Understanding Corporate Vulnerabilities in Latin America

This paper analyzes the potential risks and vulnerabilities of non-financial corporates in Latin America and Canada. We quantify the impact of company-specific, countryspecific, and global factors in driving corporate spreads. Overall, we found that all these factors play a role in explaining corporate risk. In particular, country specific factors such as exchange rate and sovereign CDS spreads are significantly associated with changes in corporate spreads, underscoring the importance of solid policy frameworks. We also find that global conditions, such as the VIX, are dominant drivers of corporate spreads. In recent years, the adverse effects from deteriorating domestic conditions have been broadly offset by relatively bening global financial conditions. However, a sustained reversal in these conditions would put significant pressure on corporate risk.


This paper analyzes the potential risks and vulnerabilities of non-financial corporates in Latin America and Canada. We quantify the impact of company-specific, countryspecific, and global factors in driving corporate spreads. Overall, we found that all these factors play a role in explaining corporate risk. In particular, country specific factors such as exchange rate and sovereign CDS spreads are significantly associated with changes in corporate spreads, underscoring the importance of solid policy frameworks. We also find that global conditions, such as the VIX, are dominant drivers of corporate spreads. In recent years, the adverse effects from deteriorating domestic conditions have been broadly offset by relatively bening global financial conditions. However, a sustained reversal in these conditions would put significant pressure on corporate risk.

I. Introduction

High commodity prices and strong international liquidity have powered growth in LAC for much of the last decade, but there is now consensus that this twin-engine growth process has come to an end. As the region adjusts to a harsher external reality, several challenges and risks have surfaced. A key one relates to the health of nonfinancial corporates. This topic that has attracted growing attention from markets and policymakers, but there remains ample scope for further analysis.

Against this backdrop, this paper focuses on the drivers of corporate risk in Latin America using company CDS spreads. We put together a comprehensive dataset to study the determinants of corporate risk dynamics. In particular, we quantify the relative contributions of firm-specific fundamentals, domestic macroeconomic conditions, and global factors. This allows us to shed light on issues such as the role of exchange rate depreciations and links with the sovereign risk.

Overall, the paper finds that all three set of variables – company-specific fundamentals, country-level macroeconomic variables, and global factors – play an important role in explaining corporate risks. The firm-specific variables are divided in two different sets: a set containing company specific-fundamentals, primarily based on data reported in quarterly financial statements, and a second set containing market-based company-specific variables. We separate these two concepts as our variable of interest – implied CDS spreads – itself depends on market sentiment which is closely captured in the second set. Thus, we would like to disentangle the quantitative impact of balance sheet reported-and-accounting-based measures (i.e. “pure fundamentals”) from those of market-based explanatory variables, which themselves are affected by changes in financial market sentiments.

Among firm fundamentals, we find that measures of profitability, capitalization, leverage, and liquidity all appear to be statistically significant drivers of corporate spreads. At the country level, factors such as the exchange rate and the country’s sovereign CDS spreads play an important role at driving our corporate risk measure. Finally, global conditions, in particular, global risk aversion (proxied by the VIX) is a key factor in driving swings in corporate spreads throughout Latin America.

Our results are broadly in line with previous results found in the literature, though most of the studies focus on explaining corporate spreads in advanced countries. In particular, there is a vast literature examining U.S. CDS spreads. For instance Das et al (2009) looks at CDS spreads of non-financial corporates using a mixture of accounting-based and market-based variables as explanatory variables. They find that accounting-based variables are able to explain two-thirds of CDS spreads movements, comparable to market-based variables. Unlike bond spreads, Das et al (2009) suggest that CDS spreads are not affected by tax effects. Doshi et al (2013) find that macroeconomic and firm-specific information can explain most of the variation in CDS spreads over time and across firms, even with a parsimonious specification. Those findings are also confirmed by Tang & Yan (2013), whose findings imply that firm-level and market-wide variables have similar levels of explanatory power on CDS spreads.

Some studies use alternative measures of corporate spreads, for instance, corporate bond spreads. These measures tend to raise important issues of comparability across instruments, and require pairing yields of corporate bonds to those of government bonds of similar characteristics (i.e. both maturity and cash flows), which are seldom readily available and, thus, represent an important practical limitation for empirical analysis. Examples of this research branch include Elton et al (2001) and Avramov et al (2007). Elton et al (2001) look at the difference in the rates offered on corporate bonds and those offered on government bonds over the period 1987-1996 using monthly bond data extracted from the Lehman Brothers Fixed Income Database distributed by Warga (1998). For comparability and to ensure the no-arbitrage condition,2 the spreads used by Elton et al (2001) relate to zero-coupon bonds of the same maturity, which need to be estimated, using the procedure proposed by Nelson and Siegel (1987). Those spreads are then decomposed along three components: expected default loss (based on transition matrices estimated by S&P and by Moody’s); a tax premium (as interest payments on corporate bonds are subject to state taxes in the U.S. whilst government bonds are not); and a systemic risk premium (based on the three-factor model of Fama and French, 1993). The authors find that expected default loss and the tax premium explain less than half of the measured corporate bond spreads, whilst the systemic risk premium explains the majority of the residual risk on these corporate bonds.

Likewise, Avramov et al (2007) rely on corporate bond spreads computed by Datastream, which are constructed as the yield differential between the corporate bond and the U.S. Treasury curve, but accounting for the maturity and the compounding frequency of these bonds. The authors find that a parsimonious set of aggregate company-level variables, inspired from structural model (e.g. idiosyncratic volatility, price-to-book ratio, etc), explain large part of the variation in corporate spreads. They also find evidence of the existence of a common systemic factor in the variation in corporate-spread changes.

In the case of emerging markets, Cavallo and Valenzuela (2007) look at corporate bond spreads of foreign currency-denominated bonds for a set of six Latin American countries and four emerging Asian economies. In order to compare bonds with different cash flow characteristics on a more equal basis, these authors rely on option-adjusted spreads (OAS) from Bloomberg. They find that firm-level characteristics account for the larger share of the variance. In addition, they find that a transfer of risk from the sovereign to the private sector exists, but it is less than1 to 1. Furthermore, their findings are consistent with the popular notion that panics are common in emerging markets, owing to less informed investors that are more prone to herding.

Our paper is organized as follows. Section II presents some stylized facts regarding the evolution of corporate risk, firm fundamentals, and relevant regional trends, whereas Section III describes the dataset and the empirical methodology used for the estimation of the main model. Section IV presents and discusses the main findings, including policy implications. Section V concludes.

II. Stylized Facts: Recent Dynamics in Firm Spreads and Fundamentals

During the last decade, nonfinancial corporates from financially-integrated LAC economies have benefited from a favorable funding environment. New companies gained access to international capital markets, and many have been able to lengthen debt maturities while lowering borrowing costs. This has allowed corporates to pursue new investment plans, improve cash buffers, and pay down more expensive debt. In principle, these are all positive developments for a savings-scarce region characterized by low investment rates.

However, this has been achieved at the expenses of higher leverage, significantly fueled by foreign currency debt. For instance, foreign currency bond debt in five major financially-integrated economies of Latin America (Brazil, Chile, Colombia, Mexico and Peru; “LA5” hereafter) has increased from US$ 170 billion to US$ 383 billion between 2010 and 2015 (Table 1). Moreover, the macroeconomic adjustment unfolding across the region has implied weaker domestic currencies and lower medium-term growth rates, features that are likely to persist. Hence, the favorable funding environment over the last decade also bred vulnerabilities which are coming to the fore. Not only companies are facing a currency adjustment effect in their debt stock, but also the prospects of growing out of their debt through high-return investment have diminished.

Table 1

Corporate Bond Issuance in LA5 Economies

Non-Financial Corporates: Bond Debt

article image
article image
Source: Dealogic and IMF staff calculation.

Against this backdrop, there has been growing concern about corporate risk, with market analysts and policymakers combing through scattered information, often relying on anecdotal evidence. While this approach can lead to useful insights, it is important to complement it with attempts to develop a more systemic view about the evolution of corporate vulnerabilities.

To that end, we would a need a measure of corporate risk that retains some homogeneity across a reasonable number of firms. Credit Default Swaps (CDS) spreads would be the ideal candidate, but unfortunately they are available only for a small number of firms in LAC. Thus, we turn to implied CDS spreads which closely track their market counterpart measure and are available for a much larger set of companies (Figure 1).3 4

Figure 1.
Figure 1.

Actual Vs Implied Corporate CDS Spreads

Citation: IMF Working Papers 2016, 080; 10.5089/9781484321546.001.A001

Implied CDS spreads indicate that 2015 has indeed been a year of rising corporate risk for the average Latin American corporate. Moreover, these spreads also exhibit a more permanent deterioration component since 2011 (Figure 2). The dynamics of the last four years contrasts with what has been observed during the Global Financial Crisis, characterized by an acute and short-lived spike in corporate risk.

Figure 2.
Figure 2.

Evolution of Corporate Spreads in Latin America

Citation: IMF Working Papers 2016, 080; 10.5089/9781484321546.001.A001

Examining the evolution of implied CDS spreads across countries is revealing. The year of 2011 marks the start of diverging behavior for the median corporate across countries as Argentina and Brazil decouple from other economies, displaying persistently higher corporate risk ever since. This stylized fact suggests that from 2012 onwards, idiosyncratic factors have had a more influential role in driving risk. Since 2014, this heterogeneity has also grown amongst other countries. The econometric analysis presented later in this paper will propose a strategy for comparing competing drivers behind corporate risk dynamics.

Corporate risk deterioration has been accompanied by weakening firm fundamentals. Indeed, the data reveals that leverage, profitability, capitalization and liquidity have weakened alongside with the implied CDS spreads since 2010. Table 2 provides a snapshot of several indicators based at three particular points in time over the past five years: (i) 2010Q1, (ii) 2011Q3, and (ii) 2015Q3.

Table 2

Evolution of Implied CDS and Selected Firm Fundamentals in Latin America and Canada

article image
Source: Bloomberg and IMF staff calculations.

The deterioration has been stronger in recent years and is more visible on the dimensions of leverage and profitability. This partly reflects the combination of exchange rate depreciations, foreign-currency debt, and marked-down growth prospects. In this context, the specter of sudden crises led by sharp exchange rate corrections has re-emerged. Basically, the typical concern is that companies would have indulged in financial excesses during good times, much beyond of what their real growth potential could justify, and it is now a matter of time until a corporate bust ensues.

Such a scenario remains a possibility, but the corporate resilience to currency depreciations and medium-term growth projections revisions observed in 2015 suggests that fatalism is not in order. In fact, ex ante, many would probably have considered the macroeconomic shocks during 2015 as being sufficient to trigger widespread corporate defaults in the region. Ex post, a few factors appear to have been able to stop or at least delay turmoil. High levels of international reserves provided confidence boost and ammunition for Central Banks to stabilize short-term fluctuations of the currency–though the medium-term effects and sustainability of the interventions have remained a topic of discussion. An important part of the dollar debt build-up has been accumulated in the tradable sector and by quasi-sovereigns, so natural hedges and implicit government backing have been important mitigating factors (Table 3). Also, cash buffers have remained sizeable in recent years until now, and at the same time some Latin American companies might be making more active use of financial hedges.

Table 3

Sectoral Composition of Bond Issuance in LA5 Countries

Non-Financial Corporates: Bond Debt 2015 Breakdown

article image
Source: Dealogic, Bloomberg and IMF staff calculation. Sectoral shares and company-specific data from Bloomberg.

It would be wrong to take comfort on the 2015 corporate resilience, and fail to acknowledge that margins have been stretched thin already. The interaction of high leverage and declining profitability with a more challenging environment for macroeconomic management and longstanding growth challenges is an insidious source of corporate risk for the region. In particular, a protracted period of slower growth can lead to the erosion of existing buffers may increase short-term bias in policies without addressing long-term adjustment challenges. Hence, sound macro policies and effective growth strategies are key for mitigating risks to corporate risk.

III. Dataset and Empirical Methodology

This section is split in two sub-sections. The first one describes our dataset and the construction of the main variables used in the analysis. The second one presents the empirical methodology used to estimate our core model and derive the main results.

A. Dataset

The chapter builds a large quarterly dataset covering the period 2005–2015 containing company-specific financial information, along with country and global variables. The sources are Bloomberg, Datastream, Haver, Markit, and the World Economic Outlook Database. The sample includes over five hundred nonfinancial firms from seven Latin American countries—Argentina, Brazil, Chile, Colombia, Mexico, Panama, and Peru. 5 Furthermore, we perform additional analysis by including a similar number of firms from Canada. While the analysis centers on Latin American firms, the inclusion of Canadian companies allows the investigation of the role of common regional shocks by providing a benchmark of a commodity-exporting advanced economy located in the same hemisphere.

An important issue in the analysis of corporate spreads using both accounting-based and market-based variables is the selection of the time series frequency. Most financial reports are available only at a quarterly frequency; however market-based variables are readily available on a daily basis. Furthermore, there does not seem to be a consensus in the literature as to which frequency is preferable,6 and most empirical work do not provide an explanation backing their specific choice. Basing the analysis on a daily-frequency dataset, constructed from interpolated quarterly data when needed, has the advantage of increasing the number of observations, and hence the power of the estimations. However, the use of (linearly) interpolated daily series might give rise to important econometric drawbacks. For instance, where the presence of endogeneity is suspected, the researcher would use lags of the explanatory variables so that the information contained on the “right hand side” is predetermined vis-à-vis the dependent variable, and thus the error terms. In the case of daily data from interpolated quarterly series, every observation at a given point in time embeds information from the past as well as from the relative future, which (depending on the day in question within each quarter) might be several days ahead.7

Conversely, the use of quarterly data, constructed from averaging daily observations within each quarter, does not suffer from this potential time-related endogeneity issue, however it might “smooth away” important and interesting high-frequency dynamics in the underlying data. Weighting the pros and cons, and being mindful of the econometric drawbacks in using linearly interpolated data, we prefer to err on the cautious side and opt for using quarterly data for the estimation of our core model. Nevertheless, given that our dependent variable – market-based implied CDS spreads – is available on a daily frequency, we use the observation for the last day in every quarter instead of the simple average throughout the quarter. This ensures that our explanatory variables, which are based on averages within the quarter, precede the value of our dependent variables at every quarter in the estimation model.8

Another important aspect when estimating a relatively rich model that embeds a large number of firm-specific variables in a panel setting is that several of these variables might be highly collinear, both along the time and the cross-sectional dimensions. In order to mitigate potential estimation issues arising from this multi-collinearity, we collect firm-specific variables into conceptually-related groups. In addition, we use principal component analysis (PCA) to create variables for each of the following four concepts: profitability, capitalization, leverage, and liquidity.9 10

The set of country-level variables in the core model include the country’s sovereign CDS spread, the year-on-year rate of CPI inflation, and the depreciation rate (annual) of the bilateral exchange vis-à-vis the U.S. dollar. The commodity terms-of-trade index of Gruss (2014) is also included among the explanatory variables. Although the latter is usually viewed as being outside the control of country-specific policies (as it depends on global commodity prices), the variable itself varies from country to country.

Finally, the VIX – which can be seen as both a measure of global financial markets volatility, but also as a measure of global risk aversion (i.e. market price of risk) – represents our global conditions variables.11

Descriptive statistics of all the different variables included in the core model are presented in Table 5, whereas Table 6 exhibit the pair-wise (unconditional) correlations among these variables. These simple correlations already point at clear interesting relationships in our dataset. In particular, an increase in corporate risk appear to be strongly associated with an increase in corporate leverage, in share price volatility, in sovereign CDS spreads, in the VIX, or a depreciation of the exchange rate. Conversely, lower corporate risk appears to be strongly associated with higher profitability, capitalization, liquidity, share prices, and price-to-book ratios, as well as with higher commodity prices. Interestingly, most of the variables included in our sample also appear to be highly correlated among themselves. This reinforces the view that a richer structure, in which several measures are included simultaneously, is needed to properly assess the impact of each of these measures on corporate spreads.12

Table 4

First Principal Components of Firm-Specific Fundamentals

article image
Source: Bloomberg and IMF staff calculations.
Table 5

Descriptive Statistics of Variables included in Core Estimation Model

article image
Source: Bloomberg and IMF staff calculations.Note: Latin American Countries include Argentina, Brazil, Chile, Colombia, Mexico, Panama, and Peru.
Table 6

Unconditional Cross-Correlations of Variables included in the Estimation Model

article image
Source: Bloomberg, Haver, International Financial Statistics, and IMF staff calculations.Notes: *** denotes statistical significance at the 0.1 percent level; ** denotes statistical significance at the 1 percent level; and * denotes statistical significance at the 10 percent level.

B. Empirical Methodology

Our main estimation relies on a standard panel-data setting in which our dependent variable – market-based implied CDS spreads – is modeled as a linear function of firm-specific variables, macroeconomic factors at the country-level as well as global conditions.

Algebraically, our estimation model can be written as:

Yi,t = α + β1Fi,t + β2Mi,t + β3 Cj,t + β4 Gt + β5Drt + μi + i,t

where Yi,t denotes the log of implied CDS spread of company i at time t, our measure of corporate risk; Fi,t and Mi,t denote, respectively, firm-specific accounting-based variables (that is, ‘fundamentals’) and market-based variables; Cj,t denotes macroeconomic variables in country j at time t, whereas Gt represents the global variables; μi denotes the company-specific fixed effects;13 and ∊i,t is the error term. Drt represents time dummies for two different subperiods: financial crisis (2008:Q1 to 2010:Q4) and the subsequent period (2011:Q1 to 2015:Q3). They capture changes in dynamics induced by ‘level shifts’, beyond what could be explained by variables in our data set. In addition, these dummies are allowed to be different between Canada and the group of LAC countries (thus the subscript for the region r), allowing one to investigate common LAC-regional factors driving risk.

The estimation methodology is based on the standard fixed effects (i.e. “within groups”) estimation technique. Robust standard errors clustered by country - to account for any potential correlation of the residuals within each country – are used in the core estimation model.14

IV. Main Results and Discussion

Overall, our estimation suggest that all groups of variables – firm-specific, country-level macroeconomic and global conditions – play an important role in explaining developments in corporate spreads in Latin America, as well as in our extended sample which includes Canada (see Table 7).15

Table 7

Core Model – Estimation Results

article image
article image
Note: Robust standard errors in parentheses - *** p<0.01, ** p<0.05, * p<0.1.Source: IMF staff calculations.Notes: CAN = Canada. LAC includes Argentina, Brazil, Chile, Colombia, Mexico, Panama, and Peru. LA5 includes Brazil, Chile, Colombia, Mexico, and Peru.

In terms of firm-specific measures, we find that all the accounting-based firm fundamentals included in our model appear to be statistically significant. Indeed, higher capital ratios, higher liquidity ratios, and higher profitability all lead to a reduction in corporate risk. Conversely, higher leverage is significantly associated with higher corporate risk.

Although these results are both strong from a statistical standpoint and intuitive from an economic point of view, the elasticities obtained relate to the principal components of four “fundamental concepts”, rather than the underlying fundamentals per se. The individual elasticities cannot be obtained directly from Table 7. In order to assess the impact of each individual fundamental (e.g. ROE, ROA, debt to equity, etc) on corporate risk, the estimated elasticities can be used in conjunction with the weights of each of the underlying fundamentals in their respective principal component (as per Table 4). For instance, an increase in the ROA ratio of 1 percentage point increases the “profitability” principal component by 0.10 units,16 which in turn reduces the average corporate implied CDS spread in Latin America by about 2 basis points.17 Likewise, a 5-percentage-point increase in the capital-to-assets ratio increases the “capitalization” component by 0.08 units, thereby reducing corporate spreads by about 6 basis points.18 Although statistically speaking these elasticities are highly significant, their magnitudes are admittedly small. The relatively small magnitudes reflect the fact that these elasticities represent an average linear effect over a large sample of heterogeneous firms. Indeed, the intrinsic elasticities might be very large for a few companies and effectively zero for others. 19

Turning to market-based variables at the company level, a few of these measures appear to be statistically significant in explaining corporate risk. In particular, company-specific stock price volatility and price-to-book ratios are found to be strongly associated with changes in implied CDS spreads. Even though these variables are linked to the calibration of the implied CDS spreads itself, including them in the core regression is not tautological. In fact, the variables in Block II are also relevant in explaining actual CDS spreads dynamics.20 Moreover, these variables are incorporated with a lag in the estimation model, thus reducing any potential circularity with implied CDS spreads. Essentially, we want to know if – and by how much – other variables influence corporate risk after accounting for the market-based variables.21 The estimated results suggest the answer is yes.

In addition to firm-specific fundamentals, macroeconomic and financial conditions both at the country-level and globally are found to be statistically associated with changes in implied corporate CDS spreads. In particular, higher sovereign CDS spreads are significantly associated with higher corporate spreads. This corroborates the existence of an important nexus between corporate and sovereign risk in the region: changes in sovereign spreads tend to lead to movements in corporate spreads in their respective countries. Hence, to the extent that policy frameworks affect country risk dynamics, they also have a direct impact on corporate risk.

Similarly, corporate spreads appear to rise when the exchange rate of the country depreciates. As expected, this effect appears to be stronger for the companies that exhibit higher levels of leverage. Importantly, we found that year-on-year changes in exchange rate play a more important role in explaining corporate spreads than the exchange rate level per se. This suggests that companies are not necessarily affected by underlying trends in the level of the exchange rate (for instance, when the exchange rate is continuously depreciating, albeit smoothly), as balance sheets would tend to adjust, but would suffer from a sharp and sudden depreciation.

We also find that higher inflation is associated with higher corporate risk.22 However, other macroeconomic fundamentals such as real GDP growth do not appear to play a significant role in driving risk in our core model. Of course, the latter is likely to still have indirect effects on corporate spreads through the firm-specific fundamentals included in our model. For instance, higher output growth is significantly associated with higher profitability, which is in turn an important driver of corporate spreads.

Global factors such as commodity prices and, in particular, the VIX are dominant drivers of corporate spreads. This is not surprising given that the VIX is often considered as a proxy for global risk aversion – or the market price of risk. In other words, during periods or episodes where the VIX is high, investors require a higher return in compensation for the higher perceived risk (i.e. a risk premium), which translates into higher corporate risk. Given the importance of the commodity sector for most of the economies in our sample, it is also not surprising to see that higher commodity prices tend to be associated with lower corporate spreads. Again, more ‘fundamental’ global measures (such as global output growth) do not appear to be statistically significant drivers of corporate spreads in our core estimation model. But, as in the case of country-specific fundamentals, these global factors are highly correlated with other important explanatory variables in the regression, such as global commodity prices and the VIX itself.23

Finally, two level shifts were found to be highly significant in our regression model. The first corresponds to an increase in average level of corporate implied CDS spreads during the global financial crisis, proxied with a dummy for the period 1Q-2008 to 4Q-2010. Interestingly, the magnitude of this level shift was found to be essentially the same for all countries in our sample, except for Canada. The estimated level shift represented a sizeable increase in corporate spreads of about 77 basis points for all the Latin American countries in our sample, and 41 basis points in the case of Canada. Basically, the differential could be seen as a risk premium that Latin American corporate would need to pay relative to their Canadian counterparts in times of stress. A second level shift was found over the period starting in 1Q-2011 till the end of our sample (3Q-2015), which marked a period of continuous – albeit moderate – economic softening in the region, and in emerging markets more broadly.24 Again, this second dummy represented an additional increase of about 55 basis points in corporate spreads throughout Latin America, and about 35 basis points for Canadian corporates.

Overall, the estimation results in Table 7 reflect a set of parsimonious models, in which the different blocks comprise a few variables with similar characteristics (i.e. firm-specific, country-specific, or common to the entire sample). Furthermore, the estimated elasticities are fairly robust to the exclusion of specific blocks or the use of time effects instead of macroeconomic variables (Table 8). Nevertheless, most of these variables, in particular those that present mainly a time series variation – such as the domestic and global condition variables – tend to exhibit a relatively large degree of co-movement. Thus, the conditional elasticities represent only the direct impact that these variables have on corporate spreads, as the former might still have an indirect impact on corporate spreads through other variables that are also included in the model.

Table 8

Robustness of Estimation Results

article image
Note: Robust standard errors in parentheses - *** p<0.01, ** p<0.05, * p<0.1.Source: IMF staff calculations.Notes: CAN = Canada. LAC includes Argentina, Brazil, Chile, Colombia, Mexico, Panama, and Peru. LA5 includes Brazil, Chile, Colombia, Mexico, and Peru.

To quantify the overall (i.e. direct and indirect) effects of these domestic and global conditions on corporate spreads, we conduct further regression analysis in which we omit all other time-varying variables as well as including and excluding the firm-specific variables from our core model (Table 9). These additional elasticities might be useful, for instance, when performing stress tests based on scenarios in which each exogenous variable is individually shocked. Furthermore, as mentioned earlier, there are other variables such as the country’s real output growth, G7 real GDP growth (as a proxy of global growth), or the level of the S&P-500 (as a proxy of global equity prices), that do not appear to be statistically significant drivers of corporate spreads in our core model. This is likely due to their high correlation with other explanatory variables in our model. However, in more restricted estimation settings, or when these are introduced individually, their corresponding elasticities appear to be statistical significant. For completeness, these estimation results are presented in Table 10.25

Table 9

Impact of Domestic and Global Macroeconomic Factors - Estimation Results

article image
Note: Robust standard errors in parentheses - *** p<0.01, ** p<0.05, * p<0.1.Source: IMF staff calculations.Note: these estimation results are for the sample of Latin American firms.