BALANCE SHEET EFFECTS, EXTERNAL VOLATILITY

1、 BALANCE SHEET EFFECTS, EXTERNAL VOLATILITY, AND EMERGING MARKET SPREADSSamuel W. MaloneUniversity of the AndesThis paper studies the determinants of emerging market spreads, and thus of the cost of borrowing for emerging market sovereigns, using recent data from JP Morgans EMBI index for a panel of

2、 19 countries. Controlling for traditional spread determinants, we focus on three additional factors whose importance is suggested by recent work: external shocks, the balance sheet effect of real devaluations, and the degree of current account leverage. We find clear and strong evidence that the va

3、riables in the foregoing categories have an economically and statistically significant relationship with spreads. In particular, we find a major role for the terms-of-trade volatility and the level of current account leverage in explaining spread variation. The result on current account leverage est

4、ablishes an important link between a factor shown to make countries more vulnerable to sudden stops of capital flows, and the premium required by international investors on their foreign debt.JEL classification codes: C33, F34Key words: balance sheet effects, emerging market debt, external volatilit

5、y, country risk premium.I. IntroductionThe study of emerging market spreads, defined as the difference between the yield on emerging market bonds and the yield on US Treasury bonds with the same or similar maturity, extends back to Edwards (1984, 1986). Spreads on debt reflect several factors. In ad

6、dition to a premium for the probability of default on the underlying bond and the recovery rate that investors assume they will receive in the event of default, spreads may include an additional premium related to the liquidity of the underlying bonds, the prevalent degrees of liquidity and risk-ave

7、rsion in the market, and even tax privileges realized by the investor.1In this study, we focus on highly liquid, dollar denominated debt instruments of emerging market sovereigns that are traded in international markets and included in JP Morgans EMBI+ index.Panel data studies of emerging market spr

8、eads that focus on “classic” determinants of sovereign risk include: Rowland and Torres (2004), Adesetal(2000), Eichengreen and Mody (1998), Min (1998), and Cantor and Packer (1996). A good survey can be found in Sobrinho (2004). The “traditional” candidates for the determinants of sovereign spreads

9、, which have found to be significant statistically and/or economically in at least one of these studies, include: the economic growth rate, the debt-to-GDP ratio, the reserves-to-GDP ratio, the debt-to-exports ratio, the exports-to-GDP ratio, the ratio of debt service-to-GDP, the fiscal balance, int

10、ernational interest rates, the default history of the country, net foreign assets, and the domestic inflation rate. We try most of these candidates as control variables in the present study. Despite the contributions of the above literature to understanding the determinants of emerging market spread

11、s, however, an understanding of the importance of balance sheet effects in the presence of sudden stops of capital inflows, and macroeconomic volatility, in provoking financial and debt crises is much more recent. Important papers on the latter topics, respectively, include Calvo, Izquierdo, and Tal

12、vi (2003), Calvo, Izquierdo, and Mejia (2004), and Cato and Kapur (2006). The primary contribution of the present paper is to demonstrate that both balance sheet effects and terms-of-trade volatility have economically and statistically significant effects on the spreads of emerging market sovereigns

13、, after controlling for a variety of other factors that have been shown to affect spreads. In particular, we establish a robust link between the degree of current account leverage, a key variable in determining the impact of sudden stops of capital inflows, and emerging market spreads. We define the

14、 degree of current account leverage, following closely the work of Calvo, Izquierdo, and Mejia (2004), as the value of foreign financing of the current account measured as a percentage of the value of a countrys imports. The rationale for this variable is straightforward: given a sudden stop of fore

15、ign financing of the current account, a country will be forced to reduce its purchases of imports. The percentage of import purchases it will need to forgo in the event ofa sudden stop is an important measure of the duress caused by that event. Calvo, Izquierdo, and Mejia (2004) also emphasize that

16、current account leverage is connected to the degree of real exchange rate depreciation that a sudden stop will require. Another recent paper, by Berganza, Chang, and Garca-Herrero (2004), directly tests the effects of unexpected real devaluations on sovereign spreads, which are presumed to work thro

17、ugh negative net worth effects, as hypothesized for example by Aghion, Bacchetta and Banerjee (2004), and Cspedes, Chang and Velasco (2000). Berganza, Chang, and Garca-Herrero (2004) find that the interaction between unexpected real devaluations and the ratio of foreign debt service-to-GDP is signif

18、icantly associated with higher spreads. In the present study, we show additionally that countries with high values of current account leverage face higher spreads, even after taking into account the direct balance sheet effects due to the interaction between real devaluations and the debt burden. Th

19、us, there are other reasons besides real devaluations, perhaps related to the fungibility of foreign currency revenue for import purchases, which explain the importance of current account leverage for the cost of borrowing.Regarding our finding that higher terms-of-trade volatility is associated wit

20、h higher spreads, this is consistent with the closely related finding of Cato and Kapur (2006) that higher terms-of-trade volatility is associated with a higher number of incidences of default. This finding is intuitive, because higher terms- of-trade volatility is associated with a higher volatilit

21、y of foreign currency income, and countries with more volatile income streams face a higher probability of being in a position in which the value of their foreign debt service needs exceed the value of their foreign currency reserves and short-term revenues. Such countries, as a consequence, face hi

22、gher probabilities of default, and thus lenders can be expected to demand higher premiums on their debt. Our findings are consistent with recent work by Hilscher and Nosbusch (2007), who also find a statistically and economically significant role for the volatility of a countrys terms-of-trade in in

23、fluencing EMBIG spreads.Other recent regression studies of emerging market sovereign spreads include Westphalen (2001) and Ferrucci (2003). In particular, Westphalen (2001) finds that changes in the volatility over the last 20 trading days of the local MSCI countrystock index positively and signific

24、antly affects spreads. This result provides another indication, in addition to the results of Hilscher and Nosbusch (2007) and the present study, that the volatility of fundamentals can be an important determinant of spreads. It is worth noting that the R2 values obtained in our regressions, which r

25、ange from 71% to 91% across the various specifications tested, are noticeably higher than those obtained in the previous studies mentioned. We attribute our ability to explain a significant degree of spread variation primarily to our focus on identifying and testing simultaneously the effects of dis

26、tinct and important categories of factors that have been identified in the literature in particular the term-of-trade volatility and the balance sheet variables inspired by recent work.The rest of this paper is organized as follows. In section II, we outline a simple theoretical framework for analyz

27、ing the determinants of emerging market spreads that justifies the log-linear form of our spread regressions. Section III then describes the data and independent variables and states the baseline regression. Section IV presents the main empirical results of the paper, including robustness tests of t

28、he baseline regression. Section V analyzes the connection between our current account leverage measure, FDI and portfolio flows, and spreads. Section VI concludes.II. Basic theoretical frameworkAssuming that lenders are risk neutral, there is high capital mobility, and that the return on the soverei

29、gn bond to lenders is zero in the event of default, the sovereign spread s is determined by the following condition, which has been used by others (see Min 1998 and Nogus and Grandes 2001):where r is the risk-free rate and P is the probability of default. If the recovery rate is nonzero and there ar

30、e no fixed default costs, then the above equation generalizes to: where the return in the event of default is given by . For models with more detailed considerations of the costs of default, such as spillover costs, see for example Cato and Kapur (2006). The sovereign spread implied by the above con

31、dition isSimilar to Nogus and Grandes (2001), approaching P with a logistic function such aswhere the variables Xi determine the probability of default, leads to the following log-linear form for the spread equation:We will now describe the data and dependent variables, after which we will state the

32、 baseline regression for the paper, which we adapt from the above equation.III. Dataset and econometric specificationA. The data and independent variablesWe analyze the determinants of the country risk premium on external, dollar denominated emerging market debt using yearly data on the implied coun

33、try spread for the JP Morgan Emerging Market Bond Index Plus (EMBI+). The EMBI+, according to JP Morgan, is the “most liquid US-dollar emerging market debt benchmark, and tracks total returns for actively traded external debt instruments in emerging markets” (JP Morgan 2004). We take as our sample a

34、ll of the countries (19 in total) that have been part of the index from their first year on the index up until 2004. This time period includes both the Russian debt crisis in 1997-98 and the Argentine default of 2001, as well as the generally very positive performance in emerging market debt seen in

35、 2004. The countries and years used in the study are summarized in Table 1. The dependent variable in our study will be the logarithm of the EMBI+ spread reported by JP Morgan, denoted by log spread. For each year, we use the average monthly spread in that year. The variables used to run the main re

36、gressions discussed in the paper are defined in Table 2.The variables in the study can be grouped into three categories: traditional macroeconomic variables, variables capturing some aspect of external volatility, and variables capturing balance sheet effects. In our preferred regression, we settled

37、 for parsimony on three traditional macroeconomic variables: the ratio of debt-to-GNI (debt-to-gni), the ratio of reserves-to-GNI (res-to-gni), and the five year rolling rate of GNI growth (5year_growth_rate). We also experimented with other traditional variables, such as a measure of openness, the

38、trade surplus, the government surplus (fiscal surplus as a ratio to GDP), and the rate of inflation of the consumer price index, but these did not prove robustly significant.In the second category of variables we include the terms-of-trade (tot) and the rolling ten year terms-of-trade volatility (to

39、t-volatility). Since we include yearly dummies to account for time effects in all of our regressions, there is no scope for including variables such as the US interest rate, or the VIX implied volatility index, which exhibit only time variation.Finally, the third category of determinants consists of

40、 the currency-mismatch variable and the current account leverage variable, CA-leverage, both of which are important for assessing balance sheet effects. In the first case, the currency-mismatch variable is equal to the product of the foreign debt-service-to-GNI ratio and the year-on-year change in t

41、he real exchange rate normalized by the real exchange rate in the year 2000. Second, the CA-leverage variable is defined as the ratio of debt service, minus net exports, to imports, cf. the definition in Table 2 above. Net exports is a proxy for the current account, so this ratio can be interpreted

42、as a proxy for debt service minus the current account, divided by imports: (DS CA)/I. The fraction of debt service not financed by the current account must be financed by a combination of new debt issuance, foreign direct investment flows, other new portfolio investment flows, and income on foreign

43、assets. The first three of these four components, which normally account for the majority of the capital account, may be reduced significantly in the event of a sudden stop of capital inflows. Thus, the numerator in the current account leverage variable represents an approximate upper bound for the

44、shortfall in the receipt of foreign currency income the country would face in the event of a sudden stop.資產(chǎn)負債表的影響、外部波動和新興市場Samuel W. MaloneUniversity of the Andes本文研究新興市場利差的決定因素以及新興市場國家的借貸成本，使用的數(shù)據(jù)來源于摩根大通公司所提供的19個國家的新興市場的債券指數(shù)。為了保證傳統(tǒng)傳播的決定因素，我們集中研究那些受近期影響的較大三個額外因素：外部沖擊的影響，資產(chǎn)負債表的實際貶值以及當前賬戶的杠桿程度。我們找到明確有力

45、的證據(jù)表明，上述類別的變量與經(jīng)濟利差和統(tǒng)計學有重大的關系。尤其值得注意的是，我們發(fā)現(xiàn)了貿(mào)易條件波動和賬戶杠桿水平對利差變化性所起的重要作用。賬戶杠桿的結(jié)果是使國家更容易受到資本流動突然停止的影響與國際投資者要求對外債的風險溢價之間建立緊密聯(lián)系的一個因素.關鍵詞：資產(chǎn)負債表效應，新興市場債券，外部波動，國家風險溢價。一、引言研究新興市場利差，被定義為研究新興市場債券和美國國債的收益率相同或相似成熟之間的差異，向前可追溯到愛德華茲（1984，1986）。債券息差反映了幾個因素。除了違約概率的基本債券和投資者認為他們將獲得回收率，利差可能還包括一個額外的保費與流動性的額外債率，以防債券的流動性差和市

46、場風險，甚至投資者意識到的稅收優(yōu)惠。這項研究中，我們著重關注高流動性，以美元為計價工具的新興市場國家，都是在國際市場交易，包括摩根大通所統(tǒng)計的新興市場債券指數(shù)。面板數(shù)據(jù)所研究的新興市場利差，專注于“經(jīng)典”決定主權風險，其中包括：羅蘭和托雷斯（2004）、安德雷斯泰爾（2000）、艾肯格林（1998）、蒙蒂（1998）、民（1998）和康托和培科（1996）。好的研究可以發(fā)現(xiàn)在索布林荷（2004）中所提到的“傳統(tǒng)”是決定主權風險利差的一個候選因素，在顯著統(tǒng)計學或經(jīng)濟學中至少一個研究過，包括：經(jīng)濟增長率，債務占國內(nèi)生產(chǎn)總值的比例，儲備占國內(nèi)生產(chǎn)總值的比例，債務占出口的比率，出口占國內(nèi)生產(chǎn)總值的的比

47、例，還本利息占國內(nèi)生產(chǎn)總值的比率，財政收支平衡率，國際利率，國家的歷史比率，國外凈資產(chǎn)與國內(nèi)通貨膨脹率。我們大多數(shù)人這些候因素作為控制變量在研究。 盡管上述文獻在理解新興市場利差的決定因素做出了貢獻，然而，理解資產(chǎn)負債表在影響已存在的資本突然停止流入的重要性，與宏觀經(jīng)濟波動在引發(fā)金融債務危機是很接近的。關于后者的主題重要的文獻，包括卡爾沃，基耶多，和塔爾維（2003），卡爾沃，基耶多，和梅希亞（2004），和卡桃和卡普阿（2006）。主要貢獻本文主要貢獻表現(xiàn)在，資產(chǎn)負債表和在經(jīng)濟學和統(tǒng)計學上的波動性對新興市場國家的利差有著顯著影響，在控制了各種其他因素對利差影響。特別是，我們建立對當前賬戶杠桿

48、有著強有力的聯(lián)系的變量，即突然停止的資本流入，來查看其對新興市場的利差的影響。我們定義的當前賬戶的杠桿，緊跟卡爾沃，基耶多，和梅希亞（2004）的工作，用外匯融資當前帳戶的百分比來衡量一個國家的進口的價值。這樣做的理由很簡單：給定一個變量即突然停止對外融資當前賬戶的資本注入，一個國家將被迫減少進口。進口采購的百分比降低是受突然停止資本注入事件脅迫所造成的。卡爾沃，基耶多，和梅希亞（2004）還強調(diào)：當前賬戶的杠桿綁定到實際匯率的貶值，這樣的突然降低是必須的。另一個最近的文獻，由貝爾岡查，昌，和加西亞侯茹柔（2004），直接試驗得到的令人意想不到的實際貶值主權利差的影響想，這是假定通過負資產(chǎn)的影

49、響，例如根據(jù)巴切塔和納吉（2004），和柯怡斯布德斯，椙和貝拉斯科（2000）的研究。貝崗莎，昌，和柯怡斯布德斯（2004）發(fā)現(xiàn)他們之間的相互作用和意想不到的實際貶值比外債占國內(nèi)生產(chǎn)總值的比率有著顯著的利差。在本研究中，我們表明除此之外，國家目前的高價值賬戶杠桿面臨更高的利差，即使考慮到了直接的資產(chǎn)負債表的影響，但由于貨幣貶值和債務負擔之間的相互作用沒有考慮到。因此，也有其他的原因，除了實際貶值外，也許涉及到外匯收入進口采購的替代性，這說明現(xiàn)在賬戶的杠桿借貸成本的重要性，我們發(fā)現(xiàn)，較高的條件的波動性是由于有較高的利差，這是密切符合卡桃和卡普爾（2006）相關的發(fā)現(xiàn):較高的條件的波動性是有較高的

50、違約發(fā)生率。這一發(fā)現(xiàn)是非常直觀的，因為較高的貿(mào)易波動是與較高的外匯收入波動相聯(lián)系的，和國家的更不穩(wěn)定收入流面臨更高的風險概率，其外債服務需求的資金超過其外匯儲備和短期的收入。這樣的國家，因此，面對高額的違約概率，從而可以要求債務有更高的溢價。我們的研究結(jié)果與近期希爾瑟和諾斯布希（2007）的工作是一致的，他們也發(fā)現(xiàn)波動性條件對利差有統(tǒng)計和經(jīng)濟意義的作用。 其他最近研究新興市場主權利差包括威斯特法倫（2001）和（2003）費魯奇。特別是，威斯特法倫（2001）發(fā)現(xiàn)，變化的波動性在過去20個交易日對當?shù)氐哪Ω康だ灰桌钪笖?shù)有著積極和顯著影響。這一結(jié)果提供了一個跡象，希爾瑟和諾斯布希（2007）和本文的波動性研究，基本可以成為一個重要的決定利差因素。值得注意的是，R值在我們各種規(guī)格的回歸測試范圍從71%到91%，明顯高于那些在先前的研究所。我們致力于解釋的利差變化的顯著程度，主要是為了我們的重點識別和測試，與此同時我們獨到的見解也已確定在文中，特別我們最近的工作的啟發(fā)對長期的貿(mào)易波動性和平衡性的因素的影響。二、本文的組織在第二部分中，我們勾勒出一個簡單的分析框架的理論決定新