SOME INTERNATIONAL FINANCIAL CONTRIBUTIONS: EMPIRICAL RESULTS AND POLICY IMPLICATIONS

Transcription

1 Author: Amedeo Amato Università degli Studi di Genova, Italia SOME INTERNATIONAL FINANCIAL CONTRIBUTIONS: EMPIRICAL RESULTS AND POLICY IMPLICATIONS The contributions appearing in this issue of Economia Internazionale/International Economics deal with some relevant current questions concerning the relations between economic policy and international portfolio choices. The following topics are studied either from a general perspective or with reference to specific problems emerging in international finance: i)methodologies of assessing portfolio market risk in the BRICS economies; ii) bank competition, concentration and risk taking in banking industry; iii) the adjustment process towards equilibrium in the purchasing power parity hypothesis with particular reference to the use of linear and nonlinear attractors; iv) the accuracy of direct and indirect forecasting of cross exchange rates; v) some empirical analysis of the expectations hypothesis of the term structure of interest rates in reference to Korea after the Asian financial crises. 1. Over the past years the BRICS countries have become an attractive investment destination for asset managers and investors in search of high yield and opportunities for portfolio diversification but the volatile BRICS environment remains associated with high risks. The first contribution, Assessing Portfolio Market Risk in the BRICS Economies: Use of Multivariate GARCH Models by L. Bonga-Bonga and L. Nleya, compares the performance of the different models used to estimate portfolio value-atrisk (VaR) for portfolios that contain assets in the currency and equity markets in the BRICS economies. A large number of studies have concentrated on market risk modelling using multivariate GARCH models but few of these have applied this technique on emerging markets in general and on BRICS in particular. Moreover, none of these studies have analyzed the effects of different portfolio weights on the VaR for ECONOMIA INTERNAZIONALE / INTERNATIONAL ECONOMICS 2018 Volume 71, Issue 2 May, I-X

2 II A. Amato BRICS economies. The high volatile nature of emerging markets data raises a particular interest for VaR estimation based on GARCH models and for portfolio selection. As such this paper is the first to estimate VaR by using multivariate GARCH models and accounting for the effect of different portfolio weights on the VaR within BRICS economies. It compares the performance of three multivariate GARCH risk models, the DCC, ADCC and CCC, in estimating portfolio VaR for each of the five BRICS countries (Brazil, Russia, India, China and South Africa). In addition it investigates the effect of changing portfolio weights on VaR estimation. It constructs three different portfolios for each country and each portfolio is made up of two assets: equities and currencies. The first portfolio considers equal weighting between currency and equity, the second portfolio gives more weight to equities (80%) and less weight to currencies (20%) and the third portfolio provides less weight to equities (20%) and more weight to currencies (80%). Although the weights assigned were provided arbitrarily, nonetheless they provide information as to how different weights of the two assets within a portfolio that is constituted of equity and currency will affect the performance of the VaR measure. The performance of these models is compared with the aid of a back-testing process by making use of the quadratic probability score function, the root mean square error, the number of exceptions/prediction failures and average deviations between the VaR and the realized return series. Actually, no study has ever attempted to estimate the VaR of a portfolio that is constituted of equity and currency in order to uncover the optimal weight of the two assets that minimizes the portfolio risk. It is important to note that a portfolio that combines equity and currency not only has the ability to minimize the risk (exchange rate risk) of investing in an emerging market, but this combination of assets also provides investors with some safety to conserve the real value of their investment in the equity market. The findings of this paper will be beneficial to asset managers and investors that seek to hedge their equity exposure in the BRICS markets. 2. The second paper Bank Competition, Concentration and Risk-taking in the UAE Banking Industry by A.I. Maghyereh investigates the impact of competition and concentration on bank stability and risk-taking behavior in the UAE industry over the period 2006 to Camera di Commercio di Genova

3 Some international financial contributions: empirical results and policy implications III A substantial body of research has been addressed to the question how does competition affect bank risk-taking?. Most studies have examined this issue in developed countries, tipically the United States and Europe. More recently some studies have extended their geographical scope and investigated the influence of competition on financial stability and risk-taking behavior in international samples covering banks from countries at different development levels. The results of these studies suggest a major observation. The country-level factors play a significant role in determining the relationship between competition and bank risk-taking behavior. In particular, the quantitatively and qualitatively nature of a country s regulatory framework affect the competition-stability relationship. Furthermore, it has been shown that banks behave differently under different institutional settings, which implies that the results obtained for one country may not apply to other countries. Actually although competition may have a positive influence on bank stability through financial depth, growth and efficiency it could also lead to excessive risk taking activities and hence it may end up threatening financial stability. The aim of this paper is to provide additional insides into the influence of market structure on bank risk taking behavior and financial stability in UAE. The paper extends the existing literature and provides new evidence on relationship between competition and stability using data from the UAE commercial banks over the period The sample period helps examine the association between banks competition and risk taking in the light of the turbulent aftermath of the Global financial crisis. This period also allows identifying whether changes in the degree of bank competition have affected banks risk behavior over time. The theoretical predictions on the relationship between competition and financial stability are unclear. Some arguments and country comparisons suggest that competition in the banking market leads to higher risk-taking: i) first, market power enhances profits which in turn build up reserves in the banking sector. These reserves protect banks against adverse shocks; ii)second, the high profits and the big reserves will reduce incentives for assuming excessive risks and hence, the chance of systemic banking distress is lower compared to a competitive market; iii) the monitoring will be more efficient. Monitoring a few large banks in a concentrated system is more effective than ECONOMIA INTERNAZIONALE / INTERNATIONAL ECONOMICS 2018 Volume 71, Issue 2 May, I-X

4 IV A. Amato observing many small banks in a diffuse banking system. The effective supervision will reduce the chance of systemic distress and will enhance financial stability. At marked contrast to these, theoretical reasoning is that a more concentrated banking system is bad for financial stability. According to the competition-stability hypothesis, market power in the banking market induces borrowing firms to assume greater risks to cover the increased cost of borrowing. This raises the systemic risk in the whole banking market due to the increase in the chance of default. Other arguments have been provided by Mishkin (1999), who pointed out that in the case of a few large banks, moral hazard is substantial and this raises systemic risk. According to Mishkin, the too big to fail argument that shape the public intervention decision in bailing out troubled firms will induce banks to take excessive risk and this increases financial instability. In any case the effect of competition and bank stability have been extensively examined for advanced countries (predominantly for the U.S. and Euro area). These studies showed that country-level factors play a significant role in determining the relationship between competition and bank risk-taking behavior. Furthermore, it has been shown that banks behave differently under different institutional settings, which implies that the results obtained for a country may not apply to other countries. This paper contributes to the existing literature by investigating the impact of competition and concentration on bank stability or risk-taking behavior in the UAE banking industry over the period The Herfindahl Hirschmann Index is used as a proxy for competition, while the nonperforming loans ratio and Z-scores are used as proxies for bank risk-taking. The paper also considers whether the financial crisis has changed the direction of the relation between competition and risk-taking behavior of the UAE banking industry. The empirical results here presented suggest that the increase in competition erodes banks charter value and increases their tendency to assume additional risks with associated negative repercussions on financial stability. Moreover, while the primary focus of the paper is the competition-stability nexus, we also derive some other interesting results. For instance, we found that larger, more capitalized and more liquid banks are relatively more stable. This study provides important policy implications for regulators and supervisors. The evidence of the negative association between bank competition and bank stability Camera di Commercio di Genova

5 Some international financial contributions: empirical results and policy implications V indicates that fueling competition may have adverse unintended consequences on bank stability, especially if it is not accompanied by appropriate level of regulations. Thus, to reap the benefits of bank competition, appropriate attention needs to be paid to banking regulations. Specifically, any attempt to improve the competitive environment should be associated with strengthening regulations and supervision to ensure an eventual correction of the negative consequences of competition on stability. 3. The paper Linear and Nonlinear Attractors in Purchasing Power Parity by I.A. Moosa and M. Ma supplies an examination of the PPP hypothesis over the period where strong evidence is found for nonlinearity not only in the adjustment process towards equilibrium but also on the long run relation itself. Purchasing power parity (PPP) is one of the most debated topics and tested hypotheses in international finance and economics at large. One aspect of PPP that has been dealt with repeatedly in recent times is nonlinearity in the process relating prices and exchange rates. Most of this strand of literature is about the hypothesis of nonlinear adjustment to a linear long-run relation (the attractor), typically employing TAR and ESTAR models. Apart from the use of these models, nonlinearity can be allowed for in the PPP relation either by postulating a nonlinear attractor, or a nonlinear adjustment to a linear attractor. The use of a nonlinear attractor is discussed by Granger (1991), while nonlinearity in the error correction model is discussed by Escribano (1987). With a few exceptions, the possibility of a nonlinear attractor has not received much attention, at least in the PPP literature. The objective of this paper is to contribute to the literature on nonlinearity in PPP by using a data sample with a long span covering the period since 1973, which is the period of the post-bretton Woods floating. We can claim the following contributions of this paper to the literature. First, we use and compare between two possible ways of representing nonlinearity in PPP by using a nonlinear attractor and nonlinear adjustment to a linear attractor, which is different from the literature where only one way is employed (predominantly nonlinear adjustment to a linear attractor). The second contribution is that we take matters further by considering four different combinations of attractors and adjustment processes. The third contribution is that we test the underlying hypotheses by using the longest data span ever used under the present regime ECONOMIA INTERNAZIONALE / INTERNATIONAL ECONOMICS 2018 Volume 71, Issue 2 May, I-X

6 VI A. Amato of floating exchange rates, covering the period Since cointegration is a longrun relation, a long data span is necessary, and more important than the sample size, as confirmed by Shiller and Perron (1985) and subsequently by Lahiri and Mamingi (1995). The fourth contribution is that we use non-nested model selection tests to find out which specification is superior: (i) linear adjustment to a linear attractor, (ii) linear adjustment to a nonlinear attractor, (iii) nonlinear adjustment to a linear attractor, and (iv) nonlinear adjustment to a nonlinear attractor. 4. The objective of the paper by Moosa and Vaz Direct and Indirect Forecasting of Cross Exchange Rates is to determine whether direct forecasting is more or less accurate than indirect forecasting when applied to cross exchange rate as a defined variable. The contribution by Moosa and Vaz examines the accuracy of direct relative to indirect forecasting. Since the cross exchange rate is a defined variable e.g. the ratio of two U.S. dollar exchange rates forecasting cross rates may be direct or indirect. Direct forecasting entails the fitting of a model to the cross exchange rate and using the estimated model to generate forecasts. The indirect method requires fitting separate models to the exchange rates of the two currencies against the U.S. dollar, then using the forecasts of the two individual rates to calculate the corresponding cross rate. This is not simply an exercise in number crunching, because there is some underlying economic theory. Take for example the monetary model of exchange rates, which tells us (among other things) that a country that has faster monetary growth than other countries will experience currency depreciation. It is not clear how this model works or how the effect is transmitted from monetary growth to the exchange rate, but let us assume that the mechanism works at the market micro level. This means that if foreign exchange dealers observe rapid monetary expansion in country A relative to that in country B, they will sell or short sell the currency of country A against the currency of country B. To trade this way, foreign exchange dealers must observe what happens in a vast number of country pairs. The simpler alternative would be to observe each country against the U.S., derive implications for the U.S. dollar exchange rates and consequently for the cross rates. This line of reasoning is consistent with the fact that the cross rates quoted by foreign exchange dealers and money changers are invariably calculated from Camera di Commercio di Genova

7 Some international financial contributions: empirical results and policy implications VII the dollar exchange rates. In particular, the exchange rates of currencies that are pegged to the U.S. dollar or currency baskets (typically with a dominant dollar component) are calculated as cross rates by determining the dollar exchange rate (of the pegged currency) first. If this reasoning is valid, one would expect indirect forecasting to produce better forecasts of the cross rates than direct forecasting. The objective of this paper is to find out if indirect forecasting of cross exchange rates is indeed superior to direct forecasting by applying the flexible price monetary model to three cross exchange rates involving the Japanese yen, British pound and the Canadian dollar. Apart from the exchange-rate-specific reasoning given above for why indirect forecasting is likely to be more accurate, the literature on direct versus indirect forecasting portrays the same idea in general terms. Some economists maintain the opposite view that direct forecasting is better or at least preferable. And there is the neutral view that either can be better or worse, implying that this is an empirical issue. Actually while direct forecasting entails generating forecasts from a model fitted directly to the cross exchange rate, indirect forecasting requires the generation of forecasts for the exchange rates against the dollar then combining these forecasts to obtain forecasts for the cross rate. Several reasons have been suggested for why indirect forecasting is thought to be more accurate: (i) this proposition is implied by the theory of prediction; (ii) differences in the time series properties of the components; (iii) the utilisation of more information to enhance efficiency; (iv) certain events that affect individual components may be masked in a direct forecasting model; and (v) indirect forecasting utilises the information embodied in cross correlations. There is also a specific reason pertaining to the special case of cross exchange rates that foreign exchange market participants pay more attention to macroeconomic developments relative to the U.S. economy (hence more attention to the bilateral exchange rates against the dollar). On the other hand, views have been expressed in defence of direct forecasting, either because it produces better results or because it is preferable in terms of costs and benefits. These views include the following: (i) direct forecasting may be superior if the underlying processes are not known and have to be estimated; (ii) the information obtained from explanatory variables in the aggregate model may be more important than the information embodied in the components; (iii) high correlation enhances forecasting ECONOMIA INTERNAZIONALE / INTERNATIONAL ECONOMICS 2018 Volume 71, Issue 2 May, I-X

8 VIII A. Amato accuracy at the aggregate level; and (iv) it may be preferred because it involves less work and data requirements. It has also been suggested that the choice between the two approaches depends on whether the exercise is conducted in sample or out of sample. The mixed results produced by this strand of research support the view that the choice between the two approaches is an empirical issue that is, doing both and picking the better set of forecasts. The results of this study show that indirect forecasting is better than direct forecasting, when forecasting accuracy is measured in terms of the RMSE, for two of the three cross rates examined while the opposite is true for the third rate. However, no difference is apparent when forecasting accuracy is measured in terms of directional accuracy. 5. The following paper by M. Tronzano Does the Expectations Hypothesis of the Term Structure Hold in Korea after the Asian Financial Crisis? Some Empirical Evidence ( ) explores the validity of the Expectations Hypothesis of the Term Structure (EHTS) in Korea after the Asian financial crisis. In line with the EHTS, one common stochastic trend is found in the term structure of interest rates, although the validity of the symmetry restriction is rejected. Moreover, significant liquidity premia and a causal relationship from long to short-term interest rates are documented. There are two policy implications to be drawn from this research, both related to the shortterm interest rate rule underlying the inflation targeting approach adopted since the late 1990 s: 1. This rule should be implemented putting a greater emphasis on the expectations channel through which economic agents anticipate the future monetary policy path; 2. A gradualist approach is recommended in the management of the policy rate. The former implication hinges on one important result from causality tests, namely that the short-term interest rate assumed as a proxy for the policy rate is not weakly exogenous. This finding might undermine the monetary transmission mechanism through complicated feedback effects. In this perspective, a greater emphasis of monetary authorities on forward guidance would establish a causal relationship from short to medium and long-term assets maturities, leading to a more stable and efficient monetary transmission mechanism Camera di Commercio di Genova

9 Some international financial contributions: empirical results and policy implications IX The latter policy implication is related to the evidence from multivariate cointegration tests, namely the rejection of the symmetry restriction and the existence of significant risk premia at the long end of the maturity spectrum. Both results imply that monetary policy should be implemented following a gradualist approach, that is avoiding abrupt changes in the short-term instrument. The one-to-one low frequency relationship between short and long-term interest rates represents a basic feature of the EHTS. If this one-to-one equilibrium relationship does not hold, the long-run effects of monetary impulses become more difficult to quantify. The rejection of the symmetry restriction requires therefore a gradualist approach in the management of the policy rate, in order to smooth out unforeseen effects of monetary policy. The existence of term premia components at longer maturities might further complicate the transmission of monetary impulses along the yield curve. If these term premia are maturity-dependent but time-invariant, as assumed in the Liquidity Premium theory, there are no additional problems in the implementation of monetary policy. However, as suggested by a large strand of literature, the assumption of constant risk premia is quite unrealistic, and this generates additional uncertainty in the monetary transmission mechanism, further reinforcing the case for a gradualist approach. Although this paper provides interesting insights about the validity of the EHTS in Korea, there are many other research directions, not covered in the present empirical investigation for space reasons, which are left for future research. Some straightforward extensions include the use of data sampled at different frequencies, an increase in the spectrum of assets maturities, and the analysis of nominal yields relative to other financial instruments. The existence of significant risk premia at longer maturities calls for further investigation about the possible time-varying nature of these components and their relationships with macroeconomic fundamentals. Since many Asian economies have recently introduced financial liberalization measures, another fruitful research line is represented by a joint analysis of South Korea and other Asian countries in a panel framework. A panel approach provides actually a significant ECONOMIA INTERNAZIONALE / INTERNATIONAL ECONOMICS 2018 Volume 71, Issue 2 May, I-X

10 X A. Amato increase in test power with respect to single-country studies when assessing the validity of the EHTS (see, e.g. Holmes et al., 2011). Other relevant extensions, finally, are represented by empirical tests focusing on potential structural breaks, on the role of foreign interest rates in the cointegrating relationships, and on the problems raised by the zero lower bound on nominal rates Camera di Commercio di Genova