Benchmarks

As mentioned earlier it is highly relevant to investigate whether the conclusions drawn changes when using local indices as benchmarks instead of the global benchmarks used for far. There are both arguments for and against using these, while the main argument for using global indices as

103 benchmarks is that it aligns with the theory behind the CAPM and the Fama-French Three-Factor model, as these suggests all investors should invest in the market portfolio, which is typically

regarded as e.g. the MSCI world index. Throughout this paper, it is also known as global benchmarks, although it is generally referred to simply as benchmarks.

This investigation will only look at the regressions which is are geographically focused on other regions than global markets and only includes equity-focused funds. The reason for this is simple, as changing to local benchmarks, which are dependent on the geographical focus of the regression only makes sense when the geographic focus is different of that already used. This means that the regression focusing on global markets already uses the "local" benchmark, as that is the global benchmark. The reason why it is only the equity-focused regressions which will be investigated here is that it is these funds which are the main focus area, and that equity-funds have more locally focused benchmarks in general compared to the other asset classes.

Because of this, it is regressions 9, 10, 12 & 13 that has been reconsidered here. The table 6.4 below shows these regressions in the same format as the original overview table. In appendix 18 the full output of each regression can be found, which provides further details regarding each regression.

The main discussion is centred around the table shown below.

Table 6.4: Overview table of regression figures with local indices

What can be seen in the table above should be compared to table 5.23 in section 5.2.5 showing the results of the original analysis.

For regression 9 and 10 the local index is an ETF tracking the stock price development in emerging markets. The ETF chosen is issued by iShares by BlackRock, which is one of the major issuers in

104 the ETF market. The ETF is called “iShares MSCI Emerging Markets ETF” (n.d.). The return data has been downloaded just like what was originally done with the world market ETF.

For regression 12 and 13 the local index is an ETF tracking the Nordic stock price development.

The Nordic markets have their own index, the “OMX Nordic 40” (n.d.). Just like the ETF used for emerging markets the data series has been downloaded and are used for the Nordic funds only.

The main differences are found in the significance of the fee coefficient for emerging markets, where this regression shows the coefficient estimate to be significantly negative in regression 8 and 9, as opposed to the original regression which showed an estimate insignificantly different from zero. On the other hand, the R-squared value in this regression is just about the half of what it was in the original regression. This means that the conclusions from this goes two ways.

Firstly, the reduced R-squared values indicates that the local index for emerging markets actually fits the data worse than the global index. This could be due to that there is a lot of differences across the countries included in the definition emerging markets. Some are more correlated with the global market where US and Europe are the main markets, and some are almost only correlated with themselves. An example of the first type could be Indonesia, which is generally regarded as more western-like than some of its neighbour countries. An example of a country which is almost only correlated with itself could be Russia, where the stocks are very dependent on oil and gas prices and very much on the political environment in Russia. When there is that big differences across

countries, an index of these will by definition be a very poor indicator and that is exactly what the R-squared value could suggest here. The fact that the global index actually fits the data better also confirms that the choice of using the global index in the first place was the best solution for

emerging markets.

Secondly, the other side of the results above is, that there is an indication of a negative relationship between the fee of the fund and the return, which are now significantly negative. The fact that the R-squared value has decreased that much decreases the reliability of this conclusion slightly, but nevertheless it indicates this relationship. The change that the coefficient estimate is now

significantly negative for emerging markets means that passively managed funds should now be preferred to actively managed funds, when investigating equity-focused funds with a focus on

105 emerging markets. This conclusion is drawn based on the statistically significant negative

relationship between management fees and the performance of the funds.

For the Nordic markets the conclusions are very much the same as the original conclusions. There seems to be a slight positive relationship between fee and return, but it is far from significant with p-values as high as 0.77 and 0.78, which can be seen in appendix 18. Also, the R-squared p-values are much lower for the Nordic markets now, implying the same as for emerging markets above, that using the global index was the right decision in the first place, as the model estimated using this index has significantly higher R-squared values, meaning it explains the variation in the data much better.