Investing in New Zealand: a Review

The New Zealand Superannuation Fund (NZSF) is in the process of deciding upon the allocation of its portfolio, across asset types and countries. In response to this, the New Zealand Stock Exchange (NZX, 2003) has argued that a minimum of 20% of the NZSF portfolio should be invested into New Zealand equities. This paper has examined the arguments presented by the NZX. These arguments, and my responses to them, are as follows. The first argument is that application of the Markowitz model for portfolio selection is subject to errors in estimating parameters (" estimation risk "), and this justifies tilting towards the home market. No arguments linking the premise to the conclusion are offered here. The premise is acknowledged, and a number of possible responses to the problem are examined. Contrary to the NZX suggestion, they suggest that the optimal portfolio of risky assets is the world market portfolio. The second argument is that traditional studies in support of international diversification understate the currently prevailing correlation coefficients between markets, and therefore overstate the benefits of international diversification. However, it is shown that, even if this reasoning is correct, it does not justify tilting away from the world market portfolio of risky assets. The third argument is that correlations rise in significant market downturns, implying that the benefits of international diversification are virtually zero. Accordingly, a significant tilt towards the home market is warranted. However, it is shown that unless the correlation coefficient reaches 1 (and this is not asserted by the NZX), then there is no justification for tilting away from the world market portfolio of risky assets. The fourth argument offers the theoretical analysis of Hasan and Simaan (2000) in support of the proposition that " estimation risk " justifies tilting towards the local market. However their analysis compares only the local portfolio with an international portfolio, with the latter selected in accordance with the Markowitz mathematics and with estimation risk contaminating the parameter estimates. 3 Furthermore, by allowing short-selling, these estimation risks are significantly aggravated. Had the world market portfolio been included in the comparison, it is likely to have outperformed the two possibilities considered. Thus, in my view, the Hasan and Simaan analysis does not support tilting away from the world market portfolio of risky assets. The fifth argument is that the Markowitz methodology assumes that returns are normally distributed, when they are not, and this justifies …