RESEARCH OUTPUT AND FINDINGS
4.6 Stability and Events Analysis
4.6.3 Risk and Return Relationship
As observed in Graph 1.1 in Chapter 1, investors tend to change their attitude towards risk during specific events like LTCM near financial collapse and Asian crisis turmoil. Using this same analogy that risk can be proxied as standard deviation and variance, the actual return R is regressed against standard deviation and then against t
variance as follows: + =ϕ0 t R ϕ1
σ
t + tε
(4.22)136 The net positions of hedgers for coffee had to be differenced for stationarity. Therefore, the estimated
187 +
=ϕ0 t
R ϕ1
σ
t2+ε
t (4.23)where
σ
t is the standard deviation from PARCH model (Equation 4.14.2) andσ
t2is the variance from GARCH model (Equation 4.13). The pattern of the recursive coefficients (σ
t andσ
t2) show whether there is any relationship between the measure of risk and return. The recursive estimated coefficients ofσ
t andσ
t2also help in finding whether any significant break can be attributed due to the occurrence of a major macroeconomic event, which changes the attitude of large hedgers and large speculators towards risk. Any structural break in the relationship between return and risk which is matched with any of the eight events is reported in Appendix 6.16 (Table 4.23).Results from Panel A show a positive significant relationship between hedgers’ risk (standard deviation) and return for soybean oil, gold, coffee and soybean, and a significant negative relationship for live cattle; and a positive significant relationship between speculators’ risk (standard deviation) and return for gold, coffee, live hogs, sugar and S&P500, and a significant negative relationship for live cattle and silver at 10% significance level. From Panel B, there is a different mixture of findings again due to the different sensitivity of the proxy of risk over return. Panel B shows a positive significant relationship between hedgers’ risk (variance) and return for soybean oil, coffee, wheat (Minnesota) and platinum; a negative significant relationship for gold and wheat (Chicago, Kansas); a significant positive relationship between speculators’ risk (variance) and return for feeder cattle, coffee, platinum and sugar; and a significant negative relationship for gold, copper and Treasury bonds. While the findings of a positive relationship between risk and return supports portfolio theory that a higher risk is compensated with a higher return and vice versa, the findings of a significant relationship between risk and return can be explained by Glosten, Jagannathan and Runkle (1993) who discussed special circumstances that would make it possible to observe a negative correlation between current returns and current measures of risk. For instance, investors may not demand high risk premium if they are better able to bear risk at times of particular volatility. Moreover, if the future seems risky, the investors may want to save
188 more in the present, thus lowering the need for larger premium. And, if transferring income to future is risky and the opportunity of investment in a risk-free asset is absent, then the price of a risky asset may increase considerably, hence reducing the risk premium. In addition to Glosten et al. (1993) who argued that both positive and negative relationships between current returns and current variances (risk) are possible, the study adds contribution by also finding more negative relationships between current returns and current standard deviation (risk). The higher number of negative significant relationships is due to derivatives prices being more proportional to standard deviation than variance, hence the higher sensitivity as supported by Poon and Granger (2003).
Panel A shows that speculators’ returns are affected with seven structural breaks in risk that occurred during the listed macroeconomic events at 10% significance level. These structural breaks in the return and risk relationship of Equation 4.22 occur in cotton, feeder cattle, Japanese yen, coffee, live hogs, soybean and Treasury bonds for speculators; and soybean, crude oil, cotton and copper for hedgers’ attitude towards risk. Since speculators’ attitude towards risk are more affected than hedgers’, this suggests that speculators not only bear more risk than hedgers, but also that speculators’ returns are more affected during major macroeconomic events. However, generalization about this suggestion is questionable since only soybean and Treasury bonds have significant risk coefficient estimates before and after the event. This is consistent with Flood and Rose (1999) who demonstrated that exchange rate volatility cannot be linked to changes in underlying fundamentals. The jump of the effect of hedgers’ risk on return for the soybean futures market has been occurring after the end of the long period of US tightening interest rates. This can be explained by hedgers in the soybean futures market taking more risk towards obtaining their return, due to the instability of US interest rates that eased after a long period of tightening. On the other hand, the fall of the effect of speculators’ risk on return for the Treasury bonds market has been occurring at the start of the temporary revival from the Japanese recession. This can be explained by speculators using less risk to obtain a desired return, due to the stability regained in the global economy after the temporary recovery of the Japanese recession. Overall, Panel A supports that the major global economic events named in Table 4.19 did not have much
189 effect on the risk and return relationship, except for soybean for hedgers’ return and Treasury bonds for speculators’ return.
In contrast to Panel A, Panel B shows that there is a smaller occurrence of structural breaks that occurred during the major economic events used in the study. Speculators’ attitude towards risk changed in copper, Japanese yen, wheat (Kansas, Chicago) and Treasury bonds, while hedgers’ attitude towards risk changed only in wheat (Kansas). The lower number of breaks in Panel B can be explained since many of the recursive coefficient estimates of standard deviation from Panel A were larger in magnitude than their recursive coefficient estimates of variance. This is supported by Poon and Granger (2003) who found that derivative prices are roughly proportional to standard deviation. The only structural break, due to the same economic event, where risk is measured as variance and standard deviation, occurs in Japanese yen, where speculators’ return was more negatively affected by the end of the long period of US tightening interest rates in 1995. However, more importantly, none of the structural breaks in Panel B significantly affected the risk and return relationship in the futures markets. Either measurement of risk (standard deviation and variance) tends to return to their stable long-run estimate very shortly after the macroeconomic event has disturbed the risk/return relationship in all the 29 futures markets. This is inconsistent with Christie and Chaudhry (1999) who showed that volatility persists following macroeconomic events, particularly for liquid financial markets. The study adds contribution to BIS (1999) reports that events like the Russian crisis, LTCM near financial collapse, Asian crisis, and Mexico crisis did not have significant effect upon the attitude towards risk of large speculators and even lesser significance for large hedgers.