Built to Love
上QQ阅读APP看书,第一时间看更新

The Index

In our study, we first surveyed the marketplace to identify firms that provide more emotion and those that provide less. We used stock market performance as our basis for comparison. Stock performance is the ultimate measure of performance for a publicly owned firm, scrutinized by those inside and outside the firm.

After collecting stock returns, we analyzed returns of the group of high-emotion firms relative to major market indices, namely the Dow, NASDAQ, and the S&P 500. In what follows, we describe the detailed results. The bottom line is that the results corroborate our theory—emotion pays off handsomely!

To compose the index we analyzed a promising set of firms. We first identified companies that, ahead of time, seemed to be likely providers of emotion. These firms were either found to have the strongest global brands (on the list of the 100 “Top Brands” as reported by BusinessWeek and Interbrand) or were among the most innovative (on the list of the 50 “World’s Most Innovative Companies,” a ranking compiled by BusinessWeek and Boston Consulting). Together, these lists yielded 121 unique companies. Of these, we retained consumer brands that are part of the S&P 500. For this study, only consumer product companies were considered so that we could be confident that study participants would know the companies. The result was 40 innovative companies considered strong global brands, with products and services readily recognized by U.S. consumers.

To identify which companies engender higher levels of valued emotions to their customers, we surveyed consumers to find out which of those 40 firms they believe are the ones that provide emotion. In the study, 109 respondents were asked to rate their emotional response to each firm relative to its competition. Firms were rated on a 13-point scale from weak (0) to strong (13) engendered emotion. Although all of the firms have been found either to have strong brands or to be among the most innovative (or both), the firms vary with the level of emotion that they engender in the respondents. The average emotion score was 8.2/13. The high-emotion firms were those for which the scores on the emotion scale were statistically greater than average, meaning that these firms delivered greater emotional benefits than the average firm that we tested. Similarly, low-emotion firms’ scores were statistically lower than average, those that delivered less emotional benefit than the average firm. Both the top set and the bottom set contain firms across a range of industries: retail, software, high-technology products, fashion, and consumer packaged goods, among others.

In the analysis of stock performance, we compare the performance of higher-emotion companies to lower-emotion companies that are with in the same set of outstanding companies (S&P 500 companies that are most innovative and/or top brands). All of these firms are high performers, so we expect them to do quite well relative to market averages. If the higher-emotion companies outperform the others (and we find that they do), they are outperforming some of the best companies in the marketplace. Thus, we have an extremely high standard of comparison for our analysis.

This study was conducted in July 2007, so the ending point of the historical data is the end of June 2007. (For those wondering what happened during the troubled economic times that came later, we present a follow-up study later in this chapter.) The first striking aspect of the graph of three-year returns ending June 2007 (shown in Figure 3.1) is that we are remiss that we did not invest in our high-emotion index a few years before!

FIGURE 3.1 image Three-year comparison of high-emotion index to others.

Of course, investing is always successful in hindsight, because it is so very easy to concoct sets of firms that performed well in the past. What we’ve done is different. We did not work to find a set of firms that performed well in the past. We did not use past analysis to predict future success. We developed a theory, and we are using past data to see if our theory is consistent with the stock market data. If our theory is right, higher-emotion stocks will outperform lower-emotion stocks. And that is exactly what the data shows. The high-emotion stock index exceedingly outperforms the other indices for this three-year period, by a wide margin. While the market indices (Dow Jones, S&P 500, NASDAQ) returned less than 40 percent over the three years, the high-emotion index returned greater than 100 percent.

FIGURE 3.2 image Ten-year comparison of high-emotion index to others.

Another interesting and related result is that the low-emotion stock index underperformed the others in this three-year period, returning less than 20 percent. As with the high-emotion index, the low-emotion index is comprised of S&P 500 firms that are among the top 50 innovative companies and/or among the top 100 brands. Put differently, these are widely recognized to be outstanding companies that many people would expect to outperform market averages. In looking over a period of 10 years rather than three (see Figure 3.2), the low-emotion index did perform very well, outperforming market averages. While the low-emotion index outperformed the market averages, returning 152 percent over 10 years, the high-emotion index had exceedingly high returns of greater than 1000 percent over the same 10-year period!

These results, outstanding as they are, are worrisomely strong in their support of the value of emotion. Even though the results are consistent with what one would expect after thinking carefully about the human appreciation for emotion, are they too good to be true? Certain companies in our high-emotion index are superstar stocks, in particular Apple and Google. To see if these results are due to those key firms, we also looked at the performance of the high-emotion index after omitting the superstar stocks. We found the results are not driven by a few firms, because the next 10-year performance graph (see Figure 3.3) omits the firms which appear to be in a class of their own with respect to stock performance. In Figure 3.3, the reduced high-emotion index again considerably outperformed market averages, returning around 800 percent in 10 years relative to the returns of around 70 percent from the major indices.

FIGURE 3.3 image Ten-year comparison of high-emotion index to others, without its highest performers.

Even after omitting the high-flying stocks of Apple and Google, the performance of the high-emotion index remains astoundingly ahead of standard indices. No index can be such a sure bet, right? Maybe it was the timing of our study that led to such outstanding performance? The graphs in Figures 3.13.3 end mid-summer 2007, at that time a stock market peak. What if we did not just change the starting point (one year prior, two years prior, and so on) but also varied the ending point for the holding periods?

To change starting points and ending points, we looked at returns calculated for random buy-and-sell dates. Random buy-and-sell dates mimic what investors do in reality, because investors buy and sell on different days after holding for different amounts of time. From this viewpoint, randomizing a thousand buy-and-sell dates is the same as evaluating returns for a thousand individual investors who put their money in the market, where each investor buys and sells at random times over a 15-year period of time.

To fairly compare the performance of the high-emotion index and other standard indices, we acted as if each investor put ¼ of their money into each of the indexes: ¼ in the high-emotion index fund, ¼ in a Dow index fund, ¼ in a NASDAQ index fund, and ¼ in an S&P 500 index fund. Each individual chose his or her own buy date (a random buy date) and then bought all four funds the same day. Each individual likewise chose his or her own sell date, and sold everything at the same time. Looking at all one thousand investors with random buy-and-sell dates, we could see what percentage of investors made more money in the high-emotion index than in the standard indices.

The percentages of investors who “won” with the high-emotion index are shown in Table 3.1. It turns out that 84 percent of investors beat the Dow, 85 percent beat the NASDAQ, and 86 percent beat the S&P 500. Most investors won by investing in the high-emotion index instead of the standard indices. At the same time, these numbers are reasonable in that they show that investment in high-emotion companies is not a sure bet; for example, 16 percent of these individuals made more by investing in an index that tracks the Dow than by investing in the high-emotion companies. As must be the case, the emotion index isn’t impossibly good; it does not offer guaranteed riches. Instead, we found that the high-emotion index performs very well for its investors most of the time.