While there has been lot of study done in the US and European markets to understand how the recommendations of equity analysts have performed, we were not able to find any research paper pertaining to the Indian equity markets. Hence the focus of our stud

Further we estimate the market adjusted returns based on beta values using CAPM model. Also we perform analysis based on characteristic adjusted returns (market cap and P/B multiple) and based on momentum of the stocks.

Finally we perform hypothesis testing (using t-statistic and regression analysis) for our various findings.

Stage IV – Conclusions and inferences from our study

1. DATA COLLECTION METHODOLOGY

Period of study -

The period of study is 2007-2009. The reason for choosing this period is because it captures the crest and trough of the economic cycle. 2007 was a boom year with stock market riding new highs. It was followed by the slump in later half of 2008 which battered the markets. Finally the recovery phase from second half of 2009 post the monetary stimulus actions by federal banks around the world.

The distribution of the number of recommendations during the 3 year period offers an insight into the condition of the markets during the years.

Total number of recommendations

        *Until June 2009

For Economic Times (which reports brokerage house recommendations), we can observe the following -

The year 2007 was characterised by a large number of recommendations, especially ‘Buy’ ones. With market climbing to new levels, most of the recommendations were in tune with the forward momentum.

2008 saw a sharp decline in the recommendations. Retail investors stayed away from the market. The market itself was reigned by confusion and mayhem. There were huge swings of volatility day to day and no analyst was willing to make a bet on the market.

2009 again saw a positive increase in recommendations. Most of the equities were trading at their all time lows offering opportunities for bargain buying. The key was to identify the right stocks which would recover quickly out of the crisis.

Business Line however has been consistent in the number of recommendations. It has been steadily increasing year on year.

Source of study –

For the recommendations, we chose two different types of recommendations.

1. Economic Times – It published recommendations made by brokerage houses. In effect it was passing on second hand information. However considering the wide reader base of ET, we expected this second hand information to have some investment value.

Further, ET also offered us a wide array of brokerage houses recommendations, which we would use to determine which of them were successful overall in their recommendations.

2. Business Line – The feature of BL recommendations was that it was generated from in-house financial columnists. Hence any possible conflicts of interest arising between Buy and Sell side divisions of Analyst firms could be avoided here.

The journalists making these recommendations have financial backgrounds similar to that of equity analysts, so we can dispel any doubts regarding their capabilities.

Data parameters –

For analysis we collected the following values from BSE listed companies from Bloomberg terminal -

• Adjusted Stock price – To ensure that any corporate actions like Dividends, stock splits, bonus issues are factored into the equity price and is easier to compare returns between two periods
• P/B and P/E multiples – This will give an idea of how much premium the market is willing to pay. This helps us classify the stocks into Value stocks and Glamour stocks
• Market Capitalisation – To classify into Large Cap, Medium Cap and Small Cap industries
• Beta value – We took the raw beta values of each company to be used for calculating the expected returns for the stock in tune with overall market returns
• Index returns – We collected the SENSEX index returns over three years to capture the market returns and find the market premium (rm) in the CAPM model

2. DATA CLASSIFICATION

The recommendations themselves are quite varied – Overweight, Accumulate, Outperformer, Buy, Add, Hold, Neutral, Underweight, Underperformer and Sell. For the sake of ease and consistency, we classify various stock recommendations into primarily BUY and SELL.

BUY includes – Overweight, Accumulate, Outperformer, Buy, Invest

SELL includes – Hold, Neutral, underweight, Underperformer, Book Profit, Sell

The reason for classifying Hold and Neutral as SELL is because analysts employed at investment banking houses earn large fees from corporate transactions. Either the company may be an existing or a potential client which the firm does not want to antagonise by giving an explicit SELL recommendation or it may be a large influential corporate firm that the brokerage house will not want to remain in its bad books. In these cases, a Hold or Neutral is given which the intelligent lot of investors perceive as SELL.

Buy and Sell Recommendation

           

Why more Buy recommendations than Sell?

Reason 1 - Although analysts are essential for information dissemination for the efficient functioning of capital markets, in recent past strong doubts have been expressed about their credibility and conflicts of interest in recommendations.

The sell-side analyst recommendations were overly optimistic. We can see that the number of Buy recommendations is quite high compared to Sell recommendations, the primary reason being that their i-bank employers earn significant amount in fees from the large companies getting recommended.

Further, the incentives of analysts mean that such irrational forecasts are used to get closer to company management so that they can get privy to inside information.

In April 2003, the ‘Global Analyst Research Settlement’ was reached between the top ten US brokerage firms and the SEC, New York Stock Exchange (NYSE), NASD and the New York Attorney General. This led, inter alia, to these brokerage firms paying $1.4 billion in penalties for alleged misconduct resulting in investors losing large sums of money from trading on their analysts’ stock recommendations during the technology bubble.

Reason 2 - Another reason for disparity among Buy and Sell numbers lies in the inherent effect of a Sell recommendation. A sale of stock incurs capital gains taxes for investor. Further, in absence of short selling, an investor can realise the value of a Sell recommendation only if the stock already exists in his portfolio. Due to these factors, a Sell recommendation has a smaller impact than a Buy.

Secondly, we classify the companies based on market cap and Price/Book ratio. We take the market cap at the beginning of the year for the stock as the reference point.

Using the NSE methodology to classify large, mid-cap and small cap companies –

Likewise, for determining the benchmark P/B values we took the S&P CNX 500 which represented about 92.5% of total market cap and 91% of total turnover in NSE. Based on the arithmetic average for each of the years, the benchmarks used are

Hence P/B values below the market cut-off will be classified as VALUE stocks and those above it will be GLAMOUR stocks.

Why are there more recommendations of Large cap and Glamour stocks?

The logic behind such a phenomenon is that most sell-side analysts are employed by brokerage firms whose primary source of revenues is from investment banking business and sales and trading desks. The Research department does not have any substantial revenue source of its own.

Growth firms and firms with higher trading volumes constitute as attractive i-banking clients. Also, the shares of these companies are widely held by institutional and retail investors, who execute trades via these brokerage houses.

Thus analysts have significant advantages to initiate coverage for this category of stocks and publicly recommend high growth stocks.

Sector wise analysis –

We analysed the recommendations based on various industrial sectors. Intuitively, analyzing stock prices with respect to sector is often more appropriate compared to market index. Analysts themselves are classified according to sectors with each specializing in depth in 2 or more sectors. They analyse firms within their industry context and often we come across reports with coverage of the entire industry. Hence analysts follow industry trends keenly while making stock recommendations for specific companies within the industry.

3. Hypothesis Testing

Calculation of Returns -

We define the date of stock recommendation in the newspaper as Event date (Day zero). We pick up adjusted stock prices for:

1. Short term movements – E+1, E+3, E+7 and E+14 days
2. Medium term movements – E+1, E+3 months
3. Long term movements – E+6 and E+12 months

We also symmetrically collected the past price movements i.e. prior to the event date up to 1 year prior to recommendation date.

In case the markets were closed on any particular day, the next working day prices were taken.

Based on the prices, the percentage changes were calculated for different time horizons.

1. Market adjusted returns –

We adjusted the actual returns of the stock by the overall market returns. We have used the SENSEX as a proxy for the market. Using the CAPM model, we determine the expected returns on the stock based on its β value –

Expected Return on the stock = rf + β(rm – rf)

The SENSEX returns have been used as rm while the yield on 10 year G-sec has been used as rf.

Hence the market adjusted excess returns has been found as –

Abnormal returns on the stock = (Actual returns – Expected returns)

2. Characteristic adjusted returns –

Based on portfolios created out of the recommendations, we determined the returns based on market cap and P/B multiple

3. Momentum factor –

Additionally, we also determined if the recommendations were dependent on the momentum factor of the stocks. We look at the past performance of stocks to check if the analyst gave a Buy recommendation if it has been generating positive returns over past 6 months (POSPERF). Similarly for a Sell recommendation, we analyse if the returns have been negative over the past 6 months (NEGPERF).

We compare the number of Buy and Sell recommendations for stocks classified as POSPERF and NEGPERF to understand the momentum bias by the analysts.

Hypothesis Testing –

Using ϗ-square test, we estimate the extent of validity of Buy-Sell recommendations.

H0 – There is no association between Analyst recommendation and actual stock behaviour

Test statistic – Ʃ(F0-FE)2/FE

Where, F0 – Observed frequency

        Fe – Expected frequency = (Nr * Nc)/N

Where Nr – Total sum of rows

        Nc – Total sum of columns

        N – Total observations

Calculated chi square value ET – 2.854 is lesser than the critical value 3.841 and hence we do not reject the null hypothesis. Similar analysis is done for Business Line.

We also estimate the Hit ratio, i.e. the number of recommendations made correctly by the two newspapers.

4. CONCLUSIONS

According to Efficient Market Hypothesis, there should not be observable shifts in stock prices subsequent to release of recommendations; however we do see significant movements in the short run.

Especially near the Event date we see significant stock activity and price movement. This implies that the columnists could have traded and made profits based on column’s information prior to publication date. The effect on returns implies that not all information in the public domain was fully captured in the stock price and the journalist recommendations enabled market to adjust itself to the disseminated information.

The heavy buying by investors put a price pressure on the event date. However the trend of upward movement prior to publication date does clearly imply that the information was either leaked before or planted unscrupulously so as to trigger price movement. If analyst recommendations in actuality convey no inside information, but naive investors act on them, then they become self-fulfilling prophecies. But on the flip side, since fundamentally a firm’s stock prices are dependent on future cash flows, any abnormal price change that results due to an incorrect recommendation will mean that in the longer run the stock prices correct themselves. Hence in long term scenario, the returns generated due to the recommendation get cancelled out and hence there will be no investment value for the recommendation.

Further, if the recommendations are indeed correct, then we see that the investors who act quickly and follow the recommendations stand to gain better. A delay by a week or by a month decreases average gross annual abnormal returns.

Finally, in this study we have not taken into account the transaction costs involved with trading the recommended stocks. In Indian context, the transaction costs are as under –

Adding to that there are various charges like Securities transaction Tax, duties, levies etc for every trade executed.

By considering all the above factors, we cannot objectively state the usefulness of analyst recommendations in generating investment value in the long run.

---