ESSAYS IN MACROECONOMICS AND INTERNATIONAL FINANCE
Emiru, Tilahun Molla
Only about 2.5% of the total external debt of emerging markets was issued in local currencies by the late 1990s (see Eichengreen et al., 1999). The rest was denominated in foreign currencies, predominantly in U.S. dollars. Recent data, however, shows that about a quarter of the total external debt of emerging markets is denominated in their own local currencies (see Arslanalp and Tsuda, 2014). Does this change in the currency of denomination of external debt alter the default incentives of emerging markets? The first two chapters of this dissertation provide an answer to this question. The third chapter of this dissertation examines another question that has gained considerable attention in the past decade – has monetary policy been one of the causes of rising inequality in the United States? A few decades ago, virtually all of the external debt of emerging markets was denominated in foreign currencies. Located at the heart of many emerging-market crises in the 1990s, this skewness in the currency of denomination has been termed ‘the original sin’, Eichengreen et al. (1999). Nowadays, many emerging markets are able to issue external debt in their own currency. How does this change in the currency of denomination alter default incentives? I show, empirically and theoretically, that an increase in local currency debt decreases the likelihood of default. The first chapter begins by showing that there is a statistically significant positive correlation between higher local currency external debt and better credit ratings, a proxy for the probability of default, and hence lower likelihood of default. This relationship between local currency debt and credit ratings could, however, suffer from endogeneity issues. One potential source of endogenity is reverse causation – do countries issue more local currency debt because they have a lower likelihood of default or do they have a lower likelihood of default because they issue more debt in the local currency? Hence, estimating the causal effect of local currency debt on default probabilities requires a more solid identification strategy. To study the causal effect of local currency debt on the likelihood of default, I estimate a tri-variate Structural Vector Autoregression (SVAR) model. Two rather mild assumptions are needed to identify the structural shocks: that the U.S. interest rate does not respond to the interest rate as well as the local currency external debt of emerging markets, and that the local currency external debt of emerging markets does not contemporaneously respond to the U.S. interest rate. Having imposed these identifying assumptions, I find that sovereign spreads shrink by about 40 to 60 basis points in response to a positive, one standard deviation shock to the local currency share of external debt indicating that the likelihood of default falls. The second chapter studies the relationship between local currency external debt and default likelihoods theoretically. I build a model of sovereign default where there are two currencies of denomination - the home currency and the foreign currency. A bond denominated in the foreign currency is an I.O.U to repay one unit of the foreign currency the next period whereas a bond denominated in local currency pays one unit of the home currency the next period. A benevolent government issues these bonds in the international capital market to smooth the consumption of a representative household that faces shocks to the endowment of traded goods. Moreover, the government prints money which it can use to repay local currency denominated external debt. Default occurs in response to a sufficiently large negative endowment shock, and when it does, the government defaults on all of its debt (i.e., both denominations). There are two costs of default – exclusion from international markets and output costs. Exclusion from international markets lasts for a random number of periods, and upon re-entry, the sovereign’s entire stock of debt is erased off the books. I then solve the model and simulate two model economies that are otherwise identical except for the difference in the size of their local currency external debt. One economy has 90% of its external debt in the local currency while the other economy has only 10% of its external debt in the local currency. The default rates, as well as other business cycle statistics of the simulated models, are then compared. The results point to one major conclusion – an economy’s stock local currency external debt is an important determinant of whether it will experience a debt crisis. In the model, an increase in the share of local currency debt from 10% to 90% leads to a decrease in the default rate from 4.6% to 0.3%. An important mechanism is behind this result. When there is sizable local currency debt, the government resorts to costly deficit monetization in response to negative output shocks before it exercises the option to default. However, given that inflation engenders welfare cost, outright default will still occur even when most of the external debt is denominated in the local currency. Thus, while the presence of local currency leads to lower default probabilities, it does not imply that outright default completely disappears. In addition to its prediction about default probabilities, the quantitative model is able to replicate some stylized facts about emerging market business cycles. In particular, interest rates and net exports are counter-cyclical, interest rates and net exports are positively correlated, and the standard deviation of consumption is higher than the standard deviation of income, which are all standard business cycle features of emerging markets. The third chapter of this dissertation empirically studies the effect, or the lack thereof, of monetary policy on inequality. The increase in inequality in the United States, while it is part of a trend that began decades ago, has attracted a renewed interest recently. With this attention has come the assertion that monetary policy has been one of the prime drivers of inequality in income and wealth. There are various channels through which monetary policy potentially affects inequality in income and consumption, and some of these channels conceivably work in opposite directions making a comprehensive empirical study on the topic even more important. This chapter provides one such study. The data come from Consumer Expenditure Survey (CEX) and other secondary data sources and run from the first quarter of 1984 to the last quarter of 2014. I construct three measures of income and consumption inequality – Gini coefficient, cross-sectional standard deviation, and percentile ratios – and use them in Vector Autoregression (VAR), Jorda’s local projections, and proxy VAR models. Monetary policy in this study covers both conventional and unconventional types. To study the effects of these two policies, I use fed funds rate and rates on two-year treasury notes, respectively. In the local projections and proxy VAR exercises, I also use exogenous monetary policy shocks, proposed by Romer and Romer (2004) and Gertler and Karadi (2015). Estimation of the three models results in qualitatively similar results. First, The impulse responses and the associated confidence bands from the three analyses indicate that while there are some borderline significant effects on income and earnings, the effect of monetary policy, both conventional and unconventional, on income and consumption inequality is generally very weak, if there is any. Second, the results from local projections estimation indicate that monetary policy tends to increase income inequality in the upper half of the distribution and not in the bottom half of the distribution. However, neither in the top nor in the bottom of the consumption distribution does monetary policy have any significant effect. While I believe that this study will contribute to the growing literature on monetary policy and inequality by empirically documenting the relationship between monetary policy and inequality, at least one caveat is in order. Recent studies have shown that data from the CEX suffer from measurement errors. Thus, the inequality measures based on the CEX data are to be taken with caution. Moreover, earlier theoretical work on the topic focused on one particular link between monetary policy and inequality. A more general theoretical framework would be a valuable addition to the literature.
Inequality; International Finance; Macroeconomics; Monetary Policy; Sovereign Default
Prasad, Eswar; Mertens, Karel
Ph. D., Economics
Doctor of Philosophy
dissertation or thesis