## Basics of the Research

This series of articles examines and compares the stocks of Johnson & Johnson (NYSE:JNJ), Coca Cola Co. (NYSE:KO), Procter & Gamble Co. (NYSE:PG), Chevron Corp. (NYSE:CVX) and PepsiCo Inc. (NYSE:PEP). The first article includes an introduction and a coverage of Johnson & Johnson. Near the end it presents several strategies of protecting an investment in JNJ, including one which counts on the correlation of returns between JNJ and the SPDR S&P500 ETF (NYSEARCA:SPY).

The main goal of the series is to provide information about the historical risk, and especially the tail risk, which could significantly affect the expected total return of each stock, or a combination of them in a portfolio.

The total return is calculated as a dividend and price return over the holding period. Thus, for examining the risk associated with each stock, the prices adjusted for dividends and splits will be used. Moreover, as the total return is a combination of dividends and prices, the risk on dividends will also be on focus.

Tail risk, or the risk of getting returns which stand as far as several standard deviations from the mean, is represented by two characteristics of the distribution of returns - skewness and kurtosis. These are among the main characteristics the current analysis is focused on. They may seem a bit strange to the reader but in fact are relatively easy to grasp.

*Skewness* can be defined as the degree to which a certain returns distribution differs from the normal one. A distribution is said to have a negative skewness if it shows a higher than the normal concentration of large negative returns during the examined period which are not compensated by similar by size positive returns. Thus a distribution with a negative skewness could represent a higher risk of facing relatively rare but large negative returns while continuing to collect a lot of small but higher than the mean returns.

A distribution with a positive skewness is sometimes a more preferable one concerning the risk of a loss. The reason is that even if such a distribution might present the investor small returns below the mean more often than the normal one, it also faces the possibility of having relatively rare but large positive returns.

For the *kurtosis* as a measure of risk, the reader should be aware that a value different than zero shows an increased probability of having returns far from the mean, both positive and negative.

Even though in general portfolio management the annualized values of returns and deviations are preferred as means of comparing investments, the author believes that in order to better account for risk and eventually rebalance the assets, the use of monthly coefficients would be more appropriate. The reasoning behind this, as argued by Eugene F. Fama and Keneth R. French, the authors of the famous Fama-French three factor model of describing stock returns, is that the long-run returns tend to be more volatile (and thus more uncertain) than the short-run ones. That is why the tail risk, standard deviations, correlations, betas and some of the average returns are expressed here on a monthly basis.

In the current analysis the author uses historical mean as the expected return for the sake of simplicity, among other reasons. A recent study by Mirko Cardinale, Marco Navone and Andrzej Pioch, which was published by the CFA Institute in its November edition of the "Investment Risk and Performance" newsletter, states that the

"... correlation between subsequent 10-year returns and 10-year historical returns is statistically significant but negative, which suggests some degree of long-run mean reversion."

In plain language this would mean that if we have ten exceptionally good years, they could be followed by ten not so bright ones because of the tendency of returns to revert to their true mean value.

The mean reversion in our case is fine because the average monthly values are derived from a large amount of sample data. Hence, because of these long periods of time the research is based on (at least 35 years of monthly data), the calculated long-run coefficients cover several decades which include different economic environments and cycles, both good and bad ones. As such, we could expect the coefficients to be relatively free from sampling biases and closer to the true descriptors of each stock's population of returns. In fact, the longer the sample size goes, the less volatile (more certain) the mean gets. Thus, using historical data should be fine with the purposes of the research.

Nevertheless, one should keep in mind that historical performance does not guarantee future results. The performance in the more recent periods, i.e. the last 12 months, could be more important and useful for the moment but it tends to get closer to the mean characteristics of the longer time frame periods, *given that no significant change in the economic environment or the company itself has happened*.

Having said all the above, we could step into the research itself.

**The** **Selected** **Companies**

All of the examined companies have a long history of distributing dividends. Moreover, they have increased the dividend size in almost each of the last 40 years. Chevron has suspended dividends distribution in the period 1976 - 1984.

Each of the companies has bought back amounts of its shares during the last 5 years. The extent to which they have done it varies. For instance, Procter & Gamble generally is buying back shares for more money than it distributes as dividends and Chevron did not repurchase any shares in 2009. These repurchases should be kept in mind and included in a dividend analysis as in most of the cases they could be considered a tool of distributing money to shareholders.

## Stocks Characteristics and Analysis

The current article examines the risk and reward characteristics of the stock of *Johnson & Johnson*, an U.S. based company whose business operations are in the healthcare field.

## Johnson & Johnson

For the whole examined period the stock experienced a compound annual growth rate (OTCPK:CAGR) of 12.6% which is twice as high as the CAGR of S&P500 (SPX) for the same period (6.9%). The above graph of the monthly price, adjusted for dividends and splits, shows a clear uptrend which in the recent two years intensified again after the slower 2010. In 2012, 2011 and 2010 the adjusted stock price marked increases of 10%, 9% and 2%, respectively, on a TTM basis.

The company has split its stock 6 times in history - in May 1970, May 1981, May 1989, June 1992, June 1996, and June 2001. The most common price at which the stock experienced a split was just below $100. Since the split in 2001 the stock has not traded higher than $73. The first time it traded this high was in September 2008 and the second one was this past October, 2012. Consequently, the **stock now trades close to a major resistance level** which increases the risk of unexpectedly large returns, both of a positive and negative nature.

Concerning the stock price return, the stock shows better risk-adjusted returns on a monthly basis than the S&P500 index, with the Sharpe ratio of JNJ being 0.19, against 0.16 of SPX.

As mentioned in the beginning of the article, the total return is a sum of a dividend and a price return. So now we turn to the dividend portion of the total return.

The company managed to increase its dividend size in each of the last 40 years. For the 2007-2012 period the mean average annual increase of the dividend is 9.2%. A weighted average dividend growth, which gives the biggest weight to the last 3 years, a lower weight to the previous 2 years and the lowest weight to the last two years of the 2007-2012 span, is equal to 8.5%. This is almost half of the average dividend growth of JNJ for the last 40 years (14.9%). As currently the dividend grows by such a small portion of the average dividend growth rate, one could expect that an increase is in the way. However, below is explained by the use of financial statements data and some historical estimates, that an increase of the current JNJ dividend growth in the following year or two is highly improbable.

We turn to some of the fundamental characteristics of the company to get a better understanding of its dividend practice. We see that during 2012 it would be distributing about 70% (see the table above) of its 2011 earnings as dividends. The value is higher than the average JNJ payout ratio of 50.5% (which actually includes this estimate) for the last five years. This represents a risk on the estimated dividend range for 2013 ($2.70 - $2.80). The range implies an increase between 10% and 15% which, if accomplished, should most probably translate into an increase of the payout ratio because the number of basic shares is already higher by 2.5% than it was at the end of 2011. For the payout ratio to come back to more sustainable levels, the earnings of JNJ should increase substantially. For instance, a payout ratio of 45%, which is the average one for JNJ for the years 2007-2010, would translate into 2012 earnings of $16.9B, if the increase of dividend is at the lowest end of the above mentioned range. Such a net income seems highly improbable when we look at the 2012 results till date and the earnings from the previous years. Given that number of shares stay the same, the logical conclusion is that the dividend will be increased at a rate lower than the mean for JNJ or the payout ratio will go higher. In the latter case, the risk on continuing the dividend increases gets further elevated.

## Other Risk Characteristics

Monthly coefficients:

| Last 12 months | Last 5 years | Whole period |

Standard Deviation | 2.86% | 4.35% | 6.09% |

Skewness | 1.2 | -0.35 | 0.03 |

Kurtosis | 2.46 | 1.25 | 0.39 |

Correlation a | 0.5 | 0.68 | 0.53 |

Beta b | 0.48 | 0.54 | 0.72 |

*a. Correlation is with S&P500;*

*b. Beta is measured towards S&P500*

For the whole period the stock exhibits almost normal risk characteristics. The slight kurtosis (0.39) speaks of a bit higher than the normal probability of having extremely large returns but the practically close to zero skewness (0.03) means that they are relatively equally distributed on the positive and negative side of the mean.

If we look at closer time frames, and especially the last 12 months period, the characteristics change. The kurtosis (2.46) is significantly higher so basically the stock exhibited higher probabilities of facing extreme returns. During the last 12 months the monthly distribution of returns has a significant positive skewness (1.2). This is because of the single month return of 8% in June, due most probably to the company's repurchase of its own stock for $12.85B. If the long-run coefficients are closer to the true population descriptors than the more recent ones, we should expect the recent risk characteristics of JNJ to revert to the longer-run values, i.e. to become again a close to the normal distribution with relatively low probability of large negative or positive returns.

## Protecting the Investment

Given the current situation in which the stock is close to its highest level since 2001 and the political uncertainty about the fiscal cliff, engaging in a strategy to protect the price component of the total return from the JNJ stock might prove to be a valuable idea. This could be done by buying puts on the stock or writing (selling) calls. The latter has the advantage that no additional money would be invested but carries the risk that the investor might have to deliver the stocks for which the calls are written if the stock price appreciates above the strike price before expiration time. This way the investor would also lose the dividend portion of the total return which might not be a desirable outcome.

Another, and a more unconventional way, would be to take advantage of the low Beta the stock has towards the S&P500 index. In general it means that the stock price moves slower than the index value. If the index rises, the stock would rise slower. If the index declines, the stock should decline slower. For this strategy the SPDR S&P500 ETF could be used. The SPY ETF tracks the S&P500 index and has an expense ratio of 0.09. The correlation and Beta between SPY and S&P500 index since the beginning of the ETF are 0.9969 and 0.9921, respectively, so basically their returns are perfectly correlated and move by the same magnitude. This way by selling short the SPY ETF investors could possibly make a profit while protecting the price return of their investment in JNJ, given the general market starts to move south.

The second article of the series would examine the risk and reward characteristics of the Coca Cola stock.

Comments()