# (Hebert and Schreger, 2017) --- * **Title:** The Costs of Sovereign Default: Evidence from Argentina * **Authors:** Hebert, B. and J. Schreger * **Journal:** American Economic Review * **Year, Volume, Issue, Pages:** 2017, 107, 10, 3119-3145 * **Summary Created On:** April 4, 2020 ###### tags: `papersummary` `macro` `sovereign default` `kunal` `readinggroup` --- ## Thoughts * Other fascinating results from the literature: output falls in anticipation of a sovereign default and the default itself tends to mark the beginning of recovery * Bilateral trade between creditor and debtor countries falls after default * During a sovereign default, external credit to the private sector is reduced * What is the output cost of default? Seems to be the major question * Figure 1, Panel A: Why are the non-event dots downward sloping? Presumably the grey dots are tracking changes in risk neutral default probabilities that occur from other sources... what other sources? * Even though there are 15 court rulings that the authors consider, these rulings are not made equal: those from the district court or circuit court are much less "final" for one ## Question * Does sovereign default itself cause harm to an economy? * It's a hard question to answer because of endogeneity concerns: poor condition of the economy may lead a country to default on its debt ## Results * A 10% increase in the risk neutral probability of default leads to a 6% reduction in log equity returns of Argentian securities (measured by a value weighted index of ADRs on a few large Argentinian firms...but other classes of securities also studied) * Significant heterogeneity by industry--banks are more likely to be affected, real estate and non-financial firms less so * **Mechanisms**: study characteristics of firms most affected to find the main channel through which sovereign default affects equity return * Firms that export (foreign creditors may interfere with exports) * Firms that rely on imported intermediate goods (may not be able to secure financing to import, so have to rely on domestic intermediate goods) * Balance sheets of banks * Foreign-owned firms * Authors find that foreign-owned firms underperform (react more negatively to the default probability shock than would be predicted by the exchange rate or other market forces). This hints to investors losing confidence (...?) * Firms whose primary industry is export-intensive underperform... large firms underperform more than small firms... but the latter (and actually, the former if authors are not controlling for size), may be because the equity for smaller firms may be more illiquid ## Data and Identification * Outcome: Value of Argentinian equity * IV: legal rulings (on whether Argentina would be forced to pay NLM alongside its restructured debt) are exogenous shocks to the risk-neutral probability of default * Identification assumption: information revealed by legal rulings affects equity value only through its impact on the risk neutral probability of default * The identification assumption requires that US judges do not have private information about the Argentinian private economy * 15 court rulings between December 2011 and July 2014 that potentially changed the risk neutral probability of default * Threat to identification: stock returns might affect the risk neutral probability of default OR that common shocks may affect both * This is particularly relevant because we are only considering stock returns of large firms, and large firms contribute a ton to the economy * Two-day event windows: measure changes in risk neutral default probability and value of ADRs ## Theory (Main Model) * **Heteroskedasticity-based identification strategy** * Consider the two relevant equations: $$ \Delta D_t = \gamma r_t + \kappa_D F_t + \varepsilon_t \\ r_t = \alpha \Delta D_t + \kappa F_t + \eta_t $$ * Here, court rulings are modeled as shocks to $\varepsilon_t$. Parameter of interest is $\alpha$ * The heteroskedasticity-based identification strategy does not require the complete absence of (1) common shocks and (2) idiosyncratic shocks (the two threats to identification) * Rather, it relies on the following identification assumption: Variance of the common shocks $F_t$ and equity return shocks $\eta_t$ -- same on event and non-event days * However, the variance of $\varepsilon_t$ is higher on event days than non-event days * Identify $\alpha$ by comparing the covariance matrices of abnormal returns and abnormal default probability changes on event and non-event days * ![](https://i.imgur.com/F5BZWIc.png) * The identification is measured off from differences in the slopes... because of the cov/ var terms -- Figure 1 is key for the econometric identification * The linearity of the relationship between default probability and equty size is probably overstated? ## Background and Facts * American Depository Receipts: allow American investors to invest in foreign equity; US depository banks buy foreign equity and then give certificates to domestic buyers * The Argentenian ADRs are traded on the NYSE and the NASDAQ and are relatively liquid... but only 12 of the large Argentian firms represented. So the analysis also looks at the domestic stock exchange in Argentina * A lot of assumptions here that I do not have the expertise to understand: why the cumulative default probability for 5 years, for example, although footnote 14 is useful * A Credit Default Swap is a financial security. The seller of the swap agrees to insure the buyer if the issuer of debt defaults on its obligations. * Sovereign defaults also risk currency value * Peso was a non-convertible currency (like INR) between 2011-2014 (our period of analysis)... so the exchange rate is measured as the one prevailing on the black market for USD * Figure 3: Wtf... the risk neutral probability of default for Europe (benchmark here) is 66%???