ST 九发 ST 珠峰 ST 香梨 ST 筑信 *ST 北生 *ST 白猫 S*ST北亚 ST 马龙 ST 春兰 *ST 中房 *ST 三联 *ST 宝龙 *ST 琼花 *ST张铜 0.1618 0.2207 0.0424 0.0758 0.2162 0.3447 0.0646 0.1258 0.3294 0.3271 0.0897 0.4068 0.6199 0.5438 0.0701 0.6009 0.0473 0.2365 0.1386 0.3645 0.4728 0.5457 0.0572 0.2048 0.1333 0.3143 0.0646 0.0065 0.5502 0.6489 0.1156 -0.2451 0.0618 0.4410 0.0877 0.1152 0.0685 -2.8044 0.1848 -3.3019 0.3721 -0.7126 -0.0707 -1.0447 0.2256 0.0817 0.0702 -0.0694 0.1704 -25.7771 0.0865 -1.2504 -0.2067 0.0306 0.0679 0.2322 0.2576 0.8227 0.0438 -0.0559 0.0038 -0.0858 0.0166 0.0129 0.7184 0.3577 0.0947 0.0274 0.1210 0.0196 0.0294 0.0396 -2.0112 -7.7592 -0.6658 -0.0066 0.6550 -4.0864 -0.0197 0.0002 0.0000 -0.3935 1.4626 0.0003 0.0264 -0.4175 0.0368 0.0283 -11.9657 -16.7021 0.9337 0.0034 -0.0562 0.5586 0.4493 -0.0137 -2.4977 -2.2715 0.2045 0.0000 0.0037 -0.4800 0.0053 1.1000 -0.4686 -0.3101 0.0243 0.0610 0.0298 0.4064 0.0160 0.6650 -0.7576 -0.2752 0.1400 0.1210 -1.8886 -13.3213 0.1394 -0.5744 0.1856 -0.1535 0.0219 0.1933 0.5690 0.4519 0.9886 2.5300 -0.8098 -0.5380 -0.1155 -0.1063 0.2892 0.5618 -0.1808 -0.2438 0.2637 -0.5933 -0.0474 -0.2880 0.0071 -0.1148 0.0473 0.2338 -0.7345 -1.7691 0.1882 0.0787 -0.4618 -0.2713 -0.1862 0.0056 -0.1294 0.1955 0.0934 0.3346 0.6433 0.5341 0.3747 0.5219 1.5152 0.1705 1.2310 1.1787 0.3343 0.0000 0.0307 0.1564 0.2642 0.1733 0.1861 0.0699 0.9125 0.9715 0.0000 0.0000 0.0020 0.1193 2.1644 0.6656 0.0089 0.0000 1.9781 0.1849 0.5303 0.0019 1.9011 0.0000 1.2924 0.4327 0.0221 1.7512 0.8402 0.7209 1.0850 1.3252 1.5594 1.7642 2.6137 2.5640 1.8116 2.0356 2.9700 1.3812 -5.2154 -1.3810 0.7301 -36.0869 0.3298 1.7562 0.0162 2.5952 1.2551 -15.4773 -4.9999 4.2777 -0.2960 -32.5159 2.8482 -5.4938 0.0100 1.0786 1.2397 -0.2345 -20.7990 2.0781 6.0959 -0.8903 0.8191 -0.8215 1.8780 -1.8580 -0.8312 1.7015
- 29 -
附录
附录4 外文文献原文
This paper develops a new early warning system (EWS) model, based on a multinomial logit model, for predicting financial crises. It is shown that commonly used EWS approaches, which use binomial discrete-dependent-variable models, are subject to what we call a post-crisis bias. This bias arises when no distinction is made between tranquil periods, when economic fundamentals are largely sound and sustainable, and crisis/post-crisis periods, when economic variables go through an adjustment process before reaching a more sustainable level or growth path. We show that applying a multinomial logit model, which allows distinguishing between more than two states, is a valid way of solving this problem and constitutes a substantial improvement in the ability to forecast financial crises. The empirical results reveal that, for a set of 20 open emerging markets for the period 1993e2001, the model would have correctly predicted a large majority of crises in emerging markets.
The last decade saw a large number of financial crises in emerging market economies (EMEs) with often devastating economic, social and political consequences. These financial crises were in many cases not confined to individual economies but spread contagiously to other markets as well. In particular, the Latin American crisis of 1994e1995 and the Asian crisis of 1997e1998 affected a wide group of countries and had systemic repercussions for the international financial system as a whole.
As a result, international organizations and also private sector institutions have begun to develop early warning system (EWS) models with the aim of anticipating whether and when individual countries may be affected by a financial crisis. The IMF has taken a lead in putting significant effort into developing EWS models for EMEs, resulting in influential papers by Kaminsky et al. (1998) and Berg and Pattillo (1999b). But also many central banks, such as the US Federal Reserve (Kamin et al., 2001; Kamin and Babson, 1999) and the Bundesbank (Schnatz, 1998, 1999), academics and various private sector institutions (JP Morgan, 1998; Goldmane Sachs, 1998; Deutsche Bank, 2000; Credit Suisse First Boston, 2001; Morgan Stanley Dean Witter, 2001) have developed models in recent years.
EWS models can have substantial value to policy makers by allowing them
- 30 -
附录
to detect underlying economic weaknesses and vulnerabilities, and possibly taking pre-emptive steps to reduce the risks of experiencing a crisis. The central concern is, however, that these models have been shown to perform only modestly well in predicting crises ex ante (Berg and Pattillo, 1999a).
The contribution this paper aims to make to the literature is to identify a bias in existing EWS models e what we call the post-crisis bias e and to develop a new type of EWS model, based on a multinomial discrete-dependent-variable approach, that solves for this bias. The post-crisis bias implies that existingEWSmodels fail to distinguish between tranquil periods, when economic fundamentals are largely sound and sustainable, and post-crisis/recovery periods, wheneconomic variables go through an adjustment process before reaching a more sustainable level or growth path.
We show that making this distinction by using a multinomial logit model with three regimes (a tranquil regime, a pre-crisis regime, and post-crisis/recovery regime) constitutes a substantial improvement in the forecasting ability of EWS models. Our empirical model is based on the analysis of 20 openEMEs using a monthly frequency for the period 1993e2001. In particular, the use of the multinomial logit model reduces substantially the number of false alarms (i.e., the number of times the model indicates that a crisis is likely to occur but no crisis actually happens) and of missed crises (i.e., when the model issues no signal but a crisis occurs) as compared to the binomial logit model. Overall, the EWS model based on the multinomial logit would have predicted most EME financial crises since the early 1990s, while entirely missing a crisis only in one case (Singapore in 1997). Moreover, the out-of-sample performance of the multinomial EWS model is robust and would have allowed the correct anticipation of most emerging market crises of the 1990s.
The paper proceeds in Section 2 by outlining our definition of a financial crisis and by reviewing the two most commonly used approaches for EWS models, the leading indicator approach and the discrete-dependent-variable approach. Section 3 presents the results obtained from a simple binomial logit model, which is comparable to existing EWS models in the literature. Section 4 then discusses the post-crisis bias and the multinomial logit as a way of solving it. Section 5 presents the results from the multinomial logit and compares them with
- 31 -
附录
alternative models; it also presents robustness tests and out-of-sample estimates. A discussion of the findings concludes the paper in Section 6.
There are various types of financial crises: currency crises, banking crises, sovereign debt crises, private sector debt crises, equity market crises. The EWS model in this paper focuses primarily on currency crises. Currency crises often coincide or occur in quick succession with other types of crises, for instance together with banking crises in what has been dubbed the ‘‘twin crises’’ (Kaminsky and Reinhart, 1999). More specifically, our EWS model employs the commonly used exchange market pressure (EMPi,t) variable for defining a currency crisis for each country i and period t:
EMPi,t is a weighted average of the change of the real effective exchange rate (REER), the change in the interest rate (r) and the change in foreign exchange reserves (res). Taking the real variables for the exchange rate and the interest rate accounts for differences in inflation rates across countries and over time. The weights uREER, ur and ures are the relative precision of each variable so as to give a larger weight to the variables with less volatility. Precision is defined as the inverse of the variance of each variable for all countries over the full sample period 1993e2001. The key advantage of the EMP measure is that it allows capturing both successful and unsuccessful speculative attacks.
As a next step, we define a currency crisis (CCi,t) as the event when the exchange market pressure (EMPi,t) variable is two standard deviations (SD) or more above its country average EMPi:
This is the definition of currency crises that will be used below in our econometric analysis. It is identical or quite similar to the measures commonly used in the literature.1 The next crucial question is what we are trying to predict: the timing of a currency crisis or merely its occurrence. As the state of the literature on EWS models for financial crises shows, predicting not only whether a currency crisis happens but also the timing when it will happen (the precise month) is a highly ambitious, if not unrealistic goal. The objective of our EWS is therefore not to predict the exact timing of a crisis, but to predict whether a crisis occurs within a specific time horizon. Our approach consists in transforming the contemporaneous variable CCt into a forward-looking variable Yi,t
In other words, our model attempts to predict whether a crisis will occur
- 32 -
